CN112342050A - Method and device for optimizing light oil yield of catalytic cracking unit and storage medium - Google Patents

Method and device for optimizing light oil yield of catalytic cracking unit and storage medium Download PDF

Info

Publication number
CN112342050A
CN112342050A CN202011175691.0A CN202011175691A CN112342050A CN 112342050 A CN112342050 A CN 112342050A CN 202011175691 A CN202011175691 A CN 202011175691A CN 112342050 A CN112342050 A CN 112342050A
Authority
CN
China
Prior art keywords
yield
catalytic cracking
light oil
parameters
layer
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.)
Granted
Application number
CN202011175691.0A
Other languages
Chinese (zh)
Other versions
CN112342050B (en
Inventor
杨庆伟
邵杰锋
肖丰斌
彭芳
侯晓宇
齐文峰
赵文宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Petroleum and Chemical Corp
Original Assignee
China Petroleum and Chemical Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Petroleum and Chemical Corp filed Critical China Petroleum and Chemical Corp
Priority to CN202011175691.0A priority Critical patent/CN112342050B/en
Publication of CN112342050A publication Critical patent/CN112342050A/en
Application granted granted Critical
Publication of CN112342050B publication Critical patent/CN112342050B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • C—CHEMISTRY; METALLURGY
    • C10—PETROLEUM, GAS OR COKE INDUSTRIES; TECHNICAL GASES CONTAINING CARBON MONOXIDE; FUELS; LUBRICANTS; PEAT
    • C10G—CRACKING HYDROCARBON OILS; PRODUCTION OF LIQUID HYDROCARBON MIXTURES, e.g. BY DESTRUCTIVE HYDROGENATION, OLIGOMERISATION, POLYMERISATION; RECOVERY OF HYDROCARBON OILS FROM OIL-SHALE, OIL-SAND, OR GASES; REFINING MIXTURES MAINLY CONSISTING OF HYDROCARBONS; REFORMING OF NAPHTHA; MINERAL WAXES
    • C10G11/00—Catalytic cracking, in the absence of hydrogen, of hydrocarbon oils
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • C—CHEMISTRY; METALLURGY
    • C10—PETROLEUM, GAS OR COKE INDUSTRIES; TECHNICAL GASES CONTAINING CARBON MONOXIDE; FUELS; LUBRICANTS; PEAT
    • C10G—CRACKING HYDROCARBON OILS; PRODUCTION OF LIQUID HYDROCARBON MIXTURES, e.g. BY DESTRUCTIVE HYDROGENATION, OLIGOMERISATION, POLYMERISATION; RECOVERY OF HYDROCARBON OILS FROM OIL-SHALE, OIL-SAND, OR GASES; REFINING MIXTURES MAINLY CONSISTING OF HYDROCARBONS; REFORMING OF NAPHTHA; MINERAL WAXES
    • C10G2300/00—Aspects relating to hydrocarbon processing covered by groups C10G1/00 - C10G99/00
    • C10G2300/40—Characteristics of the process deviating from typical ways of processing
    • C10G2300/4006—Temperature
    • C—CHEMISTRY; METALLURGY
    • C10—PETROLEUM, GAS OR COKE INDUSTRIES; TECHNICAL GASES CONTAINING CARBON MONOXIDE; FUELS; LUBRICANTS; PEAT
    • C10G—CRACKING HYDROCARBON OILS; PRODUCTION OF LIQUID HYDROCARBON MIXTURES, e.g. BY DESTRUCTIVE HYDROGENATION, OLIGOMERISATION, POLYMERISATION; RECOVERY OF HYDROCARBON OILS FROM OIL-SHALE, OIL-SAND, OR GASES; REFINING MIXTURES MAINLY CONSISTING OF HYDROCARBONS; REFORMING OF NAPHTHA; MINERAL WAXES
    • C10G2300/00—Aspects relating to hydrocarbon processing covered by groups C10G1/00 - C10G99/00
    • C10G2300/40—Characteristics of the process deviating from typical ways of processing
    • C10G2300/4012—Pressure

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Oil, Petroleum & Natural Gas (AREA)
  • General Chemical & Material Sciences (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Organic Chemistry (AREA)
  • Production Of Liquid Hydrocarbon Mixture For Refining Petroleum (AREA)

