Development Of Petrochemical Pump Room Equipment Abnormal Detection And Disposal System

As industrial robots gradually become more intelligent,more and more industries choose robots instead of manual operations.Robots can not only help enterprises improve production safety,but also improve production efficiency and reduce management costs.In the petrochemical industry,production activities are mostly carried out under high temperature,high pressure,flammable and explosive environments.In order to ensure production safety,abnormal detection and emergency operations of equipment in the pump room are particularly important.The abnormal detection task of the equipment in the pump room mainly includes two aspects:reading the indication of the industrial pointer meter to judge the internal medium state of the equipment;detecting the abnormal area of the equipment surface temperature to determine the equipment operating state.The main purpose of the emergency treatment measures in the display pump room is to allow the robot to replace the person to perform the valve screwing operation in an abnormal state,etc.,to control the on-site state in time to prevent the danger from further spreading.This paper focuses on the problems of abnormal detection and emergency treatment of equipment in the pump room,and solves the problems of the accuracy of abnormal detection and the stability of emergency operations.The main work is as follows:First of all,the research on the reading of the instrument representation number in the pump room environment was carried out.The instrument image data set under the pump room environment is constructed by myself,and the features of the grayscale change of the image are used to construct the features.A feature screening method based on the difference in feature distribution between positive and negative samples is proposed to remove some redundant features.Then,based on integrated learning,a dial area target detection algorithm is established to solve the interference caused by the pump room environment background to the meter pointer detection.Compared with the existing algorithms,the feature screening method proposed in this paper can effectively reduce the problem of overfitting,improve the detection accuracy of the algorithm and improve the detection speed.A dial image pointer recognition algorithm based on geometric detection is proposed,the pointer deflection anglle is calculated,the positioner indicates the number,and the algorithm detection result is compared with the manual interpretation result to verify the accuracy and stability of the detection algorithm.Then,based on the coupling of the RGB image and the temperature image,an algorithm for detecting the temperature abnormality of the pump room equipment is studied.The RGB image and temperature image training sets of the equipment in the pump room are constructed,and the target equipment is marked.A method for extracting the distribution features of the target area based on the Gaussian mixture model is proposed to achieve the initial segmentation of the image to be detected and avoid manual intervention in the image segmentation process.Further,taking the pixel points of the image to be detected as the network nodes,the probability that the pixel points belong to the target area and the similarity between the neighboring pixels are the edge weights,the graph maximum flow segmentation is performed,and the image is divided into the target area and the background area.Correspond to the segmentation result in the temperature image,extract the temperature data of the target device surface,and identify the abnormal temperature area.The experimental results show that the temperature anomaly detection algorithm proposed in this paper can identify the abnormal surface temperature area of the target device,and has good accuracy and robustness.Finally,the emergency operations such as the screwing of the mechanical arm valve in the pump room environment are studied.A valve handwheel positioning algorithm based on machine vision is proposed to identify the position of the valve handwheel in the robot's field of view,and at the same time generate the target posture for the screwing operation of the mechanical arm.A valve screwing strategy based on position impedance is proposed,and the end of the manipulator arm is made flexible by the impedance controller,and the position of the manipulator arm to the axial displacement of the valve during the screwing process is tracked.A mechanical arm valve screwing experimental platform was built to verify the accuracy of the valve positioning algorithm and the effect of the impedance control system on the position tracking and force tracking of the valve.