Design and investigation of an IIOT-enable device for remote monitoring and predictive maintenance of hydraulic hammers

Hydraulic hammers, also known as breakers and peckers, are utilized in a wide variety of applications to demolish a structure and break rocks into smaller sizes. These tools and equipment are extremely sensitive and operate in harsh environments. As a result, there is a widespread requirement for remote control and monitoring of equipment and machines. In addition, given the technological advances in sensors, data transmission, and data collection via the Internet of Things, as well as the high demand for data analytics and the importance of maintenance in the fourth industrial revolution, artificial intelligence is being used as a powerful tool. Thus, remote monitoring of industrial equipment such as hydraulic hammers has become a critical feature of Industry 4.0 and Internet of Things technologies. Data collection has also recently received a lot of attention to improve machines' ability to make future decisions based on the collected data and increase efficiency. However, a major challenge is to ensure the lifetime of equipment and machines and reduce the time and cost of maintenance, which directly affects the cost and competitiveness of the product. Therefore, machine learning, deep learning, and predictive maintenance models have become important. The first part of this study (INDECONNECT® project) involves presenting the design and development of an Internet of Things device, specifically a data logger, that aims to enhance the performance of hydraulic breakers through remote monitoring. The device is equipped with sensors for data collection, analysis, and management. By designing the platform and strategically placing sensors, the device is expected to obtain vast amounts of data (Big Data) regarding various aspects of the hydraulic hammer such as vibration, machine operation time, oil pressure, temperature, and oil flow, based on the operation conditions and type of material used. Analyzing the large amount of data collected by the Data logger directly from the hydraulic hammer during its operation can provide valuable information for adjusting process planning, implementing predictive maintenance, and establishing standard technical information for different modes of the Hydraulic hammer. Secondly, the project seeks to predict maintenance operations by utilizing artificial intelligence tools, particularly machine learning and deep learning methods, based on a dataset of various components at different time periods. In this study, we utilized machine learning and deep learning algorithms to predict machine and component failures for two different time periods - one day and seven days in the future. Nevertheless, as maintenance prediction datasets tend to be unbalanced, we employed two approaches in this study - a weighted average coefficient and a two-step method - to predict the probability of long-term failures. The two-step method is a novel technique that significantly reduces the data set imbalance and enhances the performance of neural network algorithms. The outcomes indicate that the Convolutional Neural Network is the most effective in predicting the likelihood of machines and components failing.