Fault Diagnosis by Using the Multi-class Support Vector Machine

In industrial environments, a large amount of data is generated which in turn stores the database and data from all relevant areas such as planning, process design, materials, assembly, production, quality, process control, scheduling, error detection, shutdown, relationship management. Collects with the customer, etc. Data mining has become the tool used to gain knowledge of the industrial process of iron and steel making. Due to the rapid growth of data mining, various industries have been using data mining technology to search for hidden patterns that may be more relevant to the new windshield system, which will introduce new models to improve production quality, optimal cost of productivity and maintenance, and so on. Continuous improvement of the entire steel production process due to the avoidance of quality deficiencies and associated production improvement is an essential task of the steel producer. Therefore, the zero defect strategy is popular today and several quality assurance techniques are used to maintain it. This article attempts to identify the effective state-of-the-art sensors in the system using data mining and then to obtain a suitable model using a support vector machine to predict the system status in this article, the accuracy of more than 95% of the error states is detected.

Keywords:

Group Methods , Decision Making , Patterns , Support Vector , Nearest Neighbor. , روش‌های گروهی , تصمیم‌گیری , الگوها , بردار پشتیبان , نزدیکترین همسایگی.