基于BP神经网络与改进遗传算法的泵站优化调度

通常泵组件的调节主要依靠工程师经验,难以保证水泵在所有时段内均高效运行,易造成能源浪费,对此提出了以效率模型为依据的泵站优化调度方法。为减小误差,使用泵站实际运行数据基于BP神经网络训练水泵特性。在此基础上,利用训练后的数据初始化种群,以泵站效率之和最小为目标,建立泵站优化调度数学模型,设定水量、压力和高效区等约束条件,采用改进的遗传算法求解水泵组合方案及各泵运行参数。结果表明,改进遗传算法对泵站调度的优化效果明显,可适用于泵站的优化节能运行。研究成果为泵站优化调度运行提供了一种新方法。

Optimal Scheduling of Pumping Stations Based on BP Neural Network and Improved Genetic Algorithm

In general, the adjustment of pump components mainly depends on the experience of engineers, and it is difficult to ensure that the pump runs efficiently in all periods, which can easily lead to energy waste. Aiming at this problem, optimal scheduling of pumping station is proposed based on the efficiency model. In order to reduce the error, the actual running data of the pump station is used to train the pump characteristics based on the BP neural network. On this basis, the trained data are used to initialize the population, and taking the minimum efficiency of the pumping station as the goal, optimal dispatching mathematical model of the pumping station is established. The constraints such as water volume, pressure and high efficiency zone are set up, the improved genetic algorithm is used to solve the pump combination scheme and pump operation parameters. The results show that the improved genetic algorithm has obvious effect on optimal dispatching of pump station, and can be applied to the optimization and energy-saving operation of pumping station. The research results provide a new method for pump station optimization and dispatching operation.