CN108710779B - Optimal modeling method for FCC reaction regeneration process of micro-charge interaction P system in membrane - Google Patents

Abstract

The invention discloses an optimal modeling method for an FCC reaction regeneration process of an intra-membrane micro-charge interaction P system, which aims at performing optimal modeling and rapid high-precision working condition prediction on an FCC reaction regeneration process in an oil refining process. The method comprises the following steps: 1) obtaining sampling data of the process through field operation or experiments, determining the approximate structure of each input/output submodel, and taking the sum of the square error of the estimated output and the actual output of the model as a minimized objective function; 2) ca transported by biological cell membrane ²⁺ 、Na ⁺ 、Cl ^‑ After plasma is carried out, a specific efficient optimization algorithm is abstracted by inspiring the interaction between ions in a new intracellular environment; 3) setting algorithm operation parameters; 4) and estimating unknown parameters in the reaction-regeneration model by using an algorithm through minimizing an objective function, acquiring optimal parameters and forming a mathematical model. The modeling method has the characteristics of precocity resistance, high optimization precision and quick convergence, and is also suitable for modeling other complex chemical reaction processes.

Description

Optimal modeling method for FCC reaction regeneration process of micro-charge interaction P system in membrane

Technical Field

The invention relates to an optimal modeling method, in particular to an optimal modeling method of an in-membrane micro-charge interaction P system optimization algorithm in a fluidized catalytic cracking reaction regeneration process in an oil refining process of a petroleum enterprise.

Background

Energy is a source of social progress, the popularization of automobiles speeds up the pace of urban development, and petrochemical energy, particularly gasoline and diesel oil for automobiles, are main energy sources of the automobiles. The direct stripping of crude oil in a mixed state can only obtain very small amounts of gasoline and diesel oil (about 10% -40% yield of straight run light oil), mainly because the main component in the crude oil is a heavy oil mixture. According to statistics, 78.1% of gasoline in China comes from a Fluid Catalytic Cracking (FCC) process, wherein the FCC process refers to that heavy distillate oil with poor quality and low value (or heavy distillate oil mixed with reduced pressure oil extraction) is contacted with a catalyst at high temperature (450-530 ℃) and low pressure (1-4 atm), and a series of high-value products such as light diesel oil, gasoline, combustible gas and the like are generated through cracking reaction. In consideration of the contradiction between the increasingly depleted crude oil resource supply in the world and the increasingly light product oil resource demand which is necessary to be faced by social development, the heavy oil fluid catalytic cracking is an important direction for the development of the world catalytic cracking in future.

The basic components of a conventional Fluid Catalytic Cracking Unit (FCCU) mainly include: the system comprises a reaction-regenerator, a lifting reactor, a main fractionating tower, an absorption stripping tower, a main fan, a wet air compressor and the like. As one of the most important parts, the reactor-regenerator plays a crucial role in achieving superior process control and maximizing efficiency in the FCC process. The reaction temperature, catalyst circulation rate, steam flow, prevailing air volume within the reaction-regenerator, among other factors, determine product quality and revenue sharing. However, the reaction-regenerator is a complex system with multi-parameter, nonlinear and multivariable close coupling, and the significance of establishing a dynamic model capable of reflecting the chemical reaction rule is important for the design improvement of the reaction-regenerator and the reliable and safe operation of establishing an automatic control system and an analysis control system.

Modeling methods are roughly classified into two categories: mechanism modeling and experimental modeling. Generally, people tend to model mechanisms, and the models are considered to have basic theories as guarantees and have clear physical meanings. For more complex systems, a number of simplifications are made to create a mechanism model. The experimental modeling seems to be an unfortunate method, but the experimental modeling has stronger vitality nowadays when the data processing capacity is greatly improved. The process raw material components processed by the fluidized catalytic cracking process are complex, the temperature gradient change in the physical pipeline of the reactor regenerator is large, and the heat value of each point in the fluid is dynamically changed, which brings great difficulty to the mechanism modeling of the process. Aiming at various input variables and output variables influencing the fluidized catalytic cracking process and matrix coupling among the input variables and the output variables, a step disturbance test method is adopted to obtain the representation characteristics of a response curve and the like of the Multiple Input Multiple Output (MIMO) system. According to a response curve, a scientific model structure is provided, specific sampling data in the curve are utilized, an efficient optimization algorithm is adopted to carry out global optimal estimation on unknown parameters in the model structure, the mathematical model of a research object is obtained by the system identification method, and after actual verification, the trained model can predict the output of a future period of time under the condition that the working condition is unchanged (or slightly adjusted) with higher precision.

Therefore, the model structure is reasonably assumed on the basis of field data sampling, and the parameter estimation of the FCC process model is converted into an optimization problem, so that the method is an effective and feasible scientific modeling problem. However, for the practical engineering optimization problems of complexity, constraint, nonlinearity, multiple local minimum points and the like of the oil cracking process in the fluid catalytic cracking device, the traditional optimization method is easy to fall into a local optimal value or even cannot obtain the optimal value, so that the biologically inspired intelligent optimization method is valued by people. Cells are the cornerstone of biological systems and a delicate 'machine' with a complex structure through long evolution, and the internal intricate behaviors of the cells are self-regulated in an effective manner. In the past, the computing science does not fully consider the inspiration source of the cell as a computing model, and the membrane computing is generated by adapting to the challenge. In 1998, Gheerghe P { hacek over (a) } un, academy of European sciences, the term "P" } over the first letter, membrane computing is also referred to as the P system. Biochemical reactions within or material communication between cell membranes of biological cells are considered as a computational process, and even material communication between cells can be considered as information communication between computational units; however, the scientists seem to consider only the transport of nutrients and metabolites by the cell membrane, and in fact, calcium, magnesium, sodium, carbon, etc. are more present in the cell membrane in ionic state, such as Ca ²⁺ 、Mg ²⁺ 、Na ⁺ 、

When each ion enters the membrane from the outside of the membrane in the framework of the cell membrane, the ion can perform a synergistic action on the existing charged particles in the membrane, and according to the properties of mutual repulsion of the same charges and mutual attraction of the different charges, the mechanism and internal connection of all life activities in the deeper cell can be possibly known.

