Integrating Auto-Associative Neural Networks with Hotelling T2 Control Charts for Wind Turbine Fault Detection

Abstract

This paper presents a novel methodology to detect a set of more suitable attributes that may potentially contribute to emerging faults of a wind turbine. The set of attributes were selected from one-year historical data for analysis. The methodology uses the k-means clustering method to process outlier data and verifies the clustering results by comparing quartiles of boxplots, and applies the auto-associative neural networks to implement the residual approach that transforms the data to be approximately normally distributed. Hotelling T 2 multivariate quality control charts are constructed for monitoring the turbine's performance and relative contribution of each attribute is calculated for the data points out of upper limits to determine the set of potential attributes. A case using the historical data and the alarm log is given and illustrates that our methodology has the advantage of detecting a set of susceptible attributes at the same time compared with only one independent attribute is monitored.

Using a Time Delay Neural Network Approach to Diagnose the Out-of-Control Signals for a Multivariate Normal Process with Variance Shifts

“…Similarly to the results for the other classifiers, we can observe that the AIR values were higher for the cases of ρ = 0.5 and ρ = 0.8. Types of Combination ρ = 0.2 ρ = 0.5 ρ = 0.8 C 5-1 56.67% 62.33% 61.83% C 59.67% 62.33% 58.00% C 61.67% 65.67% 53.50% C 58.83% 62.33% 58.33% C 60.50% 67.00% 56.67% C [5][6] 58.17% 67.67% 56.00% C [5][6][7] 58.33% 64.17% 73.50% C [5][6][7][8] 58.67% 65.50% 76.83% C [5][6][7][8][9] 57.17% 66.67% 76.83% C [5][6][7][8][9][10] 55.50% 63.67% 70.67% Table 9. MARS Identification results, AIR -MARS {null}, for ten combinations of faults of an MNP 9 .…”

A Scalable Predictive Maintenance Model for Detecting Wind Turbine Component Failures Based on SCADA Data

“…(i) machine learning techniques and process control tools, i.e. the residual Hotelling T2 control chart [7], [8], are wisely combined; (ii) an innovative multivariate outliers removal (MOR) method, based on k-means clustering, is proposed and applied to eliminate the abnormal samples from training instances (the proposed MOR approach generalizes the procedures already presented in [4], [9]); (iii) the status of the components is monitored by means of an original probabilistic formula defining a Key Performance Indicator (KPI): warnings of different severity are triggered based on threshold crossing rules; (iv) the Plug and Play nature of the presented approach, i.e. the fast service scalability on wind farms of increasing size, is also an added value of this work; (v) the proposed systems was realised and tested on a large number of wind turbines in different farms and geographical areas, during three years of operation.…”

A New Fault Location Approach for Acoustic Emission Techniques in Wind Turbines

“…SHM enables identifying and diagnosing the fault and its location by detecting changes in the static and dynamic features of the structure [9,10]. SHM can be remotely managed, reducing the costs of manual inspections and the time between the fault occurrences, and this has been noted [11,12]. This will lead to an increase in the productivity, reducing the potential downtimes for the wind farms and increasing the RAMS of the wind turbine [13][14][15].…”