The Intelligent Method Research Of Initial Pressure Optimization For Steam Turbine Sliding Pressure Operation

In recent years,major changes have taken place in power utilization structure,which lead to peak-valley difference of electric load between day and night increasing.Supercritical steam turbine units required to deep participating in peak regulation,the unit utilization hours decreased year by year,the low-load operation time increased generally.Therefore,thermal economy efficiency is greatly reduced for units.Meanwhile,along with the development of China's economy,energy and environmental situation,the operation of the thermal power unit energy saving and consumption reducing has become the objective demand for thermal power enterprise.Therefore,how to improve the operational economy of unit at low-load stage has to become a serious problem to be solved.The initial steam pressure of turbine must be optimized to ensure the steam turbine maintain the best running state under the variable load operation deviating from its design conditions,which can effectively reduce the heat rate of steam turbine.Inspired by the mechanisms of biological evolution and some natural phenomena,swarm intelligence optimization technique is proposed,which could solve complex systems' modeling and optimization.For the extremely complex nonlinear,multiple load cases of supercritical steam turbine' thermal properties in power plant,it is very difficult to set up models of thermal process and to achieve the initial pressure optimization.Therefore,a great deal of research would be done on shuffled frog leaping algorithm(SFLA),least squares support vector machine(LSSVM)and multi-model technology based on clustering that would be applied to the initial pressure optimization in order to make a economy running for turbine unit.The main research contents of this paper can be summarized as follows.For the optimization ability of typical SFLA is limited,an improved SFLA(mixed search SFLA,MS-SFLA)is presented to solve the optimization problem.It significantly improves the optimization performance by introducing chaotic opposition-based learning strategy,adaptive nonlinear inertia weight method and a new local disturbance strategy.This paper verifies that the MS-SFLA has better optimization performance than standard SFLA through 13 benchmark functions.Also,the MS-SFLA algorithm is employed to serve as a method for optimizing LSSVM parameters(including the regularization and kernel parameters)to improve the regression accuracy and generalization ability.The effectiveness of the modeling is verified by numerical experiments.Then,fuzzy C-means clustering algorithm is studied and applied to divide data set.In order to improve robustness of the fuzzy C-means clustering to noise and isolate data,a kernel fuzzy C-means clustering algorithm based on RBF kernel function is proposed.At the same time,a double clustering algorithm that combines the advantages of kernel fuzzy C-means clustering and GK algorithm is presented to avoid and solve the commonly existed problems in original clustering algorithms,such as clustering accuracy depends on data distribution,sensitivity to initial clustering center,being easy to be trapped in local minima,clustering number should be determined in advance and optimal clustering number difficult to determine.The feasibility of the double clustering algorithm is verified by heat rate multi-model modeling experiments.In addition,aiming at the issue that the characteristic of complex nonlinearity of heat rate for steam turbine unit which is difficult to be descript accurately by the single-model,a new multi-model modeling method based on a double clustering algorithm and LSSVM for heat rate is presented,and the MS-SFLA algorithm is introduced to adjust the model parameters.After,a 600 MW steam turbine unit is used as the identification object,and multi-model for heat rate is established.The obtained results show that the multi-model modeling method is an effective approach with better forecasting precise and generalization capability.Finally,based on heat rate multi-model,MS-SFLA is used to determine the optimal operation initial pressure within the feasible operation initial pressure range according to the optimization target of minimizing the heat rate under the certain loads of steam turbine.At the same time,the obtained optimal initial pressure is taken as the setting value of the main steam pressure,so as to achieve the goal of optimal operation of the steam turbine unit.Then,the optimal initial pressure curve is obtained after the experiments,which has more practical significance to guide the optimizing operation of steam turbine.