Abstract

The invention discloses a method for optimizing light oil yield of a catalytic cracking unit, which comprises the following steps: obtaining historical data of parameters participating in catalytic cracking yield optimization and yield actual values corresponding to the parameters, and performing abnormal value elimination and normalization operation on the obtained historical data of the parameters and the yield actual values to form sample data; training a GRNN model by using the sample data to obtain the GRNN model for predicting the yield according to the parameters; and optimizing the result of the model prediction of the parameter prediction yield by adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking unit. Therefore, the method for optimizing the light oil yield of the catalytic cracking unit can greatly improve the light oil yield, accurately describe the reaction process in terms of mechanism and improve the data accuracy.

Description

Method and device for optimizing light oil yield of catalytic cracking unit and storage medium
Technical Field
The invention relates to the field of petrochemical industry, in particular to a method and a device for optimizing light oil yield of a catalytic cracking device and a storage medium.
Background
Catalytic Cracking (FCC) is a process in which heavy oil undergoes a cracking reaction under the action of heat and a catalyst to convert into cracked gas, gasoline, diesel oil, and the like. The catalytic cracking plays an important role in the oil refining industry in China, compared with thermal cracking, the catalytic cracking device has the advantages of high light oil yield, high gasoline octane number and good diesel stability, and the byproduct of liquefied gas rich in olefin is an important means for improving economic benefit of refineries, while the light oil yield is the most important economic index for the catalytic cracking device. The core of the accurate yield model lies in accurate reaction mechanism kinetics, but the catalytic cracking process involves various reactions such as decomposition, isomerization, hydrogen transfer, aromatization, condensation, coke formation and the like, and the reaction process is difficult to be described accurately from the mechanism.
At present, for a similar complex reaction system, if starting from a reaction mechanism, a lumped method is generally adopted for kinetic analysis, so that the reaction core mechanism is reserved, a reaction network can be simplified, and the difficulty in estimating kinetic parameters is reduced. However, the centralized dynamics method has high requirements on data accuracy, and the data accuracy of the current enterprise often cannot meet the accuracy requirements of the method.
Disclosure of Invention
The invention aims to provide a method for optimizing the light oil yield of a catalytic cracking device, which can greatly improve the light oil yield, can accurately describe the reaction process from the mechanism and improve the data accuracy.
In order to achieve the above object, an embodiment of the present invention provides a method for optimizing light oil yield of a catalytic cracking unit, including the steps of:
obtaining historical data of parameters participating in catalytic cracking yield optimization and yield actual values corresponding to the parameters, and performing abnormal value elimination and normalization operation on the obtained historical data of the parameters and the yield actual values to form sample data;
training a GRNN model by using the sample data to obtain a model for predicting yield according to the parameters;
and optimizing the result of the model prediction of the parameter prediction yield by adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking unit.
In some embodiments, the parameters involved in catalytic cracking yield optimization include: the primary reaction outlet temperature, the secondary reaction outlet temperature, the reaction pressure, the remilling ratio, the agent-oil ratio, the slag mixing ratio, the inlet linear speed, the steam lifting amount, the raw material oil preheating temperature, the main air amount and the regeneration temperature.
In some embodiments, the historical data of the parameters, and the historical data of the calculated actual yield, are acquired from a real-time data acquisition system and a Lims system of the catalytic cracking unit.
In some embodiments, the outliers include 0, negative, null, and data that differs from the mean by more than 3 standard deviations.
In some embodiments, the model for parametric prediction of yield comprises: an input layer, a mode layer, a summation layer, and an output layer, corresponding to a network input of X ═ X1,x2,...,xn]The output is Y ═ Y1,y2,...,yk]；
Wherein the number of input layer neurons is equal to the dimension of the input vector in the learning sample, i.e., X ═ X1,x2,...,xn]The neurons are simple distribution units that directly pass input variables to the mode layer;
the number of pattern layer neurons is equal to the number n of learning samples, each neuron corresponds to a different learning sample, and the pattern layer neuron transfer function is:
Figure BDA0002747108140000021
the output of the neuron i is the exponential square of the squared Euclidean distance between the input variable and the corresponding sample X;
