An exemplary embodiment includes a diagnostic which can identify the source, or “root cause” of variability of process and process control parameters. A plurality of correlations is provided, each representing a possible cause of variation. One of the correlations is identified as the most likely ro
An exemplary embodiment includes a diagnostic which can identify the source, or “root cause” of variability of process and process control parameters. A plurality of correlations is provided, each representing a possible cause of variation. One of the correlations is identified as the most likely root cause of variation. The remaining possible root causes are also listed, in sequence, from most likely to least likely. The method applies to both normal and abnormal operating conditions.
대표청구항▼
1. A method to identify the root cause of variations in an industrial process having a primary reference variable and other variables, comprising the steps of: (a) looking at time-series data from a primary reference variable,(b) cross-correlating the data in step (a) with similar data from many oth
1. A method to identify the root cause of variations in an industrial process having a primary reference variable and other variables, comprising the steps of: (a) looking at time-series data from a primary reference variable,(b) cross-correlating the data in step (a) with similar data from many other variables,(c) using the results from step (b), determining the strengths of correlation and the respective lead and lag times of the variables,(d) using the correlation strengths and lead and lag times in step (c), determining which variable is the most likely root cause of variation in the primary reference variable, and(e) ranking the remaining variables in order of likelihood that they are the root cause of variation in the primary reference variable, whereby the root cause of variations is identified for a primary reference variable. 2. The method of claim 1, wherein: (a) the correlations are evaluated on a scale from zero, being no correlation, to another number, being perfect correlation: (b) the correlations are evaluated across a range of time shifts in data (c) for each variable, the peak correlation and its corresponding time shift are defined; (d) some variables are disqualified because they do not show a strong correlation; and (e) a variable is selected which has the longest lead time at its peak correlation value: whereby the variable in (c) is identified at the likely root cause of variation in the primary reference variable. 3. The method of claim 1, wherein no a priori process knowledge is used and no process models are used. 4. The method of claim 2, wherein no a priori process knowledge is used and no process models are used. 5. The method process of claim 1, applied to financial information, biological systems, weather data, discrete data sets, or continuous data sets. 6. The method of claim 2, applied to financial information, biological systems, weather data, discrete data sets, or continuous data sets. 7. The method of claim 1, further comprising analyzing historical variable data from the recent past, or a historical period, related to a particular incident in the process. 8. The method of claim 2, further comprising analyzing historical variable data from the recent past, or a historical period, related to a particular incident in the process. 9. The method of claim 7, wherein the historical data is simplified by cleansing or lumping or both. 10. The method of claim 8, wherein the historical data is simplified by cleansing or lumping or both. 11. The method of claim 1, wherein the cross-correlation analysis uses the historical data in the following equation: (f★g)(t)=def∫-∞∞f*(τ)g(t+τ)ⅆτwherein the symbol f stands for the primary reference variable, g stands for the variable being correlated as a possible root cause, t represents time, τ represents the time shift interval. 12. The method of claim 11, wherein the cross-correlation is accomplished using Fourier Transform analysis. 13. A method to identify the root cause of variations in a process having a primary reference variable and other variables, comprising the steps of: (a) selecting a reference variable,(b) select variables to investigate as possible causes,(c) retrieving historical data for all variables,(d) cross-correlating reference variables to each possible connection with another variable,(e) determining peak correlation magnitudes and lead or lag times of the variables,(f) eliminating variables with lagging correlation,(g) eliminating variables below selected correlation thresholds,(h) sorting remaining variables by lead times,(i) selecting the root cause based on the greatest lead time of the variables, and(j) displaying the root cause variable and other lead variables. 14. The method of claim 1 implemented by a computer program with or without an internet interface. 15. The method of claim 13 implemented by a computer program with or without an internet interface. 16. The method of claim 1 applied to engine operation, aircraft and nautical systems, or biological measurements. 17. The method of claim 13 applied to engine operation, aircraft and nautical systems, or biological measurements. 18. The method of claim 1 operating in real time operation of a process. 19. The method process of claim 11 operating in real time operation of a process. 20. The method of claim 13 operating in real time operation of a process.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (11)
Basu, Atanu; Hubele, Joachim, Apparatus, system and method for processing, analyzing or displaying data related to performance metrics.
Emigholz,Kenneth F.; Wang,Robert K.; Woo,Stephen S.; McLain,Richard B.; Dash,Sourabh K.; Kendi,Thomas A., System and method for abnormal event detection in the operation of continuous industrial processes.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.