Analyzing the effectiveness of MEMS sensor and IoT in predicting wave height using machine learning models

Abstract

Wave height is a critical consideration in the planning and execution of maritime projects. Wave height forecasting methods include numerical and machine learning techniques. The traditional process involves using numerical wave prediction models, which have great success but are highly complex as they re-quire adequate information on nonlinear wind–wave and wave-wave interactions such as wave ener-gy-balance equation. On the contrary, machine learning techniques can predict wave height without prior knowledge of the above-mentioned complex interactions. This research aims to predict wave height using Micro-electromechanical Systems (MEMS), Internet of Things (IoTs), and Machine Learning (ML) based approaches. A floating buoy is developed using MEMS inertial measurement unit and IoT microcontrol-ler. An experiment is conducted in which the developed buoy is subjected to different wave heights in real time. The change in 3-Axis acceleration and 3-Axis gyroscope signals are acquired in a computer via IoT. These signals are analyzed using Machine learning-based classification models to predict wave height accurately. The obtained validation accuracy of the machine learning models K-NN (K-Nearest Neighbor), SVM (Support Vector Machine), and the bagged tree is 0.9906, 0.9368, and 0.9887 respec-tively, which indicates that MEMS and IoT can be used to classify & predict wave heights in real-time accurately.

Machine Learning-Based Approach for Predicting Ocean Surface Temperature

“…In this context, it is possible to point out the advantages of random forest over the neural network. The ability to combine predictions from multiple decision trees into a single model is the main advantage of using a random forest instead of a decision tree [25]. The concept is that a single model is better than a single good model, even if it is composed of several mediocre models.…”