According to the transformation, energy transmission and transportation functions of substances in biological cells and the interaction of charged ions entering cell membranes with various original ions in the membranes, the method abstracts an in-membrane micro-charge interaction P system optimization algorithm and corresponding rules, and can be used for solving a complex nonlinear optimization problem; the advanced intra-membrane micro-charge interaction P system optimization algorithm is used for solving the problem of model parameter estimation of the fluid catalytic cracking process and obtaining a high-precision mathematical model by considering the high nonlinearity, the coupling and the complexity of the actual process, so that a more ideal effect is obtained.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides an optimal modeling method for an FCC reaction regeneration process of a P system with micro-charge interaction in a film.

The optimal modeling method for the FCC reaction regeneration process of the in-film micro-charge interaction P system comprises the following steps:

1) obtaining sampling data (containing feed preheating temperature MV) of a certain period of time in the catalytic cracking reaction process in the heavy oil lightening of petroleum enterprises through field operation or experiments ₁ And circulating oil flow MV ₂ Residual oil flow MV ₃ Feed oil flow MV ₄ The flow MV of the crude oil to the lifting pipe ₅ Lift pipe outlet temperature MV ₆ (ii) a Regenerator actual Process temperature No. 1

Regenerator actual Process temperature No. 2

Heat of reaction of riser

)；

2) Based on the characteristics of input and output data, combining with public experience, selecting a proper transfer function matrix model and planning to determine a model parameter set and a parameter search range to be determined, and taking the error square sum of the model estimated output of the FCC reaction regenerator and actual output sampling data as a minimized objective function;

3) abstracting a fluidized catalytic cracking process optimization modeling method inspired by the interaction of charged ions in the membrane according to the influence, variation and a cell membrane transport mechanism of the charged ions entering the biological cell membrane on other charged ions; inspired by the influence of charged ions entering into a biological cell membrane on other charged ions, the following rule of an optimization method of the intra-membrane charged ion interaction P system is provided: selecting rules, self-adaptive variation rules, micro-charge interaction rules, alternating current rules and termination rules;

4) setting initial environment and parameters of a P system optimization algorithm inspired by the interaction force of charged ions in the membrane: the number of layers m of the nested membrane structure of the P system is 5, the number of object sets n in each layer of membrane is 10, the number of cycle periods G is 500, and the self-adaptive mutation probability is

a _p ＝0.01，b _p ＝0.29，c _p ＝15/G，G ₀ G/2, mutual force factor K _d 0.2, the size of acceptable error ∈ 1 × 10 ^-4 And a termination rule, wherein g is a current running algebra;

5) extracting each input MV _i To output CV _j The transfer function z of (a) transforms, in the form:

6) the operating membrane internal micro-charge interaction P system optimization algorithm estimates unknown parameters in a fluidized catalytic cracking process model, and the process has 6 input MVs _i And 3 output CV _j (ii) a 1, 2, …, 6; j is 1, 2, 3; after a certain sampling period is selected, the MV is aimed at each sub-process _i →CV _j And determining unknown parameter parameters of the control variable, and simultaneously acquiring a timing sampling actual measurement data table of the control variable under given step interference for later use. Simultaneously forming a parameterized prediction output model;

7) the sum of the squares of errors (SSE) of the predicted output and the measured output under the parametric model is taken as an objective function, as follows:

wherein f is _j For SSE of jth control variable, ns is estimated for each set of CVs _j The number of sampling points used during the parameter; CV of _j，k Is the predicted output of the kth model;

is and CV of _j，k Corresponding to the real process sample value in step 6 for the same given operation.

8) Operating the provided intra-membrane micro-charge interaction P system optimization algorithm, obtaining an estimation value of an unknown parameter in a fluidized catalytic cracking reaction regeneration process model which best fits actual output data through a minimized objective function, and substituting the estimation value into the fluidized catalytic cracking reaction regeneration process model to form a mathematical model of the fluidized catalytic cracking reaction regeneration process;

the film topological structure of the film calculation optimization algorithm of the optimal modeling method for the FCC reaction regeneration process of the in-film micro-charge interaction P system is a nested film structure, except for the surface layer film and the innermost layer film, other films comprise the adjacent inner layer film and are simultaneously contained by the outer layer film, as shown in figure 1;

the optimal modeling method for the FCC reaction regeneration process of the in-film micro-charge interaction P system estimates unknown parameters in a FCC reaction regeneration process model, obtains an unknown parameter estimation value in the FCC reaction regeneration process model by minimizing an objective function, substitutes the estimation value into the FCC reaction regeneration process model, and forms a mathematical model of the FCC reaction regeneration process, and comprises the following steps of:

1) setting the initial environment of the P system optimization method and the search range of 36 parameters to be estimated in each key output variable model group of the fluid catalytic cracking model, and randomly generating an object set;

2) fluid catalytic cracking reaction from the reaction regenerator at feed preheat temperature MV ₁ Preset flow M of circulating oilV ₂ Residual oil preset flow MV ₃ And preset flow MV of feed raw oil ₄ Preset flow MV of raw oil supplied to riser ₅ Preset temperature MV of riser outlet ₆ Actual temperature value of regenerator No. 1

Actual temperature of regenerator No. 2

Heat of reaction of riser

The sum of the error squares of the model predicted values of the three variables and the corresponding sampling values of 3 output variables under 6 operating variables in the actual FCC reaction regeneration process is taken as a target function;

3) the intra-film micro-charge interaction P system optimization algorithm with the nested film structure is calculated from the innermost film, and objects in all film internal regions except the surface film sequentially execute selection rules, alternating current rules, self-adaptive variation rules and micro-charge interaction rule operations;

4) entering step 5) if all the objects in each intramembrane area are updated, or returning to step 3);

5) performing self-adaptive variation rule and selection rule operation on the object in the surface film; if the termination rule is satisfied, go to step 6); otherwise, sending part of optimal solutions, namely part of candidate objects to the innermost layer film through an exchange rule, and continuing the step 3) to finish the optimizing search of the next generation;

6) when the operation of the intra-film micro-charge interaction P system optimization algorithm reaches the termination criterion of the algorithm, the obtained optimal value is used as the estimation value of the unknown parameters of the FCC reaction regeneration process model, and the estimation value is substituted into the FCC reaction regeneration process model to form the mathematical model of the FCC reaction regeneration process.

