CN107203661B - Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable - Google Patents
Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable Download PDFInfo
- Publication number
- CN107203661B CN107203661B CN201710304898.5A CN201710304898A CN107203661B CN 107203661 B CN107203661 B CN 107203661B CN 201710304898 A CN201710304898 A CN 201710304898A CN 107203661 B CN107203661 B CN 107203661B
- Authority
- CN
- China
- Prior art keywords
- data
- mutual information
- auxiliary variable
- auxiliary
- variables
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The embodiment of the invention provides a method and a system for selecting soft measurement auxiliary variables of a catalytic cracking reaction, wherein the method comprises the following steps: acquiring basic data and coke yield data of catalytic cracking reaction, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data; acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and coke yield data by using a mutual information method, wherein the first mutual information value is more than or equal to-1 and less than or equal to 1; and screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable. The system is used for executing the method. In the embodiment of the invention, the auxiliary variable of the catalytic cracking reaction soft measurement is screened by using a mutual information method, so that the number of the auxiliary variables in the catalytic cracking reaction soft measurement is reduced, the soft measurement process is simplified, and the accuracy of selecting the auxiliary variable of the catalytic cracking reaction soft measurement is improved.
Description
Technical Field
The embodiment of the invention relates to the technical field of petrochemical industry, in particular to a method and a system for selecting soft measurement auxiliary variables of catalytic cracking reaction.
Background
Heavy oil catalytic cracking (RFCC) plays an important role in the oil refining industry of China and is an important means for improving economic benefits of refineries. However, along with the heavy and inferior crude oil, the slag mixing rate of RFCC raw materials in China is continuously improved, the properties of the raw materials are increasingly poor, the RFCC product distribution and the product quality are poor, the yield of light oil is low, and the yield of coke dry gas is high. This places higher demands on the real-time adjustment and optimization of the process point of the device. If the coke yield at a certain moment can be accurately known, the coke yield can be fed back to an advanced control system in real time to adjust and optimize the process operation.
However, the coke yield in the catalytic cracking device cannot be measured in real time by a sensor, and the correlation for calculating the product yield in the catalytic cracking process is generally obtained according to actual operation data and medium-sized experimental data at present, so that the method is suitable for process scheme estimation or technical economic evaluation and is difficult to be used for specifically guiding engineering design or field optimization operation. Soft measurement techniques have been developed to solve the problem of real-time measurement and control of such variables under test. The core of the method is to establish a soft measurement model between easily measured variables (generally called auxiliary variables and input variables) and difficultly measured variables (generally called main variables and output variables). Auxiliary variables in the soft measurement process are difficult to select, and the selection of the variables simply by experience often causes the failure of soft measurement. The auxiliary variables are selected too few, and important explanation factors can be omitted; the auxiliary variables are selected too much, on one hand, the difficulty and the memory space of data acquisition are increased, on the other hand, unnecessary interference and influence are caused, the overfitting condition occurs, and the model adaptability is poor. Selection of auxiliary variables in soft measurement of coke yield remains a gap in the petrochemical industry, particularly in catalytic cracking.
Therefore, how to provide a scheme can simplify the soft measurement process and further improve the accuracy of selecting the catalytic cracking soft measurement auxiliary variable becomes a problem to be solved urgently.
Disclosure of Invention
Aiming at the defects in the prior art, the embodiment of the invention provides a method and a system for selecting a catalytic cracking reaction soft measurement auxiliary variable.
In one aspect, an embodiment of the present invention provides a method for selecting a soft measurement auxiliary variable for a catalytic cracking reaction, including: obtaining basic data and coke yield data of catalytic cracking reaction, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data;
acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and the coke yield data by using a mutual information method, wherein the first mutual information value is greater than or equal to-1 and less than or equal to 1;
and screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable.
Further, the screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable includes:
sequencing the initial auxiliary variables according to the sequence of the absolute values of the first mutual information values corresponding to the initial auxiliary variables from large to small, and acquiring a preset number of initial auxiliary variables which are sequenced in the front as primary screening auxiliary variables;
and acquiring the target auxiliary variable according to the primary screening auxiliary variable.
Further, the screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable includes:
setting a first preset threshold range, and taking an initial auxiliary variable of which the absolute value of the first mutual information value is within the first threshold range as a primary screening auxiliary variable;
and acquiring the target auxiliary variable according to the primary screening auxiliary variable.
Further, the obtaining the target auxiliary variable according to the preliminary screening auxiliary variable includes:
acquiring a second mutual information value between the primary screening auxiliary variables by using a mutual information method, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than a preset threshold value, rejecting the primary screening auxiliary variables with smaller absolute values of the first mutual information values, and taking the remaining primary screening auxiliary variables as the target auxiliary variables;
and the second mutual information value is greater than or equal to-1 and less than or equal to 1.
Further, the acquiring a plurality of initial auxiliary variables of the soft measurement of the catalytic cracking reaction according to the basic data of the catalytic cracking unit comprises:
the method comprises the steps of obtaining various types of basic data preselection data of the catalytic cracking reaction within preset time, removing abnormal values from the preselection data to obtain a plurality of primary auxiliary variables, and obtaining the primary auxiliary variables according to the primary auxiliary variables, wherein the abnormal values comprise 0, a negative value, empty data and data with the difference of more than three times of standard deviation from the average value of the preselection data.
