A Development of Booming Index of Diesel SUV by using Artificial Neural Network 2012-01-1542
In today's competitive scenario, understanding mental modal map of individual customer perception plays a major role to create the brand image of vehicle. Among them “comfortable sound” is one of the important criteria for customer satisfaction, especially in case of diesel vehicle, where in-cab sound quality plays a crucial factor. Often customer perception concerning comfort in automotive industry relies on subjective comfort evaluation method. Converting the customer perception into objective measurements and to correlate them is often tough task for NVH engineers. It is because of human sensation behavior differs from persons to person, mental map, geographical location and domain knowledge. In addition acoustic & comfort relevant aspects are often subjectively evaluated based on jury trials conducted on the prototype vehicle and class competitive benchmark vehicles to get the feel & confidence of product for different gateways. This is critical in terms of time and costs since the prototype vehicle are often available late in development process. Hence there is strong need for a toll which can provide the objective status of vehicle NVH comfort which can facilitate to compare it with set program targets without affecting the jury biasness. Similar studies are done in past but mostly focus on gasoline engine vehicles. Recent scenario in Indian continent is that buyers are getting migrated from gasoline to Diesel Vehicles because of value proposition associated with diesel vehicle.
In present, study has been carried on boom phenomenon which is a perceivable irritant in the current diesel vehicles family. In this paper, the booming phenomena are simulated using the synthetic sounds in order to capture the entire boundary conditions of boom. The booming sensation is effectively related to objective parameters of sound quality like sound pressure level (A-weighted), loudness and sharpness which are taken as input to artificial neural network (ANN). The sounds perceived as boom are selected by using 6dB/octave criteria and they are subjectively evaluated by a group of 20 people. These subjective rates are given as targets to ANN. The trained ANN is validated with 7 interior sounds of vehicles at different prototype stages. The concept is to develop a booming index which will give an objective value of boom at different stages of vehicle prototype which will be also compared to competitor's vehicle in order to do gap evaluation and thus confidence to achieve the targets. For a booming index a relation between empirical measurement data and subjective sensation has been established. The proposed booming index has been successfully applied to the objective evaluation of the booming sound quality during product development cycle of diesel SUV's.