An Interference Optimization – Induced Electrical Turbine Fault

Prediction and Analysis Method

P. SENTHILKUMAR1, KASMARUDDIN CHE HUSSIN2, MOHAMAD ZAMHARI TAHIR3,

T. PADMAPRIYA4, S. V. MANIKANTHAN5

Kallakurichi,Tamil Nadu,

INDIA

Puducherry,

INDIA

Puducherry,

INDIA

for turbine operation using resistance and magnitude of the blades are accounted for each operation cycle. The

classifier performs segregation based on low and high thresholds for predicting failure cycles. Such cycles are

altered using pre-maintenance intervals and mechanical fault diagnosis at an early stage. This prevents turbine

failure regardless of external influencing factors.

Key-Words: - Classifier Tree Learning, Electrical Turbine, Fault Prediction, Threshold Condition, power

generation, operating cycle, failure cycle.

Received: October 11, 2022. Revised: September 26, 2023. Accepted: November 7, 2023. Published: December 5, 2023.

method provides necessary information for the

quality improvement process in wind power plants,

[1]. The structure healthy condition monitoring

(SHCM) technique is used in power plants to

identify the structural aspects of turbines. SHCM

technique gathers information which is related to

electrical turbines that reduce the latency in the

detection process. SHCM technique detects the

healthy condition range of turbines that produce data

for consequence and fault detection processes, [2].

evaluates the vibrations, conditions, working

capability, and behavioral patterns of electrical

turbines. The SVM algorithm method improves the

accuracy of electrical turbine fault detection which

enhances the performance range of the power plants,

[3], [4].

fault detection in wind power plants. The

The fault features are optimized and used for the

fault detection method which reduces the data

dimension level in the computation process. The

PCA-based method reduces the overall optimization

problems in the fault detection process, [6], [7]. A

novel convolutional neural network (CNN) model is

used for fault detection in power plants. The CNN

model uses a feature extraction method that extracts

the important features from turbine monitoring

systems. The extracted features produce the

The study, [9], proposed a fault detection method

for wind turbines based on a backpropagation (BP)

neural network and a pair-copula model. The pair-

copula model is implemented here to analyze the

input variables of turbines. Real-time data is

produced via a pair-copula model which provides

relevant data for the calculation process. The BP

extraction technique is used in the method which

extracts the important features which contain defect

signatures. The designed method analyses the data

which are provided by extraction and enhances the

effectiveness level of the systems. It is used as a

defect diagnosis that reduces latency in the fault

detection process. The designed method improves

the performance and feasibility range of wind

turbines.

artificial neural network (ANN) based operating

characteristics prediction approach. The main aim of

the approach is to predict the characteristics of gas

turbine combustors. ANN is mainly used here to

predict the root mean square error (RMSE) ratio in

wind turbines. RMSE contains optimal operating

characteristics for further processes. Experimental

address the critical issues in wind turbines. SCADA

is used to identify the abnormal conditions,

problems, and threats that are presented in turbines.

SCADA increases the energy-efficiency level of the

systems, [13]. The developed model improves the

effectiveness and significance range of wind

turbines, [14].

method for fault detection in wind turbine systems.

Firefly algorithm (FA), chaos map (CM), and

high-performance range in wind turbine systems.

cyclostationary blind deconvolution (ACYBD)

maximizes the accuracy in fault detection which

improves the flexibility range of wind turbines.

decomposition scheme using multi-variable

correlation extraction for wind turbines. The

designed scheme is a wind turbine fault detection

method that identifies the effective faults in

turbines. An auto encoder is used in the method

which detects the faults based on the frequencies.

The actual behavioural patterns of the turbines are

analyze the type-I fuzzy logic controlled (FLC)

faults in turbines. The actual power loss reason is

identified via a mathematical model that reduces the

energy consumption in the computation process.

voltage profile of turbines. The proposed model

improves the significance and reliability range of

wind turbine systems.

The proposed fault prediction method is designed to

maintain smooth operation and consistent power

generation output in electrical turbines. Figure 1

presents an illustration of the proposed prediction

method.

Fig. 1: Proposed Prediction Method Illustration

compute the resistance and magnitude of the blades

for each operation cycle. The deployed operation

cycles are responsible for maintaining consistent

operation from the active electrical turbine. The

input operation is based on minimum and maximum

threshold conditions for identifying failures. In this

 

󰇡

󰇡

󰇡



The above equation (6) and (7) computes the

final power generation output for predicting failures

following the maximum threshold. Here, the

threshold condition is the uncertain measure, for

which accurate and appropriate computation is

required for the diagnosis of mechanical faults.

threshold condition. The classification learning

process is illustrated in Figure 2.

identified for thresholds of max and min. These

classifications respond to the next cycle or operation

above sequence of segregating the operating cycles

for the turbine is pursued using classifier tree

classifier tree learning to identify and segregate the

minimum and maximum threshold conditions for

accurately predicting failures. The data from, [19],

provides a power generator output, verified at 10

min intervals. The active power and theoretical

prediction are analyzed based on wind speed (m/s)

estimated as in Figure 3.

for segregating min and max for which new cycles

distinct threshold features. The failing feature

meeting the power demands. Therefore the

the related works section along with the proposed

method.

