Predictive maintenance holds an important role in the automotive industry. The current study was focused on investigating of the vibrations resulting from a thermostat failure in diesel engines and their potential for predictive maintenance. Based on real time measurements of varied engine speed, the vibrations induced by diesel engine were analyzed, particularly in cases of cooling circuit thermostat dysfunction. Several statistical indicators, such as kurtosis factors, root mean square (RMS) and standard deviation were used to evaluate the vibration signal variations. This evaluation tends to detect and characterize the vibrational anomalies due to thermostat failure, which provide crucial insights into their overall health of the engine. Furthermore, the Artificial Neural Networks (ANN) method was applied for modelling the emitted vibrational signals. The ANN is an efficient method in analysing complex time series, such as engine vibrations. This study aims to enhance the performance of diesel engines by identifying the underlying vibrational mechanisms of these failures. In addition, it tends to optimize predictive maintenance strategies, thereby improving engine reliability and efficiency in the automotive industry.