Further, the obtaining the initial auxiliary variable according to the initially selected auxiliary variable includes:
according to the formula
Normalizing each of the initially selected auxiliary variables to obtain the initial auxiliary variable, wherein: pkRepresenting a normalized value, PiRepresents the ith data, P, in the primary selection auxiliary variableminRepresents the minimum value, P, of the primary selection auxiliary variablesmaxRepresenting the maximum value of the primary selected auxiliary variables.
And further, detecting the target auxiliary variable by using a BP neural network.
In another aspect, an embodiment of the present invention provides a system for selecting a soft measurement auxiliary variable of a catalytic cracking reaction, including: a data acquisition unit for acquiring basic data of catalytic cracking reaction and coke yield data, the basic data comprising: feedstock property data, catalyst property data, and operating parameter data;
a mutual information value calculating unit, configured to obtain multiple initial auxiliary variables of the soft measurement of the catalytic cracking reaction according to the basic data, and obtain a first mutual information value between each of the initial auxiliary variables and the coke yield data by using a mutual information method, where the first mutual information value is greater than or equal to-1 and less than or equal to 1;
and the auxiliary variable selecting unit is used for screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable.
In another aspect, an embodiment of the present invention provides an electronic device for selecting soft measurement auxiliary variables of a catalytic cracking reaction, including:
at least one processor; and
at least one memory communicatively coupled to the processor, wherein:
the memory stores program instructions executable by the processor, which when called by the processor are capable of performing the methods described above.
In yet another aspect, embodiments of the present invention provide a non-transitory computer-readable storage medium storing computer instructions that cause the computer to perform the above-described method.
According to the method and the system for selecting the auxiliary variable for the soft measurement of the catalytic cracking reaction, provided by the real-time example, the auxiliary variable for the soft measurement of the catalytic cracking reaction is screened by using a mutual information method, the auxiliary variable related to the coke yield of the catalytic cracking reaction can be quickly obtained, the number of the auxiliary variables in the soft measurement of the catalytic cracking reaction is reduced, the accuracy of selecting the auxiliary variable for the soft measurement of the catalytic cracking reaction is improved, the soft measurement process is simplified, and the performance and the generalization capability of the soft measurement of the catalytic cracking reaction are further improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for selecting soft measurement auxiliary variables of catalytic cracking reactions according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating soft measurements of initial auxiliary variables in an embodiment of the present invention;
FIG. 3 is a diagram illustrating soft measurement results of a target auxiliary variable according to an embodiment of the present invention;
FIG. 4 is a schematic flow chart of another method for selecting auxiliary variables for soft measurement of catalytic cracking reactions according to an embodiment of the present invention;
FIG. 5 is a schematic structural diagram of a system for selecting soft measurement auxiliary variables of a catalytic cracking reaction according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of an electronic device for soft measurement auxiliary variable selection of catalytic cracking reaction according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic flow chart of a method for selecting a catalytic cracking reaction soft measurement auxiliary variable in an embodiment of the present invention, and as shown in fig. 1, the method for selecting a catalytic cracking reaction soft measurement auxiliary variable provided in an embodiment of the present invention includes:
s1, acquiring basic data of catalytic cracking reaction and coke yield data, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data;
specifically, according to the actual situation of the catalytic cracking reaction, basic data and coke yield data of the catalytic cracking reaction are obtained, wherein the coke yield data refer to the actual value of the coke yield in the catalytic cracking reaction. The basic data for catalytic cracking reactions include: feedstock property data, catalyst property data, and operating parameter data. Feedstock property data refers to data relating to the properties of the feedstock for catalytic cracking reactions such as: raw oil saturated hydrocarbon content, raw oil aromatic hydrocarbon content, raw oil asphaltene plus colloid content, raw oil 10% distillation temperature, raw oil 50% distillation temperature, raw oil 90% distillation temperature, raw oil sodium content and the like; catalyst property data refers to data relating to the properties of the catalyst in catalytic cracking reactions such as: catalyst activity, regenerator micro-inverse activity index, regenerator fixed carbon content, balancing agent activity, regenerator sodium content, regenerator nickel and vanadium content and the like; operating parameter data refers to data relating to the operation of a catalytic cracking reaction such as: catalyst temperature, raw material temperature, reaction pressure, raw material oil mass, pre-lifting steam flow, raw material oil atomization steam flow, pre-lifting dry gas flow, recycle ratio, agent-oil ratio, secondary reaction material level, crude gasoline-to-riser reactor flow control and the like. Of course, other basic data related to the catalytic cracking reaction may be included as necessary, and the embodiment of the present invention is not particularly limited.
Wherein, the coke yield data and the basic data can be obtained by a monitoring system of the catalytic cracking device or a catalytic cracking DCS system, or obtained by other methods, and the embodiment of the invention is not particularly limited.
