Control Optimum On Supercritical Unit's Long Tune Lag Object During Dynamic Process

In the power plant thermal engineering control field, most control objects are of large thermal capacity system with strongly coupling and interference relations among them, and also have characteristics of long tune lag as well as great delay. Since traditional PID controller is designed according as stability and safety rule based on classical control theory, it can not guarantee best regulate quality of dynamic process under variable operating conditions and circumstance, although has good control character under designed conditions. Simultaneously, its abilities of self-adaptation, robusticity, and conquering strong nonlinearbility are very poor. Especially, several key parameters such as once through boiler's fuel flow and main steam flow are still difficult to accurately detect, which have recently attached many importance on optimal control of large supercritical unit under model dismatching and variable operating conditions. Three aspects were given emphasis to study as follows:Firstly, with respect to difficulties about fuel flow's direct measure and its heat calories'online detection on supercritical boiler, and by discussing questions of a certain once through boiler's fuel flow control system, a new optimized scheme about it based on RBFNN (Radial basis function neural network)-data fusion method was proposed. It reconstructed fuel heat coefficient by combining RBFNN's strong self-study and parallel computing capability with data fusion's complementarity and redundancy. This made it can exactly correct actual fuel flow online during combustion's internal disturbance and forward control fuel flow demand during external disturbance, so as to guarantee fuel flow's accordance with the instant demand of changeable unit load or internal boiler disturbance. Results indicate this scheme's celerity and precise on fuel flow control, as well as stability under various boiler operating conditions.Secondly, a new control scheme based on mixed structure RBFNN (MS-RBFNN) was presented. This MS-RBFNN can synthetically study current main relative state parameters, so as to parallel calculate the optimal control variable of firing rate-feed water ratio(FR/FW) by using least square of intermediate boiler heating surface temperature as its training signal. This kind of FR/FW can in time correct unit's firing rate and roughly regulate the main steam temperature. Experimental results indicate, comparing with traditional PID control, this scheme's advantages on better overcoming long tune lag during main steam temperature control and adaptability of variable working condition as well as better whole unit thermal efficiency.Lastly, with above-mentioned work, this paper also presented an optimized direct-firing medium-speed mill's startup control system based on a certain fuzzy controller. It optimized, by introducing fuzzy control and using its robusticity and capability of conquering nonlinearity, the control of uncertain parameters such as coal-bited time and current during mill's start-up. Results indicate this scheme's celerity and precise on fuel flow control, advantages on mill's start-up coordinated control between modulating control system (MCS) and burner management system (BMS) as well as stable transition of the combustion situation.