Research On Inference Technology Of Bearing State Of Offshore Platform Based On Real-time Monitoring Data

The bearing state of offshore platform studied in this paper represents the mass of the topside.The mass of offshore platform is a very important parameter in the process of design,construction,installation and operation,which has a very important impact on structural safety,structural health monitoring and offshore platform integrity management.Most of the existing researches focus on the change of structural mass and local equipment mass,and rarely involve the real mass of the structure.Therefore,based on the principle of structural dynamics,this paper has carried out a series of exploration and trial work to establish the real mass inference method of offshore platform.Firstly,based on the principle of structural dynamics,the basic idea of estimating structural mass from acceleration and bending strain response is proposed for a single column structure.Aiming at a simple frame structure,this paper discusses the technical obstacles of applying this principle directly.On this basis,an equivalent method of acceleration and bending strain response of frame structure to those of single column structure is proposed,and a semi explicit inference method for the mass of frame structure is established.Through the response data of simple frame structure under different random wave loads,the robustness of this method to environmental excitation is verified.For a jacket platform with complex configuration in Bohai Sea,the response equivalent mode and direction angle calibration method involved in the inference method are discussed.Based on the dynamic response data of the jacket finite element model,the mass inference results of different response equivalent modes and direction angle calibration methods are compared,and a semi explicit mass inference method which is more suitable for jacket platform is finally determined.On this basis,the practical application limitation of this method is further discussed.The method can be used to infer the mass of jacket topside,and the recognition result is reliable under the condition of wavelet height with the highest probability of occurrence.In order to break through the practical application limitation of semi explicit inference method,the neural network machine learning algorithm is introduced to explore the implicit inference method for the mass of offshore platform.The harmonic response analysis of the structure is carried out to obtain the training samples.A variety of options of hidden layer,activation function and loss function involved in the construction of the network have been tried,and an implicit inference method which can be applied to the dynamic response data of the jacket finite element model is preliminarily determined.The total mass of topside predicted by this method is more accurate than that predicted by semi explicit inference method,and it can preliminarily realize the layered estimation of topside mass of jacket platform.