Method and system for detecting difference between plural observed results

Abstract

A method and system for analyzing time series data. In an embodiment, a loop is executed and terminated upon a specified maximum number of iterations of the loop being performed or upon a difference between scores in successive iterations of the loop not being greater than a specified tolerance, wherein the score in each iteration is calculated as function of an absolute value of a difference between respective cumulative probability values of first and second cumulative probability distributions which are generated from respectively first and second time series data sets. In an embodiment, time series data is processed in a sequence of time periods, wherein a combined cumulative probability distribution is generated in each time period by combining a cumulative probability distribution of new time series data with previously combined cumulative probability distribution data according to a ratio of the number of new to previous observed values.

Description

FIELD OF THE INVENTION[0001]The present invention relates to a method and system for detecting a difference between a set of observed results, and in particular, the present invention relates to a method and system for statistically analyzing the difference between time series data sets.BACKGROUND OF THE INVENTION[0002]Detecting and analyzing anomalies of dynamic systems is an important technical challenge in various areas in the manufacturing industry. For example, anomaly detection in a production line has been of particular importance, and a lot of statistical techniques have been developed for quality control purposes. However, most of the traditional statistical quality control techniques are based on a strong assumption of multivariate normal distribution. Unless the system of interest is relatively static and stationary, the distribution of data is far from the normal distribution in general. This is especially the case in the analysis of automotives, where the system is high...

AI Technical Summary

Problems solved by technology

Detecting and analyzing anomalies of dynamic systems is an important technical challenge in various areas in the manufacturing industry.

As a result, the utility of such traditional approaches is quite limited in many cases.

Thus it is difficult to determine the reference value of a variable for detecting anomalies.

While experienced engineers may be able to make a decision on the state of the system based on such complicated numerical data, it is unrealistic to assume that enough manpower of experienced engineers is available in every phase and place of anomaly detection.

In addition, the knowledge of experienced engineers is often hard to translate to specific mathematical rules used in the limit-check routine.

To summarize, the applicability of limit-check in combination with partial human knowledge is seriously limited in general.

In general, it is difficult to make test conditions in all the runs exactly the same since the complexity of the system is too high to completely control their values.

In conventional techniques, it is hard to handle such fluctuations in experimental conditions, so that a substantial status of a diagnosis target cannot be appropriately characterized in many cases.

Moreover, since the number of variables of the system is very large, considering all combinations of the variable is computationally prohibitive.

Images