A Phase-space-reconstruction-based CC-HMM On Kiln Coal Feeding Trend Prediction

Rotary kiln is widely used in metallurgical,power generation,cement and other industrial production fields.The key part of the production of such kiln combustion process is the detection and control,it directly affect the quality of products,production of energy and air pollutants emission.The sintering process of the clinker in the kiln is a kind of typical complex object of industrial process.Currently,it mainly stabilizes the working condition by controlling the coal feed quantity in the kiln.The control of the quantity of coal in the kiln is mainly dominated by artificially seeing fire,and the subjectivity and arbitrariness of this way make the rotary kiln control become difficult.In this thesis,the influence of the chaotic time series of various thermal variables on the coal feeding trend is considered,and the trend prediction method by combining multivariate phase space reconstruction and HMM is put forward.Finally,the results are applied to the rotary kiln in the industrial field.This thesis mainly includes the following aspects:(1)Research on chaos and phase space reconstruction theory.This thesis introduces the development of chaos theory and expounds the essence and characteristics of chaos.Then,we give the three main parameters of quantitative description of chaos:the mathematical definition of Lyapunov exponent,fractal dimension and Kolmogorov entropy and its calculation method.Finally,the theory of phase space reconstruction is introduced.Aiming at the two important parameters:delay time and embedding dimension,this thesis briefly introduces the independent method and its characteristics,and focuses on the C-C algorithm of jointing delay time and embedding dimension.(2)A algorithm of prediction of CC-HMM feeding quantity based on phase space reconstruction is proposed.Firstly,the background of HMM is introduced briefly.Then,the theories of HMM,the three basic problems of HMM,the forward-backward algorithm,Baum-Welch algorithm and Viterbi algorithm are introduced respectively.The HMM application and the trend of coal feed Feasibility is analyzed.Finally,the implementation process of CC-HMM algorithm based on phase space reconstruction is introduced.(3)Analysis and application of experimental results in predicting coal feed.Firstly,the method of extracting the trend data of coal feed quantity is introduced.Then,we compare the experimental results with other prediction models.The experimental results of CC-HMM feed quantity trend prediction based on phase reconstruction are introduced and analyzed.The results show that the phase space reconstruction can predict the feed coal consumption trend and provide more accurate predict for the operation of coal feeding.The method proposed in this thesis finally apply to the rotary kiln industrial field.The research of the thesis is based on the control of the sintering temperature in the kiln coal feeding operation,so as to facilitate the effective implementation of the operation optimization in the sintering process of clinker.It provides accurate and reliable reference value for manual operation and expert control in industrial field,it also improves clinker sintering quality and energy saving.