Loader Seat Vibration Based On Neural Network Critical State Recognition

Loaders are a large number of construction machinery used in production.Due to the heavy overall weight,large load,and harsh working environment of the loader,it is difficult to guarantee the comfort of the construction machinery.How to improve the ride before leaving the factory Comfortable fault detection methods,improvement of detection level and detection efficiency,and realization of unmanned automatic offline detection have become the main concerns of manufacturers.Affecting the comfort of loader operators comes from three aspects: the operating noise and vibration of the loader,and the environmental noise of the work site.The operating noise and vibration of the loader will affect the comfort of the operators in the loader cab.This article studies the vibration state and vibration transmission path of a certain wheel loader seat of a certain manufacturer,in order to combine the traditional discrimination method with artificial neural network,replace the existing manual detection,judge the comfort by "experience",and improve the detection efficiency.And accuracy rate,to provide reference analysis ideas and methods for similar research.Aiming at the discomfort of the operator caused by the seat vibration of the loader,a seat vibration state recognition process and model based on the vibration signal and neural network are established,and the model is verified in practical applications.The design experiment plan collects the vibration signal of the loader under different working conditions,and establishes the status label for the signal through the subjective evaluation of the seat comfort by the operator’s trial ride.The time domain,frequency domain,and time-frequency features of the vibration signal are extracted as signal feature parameters,and the signal feature vector is constructed with reference to the parameter weighted acceleration recommended in the national standard GB/T 13441.1.By combing the theories and methods of neural network,construct the BP neural network model and the convolutional neural network model improved based on Alexnet,and use signal features as input to train and test them.Among them,the classification accuracy of BP network reaches more than 97%.The accuracy rate of the improved convolutional neural network reaches more than 98%,which verifies the rationality of the neural network model to replace manual detection in the recognition of seat vibration state.Finally,the seat vibration critical state recognition model is used for the factory inspection of the loaders,and the vibration state of the seats of the three loaders are classified.By collecting the up and down vibration signals of the suspension,the integrated value of the up and down vibration signals of the power train suspension and the cab suspension is calculated,and the vibration attenuation rate of the suspension is obtained.Determine whether it meets the factory requirements by referring to the factory standards,and obtain the vibration state of the loader seat.By comparing the classification results and calculation results of the seat vibration state recognition model,the effectiveness of the neural network pattern recognition is verified.