Prediction of fouling in condenser based on k-means algorithms and improved Chebyshev neural network

Many factors affect the production of condenser fouling, but there is no accurate way to predict it. In this paper, it makes good use of k-means algorithm in order to cluster data in different seasons and different phases to carry out a comprehensive improvement from the algorithm and network structure aiming at Chebyshev neural network weaknesses. Improved Chebyshev neural network in line with the basic characteristics of biological neural networks is simple and fast convergence. Using the modified Chebyshev neural networks to predict fouling factor, the results show that this method not only provides an effective fouling factor forecast method with good predictive ability, but also in the same precision under the premise of the convergence speed is superior to the general neural network, can provide scientific and rational decision making for decontamination period of condenser.