The in-membrane micro-charge interaction P system optimization algorithm process is that a nested membrane structure starts from an innermost membrane, and rule updating and communication are carried out on an individual set in the nested membrane structure, and is characterized in that a selection rule, an communication rule, a self-adaptive variation rule, a micro-charge interaction rule and a termination rule are provided, wherein the selection rule, the communication rule, the self-adaptive variation rule, the micro-charge interaction rule and the termination rule are inspired by recombination and variation of substances and energy in biological cells and charged ion interaction in membranes:

1) selection rules

According to the physical characteristic description of the cell membrane, the cell membrane is a date cake model with a lecithin bilayer and protein embedded, and the lecithin bilayer has fluidity and can selectively allow part of substances required by life to enter and exit the cell membrane through active transportation, cooperative transportation and free transportation; selecting an object defined as having a small objective function value as a candidate communication object; all objects are the combination of feasible solutions in the search space, the types of micro-charges carried currently and the objective function values of the micro-charges;

2) rules of communication

The exchange rule is expressed as: if the current film is not the outermost film, the current film sends a portion of its candidates to its adjacent outer film; if the current film is the outermost film, judging whether the algorithm meets a termination rule, and if so, directly outputting the optimal individual in the outermost film; otherwise, sending the optimal individual in the current outermost film to the innermost film to replace the equivalent worst individual in the innermost film; the scale of the communication is ceil (number of individuals in the current membrane. the probability of communication), and the ceil () function is rounded up;

3) adaptive mutation rules

In the intra-film micro-charge interaction P system optimization algorithm, an individual can generate a new individual through monomer local variation, the micro-charge interaction optimization algorithm is guided to jump out a local minimum value, and the following self-adaptive variation rule is provided:

p＝rand(m，n)ζ＝rand v＝1，2，…，n l＝1，2，…，m

in the formula (8), p is a randomly generated m × n matrix, and ζ is [0, 1 ]]M is the number of set objects in each layer, n is the number of independent variables contained in each object, p _m Respectively representing the adaptive mutation probability; the parameters in equation (9) are set as follows, respectively: a is _p ＝0.01，b _p ＝0.29，c _p ＝15/G，G ₀ G is the maximum operation algebra set by the micro-charge interaction P system optimization algorithm, and G is the current operation algebra;

4) rule of micro-charge interaction

From a cell biology point of view, cell membranes are capable of recognizing the types of various ionic species while being directed to actively transport these ions; the charged ions and the existing charged particles inside and outside the membrane have certain interaction, and according to the concepts of mutual repulsion of the same charges and mutual attraction of different charges, the micro-charge interaction rule is expressed as follows:

a. randomly assigning a charge "+" or "-" to each object at the beginning of the algorithm, where L ═ 1;

b. when the current intra-membrane individuals are all evolved, assuming that the number of excellent individuals is C (C ═ 1) through the exchange rule, aiming at the C excellent individuals, sequentially sorting the C excellent individuals according to the respective fitness values, and fusing the L-th excellent individual into the outer membrane of the current intra-membrane individual;

c. after the individuals in the outer membrane are sorted by fitness value, the original K (m) P (m) is reserved _ch ) Individuals, starting with K ═ K +1 individuals: calculating the Euclidean space distance between the current K +1 th individual and the individual entering the current film

If the polarity of the current charge carried by the current k-th individual is the same as the polarity of the charge entering the current film, according to the principle that the charges of the same kind repel each other, the current k-th individual is subjected toScanning according to the degree of freedom of variables, wherein j is 1;

d. if x _C1，j ＜x _k，j Then, then

Otherwise

K _d Is an offset coefficient of 0.5;

e. if the polarity of the current charged charge of the current kth individual is different from the polarity of the charge entering the current film, the next operation is carried out according to the principle of mutual attraction of the different charges;

f. if x _C1，j ＜x _K+1，j Then, then

Otherwise

g. If the current individual has no other dimensions, j is j +1, and the step d is returned; otherwise, negating the charged attribute of the current k individuals, wherein the charged attribute negation operation can prevent the good initial candidate solution from being always repelled or attracted and gathered by the constant current optimal solution, and meanwhile k is k + 1; if k exceeds the maximum index of the number of individuals in the single-layer film, directly entering the next step, otherwise, returning to the step c; until the k individual to the n individual are updated;

h.L, deleting the individuals with the worst fitness value in the film to avoid the increase of the number of individuals caused by the entry of the L-th individual into the film and ensure the constant number of individuals in the single-layer film; if L exceeds the index with the maximum number of C excellent individuals, directly entering the next step, and otherwise, returning to the step b;

i. finishing;

it is worth pointing out that: 1) when two comparison bodies are charged with the same kind of charge, and the repulsion force under the distance between the two comparison bodies enables a certain variable of the object to be updated to execute

During operation, the updated object may exceed the search boundary of the current variable, detection is needed in practice, and border-crossing equidistant rebounding is adopted to ensure that each dimension variable of the updated individual does not cross the border; 2) for each time the algorithm allows

All evaluating the fitness value of a new individual, and ensuring that only effective updating is reserved by an elite reservation strategy;

the algorithm under the micro-charge interaction rule enables the individual sets to be randomly distributed again in the whole search space according to the charge property among the objects and the charge interaction principle;

5) termination rule

The termination rule is expressed as the algorithm reaches the maximum running algebra or satisfies the following formula:

in the formula f ^* And f ^best Respectively representing the optimal solution currently found and the global optimal solution of the optimization problem, e ═ 1 × 10 ^-4 Is an acceptable error.