the summation layer uses two types of neuron algorithms to carry out summation;
one of the summation formulas is:
Figure BDA0002747108140000022
the summation formula performs arithmetic summation on the outputs of all the neurons in the mode layer, the connection weight of the mode layer and each neuron is 1, and the transfer function is as follows:
Figure BDA0002747108140000023
another summation formula is:
Figure BDA0002747108140000024
the summation formula carries out weighted summation on all the neurons in the mode layer, and the connection weight value between the ith neuron in the mode layer and the jth numerator summation neuron in the summation layer is the ith output sample YiThe j-th element in (2), the transfer function is:
Figure BDA0002747108140000031
the number of neurons in the output layer is equal to the dimension k of the output vector in the learning sample, each neuron divides the output of the summation layer, and the output of neuron j corresponds to the jth element of the estimation result Y (X), i.e. the output of neuron j is equal to the jth element of the estimation result Y (X)
Figure BDA0002747108140000032
In some embodiments, the genetic algorithm comprises the steps of:
s1, taking parameter values participating in catalytic cracking yield optimization at a certain moment as an initial population, setting the number of individuals in the population as M, setting an evolution algebra counter T as 0, and setting a maximum evolution algebra T;
s2, taking the prediction result of the trained parameter prediction yield model as an individual fitness value, and calculating the fitness of each individual in the population;
s3, selecting an individual with a large fitness proportion by adopting a roulette method;
s4, generating a next generation group by using cross operation and mutation operation on the individuals;
s5, judging whether T is equal to T, if so, outputting a corresponding parameter optimization value participating in the catalytic cracking yield optimization by taking the individual with the maximum fitness obtained in the evolution process as an optimal solution, and stopping calculation; if not, returning to repeat the loop to execute S2-S5;
wherein the number of the initial population is 50 (i.e., M is 50), and the number of iterations is 100 (i.e., T is 100); the prediction result of the trained model for predicting the parameter yield is used as an individual fitness value, and the calculation formula is as follows: fi＝SNj/SDAnd Fi is the fitness value of the individual i.
In some embodiments, the prediction result of the trained model for predicting the parameter yield is used as an individual fitness value, specifically:
adopting a roulette method, selecting an individual with larger fitness to enter the next generation, wherein the calculation formula is as follows:
fi＝k/Fi，pi＝fi/∑fi
wherein FiIs the fitness value of an individual i, k is a coefficient, piIs the proportion of individual fitness.
In some embodiments, the genetic algorithm uses a crossover probability of 0.6; the mutation probability of 0.05 is adopted in the mutation operation.
The embodiment of the invention also provides a device for optimizing the light oil yield of a catalytic cracking device, which is characterized by comprising the following components: the system comprises a sample data acquisition module, a parameter analysis module and a parameter analysis module, wherein the sample data acquisition module is used for acquiring historical data of parameters participating in catalytic cracking yield optimization and actual yield values corresponding to the historical data, and performing abnormal value elimination and normalization operation on the acquired historical data of the parameters and the actual yield values to form sample data;
the model obtaining module of the parameter prediction yield is used for training GRNN by utilizing the sample data to obtain a model of the parameter prediction yield;
and the optimization module is used for optimizing the parameters participating in the catalytic cracking yield optimization by using the model prediction result of the parameter prediction yield and adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking device.
The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement the method for making a signboard with load layout according to any one of the above embodiments.
Compared with the prior art, the method has the following beneficial effects:
the method for optimizing the light oil yield of the catalytic cracking unit disclosed by the embodiment of the invention is combined with GRNN and a genetic algorithm to optimize and optimize each operation parameter, a yield optimization operation scheme is given, the light oil yield is greatly improved, the reaction process can be described accurately from the mechanism, and the data accuracy is improved.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for optimizing light oil yield of a catalytic cracking unit according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a computer terminal device according to an embodiment of the present invention.
Detailed Description
