The Research Of In-pipe Robot Detection Based On Visual Sensor

As an important transmission,pipeline is widely used in oil,natural gas and other substances,and in many areas like petrochemical,construction,etc.The pipeline brings great convenience for human life and the modern industrial and agricultural development,simultaneously.Due to the complex and changing environment where the pipeline worked,pipeline is easy corroded,and the phenomena just like pipeline cracks and gaps can lead a serious accident,which caused great harm to people's lives and property safety.In view of this,pipeline inspection work gets more and more attention.But affected by the special environment,manual inspection is not high efficiency,and the testing results are unsatisfactory.There is an urgent that we need a better way to detect the completion of pipeline.So the new method using pipeline robot to replace manual inspection,it can complete the task quickly and efficiently.Nowadays,with the development and technology of machine learning,artificial intelligence and other areas becoming more and more mature,how to improve the intelligent robots becomes a research hotspot.As a branch of the robot,pipeline robot is a highly integrated mobile carrier,and it can be also equipped with a variety of specialized sensor in order to detect.In a complex pipeline environment,the in-pipe robot can better to complete testing,cleaning and maintenance and other tasks,particularly.This paper is based on the detection of pipeline gaps pipeline robot vision sensor.Firstly,the image of inner wall is collected by the robot video image acquisition system.Then,we use the new algorithm that improved BP neural network algorithm fusion based on genetic algorithm for image segmentation processing.First,we can use the genetic algorithms to search out the approximate range of weight;this value is used as the initial weights of BP neural network for training.Compared with the traditional BP algorithm and traditional BP-GA algorithm,the result of improved BP-GA algorithm is better in image segmentation.Especially,the problem of convergence speed,easy to fall into local minimum value terms has been greatly improved.At the same time,the test results show that the new method can basically solve the problem of resolution pipeline gap;it has laid a good foundation for the further study technology...