CN102980972A - Method for determining hot dangerousness of self-reactive chemical substance - Google Patents
Method for determining hot dangerousness of self-reactive chemical substance Download PDFInfo
- Publication number
- CN102980972A CN102980972A CN2012104413678A CN201210441367A CN102980972A CN 102980972 A CN102980972 A CN 102980972A CN 2012104413678 A CN2012104413678 A CN 2012104413678A CN 201210441367 A CN201210441367 A CN 201210441367A CN 102980972 A CN102980972 A CN 102980972A
- Authority
- CN
- China
- Prior art keywords
- chemical substance
- autoreaction
- sample
- parameter
- model
- 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
- 238000000034 method Methods 0.000 title claims abstract description 118
- 239000000126 substance Substances 0.000 title claims abstract description 102
- 238000012549 training Methods 0.000 claims description 33
- 238000004422 calculation algorithm Methods 0.000 claims description 31
- 238000012706 support-vector machine Methods 0.000 claims description 25
- 238000002474 experimental method Methods 0.000 claims description 21
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 10
- 238000012795 verification Methods 0.000 claims description 7
- 239000000203 mixture Substances 0.000 claims description 6
- 238000013459 approach Methods 0.000 claims description 5
- 230000000052 comparative effect Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000006467 substitution reaction Methods 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract 1
- 230000002265 prevention Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 35
- 150000001875 compounds Chemical class 0.000 description 34
- 230000000875 corresponding effect Effects 0.000 description 30
- 239000000463 material Substances 0.000 description 18
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 16
- 239000001301 oxygen Substances 0.000 description 16
- 229910052760 oxygen Inorganic materials 0.000 description 16
- 238000004880 explosion Methods 0.000 description 15
- 238000011160 research Methods 0.000 description 11
- 238000002485 combustion reaction Methods 0.000 description 10
- 238000000354 decomposition reaction Methods 0.000 description 10
- 230000000694 effects Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 6
- 230000002269 spontaneous effect Effects 0.000 description 6
- 238000002076 thermal analysis method Methods 0.000 description 6
- 230000015572 biosynthetic process Effects 0.000 description 5
- 230000002950 deficient Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 238000012512 characterization method Methods 0.000 description 4
- 238000013210 evaluation model Methods 0.000 description 4
- 239000002360 explosive Substances 0.000 description 4
- 230000003647 oxidation Effects 0.000 description 4
- 238000007254 oxidation reaction Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 230000001687 destabilization Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 239000003016 pheromone Substances 0.000 description 3
- 239000011541 reaction mixture Substances 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- PAWQVTBBRAZDMG-UHFFFAOYSA-N 2-(3-bromo-2-fluorophenyl)acetic acid Chemical compound OC(=O)CC1=CC=CC(Br)=C1F PAWQVTBBRAZDMG-UHFFFAOYSA-N 0.000 description 2
- 238000004618 QSPR study Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000000324 molecular mechanic Methods 0.000 description 2
- 150000001451 organic peroxides Chemical class 0.000 description 2
- 230000009024 positive feedback mechanism Effects 0.000 description 2
- 230000005610 quantum mechanics Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000005979 thermal decomposition reaction Methods 0.000 description 2
- 206010000372 Accident at work Diseases 0.000 description 1
- OMPJBNCRMGITSC-UHFFFAOYSA-N Benzoylperoxide Chemical compound C=1C=CC=CC=1C(=O)OOC(=O)C1=CC=CC=C1 OMPJBNCRMGITSC-UHFFFAOYSA-N 0.000 description 1
- -1 CHO class organic compound Chemical class 0.000 description 1
- 108010074864 Factor XI Proteins 0.000 description 1
- 241000257303 Hymenoptera Species 0.000 description 1
- 206010021143 Hypoxia Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- HSFWRNGVRCDJHI-UHFFFAOYSA-N alpha-acetylene Natural products C#C HSFWRNGVRCDJHI-UHFFFAOYSA-N 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 235000019400 benzoyl peroxide Nutrition 0.000 description 1
- 230000000975 bioactive effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000013522 chelant Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 125000002534 ethynyl group Chemical group [H]C#C* 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 239000002778 food additive Substances 0.000 description 1
- 235000013373 food additive Nutrition 0.000 description 1
- 230000019637 foraging behavior Effects 0.000 description 1
- 239000003721 gunpowder Substances 0.000 description 1
- 239000000383 hazardous chemical Substances 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 230000001146 hypoxic effect Effects 0.000 description 1
- 230000001771 impaired effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000010327 methods by industry Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 150000002828 nitro derivatives Chemical class 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
- 238000011002 quantification Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000009257 reactivity Effects 0.000 description 1
- 238000006479 redox reaction Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000005556 structure-activity relationship Methods 0.000 description 1
- 229920002994 synthetic fiber Polymers 0.000 description 1
- 239000002341 toxic gas Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Images
Landscapes
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a method for determining hot dangerousness of self-reactive chemical substances, including the following steps: 1. collecting self-reactive chemical substance experimental samples and hot dangerousness experimental data; 2. describing molecular structures; 3. dividing a sample set; 4. selecting a characteristic structure; 5. establishing a prediction model; 6. validating, correcting and determining the prediction model; and 7. applying the prediction model. The study of the prediction method of self-reactive chemical substance hot dangerousness is of great significance for assessment, prevention and monitoring of self-reactive chemical substance hot dangerousness. The method is simple, and high in predictive accuracy, and provides a simple, rapid, accurate and reliable method to determine the hot dangerousness of the self-reactive chemical substances.
Description
Technical field
The present invention relates to field of chemical technology, especially a kind of method of definite autoreaction chemical substance thermal hazard, specifically a kind of molecular structure information according to the autoreaction chemical substance is predicted the method for its thermal hazard.
Background technology
In today of modern chemistry industry develop rapidly, great industrial accident occurs in succession all over the world, and the accidents such as toxic gas leakage, hazardous chemical fire explosion emerge in an endless stream.In 1960 ~ 1977 years 18 years, the fire disaster explosion accident more than 360 that chemical substance occurs altogether for the U.S. and West Europe rises, dead and wounded 1979 people, and direct economic loss is above 1,000,000,000 dollars.China's chemical industry accident is frequent occurrence especially, in 1950 ~ 1999 years 50 years, 23425 of all kinds of casualty accidents occurs, dead and wounded 25714 people, the wherein fire explosion of chemical substance casualties number 4043 people.