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 only a part of the embodiments of the present invention, and not all of the embodiments. 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.
It should be understood that the step numbers used herein are for convenience of description only and are not intended as limitations on the order in which the steps are performed.
It is to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
The terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
The term "and/or" refers to and includes any and all possible combinations of one or more of the associated listed items.
Referring to fig. 1, an embodiment of the present invention provides a method for optimizing light oil yield of a catalytic cracking unit, including:
s10, obtaining historical data of parameters participating in catalytic cracking yield optimization and actual yield values corresponding to the historical data, and performing abnormal value elimination and normalization operation on the obtained historical data of the parameters and the actual yield values to form sample data;
s20, training a GRNN model by using the sample data to obtain a model for predicting yield according to the parameters;
and S30, optimizing the model prediction result of the parameter prediction yield by adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking unit.
The catalytic cracking is a complex process influenced by various highly nonlinear and strongly correlated factors, and various factors including the properties of raw oil, the properties of a reaction regenerated catalyst, reaction operating conditions and the like influence the reaction process and the product yield, and are particularly important for mathematical modeling analysis of optimization of the process and the product yield.
In this embodiment, a mathematical model for optimizing the light oil yield of the catalytic cracking unit is established, and parameters participating in the optimization of the catalytic cracking yield, including a primary reaction outlet temperature, a secondary reaction outlet temperature, a reaction pressure, a recycle ratio, a catalyst-to-oil ratio, a blending ratio, an inlet linear velocity, a steam lifting amount, a raw material oil preheating temperature, a main air amount, and a regeneration temperature, need to be selected. And acquiring historical data of the parameters and actual values of the yield corresponding to the parameters. And preprocessing the acquired historical data and the actual value of the yield, removing abnormal values and normalizing to form standard sample data.
And secondly, obtaining a model for predicting the yield according to the parameters by adopting the Generalized Regression Neural Network (GRNN). The GRNN has strong nonlinear mapping capability and learning speed and has stronger advantages than RBF, the network is finally converged in the optimized regression with more sample size aggregation, the prediction effect is good when the sample data is less, and unstable data can be processed. While GRNN does not appear to be as accurate as the radial basis, it is actually very advantageous in classification and fitting, especially when the data is relatively inaccurate.
And finally, the income of the operation variables is optimized by utilizing a genetic algorithm, so that a light oil yield model is optimized. After the genetic algorithm is deeply combined with the neural network, more reasonable parameter values can be searched for the initialization of the neural network parameters, so that the learning capability of the model is optimized, and the stability and the accuracy of the neural network are greatly improved.
In this embodiment, the historical data of the parameters and the historical data of the calculated actual yield are acquired from a real-time data acquisition system and a Lims system of the catalytic cracking unit.
In the present embodiment, the abnormal values include 0, a negative value, null data, and data having a difference of more than 3 times the standard deviation from the mean.
In this embodiment, the normalization process is: defining the sample sequence as { X(n)Normalizing the data according to a maximum-minimum method, wherein the maximum-minimum method formula is as follows:
Xk＝(Xi-Xmin)/(Xmax-Xmin)，
wherein, XkTo normalized value, XiFor the measured data value, X, obtained in step S10maxAnd XminAre each { X(n)Maximum and minimum values of. After the data is normalized, the optimization process of the optimal solution is obviously gentle, and the optimal solution is easier to be correctly converged.
In some embodiments, the model for parametric prediction of yield comprises: an input layer, a mode layer, a summation layer, and an output layer, corresponding to a network input of X ═ X1,x2,...,xn]The output is Y ═ Y1,y2,...,yk]；
Wherein the number of input layer neurons is equal to the dimension of the input vector in the learning sample, i.e., X ═ X1,x2,...,xn]The neurons are simple distribution units that directly pass input variables to the mode layer;
the number of pattern layer neurons is equal to the number n of learning samples, each neuron corresponds to a different learning sample, and the pattern layer neuron transfer function is:
Figure BDA0002747108140000061
the output of the neuron i is the exponential square of the squared Euclidean distance between the input variable and the corresponding sample X;
