Optimized stochastic resonance method for bearing fault diagnosis

Modern machinery and equipment are moving in a large, complex, and high-speed direction. Machinery and equipment typically runs in a strong noise background and it is difficult to detect incipient faults through vibration analysis. It has been an important problem for fault diagnosis to extract the weak fault signals from a strong noise environment. Stochastic resonance (SR) is a phenomenon where a signal that is normally too weak to be detected by a sensor can be boosted by adding white noise to the signal, which contains a wide spectrum of frequencies. Therefore, SR can converse noise energy to signal energy, and then it is commonly used to enhance the signal-to-noise ratio (SNR) of a system output using the unavoidable environmental noise and it is suitable to detect the weak faults of rotary components in modern machinery and equipment. However, the structural parameters of a stochastic resonance system have a great impact on its output, and each input signal will correspond to a set of optimal structural parameters. An artificial bee colony algorithm has been proposed to be a rapid developed optimization algorithm in recent years for its fast convergence speed, high accuracy, and good global search capability. To deal with the actual situation and make an accurate detection for rolling element bearings, a new adaptive stochastic resonance method was developed using an artificial bee colony algorithm and stochastic resonance theory. In order to obtain the maximum stochastic resonance output SNR, the structural parameters of the system has been adaptively optimized by an artificial bee colony algorithm using the SNR as the objective function. ABC is one of the population based algorithms, the position of a food source represents a possible solution to the optimization problem, and the nectar amount of a food source corresponds to the quality (fitness) of the associated solution. The number of the employed bees (in an ABC model, the colony consists of three groups of bees, i.e., employed bees, onlookers, and scouts) was equal to the number of solutions in the population. Based on the method, the input signal could correspond to a set of optimal structural parameters and the weak fault signals were finally detected from strong environment noises. The comparison study between the present ACB-based SR and traditional SR was performed by a numerical simulation signal of cosine function with Gaussian white noise. The result showed that the feature frequency peaks in ACB-based SR were 70 percent higher than those in traditional SR. Finally, experimental investigation of a rolling bearing with an inner race fault in a Machinery Fault Simulator-Magnum (MFS-MG) was performed. Due to the fact that the sampling frequency was 25.6 kHz, the experimental data should been preprocessed by a scale transformation and the scale transformation compression ratio R equaled to 5120 and the compression sampling frequency was 5 Hz. Finally, the fault detection results showed that the presented method was favored to detect and diagnose rolling bearing faults from a strong noise environment. The peak values in the output frequency spectrum of the present method were higher by about 20 percent more than those of the classical SR.