S2, acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and the coke yield data by using a mutual information method, wherein the first mutual information value is greater than or equal to-1 and less than or equal to 1;
specifically, after the basic data of the catalytic reaction is obtained, a plurality of initial auxiliary variables, such as temperature, gas pressure and the like, required by the catalytic cracking reaction soft measurement possibly related to the coke yield of the catalytic cracking reaction are obtained according to the basic data. After the initial auxiliary variables of the catalytic cracking reaction are obtained, a mutual information method is used for obtaining first mutual information values between the initial auxiliary variables and the coke yield of the catalytic cracking reaction, and the mutual information method can measure the correlation between the two objects. The magnitude of the first mutual information value is between-1 and 1, the first mutual information value of 0 represents no correlation between the initial auxiliary variable and the coke yield, and the first mutual information value of +/-1 represents the maximum correlation between the initial auxiliary variable and the coke yield.
The method for obtaining the mutual information value between each initial auxiliary variable and the coke yield of the catalytic cracking reaction by using the mutual information method can be obtained by using related computer software, such as MATLAB and the like, and the embodiment of the invention is not particularly limited.
And S3, screening the initial auxiliary variables according to the first mutual information value to obtain target auxiliary variables.
Specifically, the correlation degree between each initial auxiliary variable and the coke yield of the catalytic cracking reaction can be judged according to the magnitude of the first mutual information value, the initial auxiliary variables are further screened, and the initial auxiliary variables with the large correlation degree with the coke yield of the catalytic cracking reaction are obtained and serve as target auxiliary variables.
The method for selecting the auxiliary variable for the soft measurement of the catalytic cracking reaction provided by the real-time example of the invention screens the auxiliary variable for the soft measurement of the catalytic cracking reaction by using a mutual information method, can quickly acquire the auxiliary variable related to the coke yield of the catalytic cracking reaction, reduces the number of the auxiliary variables in the soft measurement of the catalytic cracking reaction, improves the accuracy of selecting the auxiliary variable for the soft measurement of the catalytic cracking reaction, simplifies the soft measurement process, and further improves the performance and generalization capability of the soft measurement of the catalytic cracking reaction.
On the basis of the above embodiment, the screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable includes:
sequencing the initial auxiliary variables according to the sequence of the absolute values of the first mutual information values corresponding to the initial auxiliary variables from large to small, and acquiring a first preset number of initial auxiliary variables in the front of the sequence as primary screening auxiliary variables;
and acquiring the target auxiliary variable according to the primary screening auxiliary variable.
Specifically, the embodiment of the invention acquires a plurality of acquired initial auxiliary variables possibly related to the coke yield of the catalytic cracking reaction, and acquires a first mutual information value of each initial auxiliary variable and the coke yield according to a mutual information method. And sequencing the initial auxiliary variables according to the sequence of the absolute values of the first mutual information values corresponding to the initial auxiliary variables from large to small, and acquiring a first preset number of initial auxiliary variables in the front of the sequence as primary screening auxiliary variables. The size of the first preset number may be set according to an actual situation, and the embodiment of the present invention is not particularly limited.
In actual use, a first preset threshold range can be set, and an initial auxiliary variable of which the absolute value of the first mutual information value is within the first threshold range is used as a primary screening auxiliary variable; and acquiring the target auxiliary variable according to the primary screening auxiliary variable. After the first mutual information values of the initial auxiliary variables and the coke yield are obtained, setting a first threshold range according to needs, taking the initial auxiliary variables of which the absolute values of the first mutual information values fall within the first threshold range as primary screening auxiliary variables, and then obtaining target auxiliary variables according to the primary screening auxiliary variables. The first threshold range may be set according to an actual use condition, different data may need different first threshold ranges, and the embodiment of the present invention is not particularly limited.
After the primary screening auxiliary variable is obtained, on the basis of the above embodiment, the obtaining the target auxiliary variable according to the primary screening auxiliary variable includes:
acquiring a second mutual information value between the primary screening auxiliary variables by using a mutual information method, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than a preset threshold value, rejecting the primary screening auxiliary variables with smaller absolute values of the first mutual information values, and taking the remaining primary screening auxiliary variables as the target auxiliary variables; and the second mutual information value is greater than or equal to-1 and less than or equal to 1.
Specifically, after primary screening is carried out according to a first mutual information value between an initial auxiliary variable and coke yield to obtain primary screening auxiliary variables, a mutual information method is utilized to obtain second mutual information values between the primary screening auxiliary variables. And if the absolute value of the second mutual information value between the two primary screening auxiliary variables is greater than the preset threshold, the correlation between the two primary screening auxiliary variables is relatively good, and the primary screening auxiliary variable with good correlation with the coke yield is stored and the other primary screening auxiliary variable is removed. Namely, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than the preset threshold, the primary screening auxiliary variable with the larger absolute value of the first mutual information value is reserved, and the primary screening auxiliary variable with the smaller absolute value of the first mutual information value is removed. For example: if the absolute value of the second mutual information between the two prescreening auxiliary variables is greater than 0.9, the one with the greater absolute value of the first mutual information with coke yield is retained and the other is discarded.
