Research On Some Single Pump Blade Angle Optimzation Methods

The South-to-North Water Diversion project is one of the greatest measures taken by our country to solve the water shortage problem, concerned with the sustainable development of the economy and society. Considering that it is a project of long routine, high lift, huge flow quantity, and it involves natural science and social science and other fields, the project can be considered as a complex system, therefore, rational decision making of management can produce great economic benefits. The decision of optimization of a sub-project of the east routine is presented in this paper.Jiangdu Pump Station is the source water pump station of South-to-North Water Division Project, which is affected by the periodical tides and ebbs of the ocean nearby and the variation of the electrical price in a day. How to make the decision of the best set of the blade angle in every section of each day considering the water quantity constraint, the efficiency constraint, and power constraint is the major issue discussed in this paper.As for the feature modeling:Based on the existing data of NO.4Jiangdu Station, and already known the model of single pump blade angle, the single decision variation and nonlinear optimization problem is analyzed, and the related numerical tools is illustrated and demonstrated in the paper.It is demonstrated and feature modeled in the electricity and water quantity matrixes and a corresponding multi-section chart, with the constraints of the water quantity, efficiency, and power considered.On how to optimization methods:(1) Based on the conception of directional derivative, using the incremental cost of cost-water quantity as an index, the slowest increment routine is found in the paper to finally achieve a solution with approximated lowest cost while meeting the water quantity need in a day. Finding effective and feasible direction, a new optimized solution is presented in this paper, which solved the nonlinear function with discrete constraints. This solution is explained in a detailed way.(2) Assuming the set of pump blade angle in every different section of a day including the information of the correspondent daily water quantity and the cost as an individual, the genetic algorithm is applied to solve the model. The Excellent individuals are picked up among the random produced seeds as the ancestors, the multi-individual cross is used, and the uniform probability distribution of mutation is taken, so as to have the excellent quality of the ancestors to be effectively inherited generation by generation, that is to have the low cost and feasible water quantity. For all the considered constraints, the characteristics of the constraints are all projected into the individuals by altering the parameters in an individual. After experiments, the approximate globally optimization solution can be found.(3) Parallel optimizing solution is successfully applied in solving the model. Same to the ancestor producing process, then the ancestors are taken as the original starting lines to each search a better individual surrounding in a parallel way, until no individual can find a better individual.In the paper, the algorithm design, the experiment and the execution process, and the demonstration of C program is given clearly, and the results and the solving processes are illustrated in a detailed way. The convergence proving, the program complexity, the space of the data, and the execution time are all compared, analyzed and reported. The shortage and advantages of the three algorithms and the essences of each algorithm are all explained. Where need to be improved is also presented.The next step work and the new discovered problem are written.