Method and apparatus for predicting a fault condition using non-linear curve fitting techniques

The invention relates generally to a device for predicting a fault condition associated with a machine, and more particularly, to a method and apparatus for predicting a fault condition in response to the trend of a machine parameter using non-linear curve fitting techniques. An apparatus for predicting a fault condition for a machine is disclosed. The machine has a plurality of parameters being dependent upon machine performance. A sensor is connected to the machine and is adapted to produce an electrical signal in response to one of the plurality of machine parameters. The apparatus further includes means for determining a non-linear data trend of the parameter in response to the electrical signal, and means for predicting the fault condition as a function of the trend and a threshold value indicative of the fault condition.In a second aspect of the invention, a method for predicting a fault condition is provided. The method includes the steps of sensing a parameter having a level being dependent upon machine performance and responsively producing an electrical signal, determining a non-linear data trend of the parameter in response to the electrical signal, and predicting the fault condition as a function of the trend and threshold value indicative of the fault condition.