تشخيص عيوب فن سانتريفيوژ با استفاده از يادگيري عميق و تركيب سيگنال‌هاي ارتعاشي و صوتي

Centrifugal fan fault detection using a deep learning an‎d combined sound& vibration signals

مهندسي مكانيك

Abstract Due to the high repair costs an‎d technological advancements, the countryʹs industries have taken positive approaches to the science of preventive maintenance, an‎d as a result, CM has expan‎ded significantly in Iran in recent decades. Industrial fans play an important role in production facilities, an‎d the sudden shutdown of critical fans can cause significant disruptions in production. Ensuring early, effective, an‎d accurate detection of fan failure requires understan‎ding the characteristics an‎d anomalies resulting from initial damage to rotating machinery. Until now, most existing fault diagnosis techniques have been proposed with the condition that only one fault occurs at a time. However, in industrial applications, compound faults are more common due to the connection of different components. Accurate diagnosis of compound faults is of great importance for the safe operation of industrial equipment. In this study, the diagnosis of centrifugal fan faults using a deep learning system an‎d the combination of vibration an‎d acoustic signals is investigated in order to improve the accuracy of fault diagnosis an‎d better analyze the performance of centrifugal fans. The experiments are performed on a Dedusting Fan with a capacity of 130,000 cubic meters. Both single an‎d compound faults are considered in the bearings an‎d shaft, electric motor an‎d impeller. By designing an‎d manufacturing a vibration an‎d acoustic sensor an‎d placing it in different parts of the fan, the initial data is received an‎d entered into a deep neural network for processing. .Also, different deep neural networks such as NARX, GRUs, LSTMs an‎d hybrid methods using CNN neural network have been investigated an‎d compared. The experimental results show the effectiveness of the proposed method. Keywords: Deep learning, Vibration Signal Analysis, Audio Signal Analysis, Intelligent Fault Diagnosis, Centrifugal Fan.

6 فصل

123186