Gatilova Yuliya

Decision of this problem is practically necessary in condition of the plant “Doneckstal-MZ” since at present single way of the discovery defect is a visual estimation of the type of the defect on final stage sheet production (the stage of the checking) in condition of the absence of the system formalizing parameter defect on structure, size, color feature, geometries.

At reception of the formalized description of the causal relationships of the shaping defect on stages of the reception continuously cast stocking up and sheet renting questions diagnosticses move over to practical plane: reveal and avoid the reasons of the appearance defect by by contributing the changes to technological instructions.

1.4 Diagnostics surrface defect sheet renting

Screening on defect of the steel-melting origin on stage of sheet renting by reason of discovery surrface defect forms 1,5% [2].

So the most important stage of the diagnostics to surfaces of sheet renting on presence spume is a separation steel-melting spume from the other type given defect. The Givenned stage of the recognition is founded on spectral analysis of the digitated scenes to surfaces of sheet renting. Spume is presented on figure 1.

Figure 1 – Spume

Stages of the diagnostics surrface defect:

1. Revealing the centre of the surrface defect on its digitated scene.

2. Determination of the physical features of the surrface defect.

3. Preliminary categorization of the surrface defect.

4. Comparison with base master digitated scenes surrface defect [2].

1.5 Digital image processing

Preprocessing to information is concluded in undertaking stage to segmentings, correction, standardizations of the scale to brightness, filtering, geometric transformations [3].

Algorithms to segmentings are based on two fundamental principles: breakup and resemblances [4]. In step of segmentings scene divides on component parts or objects [5]. Segmenting of the digital scene belongs to to count;calculate;list most hard nut to cracks of the processing [5].

Determination of the borders object scenes is executed on following scheme: colour scene is translated in monochrome halftone and is smoothed, realized spatial differentiation - is calculated gradient to functions to intensities in each point of the scene, and are suppressed importances less installed threshold [3].

Necessary to build the contour scene, most clearly selecting sidebars of the source scene. Contour scene - an undirected earl, being partial subgraph of the scene, which is built on base gradient [3].

Necessary to produce correction a mistake for removing false object [4].

Figure 2 - Origin false object

The Problem of the automatic determination surrface defect continuously cast stocking up actual for majority metallurgical plant [2]. The most important stage of the diagnostics surrface defect second redistribution is a determination sufficient level to physical sensitivity of the methods [2].

Overview part

2.1 Local review

Studies in the field of processing the digital scene are employee, graduate student and student of the faculty computer information technology and automations, within the framework of development of the direction “Computer systems medical and technical diagnostics”. Develop such directions as network, built on base neuron, the genetic algorithms, genetic programming, decision of the combinatorial problems, segmenting of the texture.

My work is directed on creation automated computer system to categorizations surrface defect sheet renting, got from continuously cast stocking up. Within the framework of given work are researched methods of the spectral analysis of the scene, methods of the separation by sign of the texture.

2.2 National review

Processing the digital scenes for the reason checking quality to metallurgical product specialists concern with on different stages of the technological process leading metallurgical enterprise of the Ukraine.

Taranenko Pavel Petrovich, chief of the central exploratory laboratory and Velikij Nikolay Ivanovich, main specialist of the plant “Doneckstal-MZ”, which have allowed me possibility to work with such project as “Diagnostics surrface defect sheet renting, got from continuously cast stocking up”.

2.3 World review

Automation of the systems of the checking quality to metallurgical product develops in scientific and exploratory laboratory, in high educational institutions, which are found in direct vicinity with large metallurgical complex on the whole world.

Postlude

3.1 The Planned results.

On given moment work on subject master's work is unfinished. The Final results is planned get to December 2007. For material and result on subject “Specialized computer system of the diagnostics surrface defect sheet renting, got from continuously cast stocking up” on Email: yuliya.s.g@gmail.com, doneckshipovnik@yandex.ru

As facilities of the decision of the worded problem is expected use the optical facilities fixing scenes separate defect and their locations on surfaces of the sheet as a whole, specialized software programs and computer system.

The initial stage for decision of the delivered problems is a description of the application domain of the object of the study and object of the diagnostics i.e. information field for description of the problem, separation of the component and their structured relationships, formal presentation of the object in the manner of, suitable to searching for of the mathematical methods of the decision and for computer (programme) of the presentation of the diagnostic system.

Literature

1. Rubenchik A.E., Sorokin M.I. – Steel, 1982, №2, p.72-75

2. Gatilova Y.S. Specialized computer system of the diagnostics surface defect sheet renting// The automation technological objects and processes. Searching for young. Collection of the scientific work VII International research conference graduate student and student in Donetsk, April 26-28 2007 - Donetsk, DonNTU, 2007.- P.251-253. 3 p.

3. Bakut P.A., Vornovickiy I.E., Kolmogorov G.S., Labunec V.G. The Methods to segmentings of the scene. Uraliskiy pollytechnic institute of the name S.M. Kirova, 1986. - P.127.

4. Prett U. Digital image processing. – B. 1,2. – M.: Science, 2000. – 1024p.

5. Gonsales R., Vuds R. Digital image processing. - M.: Tehnosfera, 2005. - 1072p.
   (chapter 1, part 4 - P. 56-60)