For example: if n initial auxiliary variables are pre-selected, a first mutual information value of the n initial auxiliary variables and the coke yield is obtained. And sequencing the n initial auxiliary variables according to the descending order of the absolute values of the first mutual information values corresponding to the n initial auxiliary variables, and acquiring p initial auxiliary variables with the top sequence as primary screening auxiliary variables. And then acquiring second mutual information values among the p primary screening auxiliary variables, and degenerating the primary screening auxiliary variables with relatively similar second mutual information value distribution according to the absolute values of the second mutual information values among the p primary screening auxiliary variables. Namely, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than the preset threshold, the primary screening auxiliary variable with the larger absolute value of the first mutual information value is reserved, the primary screening auxiliary variable with the smaller absolute value of the first mutual information value is removed, and q primary screening auxiliary variables are obtained and serve as final target auxiliary variables.
The method for selecting the auxiliary variable of the catalytic cracking reaction soft measurement provided by the real-time example of the invention screens the auxiliary variable of the catalytic cracking reaction soft measurement by using a mutual information method, obtains the auxiliary variable, the coke yield and the mutual information value between the auxiliary variables, analyzes the correlation between the auxiliary variable, the coke yield and the auxiliary variable, screens the auxiliary variable which has larger correlation with the coke yield of the catalytic cracking reaction, reduces the number of the auxiliary variables in the catalytic cracking reaction soft measurement, improves the accuracy of selecting the auxiliary variable of the catalytic cracking reaction soft measurement, simplifies the soft measurement process, and further improves the performance and generalization capability of the catalytic cracking reaction soft measurement.
On the basis of the above embodiment, the obtaining of a plurality of initial auxiliary variables of the soft measurement of the catalytic cracking reaction according to the basic data of the catalytic cracking unit includes:
the method comprises the steps of obtaining various types of basic data preselection data of the catalytic cracking reaction within preset time, removing abnormal values from the preselection data to obtain a primary auxiliary variable, and obtaining the primary auxiliary variable according to the primary auxiliary variable, wherein the abnormal values comprise 0, a negative value, empty data and data with the difference of three times of standard deviation from the average value of the preselection data.
Specifically, there is a need for time unification of the basic data of the catalytic cracking reaction, that is, the acquisition time of each basic data of the catalytic cracking reaction needs to be consistent, for example, acquiring the respective basic data of 4 months and 5 months in 2017 as pre-selected data, such as: feed property data, catalyst property data, and operating parameter data for months 4 and 5 in 2017. And then removing abnormal values of the acquired pre-selected data within the preset time, wherein the abnormal values comprise 0, a negative value, empty data and data with the difference of more than three times of standard deviation from the average value of the pre-selected data. Namely, eliminating 0, negative value and empty data in the preselected data, calculating the average value of the preselected data, comparing each preselected data with the average value, and eliminating the data with the difference of more than three times of standard deviation from the average value. In the embodiment of the present invention, a triple variance method is used to remove data having a large difference from the average value in the pre-selected data, and of course, other methods may be used as needed to remove data having a large difference from the average value in the data, for example, to remove data having a difference from the average value larger than a preset threshold value.
After the abnormal values are eliminated, the remaining pre-selected data are used as primary selection auxiliary variables, and the primary selection auxiliary variables are normalized to obtain the primary auxiliary variables.
The normalization process may specifically adopt the following method:
normalizing the data in the initially selected auxiliary variables according to a maximum and minimum method, specifically performing normalization by using the following formula (1):
in the formula: pkTo normalized values, PiFirst selection of the ith data, P, in the auxiliary variableminRepresents the minimum value, P, of the primary selection auxiliary variablesmaxIndicates the beginningThe maximum of the auxiliary variables is selected.
For example: two sets of primary selection auxiliary variables are obtained, namely a temperature auxiliary variable and a pressure auxiliary variable respectively, wherein each set of auxiliary variables comprises a plurality of sets of data, the temperature auxiliary variable is firstly normalized to obtain the maximum value P in the temperature auxiliary variablemaxAnd a minimum value PminEach data in the temperature auxiliary variable is normalized using the above equation (1). In the same way, the pressure auxiliary variable is normalized.
In actual use, the collected coke yield data can be normalized by the same method, the collected initial auxiliary variables and the coke yield data can form a sample set, the size of the sample set is m, and the data of the sample set is subjected to uniform preprocessing and normalization processing. For example: the selected n original auxiliary variables form a matrix X ═ Xi]T,
1, 2.. m, the coke production data is written as a matrix Y ═ Yi]T,
And i, 1,2,.. m, correspondingly processing the data in the two matrixes, then returning a first mutual information value of the two groups of data, and screening the initial auxiliary variable. The specific screening process is the same as the above embodiment, and is not described herein again.
On the basis of the above embodiment, the method further includes: and detecting the target auxiliary variable by using a BP neural network.
