Smart Structural Damage Monitoring For Jack-up Platform Based On The Vibration Information

Serving in severe sea condition,offshore platform is easy to endure structural components material aging,and the factors including environmental corrosion,sea biofouling,pile foundation scour,fatigue damage and damage expansion affect the security,reliability and durability of the offshore platform.Among many types of offshore platform,jack-up platform is widely used as a result of less steel,low cost,stable working ability and mobility within a certain depth.Nowadays only annual survey is offered to the offshore platforms,while during the survey off,the structural security is unknown,so it's necessary to develop a real-time intelligent monitoring system for the jack-up platform to keep tracing the safety of the main structure,fixing the structural damage and take early warning,assist timely maintenance and guarantee the drilling operation safely.The paper establishes an intelligent model based on the deep learning algor ithm and the vibration information to monitor the structural healthy and invert the damage location of jack-up platform.The main research developments are summed up as follows.(1)Research on the structural damage identification and heath monitoring for offshore platforms is did,a comparison of advantages and disadvantages between the existing damage identif ication algorithm is made and the importance of monitoring for jack-up platform is expounded.(2)The optimal placement of the sensors in the vibration test were studied,and the optimization principle were Modal Assurance Criterion and Fisher Information Matrix,while the allocation algor ithm were cumulative method based on QR decomposition and genetic algor ithm based on integer coding.Finite element model of laboratory jack-up was established by ANS YS,and the vibration results of modal analysis were taken as the input data of the calculation to set sensor optimal allocation of the laboratory jack-up model.(3)The deep learning theory was studied for feature extraction and has superior performanc e in the field of image recognition and speech recognition.Based on deep learning theory,the paper constructed the damage intelligent reversion model of the jack-up platform using BP neural network,the stacked auto-encoder network and convolution neural networks.Based on the modal analys is and transient dynamic analys is of finite element,modal vibration mode and vibration information were extracted from jack-up platform containing spud crack and local stiffness damping on pile legs.The information wer e taken as the input data for the intelligent reversion modal,and the applicability and accuracy of damage identification was compared during several kinds of intelligent network.(4)The damage identification experiment system of jack-up platform was established.Artificial damage such as spud crack,locking device looseness,lack of lateral plate and looseness of leg connection were made according to common forms of damage on jack-up platforms.Vibration signal were collected after hammering excitation on the damaged platform.Deep learning model was used to identify the structural damage state of platform and the applicability,accuracy and generalization performance were compared between several different intelligent algorithms.(5)Composed of data acquisition system,incentive system and real-time monitoring software,the structural health intelligent monitor ing system was set up based on Vibration system and deep learning algor ithm.The software was programed with VB,and called structure damage intelligent identification model in MATLAB.The software contained basic information management,sensor optimal placement,real-time monitoring,system management and exiting system realizing sensor optimal placement on jack-up platform and structural health real-time monitor ing based on deep learning damage model and acceleration information input.