Investigation On Damage Detection Of Aquaculture Net Based On Digital Twin Technology

Fishing nets are the critical component for marine aquaculture facilities,and their main functions are to delineate the range of fish activities,participate in water exchange,and provide a reliable habitat for aquaculture fish.However,under the action of offshore high sea conditions,the net is prone to be torn and damaged.If the damage of fishing nets cannot be detected in time,a large number of farmed fish will escape,which will bring huge economic losses to aquaculture farming.Therefore,it is of great significance to carry out damage detection of deep-sea aquaculture nets.Due to the low efficiency and high risk of current manual detection methods,it is urgent to develop a digital detection method that can replace manual work.At present,there are three main methods for digital damage detection of fishing net:the embedding detection method,the sonar detection method and the image analysis method.However,these methods are limited by the interference of the fishing net itself or the environment,and are not universal.In view of this,this paper proposes a damage detection method fishing net based on digital twin technology,which uses sensors to replace manual damage detection of fishing net.The method first obtains a large amount of simulated sensor data from the numerical simulation model of the fishing net,then it applies the data to the training and testing of the artificial neural network.Finally,it trains and generates a digital twin model that can detect the damage of fishing net.The digital twin model can judge whether the fishing net is damaged or not based on the data monitored by the sensor.Using this digital twin model,this paper conducts damage detection studies on two-dimensional(net panel)and three-dimensional structures(net cage).The contents involved are as follows:Firstly,a digital twin model is built.The construction of the digital twin model is mainly divided into two parts:the numerical simulation and the artificial neural network training.In the numerical simulation,various wave conditions as well as the damage of fishing net are considered to fully obtain the simulation data set.In the artificial neural network training,the significant wave height H_s,spectral peak period T_p and sensor data are used as input variables,and the complete state and damaged state of the net are used as output variables.The data used to build the digital twin model needs to be comprehensive,and the sea condition,damage level and damage locations of fishing nets/cages need to be taken into account.Secondly,damage detection is performed on the net panel.In this part,the influence of sea condition,damage level and locations,and wave-current angles on the detection results are fully considered.The hydrodynamic force of the net panel is monitored by the tension sensor installed in the horizontal and vertical lines,and the sensor data is input into the digital twin model for damage detection.After testing and analysis,the average accuracy of the digital twin model in identifying whether the net panel is damaged is greater than 94.32%.The test results show that the model can perform damage detection for different damaged nets under different sea conditions.Finally,the damage detection of the net cage is carried out.This section studies only one damage situation,but considers different sea conditions.The motion response of the complete and damaged cages is monitored by the acceleration sensor installed on the collar of the floating cage,and the sensor data is input into the digital twin model for damage identification.After testing and analysis,the average accuracy of the digital twin model to identify whether the nets is damaged is 98.69%.The test results show that the detection method is also feasible in the damage detection of net cage.Overall,the digital twin model proposed in this paper can accurately detect the damage of fishing nets/cages,which provides a new solution for the damage detection of deep-sea cages.