CN114036670A - Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning - Google Patents
Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning Download PDFInfo
- Publication number
- CN114036670A CN114036670A CN202111323572.XA CN202111323572A CN114036670A CN 114036670 A CN114036670 A CN 114036670A CN 202111323572 A CN202111323572 A CN 202111323572A CN 114036670 A CN114036670 A CN 114036670A
- Authority
- CN
- China
- Prior art keywords
- liquid
- hydrocyclone
- solid
- point
- transfer learning
- 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
Links
- 239000007787 solid Substances 0.000 title claims abstract description 105
- 238000000926 separation method Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000013526 transfer learning Methods 0.000 title claims abstract description 49
- 238000004088 simulation Methods 0.000 claims abstract description 40
- 239000007788 liquid Substances 0.000 claims abstract description 17
- 230000005514 two-phase flow Effects 0.000 claims abstract description 11
- 239000007790 solid phase Substances 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000009826 distribution Methods 0.000 claims description 18
- 238000013508 migration Methods 0.000 claims description 14
- 230000005012 migration Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 12
- 239000012071 phase Substances 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 9
- 239000002245 particle Substances 0.000 claims description 9
- 230000000704 physical effect Effects 0.000 claims description 8
- 239000012530 fluid Substances 0.000 claims description 7
- 230000008878 coupling Effects 0.000 claims description 6
- 238000010168 coupling process Methods 0.000 claims description 6
- 238000005859 coupling reaction Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 238000010998 test method Methods 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010992 reflux Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 abstract description 2
- 238000012360 testing method Methods 0.000 description 17
- 238000005457 optimization Methods 0.000 description 16
- 238000013461 design Methods 0.000 description 12
- 238000004519 manufacturing process Methods 0.000 description 9
- 230000008901 benefit Effects 0.000 description 4
- 239000012141 concentrate Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000004836 empirical method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 235000013305 food Nutrition 0.000 description 2
- 238000003780 insertion Methods 0.000 description 2
- 230000037431 insertion Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 239000003973 paint Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004062 sedimentation Methods 0.000 description 2
- 238000005406 washing Methods 0.000 description 2
- 229920002472 Starch Polymers 0.000 description 1
- 238000005352 clarification Methods 0.000 description 1
- 239000011362 coarse particle Substances 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000010840 domestic wastewater Substances 0.000 description 1
- 239000010419 fine particle Substances 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 239000010842 industrial wastewater Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000005191 phase separation Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 235000019698 starch Nutrition 0.000 description 1
- 239000008107 starch Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002351 wastewater Substances 0.000 description 1
- 238000004065 wastewater treatment Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Mathematics (AREA)
- Artificial Intelligence (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Fluid Mechanics (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method for optimizing the separation performance of a liquid-solid hydrocyclone based on transfer learning, which comprises the following steps: training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, and modeling the working condition under an unknown mode through transfer learning to determine characteristic parameters of the liquid-solid hydrocyclone; acquiring empirical characteristic parameters of the liquid-solid hydrocyclone; performing numerical simulation according to the characteristic parameters of the liquid-solid hydrocyclone and the acquired empirical characteristic parameters; determining and obtaining a geometric model of the liquid-solid hydrocyclone determined by transfer learning through numerical simulation; calculating the flow characteristic, the air column characteristic and the solid phase flow characteristic of the gas-liquid two-phase flow and the simulated separation efficiency of the liquid-solid hydrocyclone; and then calculating to obtain the optimal separation efficiency value of the liquid-solid hydrocyclone. The invention can efficiently realize the improvement of the separation performance of the liquid-solid separation hydrocyclone.
Description
Technical Field
The invention relates to the technical field of design of hydrocyclones, in particular to a method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning.
Background
The hydrocyclone is a commonly used two-phase separation device in industrial production and has the advantages of simple structure, convenient operation, high production capacity, high separation efficiency, small occupied area and the like. The application of the hydrocyclone comprises solid-liquid separation, liquid-gas separation, solid-solid separation, liquid-liquid separation, liquid-gas-solid three-phase simultaneous separation and the like. The hydrocyclone is used as a separation and classification device, the basic working principle is based on the centrifugal sedimentation effect, when two-phase mixed liquid to be separated enters a device from an inlet of the hydrocyclone under certain pressure, strong rotary motion is generated, because the density difference exists between a light phase and a heavy phase, the centrifugal force, the centripetal buoyancy and the fluid drag force are different, under the centrifugal sedimentation effect, most of the heavy phase is discharged through a bottom flow port of the hydrocyclone, and most of the light phase is discharged from an overflow port, so that the separation purpose is achieved.
