Performing Analysis without Buffers of Data

Analysis functions yield solutions that characterize the behavior of a data set. In array-based data acquisition and analysis, you might analyze a large set of data by dividing the data into 10 smaller buffers. Analyzing those 10 sets of data yields 10 solutions. You can further resolve those 10 solutions into one solution that characterizes the behavior of the entire data set.

In point-by-point analysis, you analyze an entire data set in real-time. A sample unit of a specific length replaces a buffer. The point-by-point sample unit can have a length that matches the length of a significant event in the data set that you are analyzing. For example, the application in a case study acquires a few thousand samples per second to detect defective train wheels. The input data for the train wheel application comes from the signal generated by a train that is moving at 60 km to 70 km per hour. The sample length corresponds to the minimum distance between wheels.

A typical point-by-point analysis application analyzes a long series of sample units, but you are likely to have interest in only a few of those sample units. To identify those crucial samples of interest, the point-by-point application focuses on transitions, such as the end of the relevant signal.

The train wheel detection application in the case study uses the end of a signal to identify crucial samples of interest. The instant the application identifies the transition point, it captures the maximum amplitude reading of the current sample unit. This particular amplitude reading corresponds to the complete signal for the wheel on the train whose signal has just ended. You can use this real-time amplitude reading to generate an event or a report about that wheel and that train.