The invention simulates the recombination and variation of the substance and energy in the biological cell and the influence of charged particles entering the cell membrane on other charged ions, provides the characteristics of repelling the same charges between the charged ions and other ions in the membrane, attracting the different charges, attenuating the acting force between the charged ions along with the space distance, and the like, extracts the corresponding mechanism and provides a novel P system optimization algorithm inspired by the mechanism, wherein the like charge mutual repulsion effect disperses a large amount of individuals gathered around the optimal individual by repulsion, the diversity of population in the optimization process is increased, the heterogeneous charge mutual attraction effect utilizes the optimal individual to guide other parts of individuals, by the retention of elite and the inheritance of excellent experience, the algorithm has good global search capability and good local search capability, and the convergence speed and precision of the algorithm are both better.

Drawings

FIG. 1 is a schematic diagram of a P system with a nested membrane configuration;

FIG. 2 is a reaction diagram of a fluidized catalytic cracking reaction regeneration process;

FIG. 3 is a matrix diagram of the input-output relationship of the FCC reaction regeneration process

FIG. 4 manipulated variables MV ₆ Regenerator number 1 temperature response plot under step disturbance

FIG. 5 regenerator temperature response plot for 1 st co-step disturbance

FIG. 6 is a graph comparing the modeling results of different methods for the temperature of regenerator No. 1

Detailed Description

The optimal modeling method for the FCC reaction regeneration process by utilizing the intra-film micro-charge interaction P system comprises the following steps of:

1) after the regeneration process of the fluidized catalytic cracking reaction is in a steady state, the temperature CV of the No. 1 reactor is subjected to disturbance of 5% -10% of each of 6 operating variables of the regeneration process of the fluidized catalytic cracking reaction (or new setting is given to each operating variable of the original FCC reaction at a certain moment after the regeneration process of the FCC reaction is in a steady state) through field operation ₁ Temperature CV of reactor No. 2 ₂ Heat of reaction CV of riser tube ₃ And respectively carrying out timing sampling measurement to obtain 3 batches of sampling data.

2) Respectively drawing a mixed time domain response curve of each output variable by using a point drawing method and taking a time axis as a horizontal axis, comparing an observation curve with a known general mathematical model structure, and confirming the structure of the model and the number and the definition domain of parameters with optimization in the model again.

3) MV of fluid catalytic cracking process for the same output variable ₁ ～MV ₆ The error square sum of the estimated output variable response of the fluid catalytic cracking process model and the actual sampling output data of the fluid catalytic cracking process is taken as a target function;

4) according to the mechanisms of recombination and variation of substances and energy in biological cells and the influence of charged ions entering cell membranes on other charged ions and the like, the rule of the intra-membrane micro-charge interaction P system optimization algorithm is checked by combining a structural framework of a standard P system optimization algorithm: a selection rule, an alternating current rule, a self-adaptive variation rule, a micro-charge interaction rule and a termination rule;

5) setting initial environment and parameters of an intra-film micro-charge interaction P system optimization algorithm: topology, evolution algebra G is 500, and burst mutation probability is

Alternating current probability pc is 0.2 and interaction factor K _d ＝0.2、p _a 0.8, the size of acceptable error ∈ 1 × 10 ^-4 And a termination rule, wherein a _p ＝0.001，b _p ＝0.099，c _p ＝15/G，G ₀ G is G/2, and G is the current running algebra;

6) the proposed intra-membrane micro-charge interaction P system optimization algorithm is operated to estimate the temperature CV of 6 inputs to the No. 1 regenerator in the regeneration process of the fluid catalytic cracking reaction ₁ 36 unknown parameters in the model of (1): wherein, the dynamic data of the actual response process comes from the field experiment measurement, and 6 input variables enter the steady state S in the original process ₁ After that, from the specified time t ₁ Step interference is applied to the data at the same time when the data is 0, a series of temperature values in the FCC reaction regeneration process are recorded at the same time, the sampling period is kept consistent, and all data sampling is synchronous; obtaining an estimated value of an unknown parameter in an FCC reaction regeneration process model by minimizing an objective function, substituting the estimated value into the FCC reaction regeneration process model to form an optimal high-precision mathematical model of the FCC reaction regeneration process, and providing scientific and reliable guidance for next working condition prediction;

examples of the embodiments

As shown in FIG. 2, a 140-million-ton heavy oil FCC reaction regeneration device in a certain refinery is shown, wherein feed oil is mixed with residual oil and then mixed with circulating oil, and a regenerated catalyst is added to the mixture and then enters the bottom of a riser to start circulation. In the riser, under the action of catalyst, the crude oil is cracked into dense phase mixture of hydrocarbon steam and carbon, and simultaneously the elements or impurities of sulfur, nitrogen, carbon, nickel, vanadium and the like which are not removed in the crude oil cause catalyst poisoning and deactivation. In the separation column, the hydrocarbonThe like dilute phase gas rises and is sent to a fractionating device to obtain products such as gasoline, diesel oil, fuel gas and the like, the waste catalyst containing a large amount of coke settles to the bottom of the separating tower by gravity and flows out to a No. 1 regenerator, under the introduction of pressurized air, a large amount of coke in the dense phase fluid is continuously ignited and combusted to generate CO and CO ₂ And (3) waiting for the flue gas to escape, discharging the rest waste catalyst containing a small amount of coke from the bottom of the 1# regenerator, and conveying the waste catalyst to the 2# regenerator again, wherein the small amount of coke is completely combusted to become gas to escape, thereby obtaining the regenerated catalyst. The heat generated by combustion during catalyst regeneration provides the desired temperature environment for the cracking reaction within the riser. The regenerated catalyst is mixed with residual oil and circulating oil and then enters the riser to enter the next cycle.

