| Abstract [eng] |
The use of biomass for energy production is becoming increasingly popular in Lithuania. There are tens of already operating power plants, and the expansion of this energy sector is still present. Because biomass is not of volatile nature, the machinery required for automated burning process is more complex compared to traditional fossil fuels. In order to ensure proper function of these plants, regular checks and maintenance must be performed. The goal of this paper is to create an accurate model of the biomass furnace, working in optimal conditions, which could be compared later, to the same furnace, during its operation. This would give the possibility for the responsible personnel to look for deviation in output parameters, which would allow early diagnosis of upcoming problems, thus providing an addition to existing maintenance procedures. The paper describes several possible modeling methods and provides analysis of their suitability for the task. The analysed plant features a condensing economizer and the maximum power capacity is 5,3MWh. Heat is produced all year round, with varying loads on the system. The most important technological parameters, which describe operation of the furnace, are selected and their historical values are collected. A neural network based model is created using the data, together with a special system, which informs the user of parameter deviations. Furthermore, there is a possibility to identify the input parameters, which could have caused the deviation, using evolutional programming method based approach. |