the summing layer sums using two types of neurons;
one of the summation formulas is:
Figure BDA0002747108140000062
the summation formula performs arithmetic summation on the outputs of all the neuron algorithms of the mode layer, the connection weight of the mode layer and each neuron is 1, and the transfer function is as follows:
Figure BDA0002747108140000063
another summation formula is:
Figure BDA0002747108140000064
the summation formula carries out weighted summation on all the neurons in the mode layer, and the connection weight value between the ith neuron in the mode layer and the jth numerator summation neuron in the summation layer is the ith output sample YiThe j-th element in (2), the transfer function is:
Figure BDA0002747108140000071
the number of neurons in the output layer is equal to the dimension k of the output vector in the learning sample, each neuron divides the output of the summation layer, and the output of neuron j corresponds to the jth element of the estimation result Y (X), i.e. the output of neuron j is equal to the jth element of the estimation result Y (X)
Figure BDA0002747108140000072
Specifically, in the GRNN model, the number of neurons in the output layer is equal to the dimensionality of the output vector in the learning sample; in the GRNN model training, the number of radial basis neurons and the number of linear neurons are the same as the number of input vectors in an input training sample.
The purpose of the GRNN neural network training is to generate a proper weight matrix R1.1、R2.1And a threshold vector b1The training is specifically as follows: the input vector of the input layer is transmitted to a hidden layer, the hidden layer is provided with M neurons, a node function is a Gaussian function, and the weight matrix is input to be R1.1The threshold vector is b1(ii) a The signal output layer is a specific linear output layer and also comprises M neurons, the weight function is a normalized dot product weight function, the node function is a pure linear function, and the corresponding weight matrix is R2.1The output of the network can be represented by the following expression:
y＝a2＝purelin(R2.1×a1/sum(a1)；
the specific linear output layer has the following characteristics: the hidden layer output is not directly used as the input of the linear neuron, but the output of the hidden layer and the weight matrix R of the layer are firstly used2.1And after normalized dot product operation is carried out, the normalized dot product operation is taken as weight input and then is sent to the transfer function, wherein the transfer function is a linear function.
The GRNN neural network prediction process can obtain an optimal parameter, namely an optimal spread value according to the characteristics of a program of the GRNN neural network prediction process, and then a GRNN model is constructed by adopting an optimal method, so that the predicted value of the dependent variable is very close to the corresponding dependent variable in the sample. And when the optimal spread value is equal to 0.7, obtaining an output value, and finishing the model construction.
In this embodiment, the genetic algorithm comprises the steps of:
s1, taking parameter values participating in catalytic cracking yield optimization at a certain moment as an initial population, setting the number of individuals in the population as M, setting an evolution algebra counter T as 0, and setting a maximum evolution algebra T;
s2, taking the prediction result of the trained parameter prediction yield model as an individual fitness value, and calculating the fitness of each individual in the population;
s3, selecting an individual with a large fitness proportion by adopting a roulette method;
s4, generating a next generation group by using cross operation and mutation operation on the individuals;
s5, judging whether T is equal to T, if so, outputting a corresponding parameter optimization value participating in the catalytic cracking yield optimization by taking the individual with the maximum fitness obtained in the evolution process as an optimal solution, and stopping calculation; if not, returning to repeat the loop to execute S2-S5;
wherein the number of the initial population is 50, and the number of iterations is 100; the prediction result of the trained model for predicting the parameter yield is used as an individual fitness value, and the calculation formula is as follows: fi＝SNj/SDAnd Fi is the fitness value of the individual i.