In the fire explosion of numerous chemical substances, the accident that causes is one of common accident form because of the thermal hazard (hot spontaneous combustion, thermal decomposition, thermal explosion) of autoreaction chemical substance.The autoreaction chemical substance refers generally to itself to have certain energy, need to just can not carry out by the oxygen in the external world in the molecule decomposing, in the molecule or the chemical substance of intermolecular redox reaction.It is the security incidents such as easily breaking out of fire, blast under extraneous energy not only, and even without the effect of outside energy, chemical reaction in various degree can occur also under field conditions (factors), emit heat.Usually said autoreaction chemical substance comprises organic peroxide, oxygenant, itrated compound, gunpowder, explosive etc.If the chemical heat release speed in the system that is comprised of these materials greater than the radiating rate of this system to environment, will cause the heat accumulation in the system, finally cause hot spontaneous combustion or thermal explosion accident.In commercial production and the development of the national economy, fire, explosion accident frequent occurrence because hot spontaneous combustion of autoreaction chemical substance etc. causes have brought huge loss to the people's lives and property.September 21 calendar year 2001, big bang has caused because of the hot spontaneous combustion of ammonium nitrate in the production run in the factory of the manufacturing ammonium nitrate fertilizer of South of France city Te Lushi, the explosion scene has stayed the fried hole of diameter 50m, deeply about 15m, all is subject to destruction in various degree from explosion scene 5km house in addition.This plays the blast accident and causes altogether 30 people dead, and 2442 people are injured, and the most of house of factory is wracked, and the private residence more than 4000 and the schoolhouse more than 80 are impaired.In February, 2003, Wujin County, Jiangsu Province Wujin, Jiangsu Province Jiao Xi food additives factory blasts when producing dibenzoyl peroxide, causes 5 people dead, and 1 people is injured.In July, 2003, the Guo Xi of Xinji City, Hebei province fireworks and firecrackers factory is because the medicine ball temperature is high, high humidity causes hot spontaneous combustion and causes together especially big explosion accident, and accident causes 35 people dead altogether, and 103 people are injured, and the window in house all is shattered beyond the 3km.
Therefore, in order to ensure this security of class autoreaction chemical substance in the processes such as production, storage, transportation and use, must have sufficient understanding to its response characteristic, its potential thermal hazard be carried out the evaluation study of science.At present, the application experiment assay method is to estimate and prediction autoreaction chemical substance thermal hazard effective method the most directly perceived, but experimental technique also exists following defective and deficiency: (1) experimental technique not only requires to possess good experimental facilities, and workload is huge; (2) because the hazard property difference of autoreaction chemical substance, the experimental apparatus that possesses is difficult to all kinds of autoreaction chemical substances are estimated, must consider simultaneously the characteristic of instrument and material, experimental technique to them effectively makes up, and it is impossible therefore will testing one by one them; (3) consider safety problem in the experimentation, general experimental study can only be on a small scale, undersized experiment, there is no method with the simulated experiment of first approximation and embody preferably scale effect; (4) poisonous, volatile, explosive for those or the material of radiation is arranged, exist certain difficulty in the measurement; (5) for those not yet synthetic materials, and some is own highly unstable, or can produce the chemical substance of very unsettled intermediate or high malicious product in the process of the test, also can't determine its danger based on experiment.
This shows, merely Applied experimental study is estimated and is predicted that the thermal hazard of autoreaction chemical substance is worthless, be necessary by theoretical calculation method the thermal hazard of autoreaction chemical substance to be carried out theoretical calculation or prediction, to remedy the shortcomings and deficiencies of experiment research.
There are some researches show the impact of the atomic group that the heat release behavior of autoreaction chemical substance is subject to comprising in the material.Can infer thus, between the molecular structure and thermal hazard of autoreaction chemical substance, certainly exist some intrinsic inner link.According to this rule, in recent years, different researchers has proposed some methods according to molecular structure qualitative evaluation autoreaction chemical substance thermal hazard from the molecular structure angle.
(1) oxygen balance method
Find by the research to the different classes of organic explosive substance of kind more than 300, have certain intrinsic contacting between the oxygen balance of chemical substance and its explosion property.Oxidation element in the oxygen balance of chemical substance (oxygen balance, OB) the expression material molecule is used for the contained combustible element of complete oxidation remaining or not enough oxidation amount of element when being the complete oxidation product itself.For the chemical substance that consists of CaHbOcNd, its
In the formula, M is the material relative molecular weight.OB〉0 o'clock claim positive oxygen balance, claim zero oxygen balance during OB=0, OB＜0 o'clock claims negative oxygen balance.The method often is used to the potential danger classes of evaluating chemical material, and its oxygen balance numerical values recited and corresponding danger classes see Table 1.
Table 1 oxygen balance and danger classes
| Oxygen balance (OB) | Danger classes |
| OB≥160 | Low |
| 80≤OB<160 | Medium |
| -120≤OB<80 | High |
| -240≤OB<-120 | Medium |
| OB<-240 | Low |
The advantage of the method is effectively to predict the explosion hazard of the organic nitro-compound that contains the elements such as CHON, and is widely used in explosive industry.Yet existing research also shows, does not have inevitable contact between the oxygen balance of compound and the autoreaction.For example, the oxygen balance value of water is 0, and according to this method, its corresponding danger classes should be " height ", does not obviously square with the fact; In addition, less than normal because of its calculating OB value for those hypoxic compounds, draw easily corresponding danger classes wrong conclusion on the low side, and in fact these compounds often have stronger thermal hazard; Simultaneously, do not contain the dangerous substance of aerobic for those, such as acetylene etc., the method is also inapplicable.
(2) CHETAH method
CHETAH is U.S. material and the assessment process that is used for the prediction reactivity hazard of testing association (ASTM) exploitation.This program has two kinds of major functions, and the one, according to the molecular structure of compound, the thermodynamics parameters such as thermal capacitance, Heat of Formation and heating power when utilizing the Benson group to add with method reckoning compound gaseous state; The 2nd, utilize these thermodynamics parameters to calculate the some characterisitic parameters relevant with thermal hazard, and estimate its thermal hazard by the criterion of drafting.The thermal hazard criterion that the CHETAH program comprises mainly contains 5, but practical application maximum be maximum decomposition heat standard and oxygen balance standard.According to maximum decomposition heat standard, if the maximum decomposition heat 〉=700cal/g of something, then its danger classes is high; If maximum decomposition heat≤300cal/g, then its danger classes is low; In then danger classes is between the two.About the oxygen balance standard, in CHETAH, judge that with oxygen balance dangerous standard can be referring to table 1.The detailed using method of above-mentioned standard and all the other 3 standards can be referring to the CHETAH user manual.Wherein, maximum decomposition heat standard is proved to be has stronger reliability, can be effectively applied in judging the potential thermal hazard of material; All the other indexs then do not have general applicability.
The advantage of CHETAH method is only to know the structural formula of compound, just can not estimate its danger by experiment.And this software interface is friendly, and is easy to use, do not need complicated calculating, just can realize the fast prediction to compound danger.
Yet the CHETAH method also exists some defectives and deficiency: at first, this software does not also have unified value for the Heat of Formation of all important group.And the compound that contains the destabilization group is because destabilization own, and Heat of Formation generally is difficult to actual measurement, therefore uses often to lack necessary data when the CHETAH method is estimated these destabilization materials.Simultaneously, CHETAH method employing Benson group adds the Heat of Formation with the method predictive compound, and the Benson method can not embody steric hindrance, the influence factors such as annular strain and generation chelate, although can consider vucubak effectl alpha effect and the annular strain of typical compound, for the prediction noval chemical compound, be nowhere near.In addition, the CHETAH method is mainly judged having or not of its danger or size by the size of calculating compound decomposition heat, but for causing decomposition reaction under which kind of condition, at present can't perfect forecast, and method is determined by experiment.
