Diagnosis and Classification of Diesel Engine Components Faults Using Time–Frequency and Machine Learning Approach

Diagnosis of engine component faults is a challenging task for every researcher due to the complexity involved in the engine operations. The developed faults on the engine components subsequently reduce their performance and cause higher maintenance costs. Hence, an effective condition monitoring technique should be implemented to diagnose engine component faults. Therefore, in this work, potential fault diagnosis techniques are presented to detect and diagnose the scuffing faults developed on the diesel engine components. Condition monitoring techniques such as vibration and acoustic emission analyses were employed to acquire the fault-related signals. These signals were analyzed in the time-domain, frequency-domain, and time–frequency domain using signal processing methods viz. fast Fourier transform (FFT) and short-time Fourier transform (STFT). The statistical feature parameters were also extracted from the acquired signals to diagnose the severity of the faults. Further, the artificial neural network (ANN) models were developed to predict and classify the scuffing faults developed on the engine components. The results showed that the FFT and STFT techniques provide better fault diagnostic information. The developed neural network models have effectively classified the scuffing faults on engine components with an accuracy of 100%.

ANN:

Artificial neural network

\(b_{k}\) :

Bias to ANN

c :

Damping coefficient

C :

Sound velocity (\(\text m/s\))

g :

Acceleration due to gravity (\(\text m/s^{2}\))

FFT:

Fast Fourier transform

f :

Frequency parameter (Hz)

I :

Basic input

\(I_\text{min}\) :

Minimum value from input

\(I_\text{max}\) :

Maximum value from input

\(I_\text{norm}\) :

Normalized value

k :

Stiffness of system

MLP:

Multilayer perceptron

m :

Mass of the system (kg)

N :

Total number of samples

\(O_{i}\) :

Predicted output value

STFT:

Short-time Fourier transform

S :

Emission area (\(\text m^{2}\))

t :

Time parameter (\(\text s\))

\(T_{i}\) :

Target value

\(u_{k}\) :

Output at neuron

\(\bar{V}\) :

Spatial mean of vibration velocity

\(W^{*}\) :

Windowing function

\(w_{kj}\) :

Weights assigned to neuron

W :

Sound pressure level (dB)

\(x_{i}\) :

Measured data

\(x_{j}\) :

Input to ANN

\(\bar{x}\) :

Mean value of the signals

x(t):

Time-domain data

x :

Displacement (\(\text m\))

\(\dot{x}\) :

Velocity (\(\text m/s\))

\(\ddot{x}\) :

Acceleration (\(\text m/s^{2}\))

\(y_{k}\) :

Output at neuron

\(\tau\) :

Time variable

\(\omega\) :

Rotational frequency (\(\text rad/s\))

\(\sigma\) :

Standard deviation

\(\phi\) :

Activation function

\(\delta\) :

Downward displacement (\(\text m\))

\(\rho\) :

Specific mass (\(\text kg/m^{3}\))

\(\sigma _{rad}\) :

Radiation efficiency

Authors and Affiliations

1. Department of Mechanical Engineering, Internal Combustion Engine Laboratory, PDPM Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, 482005, India

   Sangharatna M. Ramteke & H. Chelladurai

2. Department of Mechanical Engineering, G.H. Raisoni College of Engineering, Nagpur, 440016, India

   Sangharatna M. Ramteke

3. Department of Mechanical Engineering, Tribology and Machine Dynamics Laboratory, Indian Institute of Information Technology Design and Manufacturing Jabalpur, Jabalpur, 482005, India

   M. Amarnath

Corresponding author

Correspondence to Sangharatna M. Ramteke.

Conflict of Interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Reprints and permissions