In fig. 2, the combustion heat of the 1# regenerator and the 2# regenerator must be strictly maintained within a relatively small fluctuation range, which requires the actual process to be relatively stable. Otherwise, insufficient combustion of the coke cannot provide sufficient temperature for the cracking reaction in the riser, and the violent combustion of the coke may damage the normal cracking process and even cause fire explosion accidents. From the analysis, the factors affecting the cracking process are: actual regenerator 1# temperature, actual regenerator 2# temperature and heat of reaction in the riser. And the controllable factors are: feed preheat temperature, cycle oil flow, residual oil flow, feed flow, crude oil supply flow to the riser and riser outlet oil temperature. For the whole process, the former is defined as the system control variable and CV is used ₁ 、CV ₂ And CV ₃ Represents; defining the latter as system operating variables, using MVs ₁ 、MV ₂ 、MV ₃ 、MV ₄ 、MV ₅ And MV ₆ Then the input-output diagram of the whole system can be shown in fig. 3 using the following matrix diagram.

The usual ranges of variation for system control variables and manipulated variables in the reaction regenerator were chosen as follows:

TABLE 1 operating variables and control variables in FCC reaction regenerator

Based on the input and output diagram of the fluidized catalytic reaction regeneration process, the system is analyzed firstly and the process model is established in three steps by combining the actual reaction process. An analysis system: 1) in practice, the bed temperature of the 1# regenerator and the 2# regenerator of the process must be strictly kept within a very small range on site; 2) CO in 1# regenerator in flue gas in view of instability of on-line gas composition and concentration measurement ₂ Concentration of CO and O in 2# regenerator ₂ The concentration of the gas is measured accurately off line after sampling gas with equal volume at fixed time in the dynamic process; 3) the reaction heat in the riser is directly related to the cracking degree of the feed oil, and the dynamic cracking degree cannot be expressed in the static dilute phase component, so that the reaction heat of the riser is directly used for soft measurement of the cracking degree in the actual riser. For each test, the step size selected should be as reasonable as possible to ensure that the response of the process output is clearly observed and that the process model is obtained by identifying the entire set of step test data for a parameterized model.

The first step is as follows:

a) the setpoint is normalized (in MV) when the entire system is already in steady state ₆ For example, the following steps are carried out: due to MV ₆ The set value of the outlet temperature of the riser is 507-517 ℃, the normal range after normalization is 0.98-1), and when the MV is ₆ The setpoint is adjusted up by 3.33% of the current setpoint with the remaining MVs _k All are kept unchanged, and MV can be approximately obtained by observing the actual temperature response curve of the No. 1 regenerator ₆ →CV ₁ Temperature CV of regenerator No. 1 under input step disturbance ₁ The sampling period is 1min, the process of the whole process is recorded, the total time is 90min, and the whole data is shown in table 2:

TABLE 2 temperature detection data recording table for regenerator No. 1 under interference of riser temperature step

Time

Mean value of temperature

Temperature sampling

Time

Mean value of temperature

Temperature sampling

Time

Mean value of temperature

Temperature sampling

1min

653.04℃

652.50℃

31min

538.16℃

547.61℃

61min

529.40℃

515.79℃

2min

653.03℃

668.94℃

32min

536.81℃

545.83℃

62min

529.39℃

517.90℃

3min

653.03℃

669.40℃

33min

535.67℃

555.65℃

63min

529.38℃

537.97℃

4min

653.02℃

635.44℃

34min

534.7℃

550.24℃

64min

529.38℃

514.59℃

5min

653.02℃

669.20℃

35min

533.88℃

523.20℃

65min

529.37℃

513.01℃

6min

653.01℃

653.19℃

36min

533.18℃

525.43℃

66min

529.37℃

520.35℃

7min

653.01℃

653.66℃

37min

532.59℃

526.63℃

67min

529.36℃

509.48℃

8min

653.01℃

645.77℃

38min

532.08℃

532.61℃

68min

529.36℃

525.93℃

9min

653.00℃

672.47℃

39min

531.66℃

535.30℃

69min

529.36℃

510.43℃

10min

653.00℃

652.76℃

40min

531.3℃

545.14℃

70min

529.35℃

537.75℃

11min

653.00℃

643.65℃

41min

531.00℃

527.48℃

71min

529.35℃

546.87℃

12min

650.25℃

633.88℃

42min

530.74℃

544.4℃

72min

529.35℃

518.95℃

13min

643.49℃

661.4℃

43min

530.52℃

521.30℃

73min

529.35℃

516.59℃

14min

634.54℃

617.49℃

44min

530.34℃

526.95℃

74min

529.35℃

522.05℃

15min

624.60℃

624.63℃

45min

530.18℃

531.68℃

75min

529.35℃

544.83℃

16min

614.48℃

609.84℃

46min

530.05℃

528.77℃

76min

529.35℃

535.43℃

17min

604.66℃

595.75℃

47min

529.94℃

521.43℃

77min

529.35℃

515.36℃

18min

595.45℃

612.00℃

48min

529.85℃

516.98℃

78min

529.35℃

536.60℃

19min

586.99℃

588.18℃

49min

529.77℃

515.92℃

79min

529.34℃

524.78℃

20min

579.35℃

577.93℃

50min

529.7℃

532.57℃

80min

529.34℃

524.85℃

21min

572.54℃

590.18℃

51min

529.65℃

541.74℃

81min

529.34℃

529.33℃

22min

566.52℃

548.53℃

52min

529.6℃

510.92℃

82min

529.34℃

515.24℃

23min

561.25℃

571.71℃

53min

529.56℃

530.94℃

83min

529.34℃

532.83℃

24min

556.66℃

567.47℃

54min

529.53℃

529.46℃

84min

529.34℃

543.17℃

25min

552.68℃

565.79℃

55min

529.5℃

547.71℃

85min

529.34℃

532.95℃

26min

549.25℃

534.26℃

56min

529.47℃

539.40℃

86min

529.34℃

547.56℃

27min

546.29℃

526.93℃

57min

529.45℃

531.64℃

87min

529.34℃

531.59℃

28min

543.76℃

551.30℃

58min

529.43℃

545.06℃

88min

529.34℃

515.27℃

29min

541.59℃

556.32℃

59min

529.42℃

534.41℃

89min

529.34℃

548.67℃

30min

539.74℃

544.92℃

60min

529.41℃

543.09℃

90min

529.34℃

525.69℃

The chart is shown in fig. 4, which is a graph formed according to the time axis.