Fig. 4: Fault Prediction

turbine based on threshold condition is improved.

Using the proposed method, the inferred

optimization is performed using classifier tree

learning. The proposed method used for rectifying

the turbine failures or altered operating cycles in any

instance achieves high fault prediction as presented

in Figure 4.

classification rate for computing the resistance and

magnitude of the blades in each cycle based on its

consistent operation and power generation output,

reducing the fault occurrence (Refer to Figure 5).

The failure in the electrical turbine is mitigated

based on segregating the min/max threshold using a

learning process. Therefore, regardless of the

consistent operation is maintained for reducing the

turbine failures.

Fig. 6: Prediction Time

This proposed method achieves a high

prediction time compared to the other factors as

represented in Figure 6. The identification of

failures and faults to ensure the support of the

electrical turbine prevents failures. Therefore, the

proposed method identifies the threshold condition

of the operating cycles based on the resistance and

consistent operation and power generation output in

electrical turbines for segregating the operating

cycles achieves fewer failures and computing time is

represented in Figure 7. If the failure is high in this

process, the particular cycles can be altered. The

maximum threshold reduces the turbine failures

with less altered cycles.

conditions is identified using resistance and

magnitude of the blades is addressed for each

operation cycle. The minimum and maximum

threshold conditions are classified through classifier

tree learning. This learning is used to identify power

generation interrupts and segregate the operating

cycles for the turbine. Based on the high and low

thresholds, the failure cycles are identified. The

maximum threshold detected cycles are halted and

changed using pre-maintenance intervals. The

mechanical faults are diagnosed at an early stage

using threshold conditions for identifying the

turbine failure and operation faults. This proposed

fault prediction method uses classifier tree learning

for serving smooth operation purposes based on

electrical fault prevention is generally good, there

are instances of complex problems, including those

with prolonged operating times and challenging

fault characteristics to identify in wind turbines. A

significant field for future research is investigating

more techniques that execute better in fault feature

extraction and selection for increased classification

accuracy and operating speed.

detection system for wind turbines on multiple

dimensions, including crucial component

monitoring and overall turbine effectiveness

more significant aspects. The advancement of this

technique into the prediction of electrical faults is a

potential future growth of this strategy. Faults that

happened need to be linked to particular patterns on

the control charts in the process of defining the

anomalous behavior. This helps to determine the

cause precisely and sets the stage for later

automation of the fault prediction.

References:

Deep Learning Approach for Sensor Data

Prediction and Sensor Fault Diagnosis in

Wind Turbine Blade. IEEE Access, 10,

117225-117234.

Song, W., ... & Cao, L. (2022). Prediction of

wind turbines based on a support vector

machine optimized by the sparrow search

algorithm. Ieee Access, vol.9, pp.69307-

69315.

diagnosis via a parameter-optimized feature

mode decomposition. Measurement, vol.203,

112016.

A. F., & Magagnato, F. (2021). Investigation

for micro-gas turbine based on optimized T–S

fuzzy model and nonsingular TSMO.

International Journal of Fuzzy Systems,

https://doi.org/10.1007/s11071-021-07009-7.

detection by an ensemble framework of

Extreme Gradient Boosting (XGBoost) in the

operation of offshore wind turbines.

Renewable Energy, vol.179, pp.945-962.

fault prediction method of wind turbine

gearbox based on pair-copula construction

and BP neural network. IEEE Access, vol.8,

bearing with limited samples. Renewable

Energy, vol.145, pp.642-650.

Na, S., & Choi, G. (2020). Prediction of

operating characteristics for industrial gas

turbine combustor using an optimized

artificial neural network. Energy, vol.213,

118769.

pp.5323-5333.

(2015). Building energy index: A case study

of three government office buildings in

Malaysia. Advanced Science Letters,

vol.21(6), pp.1798-1801.

Baluch, N.H., Mohtar, S. (2017). Building

energy index (BEI): A study of government

office building in Malaysian public university.

system. Journal of Electrical Engineering &

Technology, vol.16, pp.547-557.

anomaly decomposition scheme based on

multi-variable correlation extraction for wind

turbine fault detection and identification.

Applied Energy, vol.321, 119373.

(2022). Genetic Algorithm Optimized and

Type-I fuzzy logic controlled power

smoothing of mathematical modeled Type-III

DFIG based wind turbine system. Materials

Today: Proceedings, vol.56, 3355-3365.

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Conflict of Interest

Creative Commons Attribution License 4.0

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US