Real time State Estimation and health indication of Power System

Project Title

Real time State Estimation and health indication of Power System

Project Area of Specialization Artificial IntelligenceProject Summary

EMS (Energy Management System), an integral part of modern dispatch centers nowadays, consists of several applications like Power / Network / Communication applications. Network Applications (NA) provides various solutions for fast and comprehensive analysis of power system. Among various sub-applications in NA, State Estimation (SE) tool is used to provide a reliable and complete network solution from the real-time measurements. Conventional SE algorithm is based on Weighted Least Squares (WLS) method, which is prone to convergence issues and sometimes unable to estimate unobservable part of the network.

This project aims to provide a real- time power system state estimation and health monitoring tool based on Artificial Neural Network (ANN) which have strong ability to determine system’s current state and finding out the desired parameters by monitoring changes in load buses only, thus providing an attempt to solve under-determined systems.

The tool will have a Graphical User Interface having options to perform the state estimation by various types of Neural Networks in a user friendly manner. The tool will just take the input readings of real and reactive power and will estimate the state of Power System on a single click. It will then also categorize the system as normal or in alert state indicating its health status. The approach will be tested on IEEE 14 Bus System.

Project Objectives

The main objective of our project is to have a real time and user friendly application or tool that will estimate the state of power system and indicate its health using fast AI based technique of Neural Networking. The tool will also be applicable to estimate the state of under-determined power system that is un-solvable by weighted least square algorithm (conventional algorithm in use) . The tool will be efficient in terms of speed, accuracy and memory as compare to conventional algorithm.

Project Implementation Method

The implementation method is to improve the estimation technique that can work on under-determined system. For achieving this we have following tasks:

• Verify the selected system by comparing the load flow results from MATLAB code, ETAP and PSSE load flow software modules
• Make data creation code that works randomly as well as user-defined for generating the data and organize it for state estimation in terms of active power P, reactive power Q, voltage magnitude V and its angle ?
• Create neural network on MATLAB. Various ANN configurations shall be implemented to make model as accurate as possible. An algorithm will be used which automatically tests the networks performance based on various hidden layers, neurons and transfer functions. After comparing will give best model based on least mean square error
• Convert an algorithm into a tool or application that has a GUI feature to make it more user-friendly.
• Bad data detection can also be made using chi-squared method. (optional)

Benefits of the Project

This project is beneficial particularly for EHT Power Control Centers, it will help system operators (SO) in the following way

• It depicts the real time view of power system.
• Estimates actual running values of state variables as estimators may result in bad data detection.
• Useful in system modeling, such as in implementing complex mapping and system identification
• To control the process of power-generation systems and accurately monitoring tie-line flows.
• To run online system studies based on state estimation, for example online dispatcher, Power flow, contingency analysis, optimal power flow etc.

ANN technique being used to estimate power system state variables here, can be utilized for any other process control systems to predict KPIs.

Technical Details of Final Deliverable

The completion of the project will give a GUI based user friendly tool estimating the state of Power System by Artificial Neural Network. The technical details are:

• The tool will have the options to implement various types of neural network configurations on user-demand.
• The tool will just use real and reactive power injections as input and estimate the state on a single click.
• The tool will be able to solve the under-determined systems also which are un-solvable by conventional algorithms in use.
• The tool will categorize the system as in normal or alert state indicating its health status.
• The tool will be efficient in terms of memory, accuracy and time as compare to conventional algorithm (Weighted Least Square method).

Final Deliverable of the Project Software SystemType of Industry Education , Energy Technologies Artificial Intelligence(AI)Sustainable Development Goals Decent Work and Economic Growth, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	24000
stationery and printing	Miscellaneous	2	5000	10000
Softwares MATLAB, ETAP and PSSE	Equipment	4	2000	8000
Hard drives and RAM	Equipment	2	3000	6000