Research On Ash Prediction Of Heating Area Of Coal-Fired Power Station Boiler Based On Neural Network

The long-term normal operation of coal-fired power station boilers will inevitably lead to the accumulation of dust on the metal heating surface.Due to the large thermal resistance of dust,the heat transfer efficiency and economy of the heating surface will be greatly reduced.In addition,the problem of ash fouling and slagging on the heating surface is also easy to cause safety accidents such as pipe explosion.At present,most coal-fired power stations adopt the strategy of regular soot blowing per shift.This strategy requires higher soot blowing experience for the staff,and there are also potential safety hazards for long-term over-blowing and under-blowing.Nowadays,with the vigorous development of artificial intelligence and deep learning technology,it has a strong ability to process big data and large samples,and the optimal time can be obtained through intelligent computing methods at the time of blowing dust,which greatly reduces the misjudgment.Therefore,aiming at this subject,this paper conducts a series of analyses and experiments on the prediction of boiler ash deposition by taking the boiler metal heating surface of various devices as the research object.The main research contents are as follows:（1）In this paper,the Clearness Factor（CF）is used as a health indicator to measure the health of the heating surface.The time series of the clearness factor can be calculated from the heat balance formula,and the required variable parameters can also be obtained from the DCS monitoring system.In addition,data preprocessing can reduce the unnecessary difficulty of forecasting,retain the trend of ash accumulation,and lay a solid foundation for the following research on ash accumulation prediction.（2）The time series of the clearness factor of the heating surface is used as a dynamic time series.Aiming at the time series characteristics of ash accumulation,this paper proposes a comprehensive method for dynamic prediction of ash in the heating area of ??coal-fired power plant boilers.Combined with the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise（CEEMDAN）and the Nonlinear Auto Regression Neural Network（NARNN）,the ash condition of the heated area of the coal-fired boiler is predicted.In order to construct a reasonable network model,the minimum information criterion and the trial-and-error method are used to determine the number of delays and the number of hidden layers.The experimental results show that compared with conventional methods,the proposed method can achieve high-precision multi-step forecasting and different starting point forecasting.（3）With the detonation of Deep Learning（DL）models,the superiority of deep learning algorithms in time series forecasting has gradually become prominent.In order to solve the nonlinear prediction problem of dust accumulation time series,this paper proposes a hybrid model combining signal analysis and deep learning algorithm.The Ensemble Empirical Mode Decomposition（EEMD）algorithm can decouple nonlinear time series with multiple features,and the deep learning model has a good prediction effect on the decoupled high-frequency components.Finally,a variety of heating area ash data sets are used to verify the effectiveness of the hybrid model,which provides a certain reference value for the economical operation of the boiler.（4）Although the signal analysis method has important value for the prediction of nonlinear time series,it has a large number of decomposed components,and the training time of the deep learning model is long,which may cause a series of potential problems.This paper proposes a deep feature extraction model,which completes the role of input reconstruction and indirectly reduces the lengthy training time of deep learning.In addition,the adaptive sliding window can monitor the fluctuation and trend of the time series.The proposed method has proved its feasibility and superiority in various heating surface prediction and heating surface maintainability experiments.