Feature extraction and process monitoring of multi-channel data in a forging process via sensor fusion

Sensor signals acquired in the industrial process contain rich information that can be analyzed to detect system anomalies and facilitate effective monitoring of the process. In many processes, multiple signals are acquired by different sensor channels (i.e. multi-channel data) which have high-dimensional and complex cross-correlation structures. When analyzing such signal data, two main issues must be resolved: (1) feature extraction of multi-channel data to reduce the data dimensionality and improve signal analysis efficiency, and (2) the sensor fusion to achieve better monitoring of the process. It is crucial to develop a method that considers the interrelationships between different sensor channels. This paper proposes an improved multilinear feature extraction method with a feature selection strategy to improve the separability of profile data. The proposed method is applied directly to the high-dimensional multi-channel data. Features are extracted and combined with multivariate control charts to monitor multi-channel data. The effectiveness of the proposed method in quick detection of process changes is demonstrated with both the Monte Carlo simulation and a real-world case study. The real multi-channel data in the case study are recorded in a multi-operation forging process for the propose of process monitoring and fault detection.