According to CV of ₁ The MV can be approximated as a result of the disturbance response ₆ →CV ₁ The process of (a) is considered to be a 2-step system with damping, so the MV can be assumed ₆ →CV ₁ The model structure of (2) is as follows:

in view of the timing sampling, the model structure for transforming the above equation into a discrete form by z-transform is as follows:

due to MV ₆ Step perturbation of (2) is a positive rise, while CV ₁ Response is a decline in attenuation, so C ₆₁ Is negative.

b) Repeating the operation of a) sub-step in the first step, but with the manipulated variable being set by MV ₆ Modified into MV ₅ By 5 monitoring, a new MV is obtained _i，1-5 →CV ₁ 5 model structures of (1);

c) repeating the sub-steps a) and b) of the first step, but passing the control variable from the CV ₁ To CV ₂ By 6 monitoring, a new MV is obtained _i →CV ₂ 6 model structures of (a);

d) repeating the operations of a) and b) of the first step, but with the controlled variable being determined by CV ₁ To CV ₃ By 6 monitoring, a new MV is obtained _i →CV ₃ 6 model structures of (1);

the second step is that:

a) the system reverts to the original steady state while simultaneously observing the MV-CV for new set values (the disturbance amplitudes may be completely different from each other within the safety range, but are to be recorded) for 6 process operating variables ₁ Is added and coupled out, records CV ₁ Responding output data of the whole process; by CV of ₁ For example, an initial state may be selected as a cold system steady state, at MV ₁ ～MV ₆ The ends add disturbance of a certain amplitude respectively, temporarily fix the MV ₁ ～MV ₆ Are all equal in magnitude, in view of the MV _i -CV ₁ Each link comprises time lag and inertia, so in the centralized test, the sampling period is adjusted to 6min, the total test time is 4h (namely 240min), and CV is realized under the disturbance of 6 operation variables ₁ The temperature recorded values of (a) are shown in table 3.

Table 36 table of temperature sensing data of regenerator No. 1 under common step disturbance of manipulated variables

Time	Mean value of temperature	Temperature sampling	Time	Mean value of temperature	Temperature sampling
						6min	653.00℃	653.50℃	126min	661.41℃	661.67℃
12min	652.54℃	652.74℃	132min	661.38℃	661.67℃
						18min	651.58℃	651.39℃	138min	661.37℃	661.11℃
24min	651.26℃	650.97℃	144min	661.37℃	661.36℃
						30min	650.10℃	649.92℃	150min	661.37℃	661.66℃
36min	649.98℃	651.06℃	156min	661.38℃	661.62℃
						42min	647.26℃	647.24℃	162min	661.38℃	661.20℃
48min	647.33℃	647.24℃	168min	661.38℃	660.80℃
						54min	654.52℃	654.43℃	174min	661.38℃	661.50℃
60min	660.95℃	661.64℃	180min	661.38℃	661.52℃
						66min	663.83℃	663.05℃	186min	661.38℃	662.06℃
72min	663.81℃	664.31℃	192min	661.38℃	660.67℃
						78min	662.60℃	662.27℃	198min	661.38℃	661.22℃
84min	661.50℃	661.38℃	204min	661.38℃	661.57℃
						90min	660.99℃	661.3℃	210min	661.38℃	661.33℃
96min	660.98℃	660.55℃	216min	661.38℃	662.30℃
						102min	661.17℃	661.36℃	222min	661.38℃	661.21℃
108min	661.35℃	661.31℃	228min	661.38℃	660.31℃
						114min	661.43℃	660.80℃	234min	661.38℃	661.71℃
120min	661.44℃	661.53℃	240min	661.38℃	660.74℃

The chart is plotted against time as shown in figure 5.

b) Repeating the operation of a) substeps of the second step to apply the controlled variable to the CV ₁ To CV ₂ By applying step disturbance while observing MV-CV ₂ Is added and coupled out, records CV ₂ Various input and comprehensive response output data of the whole process are used for standby;

c) repeating the operation of sub-step b) of the second step to obtain the controlled variable from CV ₂ To CV ₃ By applying stepsJump interference while observing MV-CV ₃ Is added and coupled out, records CV ₃ Various input and comprehensive response output data of the whole process are used for standby;

the third step:

in view of the mutual coupling between input and output variables, respectively for CV ₁ 、CV ₂ And CV ₃ Performing disposable MV ₁ ～MV ₆ To CV ₁ ，CV ₂ ，CV ₃ It should be noted that the reference to the model structure in the first step is the basis of the number of parameters to be optimized.

The method comprises the following steps of performing model selection and parameter optimal estimation on the FCC reaction regeneration process by utilizing an intra-membrane micro-charge interaction P system optimization algorithm, and specifically comprising the following steps of:

a) first aiming at MV-CV ₁ The 6 processes are subjected to parameter estimation of batch models, 40 groups of input and output sampling data (shown as table 1) of the actual process are obtained through experiments and are used as training samples of parameter estimation, and an optimization index function is as

Wherein CV is _1，k Is the output data of the model, and the calculation formula is

The sampled values are output for the process at the same MV. The optimization index is used as an objective function in optimizing search of the optimal modeling method in the regeneration process of the in-film micro-charge interaction P system FCC reaction;

b) setting the running environment of the program running, wherein the number m of the embedded film structure of the P system is 5, the number n of object sets in each film layer is 10, the number G of cycle periods is 500, and the adaptive mutation probability is

a _p ＝0.01，b _p ＝0.29，c _p ＝15/G，G ₀ G/2, mutual force factor K _d 0.2, the size of acceptable error ∈ 1 × 10 ^-4 And a termination rule, wherein g is a current running algebra; of an algorithmThe termination criteria is as described in the previous section, and one of 2 conditions is satisfied;

c) an optimization algorithm for the P system of the micro-charge interaction in the operating film is used for estimating unknown parameters in an FCC reaction regeneration process model, and the MV parameter is aimed at a first sub-process _i →CV ₁ The 36 parameters that best fit the actual sampled data were determined to complete a model of the 6 operating variables to regenerator number 1 temperature as shown in table 4.

