明渠模型预测控制快速优化计算方法

Fast optimization calculation method for predictive control of open canal models

针对基于圣维南方程使用启发式算法进行渠道预测控制时,现有优化计算方法生成调控方案时间过长、无法满足实时控制需求的问题,提出了一种基于多目标优化算法的快速优化方法.以一处试验渠道的数值模型作为控制对象,应用此方法构建了预测控制模型以生成调控方案,并构建了一维水动力模型对调控方案进行验证.研究发现在闸门控制系统中,超出水位控制范围的最大值最小、超出水位控制范围的平均值最小、闸门动作次数最少、闸门开度总变化量最小4个目标间能够相互诱导,加速优化计算过程4倍以上.在该试验渠道的流量计划调整工况中,验证了此种方法的有效性,将单次计算时间缩短至1 s内.验证效果表明,该方法在提高了水位控制效果的同时,能够减少由单独设置超出水位控制范围的最大值最小目标导致的非必要闸门调控,总计减少69%的闸门调控次数.

A fast optimization method based on multi-objective optimization algorithm was proposed to address the problem that the existing method takes too long to calculate the control scheme and cannot meet the real-time control requirements when using the heuristic algorithm for channel predictive control based on Saint-Venant′s equation.Using a numerical model of an experimental channel as the control object,a predictive control model was constructed using this method to generate a control scheme,and a one-dimensional hydrodynamic model was constructed to verify the control scheme.It is found that in the gate control system,the four objectives of minimizing maximum value beyond the water level control range,minimizing average value beyond the water level control range,minimizing number of gate actions,and minimizing total gate opening change can be mutually induced to accelerate the optimization calculation process by more than four times.In a working condition where the flow rate of the experimental canal is adjusted according to the plan,the effectiveness of this method is verified,and the single calculation time is shortened to within 1 s.The verification results show that this method can improve the water level control effect while reducing unnecessary gate control caused by setting the maximum and minimum targets separately that exceed the water level control range,resulting in a total reduction of 69%in gate control times.