Форум

Срочно требуется помощь англофона!

Mukhin: Коллеги, я отлично понимаю, что занимаюсь злостным офтопогонством. К сожалению, в силу ряда обстоятельств, обратится больше не к кому и некуда. Завтра супруга должна отправить тезисы своего выступления на конференции, в т.ч. - на английском языке. прричём именно этот текст будет рассматриваться в первую очередь. Для неё (а значит, и для меня) очень важно, что бы этот текст не содержал каких-либо нелепых ошибок. Мне известно, что некотрые из участников форума свободно владеют аглицкой мовой. Поэтому я был бы очень благодарен, если бы кто-нибудь из таковых любезно согласились посмотреть нижеследующий текст. И, соответственно, сообщить, если там где-то неверно артикль поставлен, падежи не спрягаютси т.п. Заранее благодарен ответившему и заранее согласен с любой карой, которою наложит на меня модератор. The Application of Statistical and Neural networks Algorithms for the Classifica-tion of MFL Signals Magnetic flux leakage (MFL) methods are extensively used for in-line inspection of pipe-lines. As it moves along the pipeline under pressure, intelligent pig performs magnetization of the wall and measures of magnetic flux leakage. The topographic features of this magnetic field provide the information about characteristics and location of any specific pipeline construction elements, welds and defects. The diagnostic data are two-dimensional magnetic signals, which serve to describe magnetic field topography near the pipeline inside surface. To make the testing procedure more productive and to make its results more reliable, in-terpretation of the magnetic signals should be performed automatically, with using intellectual software designed to segmentation of the data into irrelevant parts of the pipe and regions of in-terest, classification of each segment of a signal and estimation of geometric dimensions of de-fects found. Signal classification is an important stage of the data interpretation algorithm, be-cause result of such classification is one of the factors for the defects danger estimation. This paper describes mutual comparison of three classification algorithms applicable to MFL signals – i.e., statistical hierarchic algorithm and neural network classification algorithms realized by multiplayer perceptron and Hopfield network. Statistical algorithm is built hierarchically, where each step is accompanied with check-ing as to specific class of the analyzed pattern. The hierarchical principle underlying algorithm allows to make a detailed study of the pattern and to apply a wide range of data preprocessing procedures. Fully predictable in its nature, the statistical classification algorithm has the disad-vantage of being too formal in approach. Designed to simulate human cerebral functions, neural network classification algorithms have the benefit of enhanced capabilities to recognize patterns; its disadvantages are difficulties lying in the problem of local minima by the training of percep-tron and selection of the proper Hopfield network configuration, as well as it is not always possi-ble to make reliable forecasts of pattern classification results using neural network. These three algorithms have been in use for classification of the most typical pipeline de-fects and pipeline construction elements, such as supports, pressure-gauge pipes, pipeline valves, patches as well as corrosions, risks, cracks, dents and other defects.

Ответов - 6 новых [см. все]

Читатель: Mukhin пишет: цитатападежи не спрягаютси т.п. В английском только два падежа и они не спрягаются! Исправил Application of Statistical and Neural networks Algorithms for Classifica-tion of MFL Signals Magnetic flux leakage (MFL) methods are extensively used for in-line inspection of pipelines. As it moves along a pipeline under pressure, intelligent pig performs magnetization of the wall and measures magnetic flux leakage. Topographic features of this magnetic field provide information about characteristics and location of any specific pipeline construction elements, welds and defects. The diagnostic data are two-dimensional magnetic signals, which serve to describe magnetic field topography near the pipeline inside surface. To make the testing procedure more productive and to make its results more reliable, interpretation of magnetic signals should be performed automatically, with use of intellectual software designed for segmentation of the data into irrelevant parts of the pipe and regions of interest, classification of each segment of a signal and estimation of geometric dimensions of the defects found. Signal classification is an important stage of the data interpretation algorithm, because result of such classification is one of the factors for the defects danger estimation. This paper describes mutual comparison of three classification algorithms applicable to MFL signals – i.e., statistical hierarchic algorithm and neural network classification algorithms realized by multiplayer perceptron and Hopfield network. Statistical algorithm is built hierarchically, where each step is accompanied with checking for specific class of the analyzed pattern. The hierarchical principle underlying algorithm allows to make detailed study of the pattern and to apply a wide range of data preprocessing procedures. Fully predictable in its nature, the statistical classification algorithm has a disadvantage of being too formal in approach. Designed to simulate human cerebral functions, neural network classification algorithms have a benefit of enhanced capabilities to recognize patterns; its disadvantages are difficulties lying in the problem of local minima by the training of perceptron and selection of the proper Hopfield network configuration, as well as the fact that it is not always possible to make reliable forecasts of pattern classification results using neural network. These three algorithms have been in use for classification of the most typical pipeline defects and pipeline construction elements, such as supports, pressure-gauge pipes, pipeline valves, patches as well as corrosions, risks, cracks, dents and other defects.

OlegM: Ну я конечно не лингвист но с научными текстами работаю. Сразу скажу что могу быть неправ. Читатель пишет: цитатаAs it moves along a pipeline As a result of the flux motion along a pipeline... or Following the flux motion along a pipeline... Читатель пишет: цитатаintelligent pig pig ? Читатель пишет: цитатаTopographic features of this magnetic field Topolographic features of the induced magnetic field... Читатель пишет: цитатаThe diagnostic data are two-dimensional magnetic signals, which serve to describe magnetic field topography near the pipeline inside surface. The diagnstic data, obtained in a form of two-dimensional magnetic signals, are used to describe... Читатель пишет: цитатаalgorithm allows to make algoritm allows one to make Читатель пишет: цитатаhas a disadvantage of being too formal in approach. too formal for practical use? Читатель пишет: цитатаDesigned to simulate human cerebral It was created in order to simulate... Читатель пишет: цитатаas the fact that it is not always possible to that sometimes it is impossible to З.Ы. мало артиклей... Много безличных предложений...

Pasha: По-моему, в заголовке слово "networks" должно писаться с большой буквы, как и другие слова (за исключением коротких (1-3 буквы) артиклей, союзов и предлогов). И вообще я не уверен, стоит ли писать "networks algorithms". Или "network algorithms", или "networks' algorithms".

Mukhin: Огромное спасибо всем ответившим, и модераторам, не снесшим эту тему на фиг. Мировая наука в лице альтернативной истории и Неразрушающего контроля материалов в неоплатном лице перед вами!!!

Mukhin: Тьфу, в неоплатном долгу, конечно!!!

Нико Лаич: Mukhin пишет: цитатаТьфу, в неоплатном долгу, конечно!!! Привет! Это точно! Так что подавайте-ка нам какой-нибудь новый тайм-лайн!!!