At present, the hydrocyclone is widely used in the fields of petroleum, chemical industry, mining industry, food, environmental protection and the like. For example, classification, sorting, product concentration, paint washing clarification and the like in mining and metallurgy engineering; oil-water separation in petrochemical industry; grading coarse and fine particles, removing impurities, washing paint of starch and the like in the food industry; preparation of paper coating in the paper industry, treatment of paper-making wastewater and the like; industrial and domestic wastewater treatment in environmental engineering, and the like.
Hydrocyclones are generally used for industrial scale production, and the performance of hydrocyclones has an important influence on the production capacity, economic benefit and product quality thereof. In practical applications, hydrocyclones are usually selected according to production capacity, classification particle size or existing products, or preliminary design of basic diameters and related structural parameters of the hydrocyclones is carried out by an empirical method, and the hydrocyclones determined according to the modes are difficult to realize the optimal performance under specific production requirements. Therefore, the method has important engineering significance for further optimization of the hydrocyclone.
At present, the optimization research of the hydrocyclone is more specific to the engineering application problem. (1) The influence of each factor on the performance of the hydrocyclone is quantitatively analyzed through a single-factor test method, and then the selection of key parameters of the hydrocyclone is realized. The method can intuitively obtain the influence rule of each factor on the performance of the hydrocyclone, but the test quantity is huge, and meanwhile, the influence of the interaction among the factors is ignored; (2) and (3) designing a test scheme by using a test design method, and performing statistical analysis on test results to realize parameter optimization. The method is also limited by the test cost and workload, the selection of the number of test factors is greatly limited, and only part of parameters can be optimized; (3) and the other method is to apply a numerical simulation method and analyze the influence of each influencing factor on the internal flow field characteristic and the performance of the hydrocyclone through single-factor numerical simulation. The method can perform approximate quantitative analysis on the partial performance of the hydrocyclone by selecting a proper turbulence model and simulation parameters, and simultaneously obtains the influence rule of each factor on the internal flow field, thereby analyzing the separation mechanism of the hydrocyclone. However, the numerical simulation method is greatly influenced by model selection and grid division methods, and the current numerical simulation method is mostly used for researching the influence rule of each factor on the flow field characteristics in the hydrocyclone, has a certain theoretical guidance effect on the structural design and process formulation of the hydrocyclone, but is difficult to be directly used in engineering application. (4) And the other method is to analyze the internal flow field of the hydrocyclone through numerical simulation, so as to improve the structure of the hydrocyclone according to the existing hydrocyclone separation mechanism. Obviously, the improvement is still obtained by a qualitative method, the optimization capability is limited, meanwhile, the pertinence is strong, and the reference of other engineering application problems is inconvenient.
Specifically, the separation performance of hydrocyclones is affected by a number of factors including: physical property parameters, operational parameters, and structural parameters. Wherein, the physical property parameters comprise 2 items of particle density and particle granularity; the operation parameters comprise 2 items of ore pulp concentration and inlet speed; the structural parameters comprise 8 items of basic diameter of the hydrocyclone, length of a column section, cone angle, inlet diameter, overflow port diameter, underflow port diameter, insertion depth of the overflow pipe, wall thickness of the overflow pipe and the like. In addition, the installation inclination angle of the hydrocyclone and the external environment pressure can also influence the separation performance.
In theory, the performance of a hydrocyclone can be analyzed and described by its internal flow field. However, the structure and operation of the hydrocyclone seem simple, the internal flow field of the hydrocyclone is extremely complex, and although scholars at home and abroad carry out a great deal of research on the internal flow field of the hydrocyclone and put forward a plurality of theories, methods and empirical models for designing and optimizing the hydrocyclone, various defects still exist. For example, lack of generality, based on certain specific conditions; or the accuracy is low due to the removal of various influencing factors and the interaction influence among the factors; or simply analyzing the optimization from a qualitative perspective results in a more comprehensive result; or rely on a large amount of tests to cause the problems of huge workload and cost and low working efficiency. Based on this, there is a need to provide a new method for optimizing the separation performance of a liquid-solid hydrocyclone based on migration learning, so as to solve the above technical problems.
