The process used for predictive control in artificial neural network su00ecnter machine

摘要

A process using artificial neural network for predictive control in su00ecnter.a production of sinter machine in accordance with the standards is of fundamental importance to the iron and steel industry because it depends on the economic productivity of blast furnaces and, consequently, the productivity D. And the whole plant.Although the improvement of sintering have economic meanings are of major economic importance and ecological, such as facilities of ore fines and coal tailings, exploitation of the mines and facilitation of mines whose ores has a tendency to produce large qty The ages of fine in the processes of crushing and grinding.The thermodynamic conditions of sintering process, require that the layer of pellets to be sintered to have level maintained within strict limits, which, if not obeyed, stops of slow recovery and non-compliance of materials, involving reprocessing and a u00e9rie of productivity losses.The big problem of the prior art that this patent is to advance is that the traditional controls the level of the hopper (5), aumentadora of sinter machine (6) have response times of about 250 seconds, time too long for a continuous operation and safe.The "process using artificial neural network for predictive control in machine of su00ecnter" object of this patent is the core of neuro fuzzy artificial intelligence software specific, supported, preferably by MATLAB tools and Adaline may, however, use inu00fa But other tools and platforms of the RNA.The RNA is trained to predict the level of filling of the hopper (5) 250 seconds or more, up front, in case your specific application.The artificial neural network was trained with the information process such eat the weights of materials (10) fed by the silos aumentadores of pellets (2), the density of the material (11), the production volume per unit of time (12), which led to the software specifies (9) enable The control of the system with an advance of 250 seconds or more.And this specific software (9) provides the interfaces (13) for the control panels and is connected with the database, in order to allow for learning continuous, since RNA can operate variable values that were supplied during the process of training.