Mold Temperature Measurement In The Adaptive Filtering Method

For reducing accidents of breakout in the continuous casting, it is necessary to measure the crystallizer temperature. It is an effective method to use thermocouple to get the temperature of the round mould. Because there are so many interferences in the industry, especially the electromagnetic stirring make the noise is more greater than the useful signal. Although some anti-interference hardwares have been used , the result could not be satisfactory.The common software filter method get rid of thenoise insome degree,so the new technique is used to do it.At first this paper studies the structure of the adaptive transversal filter,and analyzes the common adaptive filter algorithm(LMS,RLS and NLMS)'s characteristics and applied range,and so on.Based on adaptive noise cancellation theory, this paper analyzes these algorithms in an adaptive trap filter example.Based on the different characters between the temperature signal and noise, a new adaptive noise canceller structure is given to dispose the temperature data by the adaptive noise cancellation theory. The delayed signal of temperature is selected for adaptive filter's reference signal. For cancelling the effect of the direct current, a high-pass filter is added in front of the input. To revise these adaptive filter algorithms, the variable step-size LMS algorithm is studied in this paper. At last, a variable step-size NLMS algorithm is given. The new variable step-size NLMS algorithm is more simpler than LMS algorithm and NLMS algorithm, and the new VSNLMS algorithm is better than the LMS algorithm and NLMS algorithm in the initial convergence rate ,time-varying system tracing ability and steady offset.On the basis of neural networks technology and principles of adaptive noise cancellation system, this paper sets up an adaptive noise canceller with Back Propagation(BP) algorithm, and processes the temperature signal.The Matlab simulation examples using above methods show their effectiveness.