Using MLP‐GABP and SVM with wavelet packet transform‐based feature extraction for fault diagnosis of a centrifugal pump

Abstract

This paper explores artificial intelligent training schemes based on multilayer perceptron, considering back propagation and genetic algorithm (GA). The hybrid scheme is compared with the traditional support vector machine approach in the literature to analyze both fault and normal scenarios of a centrifugal pump. A comparative analysis of the performance of the variables was carried out using both schemes. The study used features extracted for three decomposition levels based on wavelet packet transform. In order to investigate the effectiveness of the extracted features, two mother wavelets were investigated. The salient part of this work is the optimization of the hidden layers numbers using GA. Furthermore, this optimization process was extended to the multilayer perceptron neurons. The evaluation of the model system performance used for the study shows better response of the extracted features, and hidden layers variables including the selected neurons. Moreover, the applied training algorithm used in the work was able to enhance the classifications obtained considering the hybrid artificial intelligent scheme been proposed. This work has achieved a number of contributions like GA‐based selection of hidden layers and neuron, applied in neural network of centrifugal pump condition classification. Furthermore, a hybrid training method combining GA and back propagation (BP) algorithms has been applied for condition classification of a centrifugal pump. The obtained results have shown the good ability of the proposed methods and algorithms.

A bearing fault diagnosis approach based on an improved neural network combined with transfer learning

“…This kind of methods needed a lot of human energies, and the artificially chosen hyperparameters may not be suitable for the current Unlike empirically manual tuning, this work employed the program auto-tuning technique in an attempt to improve the efficiency of model tuning through customized parameter range, random parameters selection and early termination of poor performance. The selection range of learning rate was set as [1 × 10 −6 , 1 × 10 −3 ], and the batch size was chosen among [8,16,32,64]. Forty sets of hyperparameters were randomly selected from them for training, and the upper limit of training times is 50.…”

“…This method achieves fault diagnosis by collecting bearing vibration signals and performing their feature extraction, analysis and identification. The classical methods include root mean square [5], crest factor [6], fast Fourier transform [7], wavelet transform and wavelet packet transform [8], etc. The main advantages of the methods are that they can be used for noise reduction and feature extraction with no need of actual mathematical modeling [9].…”

Multi-Objective Optimization Design of Low Specific Speed Centrifugal Pumps Based on Genetic Algorithm

“…At present, most of the methods adopted by researchers for optimizing the design of centrifugal pumps still remain semi empirical and semi theoretical. Common optimization methods include similar design method, velocity coefficient method, and area ratio method [26]. The hydraulic performance of centrifugal pump models designed through such methods usually cannot reach the optimal level.…”

TLSCA-SVM Fault Diagnosis Optimization Method Based on Transfer Learning

“…Training using analog circuits for diagnostic models in fault diagnosis is more valuable. With the continuous refinement of various data-analysis methods, the troubleshooting technology for analog circuits is also improving; common troubleshooting techniques, including PCA [1], Search Grid [2], particle swarm optimization (PSO) [3], ant colony algorithm (ACA) [4], simulated annealing (SA) [5], genetic algorithm (GA) [6], Back Propagation Neural Network (BP) [6], Self-organizing Maps (SOM) [7], Extreme Learning Machine (ELM) [8], decision tree [9], random forest [10] and SVM [11] all have good classification results to some extent [12].…”