|
The rapid development of modern automotive technology, people are increasingly demanding high quality cars, car noise, vibration, and comfort that NVH (noise, Viblalion, Harshness) is a measure of the quality of an integrated car manufacturing index. Noise in the car, the main gear transmission system is a micro car key component assembly, gear reducer as the main important parts, the vibration noise is the main noise sources. Gear vibration noise is an objective reality, but if the gear noise is too large, the quality of the car safety and security, while polluting the environment, affecting people ride comfort. Currently, the gear fault diagnosis, business practices, mostly by artificial hearing the voice of gears to judge the quality of the gear, the higher the operational requirements for workers, requires experienced master workers can play in this task. Therefore, the information carried on the gear noise, the use of related equipment for the collection, analysis of its characteristics, laws, and its use of modern methods to diagnose computer intelligence research, is a rewarding job. Gear noise to pass out information is messy, non-linear, in order to seek the law of which, the use of artificial neural networks to study it is complementary. Artificial neural network is a non-procedural, adaptability, the brain's information processing style. Especially BP neural network, a simple structure, plasticity, and in fault diagnosis field has been widely used. However, BP network is also very prominent advantages and disadvantages, and its self-learning adaptive ability, fault-tolerant, very suitable for handling complex nonlinear problems, but slow convergence, especially easy to fall into local minima, limiting its performance . Simulated annealing algorithm is suitable for solving large combinatorial optimization problems, the calculation process simple and universal, robust, suitable for parallel processing, especially in the theory proved to be a with probability 1 converge to the global optimal solution global optimization algorithms. Using simulated annealing algorithm to improve BP neural network, BP neural network to overcome local minima easily fall into the shortcomings, to further improve network performance. This article describes the gear fault diagnosis related research, specifically addressed the BP neural network and simulated annealing algorithm, its advantages and disadvantages and improved methods are analyzed, and the integration of simulated annealing with BP neural network diagnostic information processing gear noise and the diagnosis results are analyzed. The results comparison shows that the integration of simulated annealing BP neural network has better performance, higher precision gear fault diagnosis.
|