Prognostic Maintenance of Hydro Power Sub System Using Machine Learning Algorithm

Project Title

Prognostic Maintenance of Hydro Power Sub System Using Machine Learning Algorithm

Project Area of Specialization Artificial IntelligenceProject Summary

Condition monitoring or system health monitoring includes the set of processes that are performed in order to maintain the system in its optimal state. Monitoring is usually limited to observing the current states of the system and prompt repair actions based solely on these current states. Alternatively, when monitoring of these current states of the system is augmented with the future operating states and future failure states, the process then leads to the Predictive Diagnostics. The Predictive Diagnosis or Prognosis is vital for the complex systems so as to optimize the performance of the equipment and reduce the frequency of scheduled maintenances.

This project will therefore be focused on developing a condition monitioring and progrnostics system that will be used to predict the equipment faults, failures and degradation. A model based on machine learning techniques shall be developed and trained using the historical data.

Project Objectives

The primary objectives of the project are:

1)    To proactively diagnose the faults occurence 

2)    To develop a cost and time effective predictive maintenance solution

3)    To mitigate the faults occurence and enhance the Machine's life expectancy

4)    To improve system's reliability and maintainability

Project Implementation Method

A robust prognostic model shall be developed duing the project. The success of the Prognotic Model will depend on three main components, i.e. 1) having the right data, 2) framing the problem appropriately and 3) evaluating the predictions properly. Following will be the main steps in developing the proposed prognostic model.

1. Data Acquisition

2. Data preprocessing

3. Feature extraction/Data identification

4. Detecting relationships in data

5. Selecting appropriate Machine Learning techniques

6. Designing and training the model

7. Identifying key trends in the data

8. Developing maintenance techniques

Benefits of the Project

1. To increase the Remaining Useful Life (RUL) of the machine

2. To reduce the machine down time 

3. To lower the maintenance cost

4. To improve efficiency of the overall system

Technical Details of Final Deliverable

A software based prognostic model would be developed. Following are the technical details of the model:

1) Python libraries such as numpy and pandas would be used for data preprocessing.

2) Support Vector Machine (SVM) would be used for the classification of data where the data would be classified on the basis of features using the best possible split technique, the hyper plane.

3) A Python library sci-kit learn would be used for the training and testing of the model.

4) A Pythone library matplotlib and a data analytics tool named tableau would be used for graphical visualization of the results.

Final Deliverable of the Project Software SystemType of Industry IT , Energy Technologies Artificial Intelligence(AI)Sustainable Development Goals Affordable and Clean Energy, Industry, Innovation and Infrastructure, Responsible Consumption and ProductionRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	70000
NVIDIA Tesla K40 GPU Computing Processor Graphics Card	Equipment	1	70000	70000