过程模拟信息网络的新分解法

本文提出的寻找最优断开线集的新方法,无须识别回路,也避免了任何计算过程,只须将网络中的节点控权重递降次序排列,把二维图化为一维节点序列,然后按照使反馈流线的权和减小的规则调整节点在序列中的位置.调整过程结束时保留的逆流线便构成一组断开线集,此时的节点序列便规定了最优计算次序.对文献上通常引用的七个网络图的分解表明,新分解法简单快速,且均给出最优解.

A New Decomposition Strategy farInformation Flow Nets

Decomposing the information-flow net is the first task in process simulation of a complex process flowsheet. Among the efforts since the 1960s dedicated to this subject for finding the best set of tear streams that guarantees fast convergence and minimal computation time, most strategies involve two stages of decomposition, namely, indentification of loops in the net, and tearing them afterwards. The optimal solution of the decomposition problem usually cannot be obtained without considerable expenditure of computation time.The new method for decomposition of nets in this paper avoids the need of identification of loops and any matrix operation. By converting two-dimensional net to a onedimensional node sequence according to the decreas ing magnitude of "node weights", which can be easily counted in terms of the adjacent matrix of the"joint nodes", each of which connects three or more streams, followed by adjusting the node positions in the sequence in such a way that the total of weights of counter-streams decreases after the adjustment, one quickly reaches the optimal set ef tear streatns and the final node sequence, which determines the computation order in simulation. The decomposition procedure and its advantage are demonstrated by seven examples commonly referred to.