Research On Fault Diagnosis Of The Control System Of Hydraulic Drilling Machine Based On Blind Source Separation

With the development of mechanical, electrical and hydraulic integration, the structure of drilling rig is becoming more and more complex, which leads to the increase of the type of fault that may occur in the drilling rig, and the possibility of the simultaneous of multiple faults. For all hydraulic drilling rig, the structure of the system is closed. In the case of single fault, there is a fuzzy correspondence between the fault phenomenon and the failure causes. In the case of multiple faults, there is a coupling relationship between different faults. A sensor of a certain point will detect the information of multiple fault sources, and the fault feature is not a simple superposition of every fault feature, so it is difficult to analyze and deal. Therefore, the difficulty to apply the diagnosis theory and methods for single fault in the multiple faults diagnosis will increase greatly, even the diagnosis theory and methods can't play a role. Aimed at the multiple faults diagnosis of drilling rig, seek an effective signal feature analysis method to comprehend the working state of the drilling rig timely, can not only ensure the safety work of drilling rig, but also avoid the fault occurs in the rig, and reduce the use cost.This paper regards the full hydraulic drilling rig as the research object. On the basis of the rig structure, there is an analysis on the failure modes and failure mechanisms of the main components of drilling rig, and it has carried on a comprehensive analysis of all kinds of fault features to determine the drilling rig multi fault diagnosis method.Based on the research background of the signal processing method-blind source separation, this paper briefly introduces the source of the problem of blind source separation and its mathematical model. On the basis of the analysis of various assumptions and uncertainties, two kinds of algorithms of blind source separation based on statistics and information theory are analyzed. The JADE algorithm and the maximum signal to noise ratio algorithm are used in the simulation of the mixed signal separation experiment, and the separation effect is compared and analyzed.Because of the high similarity between the mathematical models of independent component analysis and blind source separation, independent component analysis has become a kind of main algorithm in the blind source separation. On the basis of the fast independent component analysis method, aimed at the complexity of Jacobi matrix derivative calculation in the objective function, this paper proposes that use secant method instead of the Newton method for the optimal solution of the objective function to reduce the complexity of the original algorithm. Finally, the drilling rig experimental platform is constructed to make experiments. By testing and separating the vibration signals of the reducer and the motor, the effectiveness of Secant-Fast ICA algorithm in the separation of the full hydraulic drilling rig fault signals is verified.