Suppression of high frequency oscillations on suspension system of snow groomer with evolutionary algorithms optimization

In current engineering practice for the design and dimensioning of hydropneumatic suspension systems, the effect of main parameters is considered; this approach can be used to implement approximate models basically suitable to describe low frequency and high amplitude oscillations of the machine. The target of this study is a Snow Groomer, a tracked vehicle driven by diesel engines and equipped in front with a shovel and behind with a cutter. When the machine drives over a snowfield, it pushes snow ahead of it and, at the same time, smooths out any surface unevenness. The suspension system is the key element to ensure the driver's safety and comfort, the effectiveness of snow grooming and finally enhance the reliability of the machine components. The on-field testing had shown high frequency pressure oscillations transmitted from the sprocket to hydraulic system, propagated through the flexible hoses. Those Pressure Oscillations cause noise and can affect negatively the durability and reliability of the Machine. A lumped parameter non-linear dynamic model of the hydraulic circuit and of the machine interactions is built in Amesim environment, including Lax Wendroff wave propagation models, to make it able to catch the high frequency oscillations experienced in the test field. Most of the design parameters are fixed (such as vehicle weight and hydraulic lines length), other parameters can be varied to study the optimal solution, these parameters define the "factors" of the optimization problem. As a next step it is important to define the objectives of the optimization, in this case corresponding to various figures of merit describing the behavior of the system in different work conditions. The large number of factors included in the lumped parameter model generates an exponentially larger number of possible configurations. Moreover the relationship between factors and objective is not always possible to express with explicit mathematical models. Finally the presence of multiple and sometimes conflicting objectives forces more refined analysis methods to be adopted. For the above mentioned reasons a Multi-objective Optimization method is proposed taking advantage of Evolutionary Algorithms and Pareto Front optimization. Two different architectural solutions are analyzed and optimized using two different algorithms, Non Sorting Genetic Algorithm II (NSGAII) and Multiple Object Swarm Particle Optimization Algorithm (MOPSO). The results of the optimization belonging to the Pareto Front will be analyzed to assess the expected improvement of the suspension performance and will be chosen as candidates for a new setting of the Snow Groomer. Furthermore a comparison in terms of effectiveness and speed in finding solutions will be given for the current simulation environment.