The Active Disturbance Rejection Control And Parameter Optimization For Power’s Thermal Objects

Power’s thermal objects are large thermal capacity system with many disturbancefactors and strongly multivariable coupling, and also have long tune lag andtime-varying obviously when control them. Under variable conditions, traditional PIDcontroller often difficult to tuning parameters and gain online which was designedaccording as stability and safety rule, the regulate quality of dynamic process can notguarantee best. At the same time, its abilities of robustness and self-adaptation are verypoor. The active disturbances rejection control (ADRC) does not depend on the objectmodel, and can estimate and compensate the internal and external disturbances of thesystem, meanwhile, ADRC can inhibit the disturbances effectively and has a strongself-adaptation through the nonlinear feedback regulation based on state error vector.In this paper, two typical power’s thermal objects were researched by ADRCscheme—firing rate to feed water ratio (FR/FW) of supercritical unit and the steamgenerator (SG) water level of nuclear power plant.The control of firing rate to feed water ratio is the key and the difficulty incontrolling supercritical power units, but also restricts the whole control level of thesupercritical power units. Supercritical coal fired power units’ long tune lag andmultivariable coupling, variable load and variable burning coal quality restrict thecontrol of FR/FW. Conventional control often caused by the amount of fuel or feedwater regulation overdose, make the boiler run far deviated from the design conditions,and thus endanger the safe operation of the unit. The two inputs and two outputs modelof supercritical once-through boiler based on the amount of fuel feed and feed waterflow as input variables, the micro superheated steam enthalpy and the main steampressure as output variables. The control of FR/FW must coordinate fuel, feed water inresponse to load variation. The control system was considered be maked up by twosingle loop subsystems–micro superheated steam enthalpy loop and main steampressure loop, according to the above separate single loop, the corresponding ADRCcontroller was put forward, and the coupling between the two loops were taked for eachexternal disturbance, the ADRC system can use the ability of disturbance estimation toreal-time estimate and dynamic compensate the dynamic disturbances between couplingcircuit by a feedforward mode, and to realize the decoupling control of FR/FW. Theexperimental results show that, this scheme’s advantages are adaptability of variable working condition as well as excellent dynamic and static control performance whenrelative to traditional PID decoupling control.Malfunction of steam generator is the main cause of unexpected shutdowns andloss of capacity factor in nuclear power plants. Therefore, the ADRC scheme waspresented for steam generator water level control system. The internal and externaldisturbances of the system were estimated and dynamic compensated by the extendedstate observer; meanwhile, the contradiction of the control system between rapidity andovershoot was alleviated through the nonlinear feedback regulation based on state errorvector. However, ADRC parameters are numerous and without uniform adjustmentspecification, and are very difficult to be tuned by cut and try method with largerworkload, it also difficult to popularization and application in engineering practice. Thefalse water level make the nuclear steam generator water level difficult to identify, so, akind of second-order diagonal recurrent neural network (SDRNN) identification methodwhich has simple structure with fast and accurate identification ability was proposed,and the SDRNN is used to prompt track SG water level Jacobian information anddynamic identification the change rate of water level in order to real time optimizationADRC parameters. Besides, the simulation tests of the controller under differentdisturbances such as water level disturbance, steam disturbance, feed water disturbanceand load change were carried out, the simulation results indicate that the SG water levelcontrol scheme which was based on SDRNN-ADRC can effectively inhibit theinstantaneous false water level and also has a better quality of identification and control.