(3) CART method
CART is the english abbreviation that calculates adiabatic reaction temperature (calculated adiabatic reaction temperature).The adiabatic temperature rise that causes because of reaction is defined as:
Wherein, △ H is reaction heat, and Cp is the mean heat capacity of reaction mixture, △
TadiabaticBe adiabatic temperature rise.The CART method can realize the calculating with the adiabatic reaction temperature of minimizing of heterogeneous Gibbs free energy.According to the adiabatic temperature rise Δ T that calculates
Adiabatic, material is divided into E(represents the time can blast without constraint) and N(represent not exist when retraining known explosion hazard) two classes.The standard of above-mentioned classification be adiabatic temperature rise whether greater than 1400K, if adiabatic temperature rise greater than 1400K, then this material is divided into E; Otherwise then be N.This standard mainly is because most CO of generation
2And H
2Its critical temperature of the combustion reaction of O is all close to 1400K, and this temperature to be CO can propagate from the steady needed minimum temperature of flame.
Only consider that with the CHETAH method reaction energy compares, the CART method has been owing to taken into account the thermal capacitance of reaction mixture, thereby generally has better prediction effect.Simultaneously, higher CART value shows that toward contact material has larger ignite sensitivity and higher propagation rate.Yet, also there are following obvious defectives in the CART method: 1. the method must be with thermophysical parameters such as Δ H and Cp as input value, and these parameters generally calculate to obtain by CHETAH software, can run into equally the Benson group and add above-mentioned inherent shortcoming with method in computation process; Although 2. the CART method is good to the thermal hazard prediction effect of the compound of those easy generation combustion reactions, for those compound of combustion reaction does not occur, at definite its Δ T
AdiabaticThere is certain difficulty during critical value; 3. the CART method can't effectively be predicted the danger of organic peroxide; 4. the CART method is being calculated Δ T
AdiabaticProcess in need to use the mean heat capacity of reaction mixture, and the precise information of thermal capacitance is difficult to obtain by prediction; 5. similar with the CHETAH method, the CART method is the unpredictable generation that can cause decomposition reaction under which kind of condition also.
This shows that the thermal hazard prediction and evaluation method of existing autoreaction chemical substance rests on the qualitatively stage mostly, is all limited largely in practical engineering application.According to the knowledge of the applicant, at present from the molecular structure angle, the correlative study of quantitative forecast autoreaction chemical substance thermal hazard is still blank on document.
As everyone knows, molecular structure determines that molecular property is a basic law in the chemical research.Quantitative Structure-Property Relationship correlativity (Quantitative Structure-Property Relationship, QSPR) research is exactly a kind of advanced method according to the Molecular structure prediction molecule performance that in recent years formation and development is got up.It seeks the inherent quantitative relationship between molecular structure and the molecule performance according to compounds property and the closely-related principle of molecular structure.Its basic assumption is that organic performance and molecular structure are closely related, and molecular structure can be described with the various parameters (molecule descriptor) of reflection molecular characterization, and namely the performance of compound can represent with the function of chemical constitution.By adopting suitable statistical modeling method to carry out association to the inherent quantitative relationship between molecular structural parameter and the destination properties, thereby set up the causes between molecular structural parameter and the destination properties.In case set up reliable QSPR model, only need molecular structures information, just can predict with it the various character of compound new or that not yet synthesize.At present, QSPR research has been widely used among the forecasting research of all kinds of conventional physicochemical properties of compound, become in chemistry, medical science, life science and the environmental science study hotspot it
The superiority of QSPR method is mainly reflected in: (1) need not other empirical parameter, only just can realize the prediction of autoreaction chemical substance thermal hazard according to molecular structure; (2) the employed input parameter of forecast model is less, can guarantee the stability of institute's established model; (3) in a single day set up reliable and stable forecast model, can predict all organism according to this model in theory, applied widely.
The main thought of QSPR research is: at first calculate the structural parameters of a large amount of reflection molecular structure informations according to molecular structure, such as topological parameter, composition parameter, electrical parameter and the Quantum chemical parameters etc. of molecule; Use subsequently the characteristic variable screening technique from a large amount of structural parameters that calculate, to choose to comprise the characteristic parameter that enriches structural information as the molecule descriptor, at last for selected descriptor and the inherent quantitative relationship between the physicochemical property studied, adopt suitable statistical modeling method to carry out association, set up pervasive forecast model.
Not yet have at present on the document and use the QSPR method is determined autoreaction chemical substance thermal hazard according to molecular structure relevant report.
Summary of the invention
The objective of the invention is for mainly relying at present experiment and theoretical calculation to determine the thermal hazard of autoreaction chemical substance, cost is high, the cycle is long, dangerous large shortcoming and experimental method exists, existing evaluation method exists again and uses loaded down with trivial details, poor accuracy, is difficult to the problems such as quantifications, narrow application range, and proposed a kind of prediction effect good, applied widely, only need molecular structure information just can determine the method for autoreaction chemical substance thermal hazard.
Technical scheme of the present invention is: according to the molecular structure of autoreaction chemical substance, calculate the structural parameters that are used for the various structural informations of reflection molecule, realize the parametric description of molecular structure information; Use subsequently ant colony optimization algorithm (ant colonyoptimization, ACO) carry out characteristic variable screening, from the structural parameters that calculated in a large number, filter out closely-related with associated hot danger, comprise enrich structural information one group of parameter as Molecular structure descriptor.Adopt on this basis support vector machine method that the inherent quantitative relationship between selected molecule descriptor and the dangerous data of associated hot is carried out statistical learning, draw the quantitative function relation between autoreaction chemical substance molecular structure and the thermal hazard, set up corresponding forecast model.The thermal hazard data that can obtain being correlated with in the Molecular structure descriptor forecast model that input is set up as input parameter with the unknown autoreaction chemical substance of needs predictions.
Concrete technical scheme of the present invention is: a kind of method of definite autoreaction chemical substance thermal hazard, and only need molecular structure just can predict its thermal hazard, concrete steps are as follows:
(1) collection of autoreaction chemical substance experiment sample and thermal hazard experimental data thereof:
Select a series of autoreaction chemical substances according to statistical standard and construction standard, consist of the experiment sample collection; The condition that the autoreaction chemical substance is selected is statistical randomness, structural representativeness and comprehensive, and the availability of data; For the autoreaction chemical substance in the sample set, collect the thermal hazard data of paying close attention to; The approach of Data Collection mainly contains three kinds: measuring, various authoritative character data storehouse and handbook;
(2) description of molecular structure:
According to the molecular structure of autoreaction chemical substance in the sample set, calculate all kinds of structural parameters that obtain for the reflection molecular structure information, realize the parametric description of molecular structure information;
(3) division of sample set:
Be training set and two parts of forecast set with the sample set random division, wherein training set is used for setting up forecast model, and forecast set does not participate in modeling, is used for institute's established model is estimated and verified;
(4) choosing of feature structure:
For the training set sample, use has ant group-offset minimum binary (ACO-PLS) algorithm powerful global search function, that ant colony optimization algorithm is combined with the offset minimum binary method and carries out the characteristic variable screening, from a large amount of structural parameters that step (2) calculates, filter out the most closely-related with the target thermal hazard, comprise enrich structural information one group of parameter as the molecule descriptor of describing autoreaction chemical substance feature structure;
(5) foundation of forecast model:
For the training set sample, with the selected molecule descriptor of step (4) as input variable, corresponding thermal hazard is as output variable, use the support vector machine chemometrics method, internal relation between molecular structure and the target thermal hazard is carried out statistical modeling, the quantitative function relation that is existed is between the two set up corresponding forecast model;
(6) checking of forecast model, correction and definite:
Adopt cross verification to verify the robustness of the forecast model of building, adopt external certificate method (namely calculating the thermal hazard of forecast set sample with institute's established model) to verify the extrapolability of forecast model; According to predicting the outcome of validation-cross and external certificate, the predicted value of comparative sample and desired value, when if the mean deviation of predicted value and desired value surpasses acceptable scope (for example average relative error surpasses 10%), then reject the sample (for example Relative Error surpasses the sample of 2 times of average relative errors) that predicated error exceeds standard, return (3), again model and forecast, until the mean deviation of predicted value and desired value within the acceptable range (for example average relative error is less than 10%), thereby determine forecast model.
