Application of Partial Least-squares Regression to oil atomic emitting spectrum data of a type diesel engine

Aiming at relation between the concentrations of wearing elements of diesel engine and it's loads(X<inf>1</inf>), cylinders' clearances(X<inf>2</inf>, X<inf>3</inf> and X<inf>4</inf>) and runtime after renewing oil(X<inf>5</inf>), the Partial Least-squares Regression(PLSR) has been used to analyze the oil atomic emitting spectrum data of a 6-cylinder diesel engine. The results show that Cu concentrations variance explained by the five components is largest. These components are derived from X<inf>1</inf>, X<inf>2</inf>, X<inf>3</inf>, X<inf>4</inf> and X<inf>5</inf>. The PLSR-function concerning Cu can forecast Cu concentrations well. It has proved perfect in forecasting all the Cu concentrations of the 69 samples in the seven kinds of operating conditions. The effect of X<inf>1</inf>, X<inf>2</inf>, X<inf>3</inf>, X<inf>4</inf> and X<inf>5</inf> upon Cu concentrations has been effectively evaluated by the Variable Important in Projection (VIP). As compared with the obvious effect of cylinder's clearances(X<inf>2</inf>, X<inf>3</inf> and X<inf>4</inf>) and that of runtime(X<inf>5</inf>), the effect of the loads is small(X<inf>1</inf>).