Browse Prior Art Database

Self-adaptive Methodologies and Selector for Remote Level Wind Farm Health Evaluation with SCADA Data

IP.com Disclosure Number: IPCOM000246403D
Publication Date: 2016-Jun-06
Document File: 5 page(s) / 184K

Publishing Venue

The IP.com Prior Art Database

Related People

RongRong Yu: AUTHOR [+2]

Abstract

This disclosure has proposed 3 types of methodologies and a self-adaptive selector to realize condition monitoring and health evaluation for wind farm. The 3 types of methodologies are all designed in particular based on data mining technology. Since the 3 ones have respective advantages and limitations to be applied, the perfect choice may vary in different situations. In order to make the solution adaptive to any types of wind farms, a self-adaptive selector is further designed in order to automatically identify the perfect method for different wind farms. With the proposed techniques, not only health condition (normal or abnormal) of wind turbines can be given, but also remaining useful lifetime can be evaluated for the wind turbines detected as abnormal (not all the 3 methods can do it).

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

Page 01 of 5

Disclosure Title:

Self-adaptive methodologies and selector for remote level wind farm health evaluation with SCADA data

Abstract of Disclosure:

This disclosure has proposed 3 types of methodologies and a self-adaptive selector to realize condition monitoring and health evaluation for wind farm. The 3 types of methodologies are all designed in particular based on data mining technology. Since the 3 ones have respective advantages and limitations to be applied, the perfect choice may vary in different situations. In order to make the solution adaptive to any types of wind farms, a self-adaptive selector is further designed in order to automatically identify the perfect method for different wind farms. With the proposed techniques, not only health condition (normal or abnormal) of wind turbines can be given, but also remaining useful lifetime can be evaluated for the wind turbines detected as abnormal (not all the 3 methods can do it).

Background of the Disclosure:

This disclosure makes improvement on Pat. No. WO2016/077997 "Wind turbine condition monitoring method and system". In later one, a kind of cost-effective remote level wind farm condition monitoring system purely based on SCADA data with no need of sensor installation has been proposed. In former one, the proposed methodologies and selector are designed in particular to implement the SCADA based procedure so as to make it a more complete solution.

Theoretically the technique is more suitable for those with complex component composition for example wind turbine (with both mechanical components and electrical components) because this is what data mining technology is good at. For pure mechanical components for example gearbox, it also works according to our prior test in project ACM4W but with slightly lower detection rate (due to limited signals). For electrical components for example wind converter, the performance is not demonstrated yet.

Statement of the Problem:

This invention disclosure aims to solve two problems to implement SCADA based condition monitoring solution:

- How to design the methodologies to make the SCADA based condition monitoring solution applicable for all sorts of wind farms.


- How to design the rules to automatically select the perfect method for different wind farms.

Description of the Disclosure:

The basic flowchart for the condition monitoring solution is shown in Figure 1, which mainly includes following parts:

We reserve all rights in this document and in the information contained therein. Reproduction, use or disclosure to third parties without express authority is strictly forbidden. Ó 2016 ABB Ltd.


Page 02 of 5

Figure 1 Diagram of the basic concept

Method1: Design individual health evaluation model for individual wind turbine

The core logics of Method1 is to use machine learning algorithm to train a health-evaluation model for individual wind turbine with adoption of historical operating SCADA data of particular wind turbine. With the traine...