(7) application of forecast model:
Molecular structure for unknown autoreaction chemical substance, the molecule descriptor that screening is determined according to step (4), calculate corresponding numerical value, the determined corresponding forecast model of its substitution step (6) is calculated, namely obtain the corresponding thermal hazard data of this autoreaction chemical substance.
All kinds of structural parameters described in the above-mentioned steps (2) are topological parameter, composition parameter, geometric parameter, electrical parameter, electrical topological parameter and the Quantum chemical parameters of molecule.
Described thermal hazard is that reaction heat, themopositive reaction begin temperature, SADT and maximum temperature rise rate time.
Details are as follows:
Main points of the present invention are on the basis of extracting fully and effectively molecular structure information, adopt characteristic variable triage techniques ant group algorithm and statistical learning method support vector machine method, take lot of experimental data as the basis, statistical learning reaches by the molecular structure of existing autoreaction chemical substance sample and the inherent quantitative relationship between the dangerous data of associated hot are carried out.
At first, want to realize only just realizing according to the molecular structure of autoreaction chemical substance the prediction of its thermal hazard, just must effectively extract comprehensive molecular structure information, realize the parametrization of molecular structure.The present invention uses molecule simulation method and makes up correct two dimension or three-dimensional molecular structure, adopt the methods such as molecular mechanics, conformational analysis to obtain optimized conformation, adopt topology method, quantum mechanics method etc. to calculate the structural parameters that are used for the various structural informations of reflection molecule, the topological parameter, composition parameter, geometric parameter, electrical parameter, electrical topological parameter and the Quantum chemical parameters that comprise molecule, to obtain molecular structures information, realize the parametric description of molecular structure.
Secondly, the foundation of forecast model requires to select with the most closely-related structural parameters of associated hot danger as the molecule descriptor that characterizes molecular characterization.In order to describe the molecular structures feature, forefathers are own through proposing can be used in a large number the molecular structural parameter of sign molecular characterization, and these structural parameters can reflect many-sided structural informations such as composition, topology and electronic structure of molecule.But how selecting in the middle of numerous parameters with the most closely-related structural parameters of thermal hazard of studying is very crucial problems.The quality of forecast model depends on selected parameter to a great extent, from angle of statistics, wish to characterize structural information as much as possible with the least possible variable, because too much variable not only can increase calculated amount, also can cause the forecast model set up unstable, make the variation that predicts the outcome of model.For these reasons, the present invention has adopted the Variable Selection method based on ant group algorithm and offset minimum binary (ACO-PLS), the a large amount of molecular structural parameter that calculate are optimized screening, with pick out relatively poor or with the irrelevant parameter of the character of being studied, find out and the most closely-related structural parameters of target thermal hazard of studying as the molecule descriptor that characterizes autoreaction chemical substance feature structure.On this basis, for particular problem, select suitable statistical modeling method that the molecule descriptor that filters out and the inherent quantitative relationship between associated hot danger are simulated, set up corresponding thermal hazard forecast model.
Ant group algorithm is the colony intelligence optimized algorithm that a class of rising in recent years has ability of searching optimum, its superior distributed Solution model, implicit parallel computation characteristic and become the sharp keen weapon that solves the combinatorial optimization problem with NP-hard characteristic based on the reinforcement learning ability of positive feedback have caused association area scholar's extensive concern.Ant group algorithm is take the ant colony foraging behavior as background, by the monoid intelligent optimization algorithm that the people such as Dorig propose, has obtained preferably effect in the combinatorial optimization problems such as solution TSP (traveling salesmanproblem).Compare with traditional optimized algorithm, ant group algorithm has following advantage: concurrency and the Distributed Calculation of (1) essence.All ants are independent, unsupervised searches for many points in the solution space simultaneously, is very suitable for Parallel Implementation, thereby is a kind of efficient searching algorithm in essence.(2) powerful global optimizing ability.Probability of use rule, rather than Deterministic rules guidance search are so that algorithm can be fled from local optimum.(3) positive feedback mechanism.During the ant selecting paths, according to the pheromones information guiding search that former ant stays, this positive feedback mechanism is conducive to the ant group and finds the more solution of good quality.(4) strong adaptability.Ant group algorithm without any specific (special) requirements, such as connectedness, convexity etc., does not need other information such as derivative to the search volume.(5) be easy to be combined with other heuritic approach.
The present invention proposes a kind of algorithm that ant group algorithm and offset minimum binary method (ACO-PLS) are combined, this algorithm combines the global optimization search capability of ACO and the ability that PLS effectively solves Problems of Multiple Synteny between variable, can the predictive variable of PLS modeling effectively be screened.At first adopt ant group algorithm that a large amount of molecular structural parameter are carried out whole optimizing, subsequently take several groups of more excellent structural parameters finding as initial value, adopt the offset minimum binary method that variable is done further " optimization ", with training precision and the forecast precision of raising model.Its basic process is as follows:
L, initialization
Initialization information initial value τ
0, ant is counted m, the parameters such as pheromones increment Q;
2, the ideal adaptation degree is estimated
Each individuality is corresponding to a feasible solution in the research system, by the value of calculating fitness function the individuality in the colony estimated, and determines the direction of search, finally reaches to comprise or near the state of optimum solution in the search volume.The formula of fitness function of the present invention (F) is as follows:
In the formula, n is the quantity of training set sample, Y
iBe the desired value of training set sample,
Be the mean value of training set sample object value,
Be the predicted value of "current" model to the training set sample.The fitness function value is larger, and then the match degree of correlation of model is higher, i.e. the probability that current parameter (descriptor) combination finally is chosen as the regression variable in the model is higher.3, while (not reaching iterations)
4, output optimum solution
In the ant group algorithm, α, β, the parameters such as ρ have a great impact algorithm performance.Wherein, the size of α value shows the valued degree of the quantity of information of staying on each node, and the α value is larger, and the possibility of the route of process was larger before ant was selected, and makes search sink into too early local minimal solution but cross conference; The size of β shows the valued degree of heuristic information, and the β value is larger, and ant selects the possibility in the city close to it also larger; ρ represents the retention rate of pheromones, and is incorrect if its value obtains, and the result who obtains understands relatively poor.
Simultaneously, successful structure-activity relationship model also depends on the validity of the statistical modeling method that adopts.The present invention selects to have the support vector machine method of strong generalization ability.Support vector machine method has powerful nonlinear fitting ability, can overcome the defective that the conventional linear homing method is not suitable for the complex nonlinear system; Simultaneously, it is based on structural risk minimization, pursue minimizing of fiducial range value, but not the minimizing of training error, can reach globally optimal solution in theory, therefore can overcome traditional neural net method and be easy to produce the shortcomings such as " crossing training ", " over-fitting ", be specially adapted to the structure activity study system of small sample, and have better Generalization Capability; In addition, in case behind the setting parameter, the solution of support vector machine also has uniqueness and repeatability, this point obviously is better than artificial neural network especially.Therefore, adopt support vector machine method that the inherent quantitative relationship between molecular structure and the thermal hazard is carried out statistical modeling among the present invention, set up corresponding forecast model.
