Modeling And Advanced Control Strategy Research Of Supercritical Unit Flexible Operation

With the increasing proportion of fluctuating renewable energy generation,the operation goal of supercritical thermal power units has changed from the pursuit of high efficiency and energy saving to the focus on improving the flexibility of units,depth peak regulation of units and rapid load lifting capacity.In order to improve the flexibility of the unit,it is necessary to put forward an advanced control algorithm,and the premise of the control algorithm is an accurate and concise mathematical model.Thus,it is of great significance to study the modeling and control strategy of supercritical unit.The coordinated control system is taken as the research object in this paper.Based on the establishment of high precision model,the advanced control algorithm is designed to improve the operation flexibility of the unit.In the modeling section,the output expression of T-S fuzzy incremental model is combined with the consequent part of the fuzzy neural network to construct a novel fuzzy neural network structure,in which the accuracy of the local linear model is greatly improved.In terms of parameter training,the improved kernel k-means++algorithm is used to train the parameters of the antecedent part.The algorithm initializes the number of fuzzy rules by using the Xie-Beni index method,which eliminates the limitation of the traditional manual selection of the rules number.The kernel space distance is then applied instead of the traditional Euclidean distance to obtain better cluster center and radius parameters.In the consequent part of the network,the supervised adaptive gradient descent method is utilized to optimize the initial parameters,and then the artificial immune particle swarm optimization algorithm is applied to re-optimize the parameters.In the part of control strategy,a two-layer hierarchical control structure is proposed in this paper,in which the upper layer is a static error free nonlinear generalized predictive controller with constraints to obtain the optimal control sequence.The lower layer is L1 adaptive controller,which realizes the optimal trajectory tracking by estimating the uncertainty.In the aspect of controller set point optimization,a set point softening operation with adaptive adjustment of the softening factor is utilized to further improve the control performance.Finally,in the simulation experiment part,the control algorithm is tested on the basis of the fuzzy neural network model which is driven by on-site data.Satisfactory tracking performance have been shown in many experiments,such as single output change experiment,flexible operation comparison experiment and disturbance rejection,among which the maximum load climbing rate reached 6%per minute of the rated load.Moreover,the control variables do not fluctuate greatly.The experimental results confirm that the controller in this paper can ensure the safe and stable operation and improve the performance of the unit to meet the requirements of flexible operation.