Research On Dynamic Response Catastrophe Forecasting Method For Mega Floating Platform

Mega floating platform is state-of-art equipment emerged in marine engineering for demands of national defense and resource exploration.It can be viewed as a nonlinear dynamic network system based on the characteristics of multi-modular configuration,rigid-flexible connection and fluid-solid interaction.This paper aims at the development of a new prediction method for catastrophes of the floating platform by using the complex network theory.The floating platform consists of multiple modules that are serially linked by elastic connectors,a typical non-autonomous network system.Due to the geometric nonlinearity of the connectors,the system illustrates strong nonlinear characteristics,including the amplitude death,jumping phenomena and synergistic effect.Jumping phenomena are the events where motions of floating modules may evolve into large oscillations rapidly.Large displacements between adjacent modules may provoke overloads on connectors,which causes the system to fall apart resulting in catastrophic accidents.In order to prevent from the occurrence of catastrophes in such super-scale floating systems,we study the prediction method of catastrophes based on the floating platform.Numerical simulation works illustrate that the responses of the floating body appear wave-frequency dependent oscillations,a state of amplitude death,which means the system is stable;while the responses of the floating body appear chaotic or sub-harmonic motions that lead to catastrophic events.This feature will be used to develop a network-based prediction method for the catastrophe incidence.In this paper,a symbolic network-based method is proposed for the prediction of catastrophes of the floating platform.Firstly,a time series of sampling data of system responses are collected and mapped into symbolic sequences by a coarse-grained method.Then the symbolic sequences are used to form the nodes of network and the links between nodes are determined by chronological order.The topological properties of the symbolic network will be employed to identify the sign of catastrophes.To improve the prediction efficiency,the parameters for symbolic mapping and nodal configuration are investigated through the comparison studies of the corresponding networks derived from different response time series.Topological properties include the average path length and the average node degree of the symbolic network,which can be used to formulate the catastrophic indicator of dynamic responses of floating modules.For the floating system,the property of the average path length outperforms over the property of the average node degree as the prediction index of disasters.It provides an obvious early warning sign before the occurrence of catastrophes of the floating platformSince the floating platform is composed of multiple modules and each module has multiple degrees of freedom,thus there are flexible options available for the collection of response time series.In order to make full use of the time series from different sampling sources as well as the feature of synergetic effects of network systems,a recurrence network method based on the average distance of adjacent nodes is proposed.The network structure is determined by the characteristics of node distribution,which can better reflect the dynamic response characteristics of multi-dimensional time series.Moreover the connection of nodes is limited,so that the method is effective for the analysis of wave-frequency dependent time series.The numerical results show that the performance of the complex network based on displacement-velocity two dimensional time series is better than that of the symbolic network based on one dimensional time series.