| In the process of industrialization in our country,rotating machinery is more and more widely used in large and complex industrial equipment.Timely and accurate fault diagnosis is the key to ensuring the safe and reliable operation of rotating machinery.Therefore,for the high-noise vibration signals of rotating machinery,this paper proposes a fault diagnosis framework based on the Ensemble Empirical Mode Decomposition(EEMD)and Output Input Hidden Feedback Elman(OIHF Elman)Adaptive Boosting-Bootstrap Aggregating(Ada Boost-Bagging)dual ensemble algorithm.In this diagnosis framework,firstly,the EEMD is used to denoise,decompose and reconstruct the original vibration signal,so as to extract the effective fault features in the time domain as the input vector of the diagnostic model.In terms of fault pattern recognition,this paper studies the OIHF Elman neural network as a weak learner,combining the Ada Boost ensemble algorithm with the Bagging ensemble algorithm to obtain the OIHF Elman Ada Boost-Bagging dual ensemble algorithm,while ensuring the generalization and stability of the diagnostic model.Among them,the OIHF Elman neural network is constructed by adding a dual feedback structure from the output layer to the hidden layer and the input layer based on the Elman neural network.This dual feedback structure improves the neural network model’s ability to process the timing characteristics of vibration signals.In addition,in order to effectively ensure the validity of the model diagnosis output,this paper defines the model output as a multi-dimensional vector and outputs the probabilities that the sample belongs to each fault mode.In order to apply this output form,this paper improves the Ada Boost ensemble algorithm.Firstly,a novel Ada Boost regression algorithm is obtained by combining the Ada Boost two classification algorithm,which is verified in the fault diagnosis of rolling bearing.In order to further improve the wide applicability of the Ada Boost ensemble algorithm,the fixed threshold in the Ada Boost ensemble algorithm is changed to an independent variable,that is,the average diagnostic error of the weak learner and the weight update coefficient of the sample is changed to adapt to the change of its threshold.In order to verify the effectiveness of the OIHF Elman Ada BoostBagging dual ensemble algorithm model,this paper conducts numerical experiments on the fault datasets of rolling bearing and planetary gearbox.Experiments prove that the proposed OIHF Elman Ada Boost-Bagging dual ensemble algorithm obtains better generalization ability and stability in fault diagnosis of rotating machinery equipment,which providing new solutions and tools for fault diagnosis of rotating machinery equipment. |