Specifically, the embodiment of the present invention further detects the selected target auxiliary variable by using a BP neural network, so as to verify the accuracy of the selected target auxiliary variable. Specifically, a BP neural network may be used to establish a soft measurement model in advance, for example: and randomly selecting 90% of data quantity data from the n initial auxiliary variable data as a training sample, setting the iteration times to be 100 times, the learning rate to be 0.2 and the target value to be 0.00004, and performing model training. And (4) using the remaining 10% of data of the n initial auxiliary variables as prediction samples, and predicting the nonlinear function output by using a trained BP neural network. After a BP neural network measurement model is trained, the selected target neural network is detected, data of q auxiliary variables are divided into three parts, 90% of data are randomly selected as training samples, and the remaining 10% of data of the test samples of the q target auxiliary variables are used as prediction samples to verify the effect. Randomly selecting 90% of data, using half as training sample to train model, using the other half as test sample to test model. And (3) carrying out model training on the training sample with the set iteration times of 100 and the learning rate of 0.1. And predicting the output of the nonlinear function by using the trained BP neural network, and comparing the coke yield output results predicted by soft measurement of the n initial auxiliary variables and the q target auxiliary variables to verify the accuracy of the selected target auxiliary variables. Fig. 2 is a schematic diagram of a soft measurement result of an initial auxiliary variable in an embodiment of the present invention, and fig. 3 is a schematic diagram of a soft measurement result of a target auxiliary variable in an embodiment of the present invention, as shown in fig. 2 and fig. 3, the number of auxiliary variables in the soft measurement in fig. 2 is 22, an average error between a soft measurement value and an actual value is 2.22%, the number of auxiliary variables in the soft measurement in fig. 3 is 16, and an average error between a soft measurement value and an actual value is 1.72%. It can be seen that after the auxiliary variables are reduced, the soft measurement result is not obviously affected, after the initial auxiliary variables are screened by the embodiment of the invention, the number of the auxiliary variables is effectively reduced, meanwhile, the prediction effect is obviously improved, and the average error between the soft measurement value and the actual value is reduced from 2.22% to 1.72%.
Fig. 4 is a schematic flow chart of a method for selecting a catalytic cracking reaction soft measurement auxiliary variable in an embodiment of the present invention, and as shown in fig. 4, a specific flow chart of the embodiment of the present invention is described below with reference to fig. 4, so as to better understand the technical solution of the present invention:
t1, data acquisition: about 13 ten thousand sets of basic data and coke yield data were collected for a year accumulated in refinery a, the collected basic data including: raw material properties (raw material oil saturated hydrocarbon content, raw material oil aromatic hydrocarbon content, raw material oil asphaltene + colloid content, raw material oil 10% distillation temperature, raw material oil 50% distillation temperature, raw material oil 90% distillation temperature, raw material oil sodium content), catalyst properties (catalyst activity, regenerator microreflection activity index, regenerator fixed carbon content, balancer activity, regenerator sodium content, regenerator nickel and vanadium content), operating parameters (catalyst temperature, raw material temperature, reaction pressure, raw material oil mass, pre-lifting steam flow, raw material oil atomization steam flow, pre-lifting dry gas flow, recycle ratio, agent-oil ratio, secondary reaction material level, crude gasoline to lifting tube reactor flow control).
T2, preprocessing of data: firstly, taking data in a common time period of each basic data; secondly, removing abnormal values of each basic data, wherein the abnormal values comprise: data containing zero, negative, null; thirdly, data which is different from the average value by a large amount, such as a difference value larger than three times of standard deviation, is removed by utilizing triple variance on each basic data. After pretreatment, about 1 ten thousand groups of data are obtained as primary selection auxiliary variables.
T3, normalization processing of data: according to
The data in the initially selected auxiliary variables are normalized, and the specific normalization processing method is the same as that in the above embodiment, and is not described here again. And normalizing the processed data as an initial auxiliary variable.
T4, selection of primary screening auxiliary variables: the method comprises the steps of obtaining n groups of initial auxiliary variables, respectively calculating first mutual information values between the n initial auxiliary variables and the coke yield by using a mutual information method, wherein the first mutual information values are between-1 and 1, 0 represents that no correlation exists between two groups of data, and +/-1 represents that the two groups of data are completely the same, and selecting p primary screening auxiliary variables closely related to the coke yield according to the absolute value of the first mutual information values. The specific method for obtaining the preliminary screening auxiliary variable is the same as that in the above embodiment, and is not described herein again.
T5, selection of target auxiliary variables: and calculating a second mutual information value among the p primary screening auxiliary variables by using a mutual information method, and degenerating the primary screening auxiliary variables according to the absolute value of the second mutual information value to obtain p target auxiliary variables. The method for degenerating the preliminary screening auxiliary variables according to the absolute value of the second mutual information value is the same as the above embodiment, and is not described herein again.
T6, BP neural network coke yield soft measurement of initial auxiliary variables: and randomly selecting 90% of data quantity data from the n initial auxiliary variable data as a training sample, setting the iteration times to be 100 times, the learning rate to be 0.2 and the target value to be 0.00004, and performing model training. And (3) taking the residual 10% of data of the n initial auxiliary variables as prediction samples, and outputting by using a trained BP neural network prediction nonlinear function to obtain soft coke yield measurement prediction results of the n initial auxiliary variables.
