Thermal Performance Calculation Based On Neural Network For Power Plant Unit

With quick improvement of process control, real-time data processing and management of the thermal system, more atention has been paid not only to the safety of power plants, but also to the economy of the whole system. But parameter sensors have been working in rigorous condition, such as high temperature and pressure, large humidity, and severe vibration for long time and some of them will be damaged to diferent extent. Once the sensor cannot work operation worker cannot get the right parameter's data. Real-time economic analysis for thermal system has a significant influence on economic operation and energy saving of steam power plan.The traditional methods of performance for thermal power system and devices are firstlt analyzed, and their advantages and disadvantages are summarized. The relations of thermal parameters are complicated and non-linear, so it is dificult to find an exact model to analyze and study. But artificial neural network has the ability to realize parameters' random map, a kind of improved artificial neural network based on the BP algorithm to simulate some key thermal parameters at real time whose sensors have been damaged is presented in the paper. Meanwhile all the factors that influence the simulation results are analyzesed and compared in details. The results show that input parameters have played important role in the simulation process and only by selecting the right parameters can we get the good simulation efect.The simulation results show that when some input data of thermal parameter has been disturbed or damaged, the analysis method of rtificial neural network can calculate correctly and output acceptableresults. But if the disturbance of the parameter exceed the bound of its limit, the calculating processe will be possibly failed. Due to the parameters of the thermal system are always variational and the phenomena of sensor failure or the error so such as absence of data transfer potentially occur in the power plant, using the analytic method of artificial neural network to calculate the performance of thermal system is effective and feasible, and has considerable engineering value.