A new parallel algorithm for optimal control problems of interconnected systems

From MaRDI portal





The paper deals with interconnected control systems of the form \(x(t+1)=f(t,x(t),u(t))\) with state \(x(t)\) composed as \((x_ 1(t),\dots,x_ N(t))^ T\) and \(f=(f_ 1,\dots,f_ N)\) described as \(f_ i(x_ i(t),u_ i(t))+C_ 1(t)z_ i(t)\), where \(z_ i(t)\) is a linear combination of \(x_ 1(t),\dots,x_ N(t)\). The performance index to be minimized is of the form \[ J=\sum^ N_{i=1}\left(\sum^{T- 1}_{t=0} g_{ii}(x_ i(t),u_ i(t),z_ i(t))+g_{iT}(x_ i(T))\right). \] The approach of the paper consists of partitioning the minimization problem as an \(N\)-fold minimization problem, for each of the \(x_ i(\cdot)\) and \(u_ i(\cdot)\) separately. These decomposed subproblems are solved parallel by means of dynamic programming. The Lagrange multipliers are selected as coordinating variables and updated to ensure global optimality. The computational complexity of the method is evaluated and some numerical issues are discussed in simulation results.











This page was built for publication: A new parallel algorithm for optimal control problems of interconnected systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4025753)