 |
 |
| High-tech Product | |
| Steel Adhesion Leakage Early Warning System for Continuous Casting | |
|
 |
| Steel Adhesion Leakage Early Warning System for Continuous Casting adopts the neural network technology of Neuralware Company of America as the predict modeling and deployment platforms for the steel adhesion leakage. By taking advantage of this comprehensive platform, Salient features of neural network in prediction, association, and mode identification are fully demonstrated in the adhesion leakage alarm modeling in combination with logic alarm modeling. |
|
| Three Salient Features | |
|
1. Intelligent analysis of interrupted thermocouple In the course of normal operation the thermocouple could be partially interrupted attributive to the possible vibration of on-site vapor and the crystal. Such incident will result in misreporting and under-reporting. In the event of partial interruption, the on-line monitoring program will start the neural system automatically to analyze the temperature relations before the occurrence of such interruption, between this thermocouple and its adjacent ones horizontally or longitudinally, establish the association model and puts it into operation in an intelligent way that the virtual temperature of the affected thermocouple built by the model will display for the adhesion alarm analysis. 2. Prediction of temperature trend By way of building a temperature trend based on that of each specific thermocouple, the adhesion leakage alarm realizes its prediction and thus avoids consequential steel leakage by a reporting delay. This system will be started when the casting machine enters normal casting state. Featured by its ahead of-time function and reliable accuracy, it is used for the neural network in alarm modeling, able to predict the temperature trend in forthcoming 3 minutes. 3. Analysis of alarm mode The alarm mode of the steel adhesion leakage forecast system can be analyzed and studied by the intelligent neural network. It enables users to study and analyze the correct and incorrect adhesion alarm models and reestablish models rapidly. The established model is used as the real time steel adhesion alarm model, applicable to the analysis alarm of various grades of steel. It requiresno professionals for modeling. This is remarkable advantage over single-model of logic prediction, which is restricted from function for the variety of steel. |
|
| Background Information | |
CARNEGIE, PA -- February 21, 2003 -- NeuralWare and WISCO-ICC, the Instrumentation and Control Company of Wuhan Iron and Steel (Group) Company announce the signing of a 10-year comprehensive Technology License Agreement that will permit WISCO-ICC to use NeuralWare's advanced neural network technology to develop analytic software applications for internal use and for resale in the People's Republic of China. The Agreement makes WISCO-ICC NeuralWare's first strategic partner in the burgeoning China marketplace, and it marks the first intellectual property license to be acquired by WISCO-ICC following China's accession to the WTO. In addition to providing access to existing core neural network technology, NeuralWare will provide training and support for WISCO-ICC engineers, and access to new analytic technologies developed by NeuralWare during the term of the license. |
|
| Back | |