The algorithm steps of support vector machine is as follows:
Suppose given training sample set { (x
i, y
i), i=1 ... n}, wherein x
i∈ R
nThe input value of i learning sample, y
i∈ R is corresponding desired value.For linear regression, use linear function
f(x)＝(w·x)+b （3）
Estimate.Smooth for assurance formula (3) 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 so minimum w just changes the minimum model complexity into, and it is equivalent to
Change into corresponding quadratic programming problem namely:
(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, 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〉0 be used for the smooth degree of balance regression function f (x) and the deviation number greater than the ε sample point.Formula (5) is based on following ε-insensitive loss function and draws, this function | ξ |
εBe expressed as follows:
When sample number is less, find the solution top support vector machine and generally adopt duality theory, 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 primal-dual optimization problem
For non-linear regression, the solution thinking of support vector machine is by a Nonlinear Mapping
Sample is mapped in the feature space of a higher-dimension and solves with the linear method of routine.Suppose sample X nonlinear function
Be mapped to higher dimensional space, then nonlinear regression problem is converted into:
Support vector machine is mapped to high-dimensional feature space by Kernel Function Transformation with sample, and kernel function K (x, x ') satisfies K (x, x ')=＜φ (x), φ (x ') 〉.Therefore formula (8) is rewritten as
The introducing of kernel function so that function find the solution and walk around feature space and directly carry out in the input space, thereby avoided the calculating Nonlinear Mapping
Support vector machine kernel function commonly used mainly contains 4 types of linear kernel, polynomial kernel, radial basis nuclear and sigmoid nuclears etc. at present.The present invention selects radial basis nuclear K (x, x
i)=exp (γ || x-x
i||
2) as kernel function.
The correlation parameter that determines the model construction of SVM 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 of the present invention is selected radial basis nuclear K (x, x
i)=exp (γ || x-x
i||
2), because it has higher learning efficiency and learning rate; Other parameter is determined by " grid search " method; The parameter search scope is as follows: 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; " stay 1/10 method " validation-cross refers to screen out 1/10 sample of training sample sum at every turn from training sample, use remaining sample, predict the character of the sample that screens out, the root-mean-square error (RMSE) that obtains like this a validation-cross is come the quality of evaluation model performance, and its computing formula is:
Wherein, y
I, predBe the predicted value of sample i, y
I, obsDesired value for sample i; By search, choose corresponding that group parameter of minimum RMSE of " staying 1/10 method " cross-verification as the optimized parameter of support vector machine; The optimized parameter that application searches goes out is set up corresponding forecast model as the input parameter of support vector machine.
In addition, the checking of model also is very important step among the present invention with revising.Only have model sane and that have the Height Prediction ability to use.The present invention estimates the predictive ability of the robustness of building forecast model and model and verifies.The robustness of model adopts " leaving-one method " (Leave-one-out, LOO) the validation-cross method is tested, " leaving-one method " validation-cross refers to screen out a compound at every turn from training set, with remaining compound modeling, predict the character that screens out compound, obtain like this multiple correlation coefficient r of a validation-cross
2(be Q
2) come the robustness of evaluation model; The checking of model prediction ability is then undertaken by the mode of using the model of setting up to calculate the thermal hazard (being the external certificate method) of forecast set sample.Subsequently, result of calculation according to validation-cross and external certificate, the predicted value of comparative sample and desired value, when if the mean deviation of predicted value and desired value surpasses acceptable scope (for example average relative error surpasses 10%), then reject the sample (for example Relative Error surpasses the sample of 2 times of average relative errors) that predicated error exceeds standard, return (3), again model and forecast, until the mean deviation of predicted value and desired value within the acceptable range (for example average relative error is less than 10%), thereby determine forecast model.The model that satisfies this requirement can come into operation.
Beneficial effect of the present invention:
Prediction effect of the present invention is good, applied widely, easy to use, only needs the molecular structure of autoreaction chemical substance just can realize its thermal hazard, begins the prediction of temperature, SADT and maximum temperature rise rate time etc. such as reaction heat, themopositive reaction.Utilize method of the present invention can be only just can dope accurately and rapidly its thermal hazard according to the molecular structure of autoreaction chemical substance, for weighing the autoreaction chemical substance complexity of the danger such as hot spontaneous combustion, thermal decomposition, thermal explosion and the basic data that hazard level provides necessity occur in the processes such as production, processing, storage and transportation, for bioactive molecule designs, process engineering and the work such as flowsheeting, safety assessment provide reference.The present invention does not need to use the autoreaction chemical substance thermal hazard sensing equipment of complexity, costliness, great many of experiments be can remove from and the inconvenience bring and loss economically measured, be specially adapted to that those experiments are difficult to carry out or the enterprises and institutions that do not possess experiment condition use, its economy is very considerable.
Description of drawings
Fig. 1 is ant colony optimization algorithm concrete operation step synoptic diagram.
Fig. 2 is the principle schematic that support vector machine is used for regression problem.
Fig. 3 is the comparison diagram of forecast model gained autoreaction chemical substance SADT predicted value and desired value.
Embodiment
The present invention is further illustrated below in conjunction with drawings and Examples.
As shown in Figure 1, 2, 3.
A kind of method of definite autoreaction chemical substance thermal hazard only needs molecular structure just can predict its thermal hazard, and concrete steps can be subdivided into following seven steps:
(1) collection of autoreaction chemical substance experiment sample and thermal hazard experimental data thereof:
Select a series of autoreaction chemical substances according to statistical standard and construction standard, consist of the experiment sample collection; The condition that the autoreaction chemical substance is selected is statistical randomness, structural representativeness and comprehensive, and the availability of data; For the autoreaction chemical substance in the sample set, collect the thermal hazard data of paying close attention to; The approach of Data Collection mainly contains three kinds: measuring, various authoritative character data storehouse and handbook; The principle of data selection is must reliable and standardization.
(2) description of molecular structure:
Carry out the input of autoreaction chemical substance molecular structure in the sample set by Chemical Software Hyperchem, use molecule simulation method and make up correct two dimension or three-dimensional molecular structure, adopt the methods such as molecular mechanics (MM+ optimization), quantum chemistry semi-empirical approach (AM1 optimization) to obtain optimized conformation.
Adopt topology method, quantum mechanics method etc. to calculate the structural parameters that are used for the various structural informations of reflection molecule, comprise topological parameter, composition parameter, geometric parameter, electrical parameter, electrical topological parameter and Quantum chemical parameters etc., to obtain molecular structures information, realize the parametric description of molecular structure.
(3) division of sample set:
Be training set and two parts of forecast set with the sample set random division, wherein training set is used for setting up forecast model, and forecast set does not participate in modeling, is used for institute's established model is estimated and verified.
(4) choosing of feature structure:
For the training set sample, ant group-offset minimum binary (ACO-PLS) algorithm that use has powerful global search function carries out the characteristic variable screening, from the numerous molecular structural parameter that calculated, find out with the most closely-related stack features structural parameters of target thermal hazard as the molecule descriptor of describing autoreaction chemical substance feature structure, the i.e. input parameter of modeling.
