Design Of Inspection Robot And Research Of Image Intelligent Recognition Technology In Multi-layer Gasification Workshop Of Chemical Plant

With the development of the construction process and scale of chemical plants,more and more attention has been paid to the problems of the traditional chemical plant,such as high daily inspection intensity,heavy inspection task,and harmful gases such as CO,H₂S endangering the health of staff in the inspection area,etc.How to accurately and effectively use image recognition technology to identify the status information of instruments and meters in a complex workshop environment has become the focus and key to solving the above problems.This paper aims at the problem of the inspection robot can not climb the stairs,the instrument detection accuracy is low,and the instrument reading accuracy is low in the traditional chemical plants.The image recognition technology and convolution neural network technology are combined to solve the above problems.The research contents are as followsFirstly,according to the field investigation of multi-layer gasification workshop in the chemical plant,combined with the special environmental requirements of chemical plant,the hardware part of inspection robot is designed,including the selection of robot hardware equipment,the setting of patrol track and image modeling operation,the construction of wireless communication network,the design and installation of elevator control equipment and the deployment of the upper computer.On the premise of ensuring the security of patrol inspection,it provides a complete hardware platform for the application of image intelligent recognition technology.Secondly,five methods based on image feature point detection and extraction are studied to solve the problem of low detection accuracy of instruments in the complex workshop environment.Through theoretical derivation and comparison of the feature point extraction experiments of laboratory instrument and workshop instrument image,the performance of the five algorithms is tested.The matching strategy of the ORB algorithm with good performance in all aspects is improved,and the RANSAC algorithm is used to eliminate the wrong matching points,which improves the accuracy and performance of image matching.Then,it analyzes the possible blocking situation of"people,objects,robots"sharing an elevator,and determines whether the robot will take the elevator by judging whether the QR code is recognized,to optimize the"machine elevator interaction"strategy.Through the improved convolutional neural network algorithm to identify and locate the QR code,improving the reliability and intelligence of the inspection robot taking the elevator,so that the robot can inspect each floor of the gasification workshop.Finally,the multi-task convolutional neural network model based on face recognition is applied to the instrument and feature point detection.Using ROI and three feature points as constraints,through the combination of digital identification and positioning technology and distance method,the reading of pointer meter’s data is carried out.