[Objectives] This paper focuses on the soot and coking monitoring of boiler heating surfaces

in order to deal with the problem that the conventional fixed-interval soot-blowing methods of boiler are no longer adequate under such dynamic conditions under the background of flexible deep peak shaving

often leading to over-blowing or under-blowing. [Methods]This study takes the low-temperature reheater of a 660 MW ultra-supercritical coal-fired boiler as the research object and develops a time-series and multi-factor ash deposition monitoring model based on a Backpropagation Neural Network (BPNN). The actual heat transfer rate is calculated from the working fluid parameters at the heating surface inlet and outlet

while a theoretical heat transfer prediction model for a clean heating surface is established by integrating time-series data and multiple influencing factors. [Results]The model achieves a root mean square error of 5635.33 and a MAPE of 0.62%

significantly outperforming traditional models. [Conclusions]Without requiring additional sensors or measurement points

the proposed method fully leverages historical data to enable online monitoring of ash deposition severity

providing reliable data support for intelligent soot-blowing in boilers.