Disclosure of Invention
Based on this, the invention aims to provide a method for optimizing the separation performance of a liquid-solid hydrocyclone based on transfer learning, so as to solve the technical problems.
The invention provides a method for optimizing the separation performance of a liquid-solid hydrocyclone based on transfer learning, wherein the method comprises the following steps:
the method comprises the following steps: training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, and modeling the working condition under an unknown mode through transfer learning to determine characteristic parameters of the liquid-solid hydrocyclone;
step two: acquiring empirical characteristic parameters of the liquid-solid hydrocyclone;
step three: respectively carrying out numerical simulation by adopting a quadratic orthogonal rotation combined test method according to the characteristic parameters of the liquid-solid hydrocyclone and the acquired empirical characteristic parameters;
step four: respectively determining and obtaining a geometric model of the liquid-solid hydrocyclone determined through transfer learning and a geometric model of the liquid-solid hydrocyclone determined through the empirical characteristic parameters through numerical simulation in the third step;
step five: by utilizing a computational fluid dynamics principle, the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by transfer learning are calculated in a simulation mode, and the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by empirical characteristic parameters are calculated in a simulation mode;
step six: and determining to obtain the optimal separation efficiency value of the liquid-solid hydrocyclone according to the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of each gas-liquid two-phase flow.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning comprises the following steps of:
training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, wherein data samples in the data source domain are provided with labels;
and taking a new working condition needing to be tested in an unknown mode as a target field, and constructing a feature subspace according to data source domain and data in the target field based on a transfer learning principle.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that in the step one, the characteristic parameters comprise physical property parameters, geometric parameters, operation parameters and separation efficiency;
wherein the physical property parameters comprise inlet flow rate, particle size, inlet density difference and particle concentration; the geometric parameters comprise the diameter of the feeding pipe, the diameter of the cylindrical section, the length of the cylindrical section, the upper diameter of the conical section, the lower diameter of the conical section, the cone angle, the diameter of the overflow pipe, the length of the overflow pipe, the diameter of the underflow pipe and the length of the underflow pipe; the operating parameters include inlet speed and reflux ratio.
In the fifth step, when performing simulation calculation, the corresponding simulation condition setting includes:
adopting pressure coupling; carrying out simulation calculation based on an implicit equation and an unsteady algorithm; constructing a fluid volume function model and a gas-liquid two-phase environment; the inlet adopts a speed inlet, and the overflow port and the low flow port both adopt pressure outlets; the pressure-velocity coupling is performed by pressure correction, and the discrete format is performed by PRESTO! And (4) format.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that after the sixth step, the method further comprises the following steps:
and inputting the obtained optimal separation efficiency value of the liquid-solid hydrocyclone into the data source domain so as to reduce the error when data migration calculation iteration is carried out.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning comprises the following steps of constructing a feature subspace according to data of a data source domain and data of a target domain based on the transfer learning principle:
and constructing the relation between the data source domain and the target domain by using an Euclidean distance formula, a cosine similarity formula, a probability mutual information formula, an information divergence formula, a JS distance formula, a maximum mean difference formula and a greedy distance formula.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that the Euclidean distance formula is expressed as follows:
wherein d isEuclideanExpressing the Euclidean distance, wherein x and y respectively express coordinates corresponding to a point x and a point y on a space model, and T is a transposed matrix operation;
the cosine similarity formula is expressed as:
wherein cos (x, y) represents the cosine similarity between point x and point y, | x |, | y | are the absolute value operations of point x and point y, respectively.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that the probability mutual information formula is as follows:
wherein I (X; Y) represents the probability mutual reliability between the point X and the point Y, p (X, Y) represents the probability that the point X and the point Y have a common feature subspace, p (X) represents the probability that the point X has an independent subspace, and p (Y) represents the probability that the point Y has an independent subspace;
the information divergence formula is expressed as:
wherein D isKL(P | | Q) represents the divergence of information between region P and region Q, P (x) represents the probability distribution of point x at region P, and Q (x) represents the probability distribution of point x at region Q.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that the JS distance formula is expressed as follows:
wherein JSD (P | | Q) represents JS distance, DKL(P M) represents the divergence of information between region P and region M, DKL(Q | | M) represents the divergence of information between region Q and region M;
the maximum mean difference formula is expressed as:
wherein, MMD2(X, Y) represents the maximum mean difference,
representing the set of mappings for point x,
represents the set of mappings to which point y corresponds,
representing the regenerated nuclear hilbert space.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning is characterized in that the greedy distance formula is expressed as follows:
wherein, Wp(P, Q) represents a greedy distance, ρ (x, y) represents a distance function between a point x and a point y in the set M, Γ (P, Q) represents a joint distribution of all points in the set M × M, the joint distribution having the region P and the region Q as an edge distribution, and d μ (x, y) represents a differential value of the distance function between the point x and the small number of points.
The method for optimizing the separation performance of the liquid-solid hydrocyclone based on the transfer learning has the following beneficial effects:
1. according to the method, based on the principle of transfer learning, according to the design theory and the empirical method of the liquid-solid hydrocyclone, the relevant structural parameters and efficiency comparison of the liquid-solid hydrocyclone are specifically and rapidly given, and according to the actual engineering situation, the structural parameters of the hydrocyclone to be optimized are determined, so that a foundation is laid for the parameter optimization of the liquid-solid hydrocyclone;
2. in the invention, a numerical simulation method is utilized, and the optimization of the operation parameters and the structural parameters is carried out step by step for optimizing the performance of the liquid-solid hydrocyclone, thereby effectively solving the problems of complex design, huge cost and low efficiency of the existing hydrocyclone. The separation performance of the liquid-solid separation hydrocyclone can be simply, economically and efficiently improved;
3. by adopting the migration data provided by the invention, the experimental method of orthogonal numerical simulation parameters is combined with the empirical data, so that the consistency of the numerical simulation result and the test result can be effectively ensured, and the feasibility of replacing the test result with the numerical simulation result is ensured; meanwhile, the influence of errors existing in numerical simulation results on optimization results can be effectively avoided by processing the orthogonal parameter test design method;
4. according to the invention, the structural design and parameter optimization of the liquid-solid hydrocyclone can be rapidly realized by using data migration according to the actual engineering requirements, the separation performance of the hydrocyclone is obviously improved, the production efficiency is improved, the production cost is reduced, and the product quality is improved; and the data optimized by the experiment is fed back to the source field, so that the accuracy and the efficiency of data migration under different working conditions can be continuously improved. The test proves that: according to the optimization result obtained by the method, high-purity overflow concentrate can be obtained while energy consumption is reduced and the concentration of the concentrate in overflow is improved.
Additional features and advantages of the disclosure will be set forth in the description which follows, or in part may be learned by the practice of the above-described techniques of the disclosure, or may be learned by practice of the disclosure.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
FIG. 1 is a schematic block diagram of a method for optimizing the separation performance of a liquid-solid hydrocyclone based on transfer learning according to the present invention;
FIG. 2 is a flow chart of a method for optimizing the separation performance of a liquid-solid hydrocyclone based on transfer learning according to the present invention;
FIG. 3 is a schematic structural view of a liquid-solid hydrocyclone in the present invention;
FIG. 4 is a basic geometric model of a liquid-solid hydrocyclone in accordance with the present invention;
FIG. 5 is a top view of an orthogonal optimized inlet configuration for a liquid-solid hydrocyclone in accordance with the present invention;
FIG. 6 is a comparison of separation efficiency curves for a liquid-solid hydrocyclone in accordance with the present invention.
Detailed Description
To facilitate an understanding of the invention, the invention will now be described more fully with reference to the accompanying drawings. Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. 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 herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
Referring to fig. 1 and fig. 2, the present invention provides a method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning, wherein the method comprises the following steps:
s101, training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, and modeling the working condition under an unknown mode through transfer learning to determine characteristic parameters of the liquid-solid hydrocyclone.
In this step, the characteristic parameters include physical parameters, geometric parameters, operational parameters and separation efficiency.
Wherein, the physical parameters comprise inlet flow rate, particle size, inlet density difference and particle concentration. The geometric parameters include feed pipe diameter, cylinder section length, cone section upper diameter, cone section lower diameter, cone angle, overflow pipe diameter, overflow pipe length, underflow pipe diameter, and underflow pipe length. Further, the operating parameters include inlet speed and reflux ratio.
Specifically, in this step, the method for training and establishing the data source domain of the liquid-solid hydrocyclone under each working condition and modeling the working condition under the unknown mode through transfer learning specifically includes the following steps:
s1011, training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, wherein data samples in the data source domain are provided with labels.
And S1012, taking the new working condition to be tested in the unknown mode as a target field, and constructing a feature subspace according to the data source domain and the data of the target field based on the transfer learning principle.
In the step of constructing the feature subspace according to the data of the data source domain and the data of the target domain, a relation between the data source domain and the target domain needs to be constructed by using an Euclidean distance formula, a cosine similarity formula, a probability mutual information formula, an information divergence formula, a JS distance formula, a maximum mean difference formula and a greedy distance formula.
Wherein, the Euclidean distance formula is expressed as:
wherein d isEuclideanAnd expressing the Euclidean distance, wherein x and y respectively express coordinates corresponding to a point x and a point y on the space model, and T is a transposed matrix operation.
Meanwhile, the cosine similarity formula is expressed as:
wherein cos (x, y) represents the cosine similarity between the point x and the point y, and | x |, | y | are the absolute value operations of the point x and the point y, respectively.
Further, in this embodiment, the two probability distributions X and Y are defined, where X belongs to X and Y belongs to Y, and the mutual probability information formula is expressed as:
wherein I (X; Y) represents the probability mutual reliability between the point X and the point Y, p (X, Y) represents the probability that the point X and the point Y have a common feature subspace, p (X) represents the probability that the point X has an independent subspace, and p (Y) represents the probability that the point Y has an independent subspace.
Further, the information divergence formula is expressed as:
wherein D isKL(P | | Q) represents the divergence of information between region P and region Q, P (x) represents the probability distribution of point x at region P, and Q (x) represents the probability distribution of point x at region Q. It should be noted that D isKL(P | Q) is used to represent the distance between probability distribution P (x) and probability distribution Q (x).
Further, the JS distance formula is expressed as:
wherein JSD (P | | Q) represents JS distance, DKL(P M) represents the divergence of information between region P and region M, DKL(Q | | M) represents the divergence of information between region Q and region M.
Further, the maximum mean difference formula is as follows:
wherein, MMD2(X, Y) represents the maximum mean difference,
representing the set of mappings for point x,
represents the set of mappings to which point y corresponds,
representing the regenerated nuclear hilbert space.
Further, the greedy Distance (Wasserstein Distance) formula is expressed as:
wherein, Wp(P, Q) represents a greedy distance, ρ (x, y) represents a distance function between the point x and the point y in the set M, Γ (P, Q) represents a joint distribution of all the points in the set M × M, the joint distribution having the region P and the region Q as an edge distribution, and d μ (x, y) represents a differential value of the distance function between the point x and the point y.
S102, acquiring empirical characteristic parameters of the liquid-solid hydrocyclone.
In this step, it should be noted that the empirical characteristic parameters are calculated by a conventional empirical calculation method.
S103, respectively carrying out numerical simulation by adopting a quadratic orthogonal rotation combined test method according to the characteristic parameters of the liquid-solid hydrocyclone and the acquired empirical characteristic parameters.
In the step, the characteristic parameters of the liquid-solid hydrocyclone determined by the secondary orthogonal rotation combined test transfer learning and the traditional method are adopted, and the parameters of the scheme are operated to carry out numerical simulation and test optimization. As described in step S101, the characteristic parameters include physical property parameters, geometric parameters, and operation parameters.
And S104, respectively determining and obtaining a geometric model of the liquid-solid hydrocyclone determined through transfer learning and a geometric model of the liquid-solid hydrocyclone determined through the empirical characteristic parameters through numerical simulation in the third step.
And S105, by utilizing the computational fluid dynamics principle, performing simulation calculation on the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by the transfer learning, and performing simulation calculation on the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by the empirical characteristic parameters.
In this step, when performing simulation calculation, the corresponding simulation condition setting includes:
adopting pressure coupling; carrying out simulation calculation based on an implicit equation and an unsteady algorithm; constructing a fluid volume function model and a gas-liquid two-phase environment; the inlet adopts a speed inlet, and the overflow port and the low flow port both adopt pressure outlets; the pressure-velocity coupling is performed by pressure correction, and the discrete format is performed by PRESTO! And (4) format.
Further, after the conditions are set, the pressure drop, the tangential velocity, the axial velocity and the radial velocity of the gas-liquid two-phase flow and the gas-liquid-solid three-phase flow of the liquid-solid hydrocyclone can be calculated and determined.
And S106, determining and obtaining the optimal separation efficiency value of the liquid-solid hydrocyclone according to the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of each gas-liquid two-phase flow.
In this step, it specifically includes:
and inputting the obtained optimal separation efficiency value of the liquid-solid hydrocyclone into a data source domain to reduce the error when data migration calculation iteration is carried out.
It should be added to the description here that characteristic parameters of the operating condition liquid-solid hydrocyclone, including physical property parameters, geometric parameters, operating parameters and separation efficiency, are fed back to the data source domain, so that the richness of data, the accuracy of migration learning and the richness of the data source domain are further improved, and finally, the error during data migration is reduced through continuous learning, so that the effect that an optimal solution can be obtained only through one migration is achieved.
In the present invention, please refer to fig. 3 and 4, for the liquid-solid hydrocyclone proposed in the present invention, wherein 1 is the length of the cylindrical section of the hydrocyclone, 2 is the insertion depth of the overflow pipe of the hydrocyclone, 3 is the diameter of the overflow port, 4 is the wall thickness of the overflow pipe, 5 is the inlet width of the hydrocyclone, 6 is the basic diameter of the hydrocyclone, 7 is the taper angle, 8 is the diameter of the underflow port, and 9 is the inlet length of the hydrocyclone.
Meanwhile, as shown in fig. 5 and fig. 6, it can be known that the method for optimizing the separation performance of the liquid-solid hydrocyclone based on the migration learning provided by the present invention has the following beneficial effects:
1. according to the method, based on the principle of transfer learning, according to the design theory and the empirical method of the liquid-solid hydrocyclone, the relevant structural parameters and efficiency comparison of the liquid-solid hydrocyclone are specifically and rapidly given, and according to the actual engineering situation, the structural parameters of the hydrocyclone to be optimized are determined, so that a foundation is laid for the parameter optimization of the liquid-solid hydrocyclone;
2. in the invention, a numerical simulation method is utilized, and the optimization of the operation parameters and the structural parameters is carried out step by step for optimizing the performance of the liquid-solid hydrocyclone, thereby effectively solving the problems of complex design, huge cost and low efficiency of the existing hydrocyclone. The separation performance of the liquid-solid separation hydrocyclone can be simply, economically and efficiently improved;
3. by adopting the migration data provided by the invention, the experimental method of orthogonal numerical simulation parameters is combined with the empirical data, so that the consistency of the numerical simulation result and the test result can be effectively ensured, and the feasibility of replacing the test result with the numerical simulation result is ensured; meanwhile, the influence of errors existing in numerical simulation results on optimization results can be effectively avoided by processing the orthogonal parameter test design method;
4. according to the invention, the structural design and parameter optimization of the liquid-solid hydrocyclone can be rapidly realized by using data migration according to the actual engineering requirements, the separation performance of the hydrocyclone is obviously improved, the production efficiency is improved, the production cost is reduced, and the product quality is improved; and the data optimized by the experiment is fed back to the source field, so that the accuracy and the efficiency of data migration under different working conditions can be continuously improved. The test proves that: according to the optimization result obtained by the method, high-purity overflow concentrate can be obtained while energy consumption is reduced and the concentration of the concentrate in overflow is improved.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present invention, which are used for illustrating the technical solutions of the present invention and not for limiting the same, and the protection scope of the present invention is not limited thereto, although the present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning is characterized by comprising the following steps:
the method comprises the following steps: training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, and modeling the working condition under an unknown mode through transfer learning to determine characteristic parameters of the liquid-solid hydrocyclone;
step two: acquiring empirical characteristic parameters of the liquid-solid hydrocyclone;
step three: respectively carrying out numerical simulation by adopting a quadratic orthogonal rotation combined test method according to the characteristic parameters of the liquid-solid hydrocyclone and the acquired empirical characteristic parameters;
step four: respectively determining and obtaining a geometric model of the liquid-solid hydrocyclone determined through transfer learning and a geometric model of the liquid-solid hydrocyclone determined through the empirical characteristic parameters through numerical simulation in the third step;
step five: by utilizing a computational fluid dynamics principle, the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by transfer learning are calculated in a simulation mode, and the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of the gas-liquid two-phase flow corresponding to the geometric model of the liquid-solid hydrocyclone determined by empirical characteristic parameters are calculated in a simulation mode;
step six: and determining to obtain the optimal separation efficiency value of the liquid-solid hydrocyclone according to the flow characteristic, the air column characteristic, the solid phase flow characteristic and the simulated separation efficiency of the liquid-solid hydrocyclone of each gas-liquid two-phase flow.
2. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 1, wherein in the step one, a data source domain of the liquid-solid hydrocyclone under each working condition is trained and established, and the method for modeling the working condition under the unknown mode through transfer learning specifically comprises the following steps:
training and establishing a data source domain of the liquid-solid hydrocyclone under each working condition, wherein data samples in the data source domain are provided with labels;
and taking a new working condition needing to be tested in an unknown mode as a target field, and constructing a feature subspace according to data source domain and data in the target field based on a transfer learning principle.
3. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 2, wherein in the first step, the characteristic parameters include physical property parameters, geometric parameters, operation parameters and separation efficiency;
wherein the physical property parameters comprise inlet flow rate, particle size, inlet density difference and particle concentration; the geometric parameters comprise the diameter of the feeding pipe, the diameter of the cylindrical section, the length of the cylindrical section, the upper diameter of the conical section, the lower diameter of the conical section, the cone angle, the diameter of the overflow pipe, the length of the overflow pipe, the diameter of the underflow pipe and the length of the underflow pipe; the operating parameters include inlet speed and reflux ratio.
4. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 2, wherein in the step five, when performing simulation calculation, the corresponding simulation condition setting includes:
adopting pressure coupling; carrying out simulation calculation based on an implicit equation and an unsteady algorithm; constructing a fluid volume function model and a gas-liquid two-phase environment; the inlet adopts a speed inlet, and the overflow port and the low flow port both adopt pressure outlets; the pressure-velocity coupling is performed by pressure correction, and the discrete format is performed by PRESTO! And (4) format.
5. The method for optimizing liquid-solid hydrocyclone separation performance based on transfer learning according to claim 2, wherein after the sixth step, the method further comprises:
and inputting the obtained optimal separation efficiency value of the liquid-solid hydrocyclone into the data source domain so as to reduce the error when data migration calculation iteration is carried out.
6. The method for optimizing separation performance of the liquid-solid hydrocyclone based on the transfer learning of claim 2, wherein in the step of constructing the feature subspace based on the transfer learning principle according to the data of the data source domain and the data of the target domain, the method comprises:
and constructing the relation between the data source domain and the target domain by using an Euclidean distance formula, a cosine similarity formula, a probability mutual information formula, an information divergence formula, a JS distance formula, a maximum mean difference formula and a greedy distance formula.
7. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 6, wherein the Euclidean distance formula is expressed as:
wherein d isEuclideanExpressing the Euclidean distance, wherein x and y respectively express coordinates corresponding to a point x and a point y on a space model, and T is a transposed matrix operation;
the cosine similarity formula is expressed as:
wherein cos (x, y) represents the cosine similarity between the point x and the point y, and | x |, | y | are the absolute value operations of the point x and the point y, respectively.
8. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 7, wherein the probability mutual information formula is expressed as:
wherein I (X; Y) represents the probability mutual reliability between the point X and the point Y, p (X, Y) represents the probability that the point X and the point Y have a common feature subspace, p (X) represents the probability that the point X has an independent subspace, and p (Y) represents the probability that the point Y has an independent subspace;
the information divergence formula is expressed as:
wherein D isKL(P | | Q) represents the divergence of information between region P and region Q, P (x) represents the probability distribution of point x at region P, and Q (x) represents the probability distribution of point x at region Q.
9. The method for optimizing liquid-solid hydrocyclone separation performance based on transfer learning of claim 8, wherein the JS distance formula is expressed as:
wherein JSD (P | | Q) represents JS distance, DKL(P M) represents the divergence of information between region P and region M, DKL(Q | | M) represents the divergence of information between region Q and region M;
the maximum mean difference formula is expressed as:
wherein, MMD2(X, Y) represents the maximum mean difference,
representing the set of mappings for point x,
represents the set of mappings to which point y corresponds,
representing the regenerated nuclear hilbert space.
10. The method for optimizing separation performance of a liquid-solid hydrocyclone based on transfer learning of claim 9, wherein the greedy distance formula is expressed as:
wherein, Wp(P, Q) represents a greedy distance, ρ (x, y) represents a distance function between the point x and the point y in the set M, Γ (P, Q) represents a joint distribution of all the points in the set M × M, the joint distribution having the region P and the region Q as an edge distribution, and d μ (x, y) represents a differential value of the distance function between the point x and the point y.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111323572.XA CN114036670B (en) | 2021-11-09 | 2021-11-09 | Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111323572.XA CN114036670B (en) | 2021-11-09 | 2021-11-09 | Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114036670A true CN114036670A (en) | 2022-02-11 |
| CN114036670B CN114036670B (en) | 2024-04-26 |
Family
ID=80143714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111323572.XA Active CN114036670B (en) | 2021-11-09 | 2021-11-09 | Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114036670B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160312552A1 (en) * | 2015-04-27 | 2016-10-27 | Baker Hughes Incorporated | Integrated modeling and monitoring of formation and well performance |
| CN112517258A (en) * | 2020-11-05 | 2021-03-19 | 上海明罗石油天然气工程有限公司 | Novel liquid-liquid separation hydrocyclone |
| CN113011114A (en) * | 2021-03-25 | 2021-06-22 | 赣江新区澳博颗粒科技研究院有限公司 | Numerical simulation method for optimizing separation performance of liquid-solid hydrocyclone |
| US20210190882A1 (en) * | 2019-12-10 | 2021-06-24 | Wuhan University | Transformer failure identification and location diagnosis method based on multi-stage transfer learning |
-
2021
- 2021-11-09 CN CN202111323572.XA patent/CN114036670B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160312552A1 (en) * | 2015-04-27 | 2016-10-27 | Baker Hughes Incorporated | Integrated modeling and monitoring of formation and well performance |
| US20210190882A1 (en) * | 2019-12-10 | 2021-06-24 | Wuhan University | Transformer failure identification and location diagnosis method based on multi-stage transfer learning |
| CN112517258A (en) * | 2020-11-05 | 2021-03-19 | 上海明罗石油天然气工程有限公司 | Novel liquid-liquid separation hydrocyclone |
| CN113011114A (en) * | 2021-03-25 | 2021-06-22 | 赣江新区澳博颗粒科技研究院有限公司 | Numerical simulation method for optimizing separation performance of liquid-solid hydrocyclone |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114036670B (en) | 2024-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yang et al. | The coordinated relationship between vortex finder parameters and performance of hydrocyclones for separating light dispersed phase | |
| Concha et al. | Thickening in the 20 th century: a historical perspective | |
| CN108536979B (en) | Underflow concentration prediction method based on thickener mechanism model | |
| Bing et al. | Numerical simulation and experimental study on internal and external characteristics of novel Hydrocyclones | |
| GAO et al. | CFD numerical simulation of flow velocity characteristics of hydrocyclone | |
| CN109868157A (en) | A kind of double-field coupled dehydration device and parameter optimization method | |
| CN105843998B (en) | The numerical simulation of optimum method of solid-solid separation hydrocyclone parameter | |
| Xu et al. | Simulation analysis of multiphase flow and performance of hydrocyclones at different atmospheric pressures | |
| Zhang et al. | Simulation analysis on the separation performance of spiral inlet hydrocyclone | |
| Wang et al. | Analysis of performance of novel hydrocyclones in ebullated bed reactor with different vortex finder structures | |
| CN114036670B (en) | Method for optimizing separation performance of liquid-solid hydrocyclone based on transfer learning | |
| Vieira et al. | Study of the Performance of a Novel Hydrocyclone Built in the Configurations of the Classical Families | |
| Ma et al. | Study and application of a cyclone for removing amine droplets from recycled hydrogen in a hydrogenation unit | |
| Podgorska | Multiphase particulate systems in turbulent flows: fluid-liquid and solid-liquid dispersions | |
| CN113011114A (en) | Numerical simulation method for optimizing separation performance of liquid-solid hydrocyclone | |
| Liang et al. | Numerical simulation on flow field of oilfield three-phase separator | |
| Svarovsky | Hydrocyclones | |
| Kołodziejczyk | Numerical simulation of coal suspension sedimentation | |
| Shen et al. | Numerical simulation of the flow field in a dense-media cyclone | |
| Su et al. | Effect of the Axially Inserted Solid Rod on the Flow Field Characteristics and Separation Performance of a Hydrocyclone | |
| Valdez et al. | Experimental evaluation of a gas liquid axial cyclone separator | |
| Li et al. | Numerical Analysis on the Effect of Combined-Curve Tapered Segment on the Flow Field and Separation Performance of Hydrocyclones | |
| CN112613252A (en) | Energy-saving operation method of absorption tower stirrer | |
| Song et al. | Structural Design and Performance Analysis of Three-Phase Hydrocyclones (3PH) | |
| Chen et al. | Effect of gas flow rate on separation efficiency at different scaling scales of cyclone separators |
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 |