The selected correlation parameter of ant group algorithm is as shown in table 2:
Table 2.ACO-PLS method parameter
(5) foundation of forecast model:
For the training set sample, with the selected molecule descriptor of step (4) as input variable, corresponding thermal hazard is as output variable, use the support vector machine chemometrics method, internal relation between molecular structure and the target thermal hazard is carried out statistical modeling, the quantitative function relation that is existed is between the two set up corresponding forecast model.
The correlation parameter that determines the support vector machine simulated performance mainly comprises: the size of ε in the parameter of kernel function, kernel function, penalty coefficient C and ε-insensitive loss function.Among the present invention, kernel function is selected radial basis nuclear K (x, x
i)=exp (γ || x-x
i||
2), because it has higher learning efficiency and learning rate.Other parameter is determined by " grid search " method.
The parameter search scope is as follows: penalty coefficient C---0-1024; The width gamma of kernel function---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." stay 1/10 method " validation-cross refers to screen out 1/10 compound of number of training at every turn from training set, with remaining compound modeling, predict the character of the compound that screens out, the root-mean-square error (RMSE) that obtains like this a validation-cross is come the quality of evaluation model performance, and its computing formula is:
Wherein, y
I, predBe the predicted value of sample i, y
I, obsDesired value for sample i.By search, choose corresponding that group parameter of minimum RMSE of " staying 1/10 method " cross-verification as the optimized parameter of support vector machine.The optimized parameter that application searches goes out is set up corresponding forecast model as the input parameter of support vector machine.
(6) checking of forecast model, correction and definite:
The predictive ability of the robustness of building forecast model and model is estimated and verified.The robustness of model adopts " leaving-one method " (Leave-one-out, LOO) the validation-cross method is tested, " leaving-one method " validation-cross refers to screen out a compound at every turn from training set, with remaining compound modeling, predict the character that screens out compound, obtain like this multiple correlation coefficient r of a validation-cross
2(be Q
2) come the robustness of evaluation model, its computing formula is:
Wherein, y
i,
With
The mean value that represents respectively desired value, predicted value and the desired value of training set sample.The checking of model prediction ability is then undertaken by the mode of using the model of setting up to calculate the thermal hazard (being the external certificate method) of forecast set sample.Subsequently, result of calculation according to validation-cross and external certificate, the predicted value of comparative sample and desired value, when if the mean deviation of predicted value and desired value surpasses acceptable scope (for example average relative error surpasses 10%), then reject the sample (for example Relative Error surpasses the sample of 2 times of average relative errors) that predicated error exceeds standard, return (3), again model and forecast, until the mean deviation of predicted value and desired value within the acceptable range (for example average relative error is less than 10%), thereby determine forecast model.Only have the model that satisfies this requirement to come into operation.
(7) application of forecast model:
Molecular structure for unknown autoreaction chemical substance, select corresponding molecule descriptor according to step (4), calculate corresponding numerical value, the determined corresponding forecast model of its substitution step (6) is calculated, can obtain the corresponding thermal hazard data of this autoreaction chemical substance.
The below is predicted as example with SADT (self-accelerating decomposition temperature, SADT), and the present invention will be further described.
Sample set comprises 41 kinds of autoreaction chemical substances altogether, and its SADT data all derive from document.Compound in this sample set all belongs to CHO class organic compound, for set up stalwartness, effectively forecast model is laid a good foundation.Subsequently, sample set is divided, selected at random 33 kinds of compounds as training set, be used for Variable Selection and set up forecast model; Select 8 kinds of compounds of residue as outside forecast set, be used for the degree of reliability and the predictive ability of institute's established model are estimated checking.
Subsequently, according to the molecular structure of autoreaction chemical substance in the sample set, to its topology, electrically, the structural parameters such as quantum chemistry calculate.On this basis, use the ant group algorithm of setting up a large amount of structural parameters that calculate are carried out the optimization screening of feature structure, obtain one group of optimum structural parameters as the input variable of modeling.These and the closely-related structural parameters of autoreaction chemical substance SADT are listed in table 3.
Table 3. ant group algorithm filter out with the closely-related molecule descriptor of SADT
Then, using support vector machine method simulates the inherent quantitative relationship between autoreaction chemical substance SADT and said structure parameter.Radial basis nuclear K (x, x are adopted in the support vector machine simulation
i)=exp (γ || x-x
i||
2) as kernel function, the sample data linear mapping to [1,1] interval, is carried out normalized; For the training set sample, adopt the grid search method to determine the optimized parameter of support vector machine, the direction of search of grid search is the lowest mean square root error (RMSE) of " staying 1/10 method " cross-verification; By search, determine that the optimized parameter that model is chosen is: penalty coefficient C=1024, the ε in ε-insensitive loss function=0.125, the width gamma of kernel function=0.03125, corresponding support vector number is 21.For the training set sample, use the optimized parameter of determining, set up corresponding forecast model.Use this model and respectively training set and forecast set sample predicted, gained predicted value and desired value relatively see accompanying drawing 3.
Listed the integral performance parameter of the forecast model of building in the table 4.For training set, forecast model has shown stronger data fitting ability, and multiple correlation coefficient is 0.975; For outside forecast set, the multiple correlation coefficient of model " leaving-one method " validation-cross is 0.971, and this illustrates that this forecast model is successfully statistically.It can also be seen that from table 2 forecast model gained estimated performance and match performance are comparatively approaching, this illustrates that this model also has stronger generalization ability and namely predicts stability.In addition, the prediction average relative error of this forecast model is 9.83, is lower than 10%, within acceptable scope.Therefore, the present invention is based on the autoreaction chemical substance SADT forecast model that molecular structure develops is successfully, can be effectively applied in the prediction of unknown autoreaction chemical substance SADT data.
The Specifeca tion speeification of table 4. established model
At present, the method for definite SADT is based on thermal analysis system more both at home and abroad.Thermal analysis system determines that the SADT of energetic material carries out Exact Solution as the basis take heat analysis method to the thermodynamics and kinetics parameter that contains the energy substance decomposition.Theoretically, thermal analysis system contains and can aspect the material thermal hazard certain advantage be arranged in evaluation, but also has some obvious defectives.Because when the research thermal analysis system is measured the SADT value of material, the levels of precision of the material autoreaction process dsc data of its acquisition directly affects the degree of accuracy of the SADT that calculates through mathematics or physical model (Semenov model, Frank-Kamenetskii model and Thomas model); Simultaneously, the approximation theory of existing model is calculated the result whether simulation rationally also directly has influence on calculating.This shows, use thermal analysis system and determine that the SADT of autoreaction chemical substance drops into practical application as a kind of method a sizable segment distance is also arranged.Compare with thermal analysis system, the present invention only just can realize the prediction of autoreaction chemical substance SADT according to molecular structure, Simple fast, and forecasting accuracy is high, applied widely, has shown that the method has good effect aspect the prediction of autoreaction chemical substance thermal hazard.
It is as follows to the concrete grammar that the SADT of unknown autoreaction chemical substance predicts to use SADT forecast model that the present invention sets up:
(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=0.03125), only need the as required molecular structure of the unknown compound of prediction during prediction, calculate the concrete numerical value of this compound 6 molecule descriptors corresponding with table 3, then as the input variable of existing SVM model, can obtain the SADT numerical value of this compound after calculating.