T7, BP neural network coke yield soft measurement of target auxiliary variables: and randomly selecting 90% of data as a training sample and the remaining 10% of data of the test sample in the q target auxiliary variables as a prediction sample to verify the effect. Randomly selecting 90% of data, using half as training sample to train model, using the other half as test sample to test model. And (3) carrying out model training on the training sample with the set iteration times of 100 and the learning rate of 0.1. And predicting the nonlinear function output by using the trained BP neural network to obtain the coke yield soft measurement prediction results of the q target auxiliary variables.
T8, soft measurement prediction result comparison: and comparing the coke yield output results predicted by the soft measurement of the n initial auxiliary variables and the q target auxiliary variables to verify the accuracy of the selected target auxiliary variables.
The method for selecting the auxiliary variable of the catalytic cracking reaction soft measurement provided by the embodiment of the invention utilizes a mutual information theory to quickly find out a few key target auxiliary variables related to the coke yield, simplifies the soft measurement process on the premise of meeting the requirement of the catalytic cracking reaction soft measurement, improves the accuracy of selecting the auxiliary variable of the catalytic cracking reaction soft measurement, further improves the performance and the generalization capability of the catalytic cracking reaction soft measurement, and provides technical support for a chemical advanced control system and refinery production.
Fig. 5 is a schematic structural diagram of a system for selecting a catalytic cracking reaction soft measurement auxiliary variable in an embodiment of the present invention, and as shown in fig. 5, the system for selecting a catalytic cracking reaction soft measurement auxiliary variable provided in an embodiment of the present invention includes: a data acquisition unit 51, a mutual information value calculation unit 52 and an auxiliary variable selection unit 53, wherein:
the data acquiring unit 51 is used for acquiring basic data of catalytic cracking reaction and coke yield data, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data; the mutual information value calculating unit 52 is configured to obtain a plurality of initial auxiliary variables of the soft measurement of the catalytic cracking reaction according to the basic data, and obtain a first mutual information value between each of the initial auxiliary variables and the coke yield data by using a mutual information method, where the first mutual information value is greater than or equal to-1 and less than or equal to 1; the auxiliary variable selecting unit 53 is configured to filter the initial auxiliary variable according to the first mutual information value, so as to obtain a target auxiliary variable.
Specifically, the data obtaining unit 51 obtains basic data of the catalytic cracking reaction and coke yield data according to the actual situation of the catalytic cracking reaction, wherein the coke yield data refers to an actual value of the coke yield in the catalytic cracking reaction. The basic data for catalytic cracking reactions include: feedstock property data, catalyst property data, and operating parameter data. After acquiring the basic data of the catalytic reaction, the data acquiring unit 51 acquires a plurality of initial auxiliary variables, such as temperature, gas pressure, etc., required for the soft measurement of the catalytic cracking reaction, which may be related to the coke yield of the catalytic cracking reaction, from the basic data. After the initial auxiliary variables of the catalytic cracking reaction are obtained, the mutual information value calculation unit 52 obtains the first mutual information values between each initial auxiliary variable and the coke yield of the catalytic cracking reaction by using a mutual information method, and the mutual information method can measure the correlation between the two objects. The magnitude of the first mutual information value is between-1 and 1, the first mutual information value of 0 represents no correlation between the initial auxiliary variable and the coke yield, and the first mutual information value of +/-1 represents the maximum correlation between the initial auxiliary variable and the coke yield. The auxiliary variable selecting unit 53 determines the correlation degree between each initial auxiliary variable and the coke yield of the catalytic cracking reaction according to the magnitude of the first mutual information value, further screens the initial auxiliary variables, and obtains the initial auxiliary variable having a large correlation degree with the coke yield of the catalytic cracking reaction as the target auxiliary variable.
On the basis of the above embodiment, the system further includes a detection unit, configured to detect the target auxiliary variable by using a BP neural network.
Specifically, the embodiment of the present invention further detects the selected target auxiliary variable by using a BP neural network through a detection unit, so as to verify the accuracy of the selected target auxiliary variable. The accuracy of the selected target auxiliary variable can be verified by comparing the coke yield output predicted by soft measurements of the initial auxiliary variable and the target auxiliary variable with the actual yield value. The method for detecting the target auxiliary variable by using the BP neural network is the same as that in the above embodiments, and details are not described here.
The system for selecting the auxiliary variable for soft measurement of catalytic cracking reaction provided by the embodiment of the invention is used for executing the method, and the specific implementation manner of the system is consistent with that of the embodiment, and is not described herein again.
The method and the system for selecting the auxiliary variable of the catalytic cracking reaction soft measurement provided by the embodiment of the invention can quickly find out a few key target auxiliary variables related to the coke yield by utilizing a mutual information theory, are specially used for catalytic cracking industrial data with large data volume and high dimensionality, simplify the process of the soft measurement on the premise of meeting the requirement of the catalytic cracking reaction soft measurement, improve the accuracy of selecting the auxiliary variable of the catalytic cracking reaction soft measurement, further improve the performance and the generalization capability of the catalytic cracking reaction soft measurement, and provide technical support for a chemical advanced control system and the production of a refinery.
