Research On The Wall-climbing Robot's Intelligent Fault Diagnosis System Based On Neural Network

Wall-climbing robot is a kind of robot for limit environment work. Due to the specificity of its working environment, safety and reliability of its working conditions becomes a very important problem. So it is significant to carry out fault detection in time. A kind of intelligent fault detecting method based on neural network is proposed and thoroughly studied in this thesis. And an intelligent fault detecting system is established.Starting from analyzing the possible faults existed; relations between the fault patterns of the wall-climbing robot and the characteristic signals are established. So the problem of fault detection of the robot is in essence a problem of pattern recognition. And based on this work, an intelligent fault detecting method is discussed.A fuzzy Hamming net is mainly studied. And during the application of this net, two deficiencies are exhibited. One is that the only tool to find the best parameters is by human experiences and it is also difficult to guarantee that these parameters are the best. The other is that when the extending areas of the samples overcross, wrong classification of the samples will occur. As for the first problem a genetic algorithm is used to improve the process of the best parameters' finding. And as for the latter a kind of improved Hamming net which uses supervised and unsupervised learning method is employed. These two methods are all tested and better results are got.A kind of two-step neural network based on the improved fuzzy Hamming net is proposed. And an intelligent fault detecting system of the robot based on this network is established in this paper. An experiment of the foot's lifting-and-dropping system of the robot is done. And the result shows that the fault detecting system can achieve the fault-detecting task successfully.