Fan vibration sensor fault diagnosis system and method

Abstract

The invention provides a fan vibration sensor fault diagnosis system and method. The system comprises a sensor, a processor and an upper computer, which are in communication connection. The sensor isused for collecting output data of a fan and selecting feature signals from the output data, wherein the feature signals comprise pneumatic torque, output power of an electric generator, low-speed shaft speed and high-speed shaft speed; the processor is used for carrying out normalization processing on the feature signals to obtain normalized feature data; and the upper computer is used for carrying out analysis on the normalized feature data through an established target neural network model, and outputting fault type corresponding to a vibration sensor of the fan, wherein the target neural network model is a BP neural network obtained by training training sample data, the training sample data comprising the feature data and fault tag corresponding to the feature data. The system and method can effectively improve efficiency of fault diagnosis of the fan vibration sensor fault and improve accuracy of the diagnosis result.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis system and method for a fan vibration sensor. Background technique [0002] The structure of the wind turbine transmission system is complex, the operating environment is harsh, and the external load is variable, which is easy to cause damage to the components of the unit, and even cause the failure of the entire unit. Therefore, it is necessary to diagnose the fault of the wind turbine. [0003] At present, scholars at home and abroad use neural networks to diagnose wind turbine faults. This method has good nonlinear fitting capabilities and can fit more complex working conditions of wind turbines. The objective function is processed, and it has excellent global optimization ability. [0004] However, most of the traditional neural network methods only improve the feature extraction when the wind turbine is faulty, and the network is easy to fall into ...

AI Technical Summary

Problems solved by technology

[0004] However, most of the traditional neural network methods only improve the feature extraction when the wind turbine is faulty, and the network is easy to fall into a local minimum, so that the expected effect cannot be achieved in the fault diagnosis of the wind turbine.

Images