The inventor in conjunction with ant group algorithm and support vector machine method, has been developed the new method of a cover according to Molecular structure prediction autoreaction chemical substance thermal hazard from the molecular structure angle among the present invention.According to the principle of structures shape character, from resolving the molecular structure angle, extraction can be described the structural parameters of molecular structure information comprehensively; Offset minimum binary and ant group algorithm are combined, designed corresponding characteristic variable screening sequence, from the structural parameters that obtain according to molecular structure in a large number, filter out for different thermal hazards respectively and its most closely-related one group of structural parameters as the molecule descriptor that characterizes the characterization of molecules structure; Take the above-mentioned characteristic parameter of known autoreaction chemical substance as input variable, take the thermal hazard of correspondence as output variable, utilize powerful machine learning algorithm support vector machine method, non-linear, the uncertain and complicated inherent quantitative relationship that exists between existing thermal hazard experimental data and its molecular structure is effectively trained and forecast, thereby set up stable, efficient SVM prediction model.Utilize the forecast model of setting up that the thermal hazard of other unknown autoreaction chemical substances is predicted, have precision of prediction height, advantage fast and easily, do not need complicated theory to derive, realized the function according to molecular structure fast prediction thermal hazard, effectively solved the problem of autoreaction chemical substance thermal hazard experimental data famine, therefore industrial process design and fire-proof and explosion-proof the grade in the work good application prospect has been arranged.In addition, 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 using value for the enterprises and institutions that do not possess combustion explosion measuring ability, economic benefit is fairly obvious.
The part that the present invention does not relate to all prior art that maybe can adopt same as the prior art is realized.
Claims (3)
1. the method for a definite autoreaction chemical substance thermal hazard, concrete steps are as follows:
(1) collection of autoreaction chemical substance experiment sample and thermal hazard experimental data thereof:
Select a series of autoreaction chemical substances according to statistical standard and construction standard, consist of the experiment sample collection; The condition that the autoreaction chemical substance is selected is statistical randomness, structural representativeness and comprehensive, and the availability of data; For the autoreaction chemical substance in the sample set, collect the thermal hazard data of paying close attention to; The approach of Data Collection mainly contains three kinds: measuring, various authoritative character data storehouse and handbook;
(2) description of molecular structure:
According to the molecular structure of autoreaction chemical substance in the sample set, calculate all kinds of structural parameters that obtain for the reflection molecular structure information, realize the parametric description of molecular structure information;
(3) division of sample set:
Be training set and two parts of forecast set with the sample set random division, wherein training set is used for setting up forecast model, and forecast set does not participate in modeling, is used for institute's established model is estimated and verified;
(4) choosing of feature structure:
For the training set sample, use has ant group-offset minimum binary (ACO-PLS) algorithm powerful global search function, that ant colony optimization algorithm is combined with the offset minimum binary method and carries out the characteristic variable screening, from a large amount of structural parameters that step (2) calculates, filter out the most closely-related with the target thermal hazard, comprise enrich structural information one group of parameter as the molecule descriptor of describing autoreaction chemical substance feature structure;
(5) foundation of forecast model:
For the training set sample, with the selected molecule descriptor of step (4) as input variable, corresponding thermal hazard is as output variable, use the support vector machine chemometrics method, internal relation between molecular structure and the target thermal hazard is carried out statistical modeling, the quantitative function relation that is existed is between the two set up corresponding forecast model;
(6) checking of forecast model, correction and definite:
Adopt cross verification to verify the robustness of the forecast model of building, adopt external certificate method (namely calculating the thermal hazard of forecast set sample with institute's established model) to verify the extrapolability of forecast model; According to predicting the outcome of validation-cross and external certificate, the predicted value of comparative sample and desired value, when if the mean deviation of predicted value and desired value surpasses acceptable scope, reject the sample that predicated error exceeds standard, return (3), again model and forecast until the mean deviation of predicted value and desired value within the acceptable range, thereby is determined forecast model;
(7) application of forecast model:
Molecular structure for unknown autoreaction chemical substance, the molecule descriptor that screening is determined according to step (4), calculate corresponding numerical value, the determined corresponding forecast model of its substitution step (6) is calculated, namely obtain the corresponding thermal hazard data of this autoreaction chemical substance.
2. method according to claim 1 is characterized in that all kinds of structural parameters described in the step (2) are topological parameter, composition parameter, geometric parameter, electrical parameter, electrical topological parameter and the Quantum chemical parameters of molecule.
3. method according to claim 1 is characterized in that described thermal hazard is that reaction heat, themopositive reaction begin temperature, SADT and maximum temperature rise rate time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210441367.8A CN102980972B (en) | 2012-11-06 | 2012-11-06 | Method for determining hot dangerousness of self-reactive chemical substance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210441367.8A CN102980972B (en) | 2012-11-06 | 2012-11-06 | Method for determining hot dangerousness of self-reactive chemical substance |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102980972A true CN102980972A (en) | 2013-03-20 |
| CN102980972B CN102980972B (en) | 2015-06-03 |
Family
ID=47855189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210441367.8A Active CN102980972B (en) | 2012-11-06 | 2012-11-06 | Method for determining hot dangerousness of self-reactive chemical substance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102980972B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344600A (en) * | 2013-06-28 | 2013-10-09 | 中国农业大学 | Characteristic wavelength selecting method for near infrared spectrum in ant colony optimization algorithm |
| CN105653795A (en) * | 2015-12-28 | 2016-06-08 | 中山大学惠州研究院 | Detection method of self-accelerating decomposition temperature of organic peroxide |
| CN106444380A (en) * | 2016-10-14 | 2017-02-22 | 中国科学院光电技术研究所 | Transfer function identification method centering on nonlinear least square method and combining with chicken swarm algorithm for fast steering mirror control system |
| CN108763855A (en) * | 2018-05-18 | 2018-11-06 | 南京工业大学 | A kind of method of determining biodiesel combustibility |
| CN109949874A (en) * | 2019-04-16 | 2019-06-28 | 南京工业大学 | A kind of risk stratification method of fine chemistry industry production process security evaluation |
| CN110414868A (en) * | 2019-08-13 | 2019-11-05 | 南京工业大学 | A method of assessment chemical engineering process thermal runaway danger level |
