摘要
为解决控制约束对一类多相流量计系统液位控制造成的影响,提出了一种基于神经动态优化的模型预测控制算法。首先构建多相流量计系统传递函数模型,通过离散化建立相应状态空间模型,进一步地提出含控制约束的模型预测控制问题,然后将带约束的模型预测控制问题转化为带约束的标准二次规划问题,并运用简化对偶神经网络模型进行实时在线优化求解,从而获得系统最优控制序列,该网络模型的神经元个数仅与不等式约束个数相等,与现有文献中相关网络模型相比规模小、计算复杂度低,充分利用神经网络并行处理的优点,以提高模型预测控制的在线优化能力。最后,通过仿真实例验证了算法的有效性和优越性。
To solve the influence of control constraints to the level control of a class of multi-phase flow device systems, this paper proposes a model predictive control algorithm based on neurodynamic optimization. First, the transfer function model of multi-phase flow device systems is constructed and the corresponding state-space model is also obtained via discretization. Then, the constrained model predictive control problem is further proposed. Consequently, the problem is described as a standard constrained quadratic programming problem. Then, a simplified dual neural network is applied to solve the quadratic programming problem online and the optimal control sequence is obtained. The number of neurons in the network model is equal to the number of inequality constraints. Meanwhile, the network model has smaller scale and lower computational complexity than that of the existing literature, where we make full use of the advantages of parallel and distributed processing in neural networks so as to improve the online optimization ability of model predictive control. Finally, a simulation example is provided to verify the effectiveness of the proposed algorithm.
出处
《控制工程》
CSCD
北大核心
2016年第6期979-986,共8页
Control Engineering of China
基金
国家自然科学基金项目(61473136)
关键词
多相流量计
模型预测控制
神经动态优化
Multi-phase flow device
model predictive control
neurodynamic optimization