Fig. 6 is a schematic structural diagram of an electronic device for selecting soft measurement auxiliary variables of a catalytic cracking reaction according to an embodiment of the present invention, and as shown in fig. 6, the apparatus may include: a processor (processor)61, a memory (memory)62 and a communication bus 63, wherein the processor 61 and the memory 62 communicate with each other via the communication bus 63. The processor 61 may call logic instructions in the memory 62 to perform the following method: obtaining basic data and coke yield data of catalytic cracking reaction, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data; acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and the coke yield data by using a mutual information method, wherein the first mutual information value is greater than or equal to-1 and less than or equal to 1; and screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable.
Furthermore, the logic instructions in the memory 62 may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Embodiments of the present invention provide a non-transitory computer-readable storage medium, which stores computer instructions, where the computer instructions cause the computer to perform the methods provided by the above method embodiments, for example, the methods include: obtaining basic data and coke yield data of catalytic cracking reaction, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data; acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and the coke yield data by using a mutual information method, wherein the first mutual information value is greater than or equal to-1 and less than or equal to 1; and screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable.
The above examples are only for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (7)
1. A method for selecting soft measurement auxiliary variables of catalytic cracking reaction is characterized by comprising the following steps:
obtaining basic data and coke yield data of catalytic cracking reaction, wherein the basic data comprises: feedstock property data, catalyst property data, and operating parameter data;
acquiring a plurality of initial auxiliary variables of the catalytic cracking reaction soft measurement according to the basic data, and acquiring a first mutual information value between each initial auxiliary variable and the coke yield data by using a mutual information method, wherein the first mutual information value is greater than or equal to-1 and less than or equal to 1;
screening the initial auxiliary variable according to the first mutual information value to obtain a target auxiliary variable;
detecting the target auxiliary variable by using a BP neural network, wherein the detection comprises the soft measurement of the coke yield of the BP neural network on the initial auxiliary variable; carrying out BP neural network coke yield soft measurement on the target auxiliary variable; comparing the coke yield output result predicted by the soft measurement of the initial auxiliary variable with the coke yield output result predicted by the soft measurement of the target auxiliary variable to verify the accuracy of the selected target auxiliary variable;
setting a first preset threshold range, and taking an initial auxiliary variable of which the absolute value of the first mutual information value is within the first preset threshold range as a primary screening auxiliary variable;
acquiring a second mutual information value between the primary screening auxiliary variables by using a mutual information method, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than a preset threshold value, rejecting the primary screening auxiliary variables with smaller absolute values of the first mutual information values, and taking the remaining primary screening auxiliary variables as the target auxiliary variables;
and the second mutual information value is greater than or equal to-1 and less than or equal to 1.
2. The method according to claim 1, wherein the screening of the initial auxiliary variables according to the first mutual information value to obtain target auxiliary variables is replaced by:
sequencing the initial auxiliary variables according to the sequence of the absolute values of the first mutual information values corresponding to the initial auxiliary variables from large to small, and acquiring a preset number of initial auxiliary variables which are sequenced in the front as primary screening auxiliary variables;
and acquiring the target auxiliary variable according to the primary screening auxiliary variable.
3. The method of claim 1, wherein the obtaining a plurality of initial auxiliary variables for the soft measurement of the catalytic cracking reaction from the base data comprises:
the method comprises the steps of obtaining various types of basic data preselection data of the catalytic cracking reaction within preset time, removing abnormal values from the preselection data to obtain a plurality of primary auxiliary variables, and obtaining the primary auxiliary variables according to the primary auxiliary variables, wherein the abnormal values comprise 0, a negative value, empty data and data with the difference of more than three times of standard deviation from the average value of the preselection data.
4. The method of claim 3, wherein said obtaining the initial auxiliary variable from the initially selected auxiliary variable comprises:
according toFormula (II)
Normalizing each of the initially selected auxiliary variables to obtain the initial auxiliary variable, wherein: pkRepresenting a normalized value, PiRepresents the ith data, P, in the primary selection auxiliary variableminRepresents the minimum value, P, of the primary selection auxiliary variablesmaxRepresenting the maximum value of the primary selected auxiliary variables.
5. A system for selecting soft measurement auxiliary variables of catalytic cracking reaction is characterized by comprising:
a data acquisition unit for acquiring basic data of catalytic cracking reaction and coke yield data, the basic data comprising: feedstock property data, catalyst property data, and operating parameter data;
a mutual information value calculating unit, configured to obtain multiple initial auxiliary variables of the soft measurement of the catalytic cracking reaction according to the basic data, and obtain a first mutual information value between each of the initial auxiliary variables and the coke yield data by using a mutual information method, where the first mutual information value is greater than or equal to-1 and less than or equal to 1;
an auxiliary variable selecting unit, configured to screen the initial auxiliary variable according to the first mutual information value, to obtain a target auxiliary variable; detecting the target auxiliary variable by using a BP neural network, wherein the detection comprises the soft measurement of the coke yield of the BP neural network on the initial auxiliary variable; carrying out BP neural network coke rate soft measurement on the target auxiliary variable; comparing the coke yield output result predicted by the soft measurement of the initial auxiliary variable with the coke yield output result predicted by the soft measurement of the target auxiliary variable to verify the accuracy of the selected target auxiliary variable;
setting a first preset threshold range, and taking an initial auxiliary variable of which the absolute value of the first mutual information value is within the first preset threshold range as a primary screening auxiliary variable;
acquiring a second mutual information value between the primary screening auxiliary variables by using a mutual information method, if the absolute value of the second mutual information value between the two primary screening auxiliary variables is larger than a preset threshold value, rejecting the primary screening auxiliary variables with smaller absolute values of the first mutual information values, and taking the remaining primary screening auxiliary variables as the target auxiliary variables;
and the second mutual information value is greater than or equal to-1 and less than or equal to 1.
