دانلود ترجمه مقاله کاربرد روش روابط ویژگی ساختار کمی در پیش بینی دمای احتراق خودکار – مجله الزویر

• فهرست مطالب:

 چکیده
۱ مقدمه
۲ روش کار
۳ نتایج و بحث
۴ نتیجه گیری

• بخشی از ترجمه:

در اين كار يك مدل QSPR با چهار توصيفگر براي پيش بيني AIT از مواد قابل اشتعال ارائه شده است. مدل پيشنهادي در عملكرد مقداري براي R=0.9 و خطاي مطلق ۳۶K را داشته كه در مقايسه با ساير تحقيقات موجود اين مدل از بزرگترين مجموعه داده ساخته شده است و با حداقل توصيف بوده و عملكرد قابل قبولي در مقايسه با دقت تجربي داشته است. به عنوان رويكرد QSPR‌فقط نياز به اطلاعات از ساختار مولكولي براي پيش بيني خواص مورد نظر است و اين مدل يك راه براي پيش بيني AIT از تركيبات در حال توسعه است كه دقتي براي ارزيابي ويژگي هاي اشتعال پذيري معقول را بوجود مي اورد.

• بخشی از مقاله انگلیسی:

Introduction As the technology progresses, more and more flammable materials are operated in process industries. Thus, explosionproof electric equipments are required in these industries to safely handle flammable materials. The allowable maximum surface temperature of these electric equipments is one of the important characteristics to classify the such equipments and the required specification on these equipments depends on the auto-ignition temperature (AIT) of the flammable materials being operated. For example, article 500.8 of NFPA 70 (also known as the National Electric Code) provides that “Class I equipment shall not have any exposed surface that operates at a temperature in excess of the ignition temperature of the specific gas or vapors.”[۱] AIT is defined as the lowest temperature at which the substance will produce hot-flame ignition in air at atmospheric pressure without the aid of an external ignition source such as spark or flame [2]. Obviously, the ability of a substance to spontaneously ignite is important to people who handle, transport, and store these flammable materials. However, although the AIT data are indispensable to safely handle and operate flammable materials, the AIT data reported in different data compilations are very much diverse. The difference between different data compilations might be up to more than 300 K for many flammable liquids. Such diversity is attributed to many experimental factors and has been discussed in the literature [3]. Besides this diversity, determining the AIT of a chemical by experimental approach is very laborious and is not always feasible[4]. In this regard, the ability to estimate the AIT of flammable materials by mathematical model will be a cost-efficient and critical aid to this discipline. One of the important approaches to predict the AIT of a flammable material is the quantitative structure property relationship (QSPR) approach [5-12]. In this category, many molecule-based parameters, which are often called as “molecular descriptors”, are directly calculated from the molecular structure of a compound, and then the relationship between the target property and these molecular descriptors are developed. As this approach does not require any existing properties of a compound, the developed model could be can easily apply to the case of predicting the FP of a novel substance. Thus, this approach has been adopted in many research to predict the AIT of flammable materials. Suzuki et al. had proposed a five-descriptor multiple linear regression (MLR) model for predicting the AIT of hydrocarbons. This model was built from a data set of fifty hydrocarbons, and the R value was reported to be 0.941 [5]. Suzuki extended aforementioned model to be a 6-descriptor model by examining 21 descriptors and the data set used to build the model also expanded to 250 hydrocarbons. The R value of the new model was reported to be 0.952 [6]. Egolf and Jurs had pointed the fitting performance is very limited if all hydrocarbons are considered to be a group, so they divided hydrocarbons into four categories: (1) low-temperature hydrocarbons; (2) high-temperature hydrocarbons; (3) alcohols; and (4) esters. These four models are a 8-descriptor model builds from 58 compounds, a 5-descriptor model builds from 46 compounds, a 4-descriptor model builds from 28 compounds and a 4-descriptor model builds from 25 compounds. Their R values are reported to be 0.975, 0.939, 0.970 and 0.963, respectively [7]. Mitchell and Jurs had divided the organic compounds into more categories to enhance the predictive performance of their model. They proposed AIT models for low-temperature hydrocarbons, hightemperature hydrocarbons, nitrogen compound, oxygen/sulfur compounds and alcohol/ester compounds, and artificial neural network models instead of the multiple linear regression models are adopted in this work [8]. Kim et al. built a 9- descriptor model from a data set of 157 organic compounds, and the R value was found to be of 0.959 [9]. Pan et al. (2008) explored the performance of the supported vector machine (SVM) approach in predicting the AIT of the flammable materials [10]. Pan et al. (2009) had proposed and compare a SVM model with the MLR model from a data set of 356 compounds [11].