TABLE 46 model parameters of the manipulated variables for the temperature of regenerator # 1

And comparing the modeling result with a well-known genetic algorithm, wherein the genetic algorithm directly adopts a ga tool box in international standard software Matlab, a matching interface is set, the search range, the initial population scale and the cycle algebra are consistent, and other default settings are selected. A comparison of the fitting results of the different algorithms is shown in fig. 6.

The fourth step:

similarly, MV-CV is loaded ₂ Obtaining a mathematical model of the temperature of the regenerator No. 2 in the FCCU reaction-regeneration process;

the fifth step:

similarly, load MV-CV ₃ Obtaining a mathematical model of the temperature of the regenerator No. 2 in the FCCU reaction-regeneration process;

as can be seen from FIG. 6, the square sum of the error between the temperature mathematical model prediction of the regenerator No. 1 obtained by the fluidized catalytic cracking process optimization modeling method inspired by the intra-membrane charged ion interaction and the sampling temperature series of the actual regenerator is the minimum, and compared with other methods, the method has higher goodness of fit and better prediction precision.

Claims (6)

1. An optimal modeling method for an FCC reaction regeneration process of an in-film micro-charge interaction P system is characterized by comprising the following steps of:

1) obtaining key sampling data of a fluidized catalytic cracking process of an oil refinery through field operation or experiments, predicting 3 main outputs of a reaction-regenerator by combining an initial non-precise model for various inputs of each group of the reaction-regenerator, and taking the error square sum of the main outputs and the actually monitored outputs of the reaction-regenerator under the equivalent input variables as a target function;

2) according to the transport effect of cell membranes on nutrients entering and exiting cells and the interaction of charged ions entering the membranes, the interaction refers to mutual repulsion of charges of the same kind and mutual attraction of charges of different kinds, and the high-efficiency optimization modeling method for the regeneration process of the petroleum fluid catalytic cracking reaction is abstracted out; inspired by the interaction of charged ions in a cell membrane, the following rule of an optimization algorithm of a micro-charge interaction P system in the cell membrane is provided: selecting rules, transferring rules, micro-charge interaction rules, self-adaptive variation rules and termination rules;

3) setting an initial environment and parameters of a P system optimization algorithm: the cycle number G of the topological structure and P system operation is 500, and the self-adaptive mutation probability is

Probability of interchange p _c 0.2, probability of transit p _sc 0.2, a hyperspace distance factor K _d 0.2, the size of acceptable error ∈ 1 × 10 ^-4 And a termination rule, wherein a _p ＝0.01，b _p ＝0.29，c _p ＝15/G，G ₀ G is G/2, and G is the current running algebra;

4) and (3) operating the extracted intra-membrane micro-charge interaction P system optimization algorithm to estimate unknown parameters in a reaction regeneration process model in the fluid catalytic cracking FCC process for lightening heavy oil products: wherein 6 sets of input variables and 3 sets of output variables have cross transfer relationship, and the z variation model of the transfer function can be expressed as the following model:

as can be seen, for each sub-process MV _i (z)→CV _j (z) as long as the parameter a is determined _ij (1)、a _ij (2)、a _ij (3)、b _ij (1)、b _ij (2) And d _ij (i ═ 1, 2, …, 6), a mathematical model of the FCC reaction regeneration process can describe the process characteristics of the subject under study; each control variable has 36 parameters to be identified;

5) to ensure the prediction accuracy of the model, we need to select an optimal set of parameters for the model under which a certain objective function value can be minimized by minimizing the following objective function:

in the formula (f) _j SSE for the jth control variable; ns is the estimation of each CV _j The number of sampling points used during the parameter; CV of _j，k (ii) a predicted output for the kth model calculated according to equation (1);

is equal to CV _j，k Corresponding real process sampling values under the same given operation; obtaining an estimation value of unknown parameters in an FCC reaction regeneration process model, substituting the estimation value into the FCC reaction regeneration process model to form a mathematical model capable of predicting three key output quantities of the FCC reaction regeneration process with high precision;

wherein, from the viewpoint of cell biology, the cell membrane can identify the types of various ion substances and carry out active transport on the ions; the charged ions and the existing charged particles inside and outside the membrane have certain interaction, and according to the concepts of mutual repulsion of the same charges and mutual attraction of the different charges, the micro-charge interaction rule has the following detailed steps:

a. randomly assigning a charge "+" or "-" to each object at the beginning of the algorithm, where L ═ 1;

b. when all the individuals in the current film have evolved, if the number of excellent individuals obtained through the rule of communication is C (C ═ 1), aiming at the C excellent individuals, sequentially sorting the C excellent individuals according to the respective fitness values, and fusing the L-th excellent individual into the outer film of the L-th excellent individual;

c. after the individuals in the outer membrane are sorted by fitness value, the original K (m) P (m) is reserved _ch ) Individuals, starting from K ═ K +1 individuals: calculating the Euclidean space distance between the current k-th individual and the individual entering the current film

If the polarity of the current charged charge of the current k-th individual is the same as the polarity of the charge entering the current film, scanning the current k-th individual according to the variable degree of freedom according to the principle that the charges of the same kind repel each other, wherein j is 1;

d. if x _C1，j ＜x _k，j Then, then

Otherwise

K _d Is an offset coefficient of 0.5;

e. if the polarity of the current charge carried by the current kth individual is different from the polarity of the charge entering the current film, the next operation is carried out according to the principle that the charges of different types attract each other;

f. if x _C1，j ＜x _k，j Then, then

Otherwise

g. If the current individual has no traversal in other dimensions, j is j +1, and the step d is returned; otherwise, negating the charged attribute of the current k individuals, wherein the charged attribute negation operation can prevent the good initial candidate solution from being always repelled or attracted and gathered by the constant current optimal solution, and k is k + 1; if k exceeds the index with the maximum number of individuals in the single-layer film, directly entering the next step, and otherwise, returning to the step c; until the kth individual to the mth individual have completed updating;

h.L, deleting the individuals with the worst fitness value in the film to avoid the increase of the number of individuals caused by the entry of the L-th individual into the film and ensure the constant number of individuals in the single-layer film; if L exceeds the index with the maximum number of C excellent individuals, directly entering the next step, and otherwise, returning to the step b;

i. ending;

it is worth pointing out that: 1) when two comparison bodies are charged with the same kind of charge, and the repulsion force under the distance between the two comparison bodies enables a certain variable of the object to be updated to execute

During operation, the updated object may exceed the search boundary of the current variable, detection is needed in practice, and boundary-crossing equidistant springback is adopted to ensure that each dimension variable of the updated individual does not cross the boundary; 2) for each case when the algorithm allows

All evaluating the fitness value of a new individual, and ensuring that only effective updating is reserved by an elite reservation strategy; the algorithm under the micro-charge interaction rule enables the individual sets to be randomly distributed again in the whole search space according to the charge property among the objects and the charge interaction principle.

2. The optimal modeling method for the FCC reaction regeneration process of the in-film micro-charge interaction P system according to claim 1, wherein the individual update rule of the in-film micro-charge interaction P system optimization algorithm is as follows: under the micro-charge force, the same charges repel each other, the different charges attract each other and the super-space distance repulsion force is nonlinear to form an individual updating rule to be optimized.

3. The method for optimally modeling the FCC reaction regeneration process in an in-membrane micro-charge interaction P system according to claim 1, wherein the FCC process is a current oil catalytic cracking reaction, a powder catalyst is used, and the catalytic cracking production device makes the catalyst in a "fluidized circulation state" in both the reaction and regeneration devices.

4. The method according to claim 1, wherein the P-system optimization algorithm has a nested film topology, and the other films include the adjacent inner film and are also included by the outer film except the surface film and the innermost film.

5. The method for optimally modeling the FCC reaction regeneration process of the P system with the micro-charge interaction in the membrane according to claim 1, wherein the P system optimization algorithm estimates the unknown parameters in the FCC process model by the following steps:

1) setting the initial environment of the P system optimization method and the search range of 36 parameters to be estimated in each key output variable model group of the fluid catalytic cracking model, and randomly generating an object set;

2) fluid catalytic cracking reaction from the reaction regenerator at feed preheat temperature MV ₁ Preset flow MV of circulating oil ₂ Residual oil preset flow MV ₃ And preset flow MV of feed raw oil ₄ Preset flow MV of raw oil supplied to riser ₅ Preset temperature MV of riser outlet ₆ Actual temperature value CV of regenerator No. 1 ₁ Actual temperature CV of regenerator No. 2 ₂ Heat of reaction CV of riser ₃ The sum of the error squares of the model predicted values of the three variables and the corresponding sampling values of 3 output variables under 6 operating variables in the actual FCC reaction regeneration process is taken as a target function;

3) the P system optimization algorithm with the nested membrane structure starts to calculate from the innermost membrane, and objects in all membrane inner regions except the surface membrane sequentially execute selection rules, alternating current rules, self-adaptive variation rules and micro-charge interaction rule operations;

4) entering step 5) if all the objects in each intramembrane area are updated, or returning to step 3);

5) performing self-adaptive variation rule and selection rule operation on the object in the surface film; if the termination rule is satisfied, go to step 6); otherwise, sending part of optimal solutions, namely part of candidate objects to the innermost layer film through an exchange rule, and continuing to the step 3) to finish the optimization search of the next generation;

6) when the intra-membrane micro-charge interaction P system optimization algorithm reaches the termination criterion of the algorithm, the obtained optimal value is used as the estimated value of the unknown parameter of the fluid catalytic cracking process model, and the estimated value is substituted into the fluid catalytic cracking process model to form the mathematical model of the fluid catalytic cracking process.

6. The optimal modeling method for the FCC reaction regeneration process of the in-film micro-charge interaction P system according to claim 1, wherein the selection rule, the alternating current rule, the adaptive variation rule and the termination rule are as follows:

1) selection rules

According to the physical characteristic description of the cell membrane, the cell membrane is a date cake model with a lecithin bilayer and protein embedded, and the lecithin bilayer has fluidity and can selectively allow part of substances required by life to enter and exit the cell membrane through active transportation, cooperative transportation and free transportation; selecting an object defined as having a small objective function value as a candidate communication object; all objects are the combination of feasible solutions in the search space, the types of micro-charges carried currently and the objective function values of the micro-charges;

2) rules of communication

The exchange rule is expressed as: if the current film is not the outermost film, the current film sends a portion of its candidates to its adjacent outer film; if the current film is the outermost film, judging whether the algorithm meets a termination rule, and if so, directly outputting the optimal individual in the outermost film; otherwise, sending the optimal individual in the current outermost film to the innermost film to replace the equivalent worst individual in the innermost film; the scale of the communication is ceil (number of individuals in the current membrane. the probability of communication), and the ceil () function is rounded up;

3) adaptive mutation rules

In the intra-film micro-charge interaction P system optimization algorithm, an individual can generate a new individual through monomer local variation, the micro-charge interaction optimization algorithm is guided to jump out a local minimum value, and the following self-adaptive variation rule is provided:

in the formula (3), p is a randomly generated m × n matrix, and ζ is [0, 1 ]]M is the number of set objects in each layer, n is the number of independent variables contained in each object, p _m Respectively representing the adaptive mutation probability; the parameters in equation (4) are set as follows: a is _p ＝0.01，b _p ＝0.29，c _p ＝15/G，G ₀ ＝G/2；

4) Termination rule

The termination rule is expressed as the algorithm reaches the maximum running algebra or satisfies the following formula:

in the formula f ^* And f ^best Respectively representing the optimal solution currently found and the global optimal solution of the optimization problem, e ═ 1 × 10 ^-4 Is an acceptable error.

Images