| CN111781230A (en) * | 2019-04-04 | 2020-10-16 | 应急管理部化学品登记中心 | Method for grading danger of thermal decomposition of nitro compound fertilizer |
| CN111781280A (en) * | 2019-04-04 | 2020-10-16 | 应急管理部化学品登记中心 | Method for predicting thermal stability of organic compound |
| CN111798935A (en) * | 2019-04-09 | 2020-10-20 | 南京药石科技股份有限公司 | Universal compound structure-property correlation prediction method based on neural network |
| CN111816266A (en) * | 2020-07-10 | 2020-10-23 | 北京迈高材云科技有限公司 | Method and system for automatically constructing material quantitative structural property model |
| CN111816260A (en) * | 2020-07-09 | 2020-10-23 | 金陵科技学院 | Method for predicting chemical substance thermal risk in industrial scale |
| CN112002380A (en) * | 2020-08-14 | 2020-11-27 | 四川大学 | Self-adaptive design method of high-heat-generation energetic material based on machine learning |
| CN116959613A (en) * | 2023-09-19 | 2023-10-27 | 烟台国工智能科技有限公司 | Compound inverse synthesis method and device based on quantum mechanical descriptor information |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339181A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound explosive characteristic prediction method based on genetic algorithm |
| CN101339180A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound explosive characteristic prediction method based on support vector machine |
| CN102608284A (en) * | 2011-12-23 | 2012-07-25 | 南京工业大学 | Method for determining explosion limit of multicomponent mixed gas |
-
2012
- 2012-11-06 CN CN201210441367.8A patent/CN102980972B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339181A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound explosive characteristic prediction method based on genetic algorithm |
| CN101339180A (en) * | 2008-08-14 | 2009-01-07 | 南京工业大学 | Organic compound explosive characteristic prediction method based on support vector machine |
| CN102608284A (en) * | 2011-12-23 | 2012-07-25 | 南京工业大学 | Method for determining explosion limit of multicomponent mixed gas |
Non-Patent Citations (1)
| Title |
|---|
| 曹洪印等: "基于蚁群算法的烃类燃烧热QSPR研究", 《工业安全与环保》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344600B (en) * | 2013-06-28 | 2015-06-03 | 中国农业大学 | Characteristic wavelength selecting method for near infrared spectrum in ant colony optimization algorithm |
| CN103344600A (en) * | 2013-06-28 | 2013-10-09 | 中国农业大学 | Characteristic wavelength selecting method for near infrared spectrum in ant colony optimization algorithm |
| CN105653795A (en) * | 2015-12-28 | 2016-06-08 | 中山大学惠州研究院 | Detection method of self-accelerating decomposition temperature of organic peroxide |
| CN106444380A (en) * | 2016-10-14 | 2017-02-22 | 中国科学院光电技术研究所 | Transfer function identification method centering on nonlinear least square method and combining with chicken swarm algorithm for fast steering mirror control system |
| CN108763855A (en) * | 2018-05-18 | 2018-11-06 | 南京工业大学 | A kind of method of determining biodiesel combustibility |
| CN108763855B (en) * | 2018-05-18 | 2022-04-26 | 南京工业大学 | Method for determining combustion performance of biodiesel |
| CN111781230A (en) * | 2019-04-04 | 2020-10-16 | 应急管理部化学品登记中心 | Method for grading danger of thermal decomposition of nitro compound fertilizer |
| CN111781280A (en) * | 2019-04-04 | 2020-10-16 | 应急管理部化学品登记中心 | Method for predicting thermal stability of organic compound |
| CN111798935A (en) * | 2019-04-09 | 2020-10-20 | 南京药石科技股份有限公司 | Universal compound structure-property correlation prediction method based on neural network |
| CN109949874B (en) * | 2019-04-16 | 2022-02-15 | 南京工业大学 | Risk grading method for safety assessment in fine chemical production process |
| CN109949874A (en) * | 2019-04-16 | 2019-06-28 | 南京工业大学 | A kind of risk stratification method of fine chemistry industry production process security evaluation |
| CN110414868B (en) * | 2019-08-13 | 2022-12-13 | 南京工业大学 | Method for evaluating thermal runaway risk degree in chemical process |
| CN110414868A (en) * | 2019-08-13 | 2019-11-05 | 南京工业大学 | A method of assessment chemical engineering process thermal runaway danger level |
| CN111816260A (en) * | 2020-07-09 | 2020-10-23 | 金陵科技学院 | Method for predicting chemical substance thermal risk in industrial scale |
| CN111816266A (en) * | 2020-07-10 | 2020-10-23 | 北京迈高材云科技有限公司 | Method and system for automatically constructing material quantitative structural property model |
| CN111816266B (en) * | 2020-07-10 | 2024-01-30 | 北京迈高材云科技有限公司 | Method and system for automatically constructing quantitative structural property model of material |
| CN112002380A (en) * | 2020-08-14 | 2020-11-27 | 四川大学 | Self-adaptive design method of high-heat-generation energetic material based on machine learning |
| CN112002380B (en) * | 2020-08-14 | 2023-05-05 | 四川大学 | Self-adaptive design method of high-generation heat energetic material based on machine learning |
| CN116959613A (en) * | 2023-09-19 | 2023-10-27 | 烟台国工智能科技有限公司 | Compound inverse synthesis method and device based on quantum mechanical descriptor information |
| CN116959613B (en) * | 2023-09-19 | 2023-12-12 | 烟台国工智能科技有限公司 | Compound inverse synthesis method and device based on quantum mechanical descriptor information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102980972B (en) | 2015-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102980972B (en) | Method for determining hot dangerousness of self-reactive chemical substance | |
| CN102608285B (en) | Prediction method for explosion characteristics of organic mixture based on support vector machine | |
| CN101339180B (en) | Organic compound explosive characteristic prediction method based on support vector machine | |
| CN101339181B (en) | Organic compound explosive characteristic prediction method based on genetic algorithm | |
| Bagh et al. | Electrical conductivity of ammonium and phosphonium based deep eutectic solvents: Measurements and artificial intelligence-based prediction | |
| CN102608284B (en) | Method for determining explosion limit of multicomponent mixed gas | |
| JP5946045B2 (en) | Compound design apparatus, compound design method, and computer program | |
| Peresan et al. | Operational earthquake forecast/prediction | |
| Gharagheizi | Prediction of upper flammability limit percent of pure compounds from their molecular structures | |
| CN103914622A (en) | Quick chemical leakage predicating and warning emergency response decision-making method | |
| CN103905560B (en) | Oil tank area flame spread prediction and analysis method based on complex networks | |
| Venkatraman et al. | Computer-aided molecular design of imidazole-based absorbents for CO2 capture | |
| Ren | Life cycle aggregated sustainability index for the prioritization of industrial systems under data uncertainties | |
| Xu et al. | Sustainability assessment framework for chemical processes selection under uncertainties: a vector-based algorithm coupled with multicriteria decision-making approaches | |
| Vaidya et al. | Quantitative structure-activity relationships: a novel approach of drug design and discovery | |
| Hosouli et al. | A Multi-Criteria decision making (MCDM) methodology for high temperature thermochemical storage material selection using graph theory and matrix approach | |
| CN102736570B (en) | Gas phase polyethylene device quality index and operational constraint on-line estimation system and method thereof | |
| CN103345544A (en) | Predicting organic chemical biodegradability according to logistic regression method | |
| Kovács et al. | A novel active parameter selection strategy for the efficient optimization of combustion mechanisms | |
| Gupta et al. | Wind ramp event prediction with parallelized gradient boosted regression trees | |
| CN103971030B (en) | Method for predicting affinity of interaction between biomarker p53 and organic phosphate fire retardant | |
| Jovanovic et al. | APPLICATION OF MCDM METHODS IN EVALUATION OF ENVIRONMENTAL IMPACTS. | |
| Ellison et al. | Unimolecular rearrangement of trans-FONO to FNO2. A possible model system for atmospheric nitrate formation | |
| CN105844085A (en) | Detection method of initial decomposition temperature of organic peroxides | |
| Marin | Kinetics and transport phenomena in the chemical decomposition of copper oxychloride in the thermochemical Cu-CI Cycle |
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 |