Research On Load Identification Method Of Rotor System Based On Motor Current

In industrial production, rotor system as the core component to drive the mechanical operation, is widely used in the steam turbine, aircraft engines,centrifugal pumps, industrial compressors and other rotating machinery and equipment. These equipment are used by people’s production and life, it plays an indispensable role in the production practice. Due to the extensive use,making in different equipment and different environment, the load of the rotor is complicated and changeable, and it is often subjected to different types of loads,such as steady state, transient state, impact, linear, nonlinear, etc.. A large number of engineering practice shows that the load condition of the rotor system will directly affect the strength and life of the rotor system, once the size of the load beyond the range of acceptable to the rotor system, the consequences will be disastrous. Meanwhile, the fault condition of the whole rotating machinery is also related to the load of rotor system, thus to find the changing regularity of the load of a rotor system have great significance to ensure the safety, high efficiency and long-term operation of the rotor system and the whole rotating machinery, thus it can avoid the occurrence of accidents and economic losses. To address this problem, this paper combined with the load identification technology and motor current monitoring technology to studied the rotor system load identification method based on motor current.Firstly, the paper described the research background and significance ofrotor system load identification, then discussed the research trends of the current domestic and international load identification technology as well as motor current diagnosis and monitoring technology.Secondly, the paper established the coupling mathematical equation of the single span rotor bearing system under the mt coordinate of the motor and the electromechanical coupling equation of rotor system of drum hoist. It obtained the time domain and frequency domain variation of the motor current under different load conditions by building the simulation model in simulink and loading different types of load in the model, so as to revealed the relationship between the load on rotor system and motor current. In order to further access to the information about the load from the motor current signal, it was proposed genetic programming optimization of time domain characteristic, the EMD empirical mode decomposition and Hilbert modulus are three effective methods for feature extraction. The purpose of load information mining in current of the rotor system under different load excitation is realized.On the basis of this, identified five kinds of loads from qualitative and quantitative aspects. In the qualitative identification, the identification method based on the wavelet packet transform and the equivalent DC of the motor stator was proposed. This method transformed the three-phase stator current into the equivalent direct current, and deal with the DC current by the wavelet packet transform, thus the type of the load can be identified in the wavelet coefficients.In the quantitative identification, two feasible methods were put forward. The first method used the electromechanical coupling model of the asynchronous motor-rotor system, and reversed the load which act on the system by the current after modification of the equation. On the basis of known load types, the second method used support vector machine regression method and the load feature extraction method achieved the purpose of quantitative identification of load.Finally, this paper designed the rotor loading test system and make a reasonable experimental plan in view of the experimental conditions andrequirements, according to this plan, the experiment of rotor system under different loads is finished. To analyzed the results of the experiment as the basis,modified the asynchronous motor-rotor system, and verified the correctness and practicability of methods of load qualitative identification and quantitative identification.