Specifically, the Genetic Algorithm (GA) is a computational model of a biological evolution process that simulates natural selection and Genetic mechanism of darwinian biological evolution theory, and is a method for searching for an optimal solution by simulating a natural evolution process. Firstly, generating 50 initial populations in a random mode, and optimizing the initial populations by using a genetic algorithm to obtain new populations; and (4) training the new population by a GRNN method, selecting individuals with high fitness value to participate in genetic operation, and eliminating the individuals with low fitness value. And performing a series of genetic operations such as copying, crossing, mutation and the like on the individuals with high fitness value until the iteration times are 100 times, and taking the individuals with the best occurrence in the offspring as the optimal execution result. The cross operation is a main method for generating new individuals, and determines the global search capability of the genetic algorithm; the mutation operation is only an auxiliary method for generating new individuals, but is also an indispensable operation step, and the mutation operation determines the local searching capability of the genetic algorithm. The cooperation of the crossover operator and the mutation operator completes the global search and the local search of the search space, so that the genetic algorithm can complete the optimization process of the optimization problem with good search performance
In some embodiments, the prediction result of the trained model for predicting the parameter yield is used as an individual fitness value, specifically: adopting a roulette method, selecting an individual with larger fitness to enter the next generation, wherein the calculation formula is as follows: f. ofi＝k/Fi，pi＝fi/∑fiIn which F isiIs the fitness value of an individual i, k is a coefficient, piIs the proportion of individual fitness.
The roulette degree selection method comprises the following basic ideas: the probability of each individual being selected is proportional to the fitness of the individual, namely the higher the fitness value is, the higher the probability of the individual being selected is, and the optimization of the selection result is further ensured.
In some embodiments, the genetic algorithm uses a crossover probability of 0.6; the mutation probability of 0.05 is adopted in the mutation operation.
The embodiment of the invention also provides a device for optimizing the light oil yield of a catalytic cracking device, which is characterized by comprising the following components: the system comprises a sample data acquisition module, a parameter analysis module and a parameter analysis module, wherein the sample data acquisition module is used for acquiring historical data of parameters participating in catalytic cracking yield optimization and actual yield values corresponding to the historical data, and performing abnormal value elimination and normalization operation on the acquired historical data of the parameters and the actual yield values to form sample data;
the model obtaining module of the parameter prediction yield is used for training GRNN by utilizing the sample data to obtain a model of the parameter prediction yield;
and the optimization module is used for optimizing the parameters participating in the catalytic cracking yield optimization by using the model prediction result of the parameter prediction yield and adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking device.
Referring to fig. 2, an embodiment of the invention provides a computer terminal device, which includes one or more processors and a memory. A memory is coupled to the processor for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement a method for optimizing the light oil yield of a catalytic cracking unit as in any one of the embodiments described above.
The processor is used for controlling the overall operation of the computer terminal equipment so as to complete all or part of the steps of the method for optimizing the light oil yield of the catalytic cracking unit. The memory is used to store various types of data to support the operation at the computer terminal device, which data may include, for example, instructions for any application or method operating on the computer terminal device, as well as application-related data. The Memory may be implemented by any type of volatile or non-volatile Memory device or combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk, or optical disk.
In an exemplary embodiment, the computer terminal Device may be implemented by one or more Application Specific 1 integrated circuits (AS 1C), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components, and is configured to perform the method for optimizing the light oil yield of the catalytic cracking unit in any of the above embodiments, and achieve the technical effects consistent with the above methods.
In another exemplary embodiment, there is also provided a computer readable storage medium including program instructions which, when executed by a processor, implement the steps of the method for optimizing light oil yield of a catalytic cracking unit in any one of the above embodiments. For example, the computer readable storage medium can be the above-mentioned memory comprising program instructions executable by the processor of the computer terminal device to perform the above-mentioned method for optimizing the light oil yield of a catalytic cracking unit, and to achieve the technical effects consistent with the above-mentioned method.
In summary, the method for optimizing the light oil yield of the catalytic cracking unit combines the GRNN and the genetic algorithm to optimize and optimize each operation parameter, provides a yield optimization operation scheme, greatly improves the light oil yield, can describe the reaction process accurately from the mechanism, and improves the data accuracy.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.

Claims (10)

1. A method for optimizing light oil yield of a catalytic cracking unit is characterized by comprising the following steps:
obtaining historical data of parameters participating in catalytic cracking yield optimization and yield actual values corresponding to the parameters, and performing abnormal value elimination and normalization operation on the obtained historical data of the parameters and the yield actual values to form sample data;
training a GRNN model by using the sample data to obtain a model for predicting yield according to the parameters;
and optimizing the result of the model prediction of the parameter prediction yield by adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking unit.
2. The method for optimizing light oil yield of a catalytic cracking unit as claimed in claim 1, wherein the parameters participating in the optimization of catalytic cracking yield include: the primary reaction outlet temperature, the secondary reaction outlet temperature, the reaction pressure, the remilling ratio, the agent-oil ratio, the slag mixing ratio, the inlet linear speed, the steam lifting amount, the raw material oil preheating temperature, the main air amount and the regeneration temperature.
3. The method of claim 1, wherein the historical data of the parameters and the historical data of the actual yield are collected from a real-time data collection system and a Lims system of the catalytic cracking unit.
4. The method of optimizing light oil yield of a catalytic cracking unit as claimed in claim 1, wherein the abnormal values include 0, negative values, null data and data having a difference of more than 3 times the standard deviation from the mean value.
5. The method for optimizing light oil yield of a catalytic cracking unit according to claim 4,
the model for predicting the yield of the parameters comprises: an input layer, a mode layer, a summation layer, and an output layer, corresponding to a network input of X ═ X1,x2,...,xn]The output is Y ═ Y1,y2,...,yk]；
Wherein the number of input layer neurons is equal to the dimension of the input vector in the learning sample, i.e., X ═ X1,x2,...,xn]The neurons are simple distribution units that directly pass input variables to the mode layer;
the number of pattern layer neurons is equal to the number n of learning samples, each neuron corresponds to a different learning sample, and the pattern layer neuron transfer function is:
Figure FDA0002747108130000011
the output of the neuron i is the exponential square of the squared Euclidean distance between the input variable and the corresponding sample X;
the summation layer uses two types of neuron algorithms to carry out summation;
one of the summation formulas is:
Figure FDA0002747108130000021
the summation formula performs arithmetic summation on the outputs of all the neurons in the mode layer, the connection weight of the mode layer and each neuron is 1, and the transfer function is as follows:
Figure FDA0002747108130000022
another summation formula is:
Figure FDA0002747108130000023
the summation formula carries out weighted summation on all the neurons in the mode layer, and the connection weight value between the ith neuron in the mode layer and the jth numerator summation neuron in the summation layer is the ith output sample YiThe j-th element in (2), the transfer function is:
Figure FDA0002747108130000024
the number of neurons in the output layer is equal to the dimension k of the output vector in the learning sample, each neuron divides the output of the summation layer, the output of neuron j corresponds to the jth element of the estimation result Y (X), namely:
Figure FDA0002747108130000025
6. the method for optimizing light oil yield of a catalytic cracking unit as claimed in claim 5, wherein the genetic algorithm comprises the steps of:
s1, taking parameter values participating in catalytic cracking yield optimization at a certain moment as an initial population, setting the number of individuals in the population as M, setting an evolution algebra counter T as 0, and setting a maximum evolution algebra T;
s2, taking the prediction result of the trained parameter prediction yield model as an individual fitness value, and calculating the fitness of each individual in the population;
s3, selecting an individual with a large fitness proportion by adopting a roulette method;
s4, generating a next generation group by using cross operation and mutation operation on the individuals;
s5, judging whether T is equal to T, if so, outputting a corresponding parameter optimization value participating in the catalytic cracking yield optimization by taking the individual with the maximum fitness obtained in the evolution process as an optimal solution, and stopping calculation; if not, returning to repeat the loop to execute S2-S5;
wherein the number of the initial population is 50, and the number of iterations is 100;
the prediction result of the trained model for predicting the parameter yield is used as an individual fitness value, and the calculation formula is as follows: fi＝SNj/SDAnd Fi is the fitness value of the individual i.
7. The method for optimizing light oil yield of a catalytic cracking unit according to claim 6, wherein the prediction result of the trained model for predicting yield of the parameters is used as an individual fitness value, and specifically comprises:
adopting a roulette method, selecting an individual with larger fitness to enter the next generation, wherein the calculation formula is as follows:
fi＝k/Fi，pi＝fi/∑fi；
wherein FiIs the fitness value of an individual i, k is a coefficient, piIs the individual fitnessThe ratio of the active ingredients to the total amount of the active ingredients.
8. The method for optimizing light oil yield of a catalytic cracking unit as claimed in claim 7, wherein the genetic algorithm is a method in which the crossover probability of crossover operation is 0.6; the mutation probability of 0.05 is adopted in the mutation operation.
9. An apparatus for optimizing the light oil yield of a catalytic cracking unit, comprising:
the system comprises a sample data acquisition module, a parameter analysis module and a parameter analysis module, wherein the sample data acquisition module is used for acquiring historical data of parameters participating in catalytic cracking yield optimization and actual yield values corresponding to the historical data, and performing abnormal value elimination and normalization operation on the acquired historical data of the parameters and the actual yield values to form sample data;
the model obtaining module of the parameter prediction yield is used for training GRNN by utilizing the sample data to obtain a model of the parameter prediction yield;
and the optimization module is used for optimizing the parameters participating in the catalytic cracking yield optimization by using the model prediction result of the parameter prediction yield and adopting a genetic algorithm so as to optimize the light oil yield of the catalytic cracking device.
10. A computer-readable storage medium, characterized in that a computer program is stored, which when being executed by a processor, implements the method for optimizing light oil yield of a catalytic cracking unit according to any one of claims 1 to 8.
CN202011175691.0A 2020-10-28 2020-10-28 Method and device for optimizing light oil yield of catalytic cracking unit and storage medium Active CN112342050B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011175691.0A CN112342050B (en) 2020-10-28 2020-10-28 Method and device for optimizing light oil yield of catalytic cracking unit and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011175691.0A CN112342050B (en) 2020-10-28 2020-10-28 Method and device for optimizing light oil yield of catalytic cracking unit and storage medium

Publications (2)

Publication Number Publication Date
CN112342050A true CN112342050A (en) 2021-02-09
CN112342050B CN112342050B (en) 2021-12-10

Family

ID=74355430

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011175691.0A Active CN112342050B (en) 2020-10-28 2020-10-28 Method and device for optimizing light oil yield of catalytic cracking unit and storage medium

Country Status (1)

Country Link
CN (1) CN112342050B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114495686A (en) * 2022-01-26 2022-05-13 华东理工大学 Real-time simulation method and system for industrial catalytic cracking device

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104789256A (en) * 2015-03-26 2015-07-22 华东理工大学 Real-time yield predicting method for catalytic cracking device
CN104804761A (en) * 2015-03-26 2015-07-29 华东理工大学 Real-time yield prediction method for hydrocracking device

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104789256A (en) * 2015-03-26 2015-07-22 华东理工大学 Real-time yield predicting method for catalytic cracking device
CN104804761A (en) * 2015-03-26 2015-07-29 华东理工大学 Real-time yield prediction method for hydrocracking device

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114495686A (en) * 2022-01-26 2022-05-13 华东理工大学 Real-time simulation method and system for industrial catalytic cracking device

Also Published As

Publication number Publication date
CN112342050B (en) 2021-12-10

Similar Documents

Publication Publication Date Title
Zhong et al. Genetic algorithm optimized double-reservoir echo state network for multi-regime time series prediction
Baughman et al. Neural networks in bioprocessing and chemical engineering
Stach et al. Numerical and linguistic prediction of time series with the use of fuzzy cognitive maps
Vega et al. Integrated design and control of chemical processes–Part I: Revision and classification
Wang et al. Melt index prediction by least squares support vector machines with an adaptive mutation fruit fly optimization algorithm
Eftekhari et al. Extracting compact fuzzy rules for nonlinear system modeling using subtractive clustering, GA and unscented filter
Sánchez et al. Mutual information-based feature selection and partition design in fuzzy rule-based classifiers from vague data
CN112342050B (en) Method and device for optimizing light oil yield of catalytic cracking unit and storage medium
Zhang et al. Task relatedness based multitask genetic programming for dynamic flexible job shop scheduling
Tian et al. Automated selection of evolutionary multi-objective optimization algorithms
Yarushev et al. Review of studies on time series forecasting based on hybrid methods, neural networks and multiple regression
Chen et al. A fuzzy-neural approach for remaining cycle time estimation in a semiconductor manufacturing factory—a simulation study
Hassan et al. Designing intuitionistic fuzzy forecasting model combined with information granules and weighted association reasoning
Guzman-Urbina et al. FIEMA, a system of fuzzy inference and emission analytics for sustainability-oriented chemical process design
CN115938502A (en) Chemical product characteristic prediction method and system based on molecular-level reaction mechanism
Paunović et al. Integrated fuzzy system and multi-expression programming techniques for supplier selection
CN113111588B (en) NO of gas turbine X Emission concentration prediction method and device
CN114492988A (en) Method and device for predicting product yield in catalytic cracking process
Li et al. Modeling and optimum operating conditions for FCCU using artificial neural network
Thao An improved fuzzy time series forecasting model using the differential evolution algorithm
CN112420132A (en) Product quality optimization control method in gasoline catalytic cracking process
Hamedi et al. Integrating artificial immune genetic algorithm and metaheuristic ant colony optimizer with two-dose vaccination and modeling for residual fluid catalytic cracking process
Orazbayev et al. Development of a linguistic model of a reforming unit of a catalytic reforming plant
Wang et al. Analysis of economic order quantity under fuzzy environments
Дебела ALGORITHM OF FUZZY PRIORITIES FOR ALTERNATIVE SOLUTIONS

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