6. An electronic device for soft measurement auxiliary variable selection for catalytic cracking reactions, comprising:
at least one processor; and
at least one memory communicatively coupled to the processor, wherein:
the memory stores program instructions executable by the processor, the processor invoking the program instructions to perform the method of any of claims 1 to 4.
7. A non-transitory computer-readable storage medium storing computer instructions that cause a computer to perform the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710304898.5A CN107203661B (en) | 2017-05-03 | 2017-05-03 | Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710304898.5A CN107203661B (en) | 2017-05-03 | 2017-05-03 | Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107203661A CN107203661A (en) | 2017-09-26 |
| CN107203661B true CN107203661B (en) | 2020-04-07 |
Family
ID=59905665
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710304898.5A Active CN107203661B (en) | 2017-05-03 | 2017-05-03 | Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107203661B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957356A (en) * | 2010-08-13 | 2011-01-26 | 北京工业大学 | Biochemical oxygen demand BOD soft measurement method based on elastic radical basis function neural network |
| CN105177199A (en) * | 2015-08-31 | 2015-12-23 | 南京南瑞继保电气有限公司 | Blast furnace gas generation amount soft measurement method |
| CN105734189A (en) * | 2016-04-12 | 2016-07-06 | 燕山大学 | Method for determining coal injection quantity of blast furnace during operation process |
-
2017
- 2017-05-03 CN CN201710304898.5A patent/CN107203661B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957356A (en) * | 2010-08-13 | 2011-01-26 | 北京工业大学 | Biochemical oxygen demand BOD soft measurement method based on elastic radical basis function neural network |
| CN105177199A (en) * | 2015-08-31 | 2015-12-23 | 南京南瑞继保电气有限公司 | Blast furnace gas generation amount soft measurement method |
| CN105734189A (en) * | 2016-04-12 | 2016-07-06 | 燕山大学 | Method for determining coal injection quantity of blast furnace during operation process |
Non-Patent Citations (2)
| Title |
|---|
| 基于互信息的软测量变量选择;杨慧中 等;《控制工程》;20120731;第19卷(第4期);正文第562-565、593页 * |
| 数据驱动的软测量及智能优化技术研究;陈志杰;《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》;20151215;B016-34 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107203661A (en) | 2017-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Komulainen et al. | An online application of dynamic PLS to a dearomatization process | |
| Zhu et al. | Toward a low cost and high performance MPC: The role of system identification | |
| CA2886695C (en) | Characterization and prediction of jet fuel quality | |
| CN110991495B (en) | Method, system, medium, and apparatus for predicting product quality in manufacturing process | |
| Guh | Robustness of the neural network based control chart pattern recognition system to non‐normality | |
| JP2009258890A (en) | Influence factor specifying device | |
| CN111639304B (en) | CSTR fault positioning method based on Xgboost regression model | |
| CN115188429A (en) | Catalytic cracking unit key index modeling method integrating time sequence feature extraction | |
| Yu et al. | A framework to identify and respond to weak signals of disastrous process incidents based on FRAM and machine learning techniques | |
| CN107203661B (en) | Method and system for selecting catalytic cracking reaction soft measurement auxiliary variable | |
| CN110021377B (en) | Method and device for predicting deactivation of hydrocracking catalyst and storage equipment | |
| CN104483958A (en) | Adaptive data driving fault diagnosis method and device in complex refining process | |
| CN106647285B (en) | A kind of catalyst activity detection method based on soft-measuring technique | |
| CN112464554A (en) | Operating parameter optimization method of gasoline refining equipment | |
| CN111863151B (en) | Polymer molecular weight distribution prediction method based on Gaussian process regression | |
| CN115856264A (en) | Method and system for judging validity of online monitoring data of dissolved gas in transformer oil | |
| CN102750454A (en) | Hydrogen consumption predicted value acquisition method, device and system | |
| Ramachandran et al. | Data analysis, modeling and control performance enhancement of an industrial fluid catalytic cracking unit | |
| CN113240246A (en) | Method and device for determining coal seam gas content main control factor based on grey correlation method | |
| CN114660931A (en) | Method and system for diagnosing and identifying petrochemical process fault | |
| CN111126694A (en) | Time series data prediction method, system, medium and device | |
| CN111242350A (en) | Continuous reforming device fractionation and extraction optimization method, system and medium | |
| WO2023119544A1 (en) | Program, information processing device, and method | |
| CN113128734A (en) | Method and device for predicting oil field yield | |
| CN115856204B (en) | Tunneling working face gas concentration prediction method based on three-dimensional echo state network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |