CN102608285A - Organic mixture blasting characteristic prediction method based on support vector machine - Google Patents
Organic mixture blasting characteristic prediction method based on support vector machine Download PDFInfo
- Publication number
- CN102608285A CN102608285A CN2012100388821A CN201210038882A CN102608285A CN 102608285 A CN102608285 A CN 102608285A CN 2012100388821 A CN2012100388821 A CN 2012100388821A CN 201210038882 A CN201210038882 A CN 201210038882A CN 102608285 A CN102608285 A CN 102608285A
- Authority
- CN
- China
- Prior art keywords
- model
- sample
- organic mixture
- vector machine
- parameter
- 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
- 239000000203 mixture Substances 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 22
- 238000005422 blasting Methods 0.000 title 1
- 238000004880 explosion Methods 0.000 claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 30
- 239000002360 explosive Substances 0.000 claims description 81
- 230000006870 function Effects 0.000 claims description 49
- 238000002474 experimental method Methods 0.000 claims description 33
- 238000010438 heat treatment Methods 0.000 claims description 12
- 238000009792 diffusion process Methods 0.000 claims description 8
- 238000002485 combustion reaction Methods 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 7
- 230000002269 spontaneous effect Effects 0.000 claims description 6
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 4
- 239000001301 oxygen Substances 0.000 claims description 4
- 229910052760 oxygen Inorganic materials 0.000 claims description 4
- 230000001105 regulatory effect Effects 0.000 claims description 4
- 238000009835 boiling Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000013210 evaluation model Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 13
- 238000004422 calculation algorithm Methods 0.000 abstract description 9
- 238000013461 design Methods 0.000 abstract description 9
- 238000010801 machine learning Methods 0.000 abstract description 8
- 230000008901 benefit Effects 0.000 abstract description 6
- 238000009776 industrial production Methods 0.000 abstract description 6
- 230000000704 physical effect Effects 0.000 abstract description 2
- 230000002265 prevention Effects 0.000 abstract 2
- 239000007789 gas Substances 0.000 description 32
- 238000011160 research Methods 0.000 description 20
- 239000000463 material Substances 0.000 description 19
- 239000000126 substance Substances 0.000 description 18
- 230000000694 effects Effects 0.000 description 5
- 238000013507 mapping Methods 0.000 description 5
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 5
- 229930195733 hydrocarbon Natural products 0.000 description 4
- 150000002430 hydrocarbons Chemical class 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000009795 derivation Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- HIXDQWDOVZUNNA-UHFFFAOYSA-N 2-(3,4-dimethoxyphenyl)-5-hydroxy-7-methoxychromen-4-one Chemical compound C=1C(OC)=CC(O)=C(C(C=2)=O)C=1OC=2C1=CC=C(OC)C(OC)=C1 HIXDQWDOVZUNNA-UHFFFAOYSA-N 0.000 description 2
- 241001269238 Data Species 0.000 description 2
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 2
- ATUOYWHBWRKTHZ-UHFFFAOYSA-N Propane Chemical compound CCC ATUOYWHBWRKTHZ-UHFFFAOYSA-N 0.000 description 2
- 239000002253 acid Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 150000001555 benzenes Chemical class 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 150000002148 esters Chemical class 0.000 description 2
- RTZKZFJDLAIYFH-UHFFFAOYSA-N ether Substances CCOCC RTZKZFJDLAIYFH-UHFFFAOYSA-N 0.000 description 2
- 239000008246 gaseous mixture Substances 0.000 description 2
- 150000002576 ketones Chemical class 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000012046 mixed solvent Substances 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 229920006395 saturated elastomer Polymers 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 229930195735 unsaturated hydrocarbon Natural products 0.000 description 2
- OYPRJOBELJOOCE-UHFFFAOYSA-N Calcium Chemical compound [Ca] OYPRJOBELJOOCE-UHFFFAOYSA-N 0.000 description 1
- 239000004215 Carbon black (E152) Substances 0.000 description 1
- VGGSQFUCUMXWEO-UHFFFAOYSA-N Ethene Chemical compound C=C VGGSQFUCUMXWEO-UHFFFAOYSA-N 0.000 description 1
- 108010074864 Factor XI Proteins 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 150000001299 aldehydes Chemical class 0.000 description 1
- HSFWRNGVRCDJHI-UHFFFAOYSA-N alpha-acetylene Natural products C#C HSFWRNGVRCDJHI-UHFFFAOYSA-N 0.000 description 1
- 238000004164 analytical calibration Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000001311 chemical methods and process Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 125000002534 ethynyl group Chemical group [H]C#C* 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 239000000383 hazardous chemical Substances 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- IJGRMHOSHXDMSA-UHFFFAOYSA-N nitrogen Substances N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000003960 organic solvent Substances 0.000 description 1
- 239000003973 paint Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 231100000614 poison Toxicity 0.000 description 1
- 230000007096 poisonous effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000001294 propane Substances 0.000 description 1
- QQONPFPTGQHPMA-UHFFFAOYSA-N propylene Natural products CC=C QQONPFPTGQHPMA-UHFFFAOYSA-N 0.000 description 1
- 125000004805 propylene group Chemical group [H]C([H])([H])C([H])([*:1])C([H])([H])[*:2] 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- IJDNQMDRQITEOD-UHFFFAOYSA-N sec-butylidene Natural products CCCC IJDNQMDRQITEOD-UHFFFAOYSA-N 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A method for predicting the explosion characteristics of organic mixture based on support vector machine features that the known contents of components and conventional experimental data on physical properties of organic mixture are used as input variables, the corresponding experimental data on explosion characteristics are used as output variables, and a strong machine learning algorithm is used to support vector machine for effectively training and predicting the non-linear, uncertain and complex internal quantitative relation between them, so creating a stable and efficient prediction model of support vector machine. The built support vector machine model is used for predicting the explosion characteristics of other unknown mixtures, and the method has the advantages of high prediction precision, rapidness and convenience. The method can realize the prediction of the explosion characteristics of the organic mixture under different components and proportions, effectively solve the problem of lack of experimental data of various mixtures in actual industrial production, and simultaneously has good application prospect in the work of industrial process design, fire prevention, explosion prevention and the like.
Description
Technical field
The present invention relates to field of chemical technology, especially a kind of method of predicting the organic mixture explosive characteristic, specifically a kind of organic mixture explosive characteristic prediction method based on SVMs.
Background technology
Along with the continuous development of chemical industry and the variation of chemical products, various chemical products have obtained using widely in national economy all departments.The chemical substance that has been found that at present and synthesize has more than 30,000,000 kinds, and wherein human chemical products used in everyday just have kind more than 80,000, and this numeral is just with nearly thousand kinds speed increase in every year.In numerous chemical substances, there are many materials to have the inflammable and explosive hazard property of Denging, in processes such as production, use, storage and transportation, exist the possibility of breaking out of fire, explosion accident, the people life property safety is caused great threat.Therefore, the combustion explosion characteristic of understanding and grasp combustible is for the safety management of strengthening dangerous substance, and the generation of fire preventing, explosion accident etc. has great importance.
Spontaneous ignition temperature, explosion limits and heating power are the important parameters that characterizes organic combustible explosive characteristic.The complexity of these parameters and combustible breaking out of fire blast is closely related; Can characterize the hazard level of organism in processes such as production, storage and transportation; Instruct carrying out of work such as engineering design and risk assessment, therefore in actual Chemical Manufacture, be with a wide range of applications.
For the explosive characteristic of pure component combustible, just can obtain concrete numerical value usually through consulting document.But, usually can run into the situation that different combustibles mix each other along with the continuous development of chemical industry and the variation of chemical products.All use mixed organic solvents in a large number like industries such as paint, coating, fine chemistry industry, pharmacy; The danger classes in these industry places all will be divided according to the flash-point of mixed solvent; And the flash-point of mixed solvent and its composition and proportioning have very big relation, are difficult to from document, directly check in.Analogue has caused the serious disappearance of required explosive characteristic data such as organic mixture flash-point, spontaneous ignition temperature, explosion limits and heating power in the actual industrial production.
Utilizing measuring is to obtain organic mixture explosive characteristic data effective method the most directly perceived.But the measuring method often exists following deficiency: (1) experimental technique not only requires to possess good experimental facilities; Expensive; And the mensuration process need pass through the series of steps such as selection, instrument calibration, experiment test, data preparation and screening of material preparation, purity evaluation, assay method and instrument, and workload is huge; (2) because the difference that exists between the material explosive characteristic; The experimental apparatus that is possessed is difficult to material of all categories is estimated; Must consider the characteristic of instrument and the explosive characteristic of material simultaneously; Experimental technique to them carries out effective combination, and therefore, it is impossible making an experiment one by one to them; (3) consider safety problem in the experimentation, general experimental study (the especially research of explosion limits) can only be on a small scale, undersized experiment, still can't embody scale effect preferably with the simulated experiment of first approximation; (4) poisonous, volatile, explosive or the material of radiation is arranged for those, exist certain difficulty in the measurement; (5) for those not synthetic as yet material and labile reactive chemicals, also can't confirm its danger based on experiment.
Therefore; Merely application experiment research confirms that the explosive characteristic of organic mixture obviously is worthless; Be necessary theoretical research and measuring are combined, combine qualitative and quantitative, existing experimental data is put in order and summarize by the theoretical prediction method; Set up simple and reliable organic mixture explosive characteristic theoretical prediction model on this basis; Enlarge the range of application and the use value of experimental data to greatest extent, solve the problem that organic mixture explosive characteristic experimental data lacks effectively, theoretical foundation and technical support are provided for chemical process designs with safe and scientific research.
From domestic and international present Research, the forecasting research of existing organic mixture explosive characteristic is mainly carried out theoretical derivation or improvement based on Le Chatelier empirical equation, and the recurrence of perhaps adopting Taylor polynomial expression etc. to carry out experimental formula is calculated.The former derivation is complicated, uses inconvenience; The latter is lacks of theoretical foundation then, and model structure lacks physical significance.Simultaneously, the forecast model that existing research institute sets up often is only applicable to specific research object, and the scope of application is single, does not have general applicability.In addition, existing research object only limits to the organic mixture that materials such as alcohol, ketone, ether, acid, ester are formed under different situations, does not see relevant report as yet for the research of industry such as benzene series thing material commonly used, and is also less to the research of hydrocarbons.Above-mentioned reason has all directly limited existing research in actual application in engineering.
SVMs (Support Vetor Machine, SVM) algorithm be Vapnik and co-worker thereof on the basis of Statistical Learning Theory, a kind of new machine learning method that proposes in nineteen ninety-five.Its basic thought is to shine upon through the linear/non-linear of prior selection input vector X is mapped to a high-dimensional feature space Z, then structure optimized data qualification/data regression hyperplane in this feature space.SVMs is a kind of nonparametric machine learning method, is cardinal rule with the structural risk minimization, has strict theoretical foundation.It seeks best compromise according to limited sample information between complicacy of the model learning accuracy of specific training sample (promptly to) and learning ability (being the ability of faultless identification arbitrary sample); In the hope of obtaining best generalization ability; Therefore solved and perplexed many problems of machine learning method in the past, obtained in a plurality of fields such as pattern-recognition, function regression using widely.Compare with traditional machine learning method, support vector machine method has following advantage:
(1) specially to limited sample situation, its target is to obtain the optimum solution under the limited information and be not only the optimal value of sample number when being tending towards infinity;
(2) algorithm transforms into a secondary optimizing problem the most at last, and in theory, what obtain will be global optimum's point, solve the unavoidable local optimum problems of method such as neural network;
(3) algorithm is transformed into high-dimensional feature space with the nonlinear transformation of practical problems through kernel function; The structure linear discriminant function is realized the Nonlinear Discriminant Function in former the having living space in higher dimensional space; Generalization ability is preferably arranged; It has solved problem of dimension dexterously simultaneously, and its algorithm complex and sample dimension are irrelevant.
In a word, the problem of model selection that exists in the conventional machines learning method, cross study and owe study, non-linear and problems such as dimension disaster problem and local minimum point, all in SVM, obtained solving preferably.In addition, SVM is applicable to and solves problem insoluble with the traditional mathematics model method, uncertainty that it can solve organic mixture explosive characteristic training sample practical problems on the low side just to the processing power of small sample.Modeling technique with traditional is compared; The SVM model only needs the parameter and the rational training data of related system; Just can avoid the many difficult problems in the organic mixture explosive characteristic theoretical prediction model performance history, for fast prediction organic mixture explosive characteristic provides a new approach.
Summary of the invention
The present invention is directed in the actual industrial production needs to confirm that through experiment the cycle that the explosive characteristic of the organic mixture of different component and different proportionings exists is long; The problem that danger is big, cost is high; Proposed a kind of explosive characteristic and be the basis with existing known organic mixture; Be equipped with a spot of conventional physical property experiment, need not the organic mixture explosive characteristic prediction method that large-scale combustion explosion experiment can obtain its explosive characteristic parameter based on SVMs.
Technical scheme of the present invention is:
A kind of Forecasting Methodology of the organic mixture explosive characteristic based on SVMs; It is characterized in that; With the component concentration of organic mixture and the explosive characteristic experimental data of conventional rerum natura and these potpourris correspondence is sample; The supporting vector machine model that foundation utilizes regression function to estimate utilizes supporting vector machine model to predict the explosive characteristic of unknown organic mixture again.
Said Forecasting Methodology may further comprise the steps:
(1) sets up sample data: collect at least 100 kinds of (or group) organic mixtures as sample; Conventional rerum natura and component concentration and corresponding explosive characteristic experimental data with these samples; As sample data; Select about 2/3rds sample data as the training sample data at random, be used to set up forecast model; Remaining about 1/3rd sample data is as the forecast sample data, is used for institute's established model is estimated and verified.
(2) set up supporting vector machine model: to the training sample data; With the component concentration of organic mixture and conventional rerum natura as input variable; Corresponding explosive characteristic is as output variable; Use support vector machine method internal relation is between the two simulated, the quantitative function relation of seeking to exist is between the two set up corresponding forecast model;
The correlation parameter of decision SVMs modeling performance mainly comprises: the size of ε in the parameter of kernel function, kernel function, penalty coefficient C and ε-insensitive loss function; Kernel function is selected radially base nuclear K (x, x for use
i)=exp (γ || x-x
i||
2), because it has higher learning efficiency and learning rate; Other parameter is confirmed through " grid search " method; The parameter search scope is following: penalty coefficient C---0-1024; The parameter of kernel function (width) γ---0-1024; ε---0-1024 in ε-insensitive loss function; The direction of search is the lowest mean square root error (RMSE) of " staying 1/10 method " cross-verification; Validation-cross is meant 1/10 sample that from training sample, at every turn screens out the training sample sum " to stay 1/10 method "; With remaining sample modeling; Come forecasting institute to screen out the character of sample; The root-mean-square error (RMSE) that obtains a validation-cross is like this come the quality of evaluation model performance, and its computing formula is:
Wherein, y
I, predBe the predicted value of sample i, y
I, expExperiment value for sample i; Through search, choose the optimum input parameter of pairing that group parameter of minimum RMSE of " staying 1/10 method " cross-verification as model;
The optimized parameter that application searches goes out is set up corresponding forecast model as the input parameter of SVMs;
(3) prediction organic mixture explosive characteristic: in component concentration in the forecast sample data and the conventional rerum natura supporting vector machine model that input is set up as input variable, calculate the explosive characteristic of forecast sample through supporting vector machine model;
(4) correction and definite forecast model: the predicted value and the experiment value of the forecast sample explosive characteristic that comparison step (3) obtains; When if the deviation of predicted value and experiment value surpasses acceptable scope; Correlation parameter numerical value to SVMs is regulated; And then train again and predict, until the deviation of predicted value and experiment value within the acceptable range, thereby confirm the SVM prediction model;
(5) application of forecast model: utilize determined SVM prediction model that the explosive characteristic of other unknown organic mixture is predicted.
The conventional rerum natura of said organic mixture comprises viscosity, relative density, vapour pressure, coefficient of thermal expansion, boiling point, intermolecular force, eccentric factor, atomic polarizability, oxygen index, Van der waals volumes, combustion rate, stoichiometric concentration, critical temperature, emergent pressure and coefficient of diffusion one of at least, and their combination in any.
The conventional rerum natura of organic mixture can adopt normal experiment to measure, and also can adopt easy theoretical calculation formula to calculate acquisition based on the conventional rerum natura and the proportioning thereof of each one-component of potpourri.
Described explosive characteristic comprises spontaneous ignition temperature, explosion limits and heating power.
Beneficial effect of the present invention:
The invention provides a kind of new method of the prediction organic mixture explosive characteristic based on support vector machine method.It is to the explosive characteristic and the problem of optimizing the industrial process design that need in the commercial production prediction organic mixture under different component and different proportionings; According to existing organic mixture experimental data; With each component concentration proportioning and the conventional rerum natura of potpourri is input variable; Explosive characteristic with correspondence is an output variable; Utilize powerful machine learning algorithm support vector machine method, the inherent quantitative relationship of non-linear, the uncertainty that exists between the two and complicacy is effectively trained and forecast, thereby set up stable, SVM prediction model efficiently.Utilize the supporting vector machine model set up that the explosive characteristic of other unknown potpourris is predicted, have precision of prediction height, advantage fast and easily.The present invention both can realize the prediction of the organic mixture explosive characteristic under different component and the proportioning; Effectively solve the problem that all kinds of potpourri experimental datas lack in the actual industrial production; For work such as process designs, flowsheeting, safety assessment provide the data necessary support; And can remove inconvenience that a large amount of measuring brings and loss economically from, and industrial process design and fire-proof and explosion-proof the grade in the work good prospects for application being arranged, its economy is very considerable.
The present invention passes through supporting vector machine model; Variation to different organic mixtures in the actual industrial production; Can dope the explosive characteristic data of organic mixture under different component and proportioning, help solving the problem that potpourri explosive characteristic basic data lacks; Simultaneously; According to forecast model research and definite chemical factors that each explosive characteristic is played a decisive role; Explore the influence rule of these factors to corresponding explosive characteristic; Can and fire-proof and explosion-proofly wait work to point the direction for potpourri configuration, thereby design and safe and scientific research provides theoretical foundation and technical support, the economic benefit of raising manufacturing enterprise for industrial process.
Description of drawings
Fig. 1 is used for the principles illustrated of regression problem for support vector machine method.
Fig. 2 is the comparison of supporting vector machine model gained mixed gas UEL predicted value and experiment value.
Embodiment
Below in conjunction with accompanying drawing and embodiment the present invention is further described.
A kind of organic mixture explosive characteristic prediction method based on SVMs; With the component concentration of known organic mixture and conventional rerum natura and explosive characteristic experimental data is sample; Utilize SVMs strong non-linear mapping ability that the explosive characteristic of organic mixture and the inherent quantitative relationship between component concentration and the conventional rerum natura are simulated; The supporting vector machine model that foundation utilizes regression function to estimate utilizes supporting vector machine model to predict the corresponding explosive characteristic of unknown organic mixture again.
Concrete steps are:
(1) set up sample data: component concentration and conventional rerum natura and the explosive characteristic experimental data of collecting at least 100 kinds of (or group) organic mixtures through experimental technique or collecting method are as in the sample data input computer program; Select about 2/3rds sample data as the training sample data at random, be used to set up forecast model; Remaining about 1/3rd sample data is as the forecast sample data, is used for institute's established model is estimated and verified.
(2) set up supporting vector machine model: to the training sample data; With the component concentration of organic mixture and conventional rerum natura as input variable; Corresponding explosive characteristic is as output variable; Use support vector machine method internal relation is between the two simulated, the quantitative function relation of seeking to exist is between the two set up corresponding forecast model;
The algorithm steps of SVMs is following:
Suppose given training sample set { (x
i, y
i), i=1 ... n}, wherein x
i∈ R
nBe the input value of i learning sample, y
i∈ R is corresponding experiment value.For linear regression, use linear function
f(x)＝(w·x)+b (1)
Estimate.Smooth for assurance formula (1) must be sought the w of a minimum.Suppose all training data (x
i, y
i) can under precision ε, use linear function fit, the problem of seeking minimum w so just changes the minimum model complexity into, and it is equivalent to
Change into corresponding quadratic programming problem promptly:
(y
i-w·x-b≤ε，w·x+b-y
i≤ε)
Consider the situation that allows error of fitting, introduce relaxation factor ξ>=0, ξ
*>=0 and penalty factor C, corresponding quadratic programming problem is rewritten as
(y
i-w·x-b≤ε+ξ
i，w·x+b-y
i≤ε+ξ
i *，ξ
i，ξ
i *≥0)
Wherein, penalty factor C＞0 is used for smooth degree and the deviation of balance regression function f (x) greater than the number of ε sample point.Formula (3) is based on following ε-insensitive loss function and draws, this function | ξ |
εRepresent as follows:
More after a little while, find the solution top SVMs and generally adopt duality theory at sample number, it is converted into quadratic programming problem.Set up following Lagrange equation:
Following formula is for parameter w, b, ξ
i, ξ
i *Partial derivative all equal 0, the substitution following formula obtains the antithesis optimization problem
For non-linear regression, the solution thinking of SVMs is sample to be mapped in the feature space of a higher-dimension and with conventional linear method through a Nonlinear Mapping
to solve.Suppose that sample X is mapped to higher dimensional space with nonlinear function
, then nonlinear regression problem is converted into:
SVMs is mapped to high-dimensional feature space through Kernel Function Transformation with sample, and kernel function K (x, x ') satisfies K (x, x ')=< φ (x), φ (x ') >.Therefore formula (6) is rewritten as
The introducing of kernel function makes function find the solution to walk around feature space directly to be carried out in the input space, thereby has avoided calculating Nonlinear Mapping
kernel function that SVMs is commonly used at present to mainly contain linear kernel, polynomial kernel, radially 4 types of base nuclear and sigmoid nuclears etc.The present invention selects radially base nuclear K (x, x for use
i)=exp (γ || x-x
i||
2) as kernel function.
The correlation parameter of decision SVMs modeling performance mainly comprises: the size of ε in the parameter of kernel function, kernel function, penalty coefficient C and ε-insensitive loss function; Kernel function is selected radially base nuclear K (x, x for use
i)=exp (γ || x-x
i||
2), because it has higher learning efficiency and learning rate; Other parameter is confirmed through " grid search " method; The parameter search scope is following: penalty coefficient C---0-1024; The parameter of kernel function (width) γ---0-1024; ε---0-1024 in ε-insensitive loss function; The direction of search is the lowest mean square root error (RMSE) of " staying 1/10 method " cross-verification; Validation-cross is meant 1/10 sample that from training sample, at every turn screens out the training sample sum " to stay 1/10 method "; With remaining sample modeling; Come forecasting institute to screen out the character of sample; The root-mean-square error (RMSE) that obtains a validation-cross is like this come the quality of evaluation model performance, and its computing formula is:
Wherein, y
I, predBe the predicted value of sample i, y
I, expExperiment value for sample i; Through search, choose the optimum input parameter of pairing that group parameter of minimum RMSE of " staying 1/10 method " cross-verification as model; The optimized parameter that application searches goes out is set up corresponding forecast model as the input parameter of SVMs.
(3) prediction organic mixture explosive characteristic: in component concentration in the forecast sample data and the conventional rerum natura supporting vector machine model that input is set up as input variable, calculate the explosive characteristic of forecast sample through supporting vector machine model.
(4) correction and definite forecast model: the predicted value and the experiment value of the forecast sample explosive characteristic that comparison step (3) obtains; If when the deviation of predicted value and experiment value surpassed acceptable scope, for example relative error surpassed 10%, and the correlation parameter numerical value of SVMs is regulated; And then train again and predict; Until the deviation of predicted value and experiment value within the acceptable range, for example relative error is less than 10%, thereby confirms the SVM prediction model.Utilize determined SVM prediction model that the explosive characteristic of other unknown organic mixture is predicted.In the model that also can the component concentration and the input of conventional rerum natura of the organic mixture that needs prediction be set up in this step predicted value that obtains and the numerical value that directly obtains through experiment are compared, and according to error amount the correlation parameter numerical value of SVMs is regulated up to error amount and to be satisfied permissible value.
The conventional rerum natura of described organic mixture is selected from viscosity, relative density, vapour pressure, coefficient of thermal expansion, boiling point, intermolecular force, eccentric factor, atomic polarizability, oxygen index, Van der waals volumes, combustion rate, stoichiometric concentration, critical temperature, emergent pressure and coefficient of diffusion one of at least, and their combination in any.For the selection of these rerum naturas, guarantee to have nothing to do between each rerum natura as far as possible.The possible array mode of organic mixture has a lot, what of component can be divided into binary, ternary even quaternary organic mixture according to; Character according to component then can be divided into the combustible components potpourri again and contain potpourri of noncombustibility component etc.Therefore, must rationally choose the research sample, both consider in the actual industrial production representativeness of potpourri commonly used, consider again the heterogeneity set of dispense than the time singularity.Can choose industry and go up materials such as hydro carbons commonly used, alcohol, ketone, aldehyde, ether, acid, ester, benzene series thing,, guarantee the popularity of sample set to contain wider material classification as research object.On this basis, from material of all categories, choose exemplary compounds respectively and form experimental subjects, what of component at first to be divided into binary and ternary organic mixture according to; Again according to the difference of constitutive property, binary and tertiary mixture further are divided into the combustible components potpourri study respectively, thereby guarantee that the forecast model of being set up has specific aim and representativeness concurrently with the potpourri that contains the noncombustibility component.
The explosive characteristic of described organic mixture comprises spontaneous ignition temperature, explosion limits and heating power.
Described SVMs program adopts the standard C language exploitation.The function of SVMs training program is through analysis, calculating and regular pattern identification to the training sample data, has formed the supporting vector machine model of similar regression formula.The SVM prediction program then is to import corresponding parameter toward supporting vector machine model, and like the component concentration and the conventional rerum natura of organic potpourri, supporting vector machine model calculates the corresponding explosive characteristic of potpourri then according to the condition of input.For kernel function and parameters of choice, be as the criterion with the error of predicted value and experiment value, select less kernel function of those errors and parameter thereof.
Each sample all comprises component concentration and the conventional rerum natura and the corresponding explosive characteristic experimental data of organic mixture.The component concentration of organic mixture and conventional rerum natura are input layer, and corresponding explosive characteristic is an output layer.The quantity of sample preferably is at least ten times of input layer.The selected conventional rerum natura difference of explosive characteristic that organic mixture is different, the rerum natura that influences the potpourri spontaneous ignition temperature generally includes relative density, intermolecular force, eccentric factor, atomic polarizability, oxygen index etc.; The rerum natura that influences explosion limits generally includes relative density, intermolecular force, atomic polarizability, Van der waals volumes, combustion rate etc.; And the rerum natura that influences heating power generally includes relative density, intermolecular force, stoichiometric concentration, critical temperature, emergent pressure and coefficient of diffusion etc.
Choose after the sample, sample data is divided into two parts.But in order to guarantee the generalization of supporting vector machine model, select about 2/3rds sample data usually at random, be used to set up forecast model as the training sample data; Remaining about 1/3rd sample data is as the forecast sample data, is used for institute's established model is estimated and verified.According to the characteristic of training sample data, select the kernel function and the parameter of suitable SVMs, let SVMs fully the inherent law of training sample data learnt and discern, set up corresponding supporting vector machine model.After the kernel function of having selected SVMs and parameter; Predict in the supporting vector machine model that the substitution of forecast sample data is set up; If the prediction average relative error of forecast sample is lower than 10%, just explain that this supporting vector machine model is qualified, can drop into application.If error is bigger, then the kernel function and the parameter of supporting vector machine model are adjusted, and then are trained again and predict, until predicated error in 10%.
Instance.
Like Fig. 1, shown in 2.
A kind of multi component mixed gas body UEL Forecasting Methodology based on SVMs, it may further comprise the steps:
The modeling sample collection comprises the binary burning mixture of the different proportionings of 140 kinds of (also can be and be not less than other kind number of 100) different components altogether, and component comprises six kinds of inflammable gass such as methane, propane, propylene, normal butane, acetylene, ethene.The UEL data of mixed gas both can derive from document (S.Kondo; K.Takizawa; A.Takahashi; Et a1.A study on flammability limits of fuel mixtures [J] .Journal of Hazardous Materials; 2008,155 (3): 440-448 and Fuman Zhao.Experimental Measurements and Modeling Prediction of Flammability Limits of Binary Hydrocarbon Mixtures [D] .Texas:Texas A&M University, 2009); If the up of gathering in the document less than modeling demand, can be confirmed the UEL (this also helps the checking to method versatility of the present invention) of part organic mixture through experimental technique.Subsequently, sample set is divided, selected 95 groups of samples at random, be used to set up forecast model as training set; Select 45 groups of samples of residue as forecast set, be used for the degree of reliability and the predictive ability of institute's established model are estimated checking.
Then, choose the conventional rerum natura such as stoichiometric concentration, coefficient of diffusion, heating power, emergent pressure of mixed gas, and the component concentration of mixed gas, as the input variable of SVMs modeling.Stoichiometric concentration (the C of potpourri
St)
Mix, heating power (Δ H
c)
Mix, emergent pressure (P
c)
MixAnd coefficient of diffusion (D)
MixCan be grouped into based on the one-tenth of mixed gas and adopt following sum formula to calculate, can adopt computes like the heating power of potpourri:
(ΔH
C)
mix＝x
1ΔH
C1+x
2ΔH
C2
Wherein, x
1And x
2Represent the ratio (x that material in the mixed gas 1 and material 2 are shared respectively
1+ x
2=1); Δ H
C1With Δ H
C2The heating power of representing material in the mixed gas 1 and material 2 respectively.Conventional rerum naturas such as the stoichiometric concentration of mixed gas, coefficient of diffusion and emergent pressure can be adopted to use the same method and calculated.Calculate the corresponding physical parameter value of needed each pure component of the above-mentioned conventional rerum natura of potpourri and see table 1.
The physical and chemical parameter of table 1 pure material
Next; To the training set sample; As input variable, corresponding UEL is used support vector machine method internal relation is between the two simulated as output variable with the component concentration of mixed gas and above-mentioned conventional rerum natura; The quantitative function relation of seeking to exist is between the two set up corresponding forecast model.Radially base nuclear K (x, x are adopted in the SVMs simulation
i)=exp (γ || x-x
i||
2) as kernel function, the sample data linear mapping to [1,1] interval, is carried out normalization and handled; To the training set sample, adopt the grid search method to confirm the model optimized parameter, the direction of search of grid search is the lowest mean square root error (RMSE) of " staying 1/10 method " cross-verification; Through search, the optimized parameter of confirming model is penalty coefficient C=1024, the g=0.125 in ε-insensitive loss function, the width gamma of kernel function=0.0625, corresponding support vector several 17.To the training set sample, the application of optimal parameter is set up corresponding supporting vector machine model.Respectively with in component concentration in training sample data and the forecast sample data and the conventional rerum natura supporting vector machine model that input is set up as input variable; Using this model returns and predicts training set sample and forecast set sample respectively; UEL predicted value to 45 samples in the forecast set is seen table 2; All 140 sample UEL predicted values and experiment value relatively see Fig. 2, the Specifeca tion speeification of model is seen table 3.
Table 2SVM model is to the predicted value of forecast set sample UEL
No matter can find out from table 3, be for training set or forecast set, and the prediction average relative error of model is all less than 10%, and the deviation of predicted value and experiment value that most samples are described is within acceptable scope.Simultaneously, all less than experiment permissible error 3% (percent by volume), this explains that all the supporting vector machine model of being built is successful to the prediction mean absolute error, and it is feasible that the present invention adopts the explosive characteristic of supporting vector machine model prediction organic mixture.In addition, it can also be seen that from table 3 that the estimated performance of model and match performance are comparatively approaching, this explanation the present invention is based on mixed gas UEL forecast model that support vector machine method sets up also to have stronger extensive performance and promptly predicts stability.In sum, predicting the outcome of its validation-cross of the supporting vector machine model that the present invention set up and external certificate all meets the predicated error requirement, and promptly the prediction average relative error of model is less than 10%.The model that satisfies this requirement can come into operation.
For similarly successfully research, do not appear in the newspapers as yet on the document.Existing multi component mixed gas body explosion limits theoretical prediction technology mainly comprises following several kinds: Le Chatelier empirical equation or its improve formula, constant adiabatic flame temperature (CAFT) method and volumetric concentration experimental formula method.External Le Chatelier has proposed the experimental formula according to the volumetric concentration of pure component and lower explosive limit prediction gaseous mixture lower explosive limit the earliest.Subsequently, people such as Kondo propose the UEL that Le Chatelier law can be used to predict some flammable mixtures equally, and have the better prediction precision.Because Le Chatelier experimental formula is to sum up out through the explosion limits of the lower potpourri of research combustible component concentration, so its prediction to the higher potpourri of combustible component concentration often only limits to some specific flammable mixed gas.Mashuga and Crowl have carried out theoretical derivation based on some hypothesis in advance to Le Chatelier law; Discovery (has bigger intermolecular force this moment) when combustible component concentration is low; These hypothesis are more consistent with actual conditions, so Le Chatelier law can carry out reasonable prediction to the lower explosive limit of mixed gas; And when UEL concentration, the hypothesis of model and actual conditions deviation are obvious, so prediction effect is not good.People such as Zhao to binary saturated/explosion limits of unsaturated hydro carbons potpourri carried out measuring research; And utilize experimental data that Le Chatelier law has been carried out the match checking, proposed simultaneously improved Le Chatelier empirical equation be used for to binary saturated/explosion limits of unsaturated hydro carbons potpourri predicts.Le Chatelier empirical equation or its improve equation and calculate comparatively loaded down with trivial details; Use inconvenience; And be only applicable to calculate the explosion limits of the combustible gas mixture that energy of activation, grammol heating power, reaction rate etc. are close; As more accurate when calculating the hydro carbons mixed gas, but then bigger deviation can appear to the calculating of other most of combustible gas mixture.Li Guoliang etc. then adopt constant adiabatic flame temperature (CAFT) method that the lower explosive limit of inflammable gas-air-nitrogen mixture and binary/ternary burning mixture has been carried out forecasting research, have obtained effect preferably.But the maximum shortcoming of this method is to calculate equally comparatively loaded down with trivial detailsly, and will depend on the software of specialty.Simultaneously, this method is the basis with the chemical thermodynamics, receives the influence of Chemical Kinetics little, therefore generally only limits to the lower explosive limit of prediction hybrid gas.People such as Wei Yongsheng then attempt coming according to the volumetric concentration of each component the explosion limits of prediction hybrid gas.They are to H2 commonly used in the Chemical Manufacture, CO, and the explosion limits of CH4 combination gas under different proportionings measured; Through a large amount of experimental datas are carried out linear regression analysis; Combination gas (H2, CO, CH4) the linear relationship model between explosion limits and each the component volumetric concentration have been set up.Subsequently, people such as Zheng Ligang propose to adopt neural net method to predict and contain H2, the explosion limits of the multi component mixed gas of CH4 and CO according to the characteristics that have nonlinear relationship between combination gas explosion limits and each the component volumetric concentration.The nonlinear model of being set up is compared predicated error with linear model have obvious reduction.The maximum shortcoming of volumetric concentration experimental formula method is the statistics rule of only considering between mixed gas explosion limits and each component volumetric concentration; Do not consider the influence of physical chemical factor to explosion limits; Cause institute's established model to lack physical significance; Prediction effect depends on the precision of experimental data, lacks of theoretical foundation fully.Simultaneously, selected specific components gaseous mixture then can't be predicted for nonspecific potpourri when the volumetric concentration empirical model was only applicable to modeling, and the scope of application is single
This shows; Support vector machine method is as a kind of novel machine learning algorithm; Owing to have strong non-linear mapping ability and good extensive performance; Under the situation that parameter is selected rationally, training method is proper, be the complex relationship that can give full expression between organic mixture explosive characteristic and conventional rerum natura and the proportioning, thereby set up effective potpourri explosive characteristic forecast model.Compare with existing multi component mixed gas body explosion limits forecast model; The forecast model that the present invention is based on support vector machine method foundation only compares the prediction that just can realize the mixed gas explosion limits according to conventional rerum natura and set of dispense; Fast simple; And forecasting accuracy is high, applied widely, has shown that this method has good effect aspect the prediction of multi component mixed gas body explosion limits.
The concrete grammar that the application binary burning mixture UEL forecast model that the present invention set up is predicted unknown multi component mixed gas body UEL is following:
(the SVM parameter that model is corresponding is C=1024 to the SVM forecast model of having set up according to preamble; ε=0.125; γ=0.0625); Only need during prediction to form and proportioning, inquire about or calculate the concrete numerical value of coefficient of diffusion, heating power, emergent pressure and the stoichiometric concentration of mixed gas, in conjunction with the set of dispense ratio of this mixed gas according to the material of this unknown mixed gas; The input variable of SVM model has been built in conduct together, can obtain the UEL numerical value of this mixed gas after calculating through supporting vector machine model.
The inventor has set up a kind of organic mixture explosive characteristic prediction method based on SVMs among the present invention.To the explosive characteristic and the problem of optimizing the industrial process design that need the prediction organic mixture in the commercial production under different component and different proportionings; According to existing organic mixture experimental data; With component of mixture content and conventional rerum natura is input variable; Explosive characteristic with correspondence is an output variable; Utilize powerful machine learning algorithm support vector machine method, non-linear, the uncertain and complicated inherent quantitative relationship that exists is between the two effectively trained and forecast, thereby set up stable, SVM prediction model efficiently.Utilize the supporting vector machine model of being set up that the explosive characteristic of other unknown potpourris is predicted; Have precision of prediction height, advantage fast and easily; Realized according to conventional rerum natura and set of dispense function than fast prediction potpourri explosive characteristic; Solved the problem that organic mixture explosive characteristic experimental data lacks effectively, therefore good prospects for application has been arranged industrial process design and fire-proof and explosion-proof the grade in the work.Use the inventive method not only to avoid buying and using the experimental facilities of complex and expensive; And reduced required a large amount of human and material resources and the time of measuring and dropped into; Have important use for the enterprises and institutions that do not possess combustion explosion measuring ability and be worth, economic benefit is fairly obvious.
The present invention does not relate to all identical with the prior art prior art that maybe can adopt of part and realizes.
Claims (4)
1. Forecasting Methodology based on the organic mixture explosive characteristic of SVMs; It is characterized in that; With the component concentration of organic mixture and the explosive characteristic experimental data of conventional rerum natura and these potpourris correspondence is sample; The supporting vector machine model that foundation utilizes regression function to estimate utilizes supporting vector machine model to predict the explosive characteristic of unknown organic mixture again.
2. the Forecasting Methodology of the organic mixture explosive characteristic based on SVMs according to claim 1 is characterized in that said Forecasting Methodology may further comprise the steps:
(1) set up sample data: the organic mixture of collecting at least 100 kinds of different components and content proportioning is as sample; Component concentration and conventional rerum natura and explosive characteristic experimental data with these samples; As sample data; Select 2/3rds sample data as the training sample data at random, be used to set up forecast model; Remaining about 1/3rd sample data is as the forecast sample data, is used for institute's established model is estimated and verified.
(2) set up supporting vector machine model: to the training sample data; With the component concentration of organic mixture and conventional rerum natura as input variable; Corresponding explosive characteristic is as output variable; Use support vector machine method internal relation is between the two simulated, the quantitative function relation of seeking to exist is between the two set up corresponding forecast model;
The correlation parameter of decision SVMs modeling performance mainly comprises: the size of ε in the parameter of kernel function, kernel function, penalty coefficient C and ε-insensitive loss function; Kernel function is selected radially base nuclear K (x, x for use
i)=exp (γ || x-x
i||
2), because it has higher learning efficiency and learning rate; Other parameter is confirmed through " grid search " method; The parameter search scope is following: penalty coefficient C---0-1024; The parameter of kernel function (width) γ---0-1024; ε---0-1024 in ε-insensitive loss function; The direction of search is the lowest mean square root error (RMSE) of " staying 1/10 method " cross-verification; Validation-cross is meant 1/10 sample that from training sample, at every turn screens out the training sample sum " to stay 1/10 method "; With remaining sample modeling; Come forecasting institute to screen out the character of sample; The root-mean-square error (RMSE) that obtains a validation-cross is like this come the quality of evaluation model performance, and its computing formula is:
Wherein, y
I, predBe the predicted value of sample i, y
I, expExperiment value for sample i; Through search, choose the optimum input parameter of pairing that group parameter of minimum RMSE of " staying 1/10 method " cross-verification as model;
The optimized parameter that application searches goes out is set up corresponding forecast model as the input parameter of SVMs;
(3) prediction organic mixture explosive characteristic: in component concentration in the forecast sample data and the conventional rerum natura supporting vector machine model that input is set up as input variable, calculate the explosive characteristic of forecast sample through supporting vector machine model;
(4) correction and definite forecast model: the predicted value and the experiment value of the forecast sample explosive characteristic that comparison step (3) obtains; When if the deviation of predicted value and experiment value surpasses acceptable scope; Correlation parameter numerical value to SVMs is regulated; And then train again and predict, until the deviation of predicted value and experiment value within the acceptable range, thereby confirm the SVM prediction model;
(5) application of forecast model: utilize determined SVM prediction model that the explosive characteristic of other unknown organic mixture is predicted.
3. Forecasting Methodology according to claim 1; It is characterized in that; The conventional rerum natura of said organic mixture comprises viscosity, relative density, vapour pressure, coefficient of thermal expansion, boiling point, intermolecular force, eccentric factor, atomic polarizability, oxygen index, Van der waals volumes, combustion rate, stoichiometric concentration, critical temperature, emergent pressure and coefficient of diffusion one of at least, and their combination in any.
4. Forecasting Methodology according to claim 1 is characterized in that described explosive characteristic comprises spontaneous ignition temperature, explosion limits and heating power.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210038882.1A CN102608285B (en) | 2012-02-21 | 2012-02-21 | Organic mixture blasting characteristic prediction method based on support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210038882.1A CN102608285B (en) | 2012-02-21 | 2012-02-21 | Organic mixture blasting characteristic prediction method based on support vector machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102608285A true CN102608285A (en) | 2012-07-25 |
| CN102608285B CN102608285B (en) | 2014-08-06 |
Family
ID=46525825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210038882.1A Active CN102608285B (en) | 2012-02-21 | 2012-02-21 | Organic mixture blasting characteristic prediction method based on support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102608285B (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102901804A (en) * | 2012-10-25 | 2013-01-30 | 四川威特龙消防设备有限公司 | Oil-gas explosion critical parameter analysis method |
| CN104573371A (en) * | 2015-01-16 | 2015-04-29 | 上海大学 | Optimal wavelet packet transformation and AR-SVM based non-stable wind speed prediction method |
| CN104964997A (en) * | 2015-03-12 | 2015-10-07 | 大连理工大学 | Method for quickly determining content of heterogeneous media in material based on physical property matching |
| CN104374867B (en) * | 2014-11-24 | 2016-01-06 | 重庆消防安全技术研究服务有限责任公司 | Fire proofing and goods oxygen index field fast detection method |
| CN105849716A (en) * | 2013-12-26 | 2016-08-10 | 道达尔炼油化学公司 | Method and device for monitoring induced properties of mixture of components, in particular emission properties |
| CN106503288A (en) * | 2016-09-18 | 2017-03-15 | 江苏大学 | Interconnection hydraulic cylinder mechanical properties prediction algorithm based on Support Vector Machines for Regression |
| CN107357966A (en) * | 2017-06-21 | 2017-11-17 | 山东科技大学 | A kind of surrounding rock of actual mining roadway stability prediction and appraisal procedure |
| CN107944070A (en) * | 2017-10-09 | 2018-04-20 | 中国电子科技集团公司第二十八研究所 | A kind of Diffusion Simulation method and system of urban atmospheres harmful influence leakage |
| CN109409533A (en) * | 2018-09-28 | 2019-03-01 | 深圳乐信软件技术有限公司 | A kind of generation method of machine learning model, device, equipment and storage medium |
| CN109522668A (en) * | 2018-11-28 | 2019-03-26 | 南京工业大学 | Gas phase protection method for inhibiting spontaneous combustion of pyrite compound on inner wall of petrochemical equipment |
| CN109596803A (en) * | 2018-12-09 | 2019-04-09 | 西安科技大学 | The principal component Multiple Regression Analysis Method of methane explosion minimum oxygen concentration Intrusion Index |
| CN111223519A (en) * | 2018-11-23 | 2020-06-02 | 中国石油化工股份有限公司 | System for determining basic physical property parameters of organic matters and application thereof |
| WO2020228215A1 (en) * | 2019-05-14 | 2020-11-19 | 北京工业大学 | Soft measurement method for dioxin emission concentration in solid waste incineration process based on multi-layer feature selection |
| CN114166773A (en) * | 2021-12-08 | 2022-03-11 | 中煤科工集团重庆研究院有限公司 | Particle swarm optimization-support vector machine-based NOx measurement method |
| CN117236198A (en) * | 2023-11-14 | 2023-12-15 | 中国石油大学(华东) | Machine learning solving method of flame propagation model of blasting under sparse barrier |
| CN117332240A (en) * | 2023-12-01 | 2024-01-02 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104933310A (en) * | 2015-06-19 | 2015-09-23 | 河海大学 | Method for predicting membrane pollution tendency in membrane distilled water processing system on the basis of GA-LSSVM (Genetic Algorithm- Least Squares Support Vector Machine) model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030176938A1 (en) * | 2002-02-04 | 2003-09-18 | Tuszynski Steve W. | Manufacturing design and process analysis system |
| CN101339180A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound combustion and explosion characteristic prediction method based on support vector machine |
-
2012
- 2012-02-21 CN CN201210038882.1A patent/CN102608285B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030176938A1 (en) * | 2002-02-04 | 2003-09-18 | Tuszynski Steve W. | Manufacturing design and process analysis system |
| CN101339180A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound combustion and explosion characteristic prediction method based on support vector machine |
Non-Patent Citations (1)
| Title |
|---|
| 时静洁等: "基于遗传算法的支持向量机预测有机物自燃点的研究", 《中国安全科学学报》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102901804A (en) * | 2012-10-25 | 2013-01-30 | 四川威特龙消防设备有限公司 | Oil-gas explosion critical parameter analysis method |
| CN105849716A (en) * | 2013-12-26 | 2016-08-10 | 道达尔炼油化学公司 | Method and device for monitoring induced properties of mixture of components, in particular emission properties |
| CN104374867B (en) * | 2014-11-24 | 2016-01-06 | 重庆消防安全技术研究服务有限责任公司 | Fire proofing and goods oxygen index field fast detection method |
| CN104573371A (en) * | 2015-01-16 | 2015-04-29 | 上海大学 | Optimal wavelet packet transformation and AR-SVM based non-stable wind speed prediction method |
| CN104964997A (en) * | 2015-03-12 | 2015-10-07 | 大连理工大学 | Method for quickly determining content of heterogeneous media in material based on physical property matching |
| CN106503288A (en) * | 2016-09-18 | 2017-03-15 | 江苏大学 | Interconnection hydraulic cylinder mechanical properties prediction algorithm based on Support Vector Machines for Regression |
| CN106503288B (en) * | 2016-09-18 | 2019-06-28 | 江苏大学 | Interconnection hydraulic cylinder mechanical property prediction method based on Support Vector Machines for Regression |
| CN107357966A (en) * | 2017-06-21 | 2017-11-17 | 山东科技大学 | A kind of surrounding rock of actual mining roadway stability prediction and appraisal procedure |
| CN107357966B (en) * | 2017-06-21 | 2020-07-03 | 山东科技大学 | Evaluation method for stability prediction of surrounding rock of stoping roadway |
| CN107944070A (en) * | 2017-10-09 | 2018-04-20 | 中国电子科技集团公司第二十八研究所 | A kind of Diffusion Simulation method and system of urban atmospheres harmful influence leakage |
| CN107944070B (en) * | 2017-10-09 | 2021-06-25 | 中国电子科技集团公司第二十八研究所 | Diffusion simulation method and system for leakage of urban gas hazardous chemical |
| CN109409533A (en) * | 2018-09-28 | 2019-03-01 | 深圳乐信软件技术有限公司 | A kind of generation method of machine learning model, device, equipment and storage medium |
| CN111223519B (en) * | 2018-11-23 | 2023-06-13 | 中国石油化工股份有限公司 | System for determining basic physical property parameters of organic matters and application thereof |
| CN111223519A (en) * | 2018-11-23 | 2020-06-02 | 中国石油化工股份有限公司 | System for determining basic physical property parameters of organic matters and application thereof |
| CN109522668A (en) * | 2018-11-28 | 2019-03-26 | 南京工业大学 | Gas phase protection method for inhibiting spontaneous combustion of pyrite compound on inner wall of petrochemical equipment |
| CN109596803A (en) * | 2018-12-09 | 2019-04-09 | 西安科技大学 | The principal component Multiple Regression Analysis Method of methane explosion minimum oxygen concentration Intrusion Index |
| CN109596803B (en) * | 2018-12-09 | 2019-08-27 | 西安科技大学 | The principal component Multiple Regression Analysis Method of methane explosion minimum oxygen concentration Intrusion Index |
| WO2020228215A1 (en) * | 2019-05-14 | 2020-11-19 | 北京工业大学 | Soft measurement method for dioxin emission concentration in solid waste incineration process based on multi-layer feature selection |
| CN114166773A (en) * | 2021-12-08 | 2022-03-11 | 中煤科工集团重庆研究院有限公司 | Particle swarm optimization-support vector machine-based NOx measurement method |
| CN117236198A (en) * | 2023-11-14 | 2023-12-15 | 中国石油大学(华东) | Machine learning solving method of flame propagation model of blasting under sparse barrier |
| CN117236198B (en) * | 2023-11-14 | 2024-02-27 | 中国石油大学(华东) | Machine learning solving method of flame propagation model of blasting under sparse barrier |
| CN117332240A (en) * | 2023-12-01 | 2024-01-02 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
| CN117332240B (en) * | 2023-12-01 | 2024-04-16 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102608285B (en) | 2014-08-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102608285B (en) | Organic mixture blasting characteristic prediction method based on support vector machine | |
| CN102608284B (en) | Method for determining explosion limit of multi-component mixed gas | |
| Jafari et al. | Exploring inherent process safety indicators and approaches for their estimation: A systematic review | |
| CN102980972B (en) | Method for determining thermal risk of self-reactive chemical substance | |
| Lowry et al. | Laminar flame speed measurements and modeling of pure alkanes and alkane blends at elevated pressures | |
| Ruiz-Mercado et al. | Sustainability indicators for chemical processes: II. Data needs | |
| Lu et al. | Systematic approach to obtain analytic solutions of quasi steady state species in reduced mechanisms | |
| CN101339180B (en) | Organic compound combustion and explosion characteristic prediction method based on support vector machine | |
| Ortiz-Espinoza et al. | Including inherent safety in the design of chemical processes | |
| Szanthoffer et al. | Testing of NH3/H2 and NH3/syngas combustion mechanisms using a large amount of experimental data | |
| Grande et al. | Process intensification in nitric acid plants by catalytic oxidation of nitric oxide | |
| Pan et al. | Predicting the net heat of combustion of organic compounds from molecular structures based on ant colony optimization | |
| Gharagheizi | Prediction of upper flammability limit percent of pure compounds from their molecular structures | |
| Liu et al. | Thermal hazard investigation and hazardous scenarios identification using thermal analysis coupled with numerical simulation for 2-(1-cyano-1-methylethyl) azocarboxamide | |
| Hu et al. | Effect of carbon dioxide on the lower flammability limit of propane in O2/CO2 atmosphere | |
| Li et al. | Extended adiabatic flame temperature method for lower flammability limits prediction of fuel-air-diluent mixture by nonstoichiometric equation and nitrogen equivalent coefficients | |
| Guo et al. | Thermal hazard evaluation conjoined with product analysis of two water-soluble azo compounds | |
| Khalil | Experimental investigation of the complex deflagration phenomena of hybrid mixtures of activated carbon dust/hydrogen/air | |
| Bhimineni et al. | Machine-learning-assisted investigation of the diffusion of hydrogen in brine by performing molecular dynamics simulation | |
| Liu et al. | Assessing the thermal properties of [Bmim] NO3 through thermokinetic calculations and the energy equilibrium method | |
| Alaghemandi et al. | Ignition in an atomistic model of hydrogen oxidation | |
| Ni et al. | Predicting both lower and upper flammability limits for fuel mixtures from molecular structures with same descriptors | |
| Bounaceur et al. | Development of an Artificial Intelligence Model to Predict Combustion Properties, With a Focus on Auto-Ignition Delay | |
| Wan et al. | Estimation of the upper flammability limits of hydrocarbons in air at elevated temperatures and atmospheric pressure | |
| Ahmad et al. | Safety assessment: predicting fatality rates in methanol plant incidents |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |