The experimental investigation of emergency valve used in heavy vehicle brake systems and modelling with artificial neural network

Since the emergency valve is a pneumatic brake valve used for manual or semi-automatic sudden braking of trailers in heavy vehicles, its dynamic performance and response time are of critical importance. Within the scope of this study, firstly, sample studies in the literature and comprehensive industrial applications related to the subject were examined in detail and parameters such as coefficients, pressure, and flow rate belonging to the springs of the valve dynamics expected to affect the braking time were investigated. The safety spring coefficient, check valve spring coefficient and hose diameter were determined as the input parameters of the system, the effect of the system parameters on the braking response time was examined using experimental design methods and a full factorial experimental design was created. In order to measure the braking response time of the system, an experimental setup was created, and comprehensive experimental studies were carried out. A mathematical model of the braking response time was obtained by regression analysis of the obtained experimental data. In order to improve the modelling performance of the system, studies were carried out on artificial neural network-based modelling of experimental data. Modelling studies were carried out to train the experimental data set by determining different neuron numbers, learning algorithms and training-test-validation percentages, and a higher accuracy model with less error values was investigated. The performances of the obtained mathematical and artificial neural network models in modelling and estimating the valve response time depending on the input parameters were comparatively examined and numerical results were presented.