The use of big data and machine learning in industrial quality control is an important emerging research area. Quality control is the process of ensuring that the product satisfies standards criteria set by regulatory authorities, companies or customers. In today’s world, quality control is an integral part of manufacturing industry. It serves multiple purposes, builds confidence in product, and identifies early on faults before the product reaches to customer. Recently, researchers have applied AI and ML to various domains. There is untapped potential in manufacturing industry using artificial intelligence in industrial control and manufacturing. Solutions have been proposed that employ Artificial Intelligence (AI) techniques for engine failure testing after the assembly. In this paper, a novel and useful approach is proposed for detecting the anomalies in engine system by combining two robust and useful machine learning algorithms which include: (1) Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and (2) decision tree algorithm. Decision Tree (DT) algorithm is used to build decision tree for prediction. The propose algorithm is tested in MATLAB. This proposed model is applied to MQ9 Rapper drone engine unit TPE331-10 failure testing as a case study and has achieved the 99.3% accuracy of detecting anomalous engine. In this research our goal was to achieve the 99.97% accuracy by using our hybrid approach. Our base paper to achieve this level of accuracy is; deep convolutional neural networks and s-transform [2, p. 8] by Guoqiang Li et al.
The use of artificial intelligence automation in manufacturing industry specially in aviation industry are relatively new and getting more attention. AI-based automation work can improve the product quality as human inspector and quality engineer can focus on another task if AI-based robot can work on complex and repeated task.
The key contribution of this research is successfully and accurately combination of DBSCAN with DT in detecting the anomalies of engine with high accuracy. It is lightweight model. Other model uses artificial intelligence models which are not much accurate as our model is, they are computationally intensive and complex to implement and do not suit to sit perfectly in such a situation. In this paper, we proposed a combined an AI-based approach using DBSCAN with DT in detecting the anomalies in UAVs engine with high accuracy.
It is lightweight model and can be successfully use in-house UAV engine testing and on the fly engines health monitoring. Other model uses artificial intelligence models which are not much accurate as our model is, they are computationally intensive and complex to implement. Our model uses DBSCAN algorithm which is suitable even we have heterogeneous data . It can form any type of cluster for given radius and minimum points values. As a future work, we plan to expand our model to include other key variables such as vibration per seconds, heat, and noise. In addition, we plan to consider real-time feeds from engine sensors.