Method for quantifying peaks in an analytical signal
A method for quantifying peaks in an analytical signal, peaks in the analytical signal being quantified by recording successive signal values and applying a peak analysis methodology to the recorded successive signal values within an interval to obtain a set of peak quantification results. Before the same peak analysis methodology is applied to the modified signal to quantify the peaks in the signal, random noise is added to the analytical signal and/or the signal is shifted within the interval to facilitate optimization of the parameters of the peak analysis methodology and to improve the robustness of the method in runtime applications. A subsequent statistical evaluation of the peak quantification results from the multiple repeated peak analyses of the original and modified signals is used to detect an occurrence of and to reduce the chance of a possible error in the peak quantification that needs to be alarmed or addressed.
1. A method, implemented in a computer, for quantifying peaks in an analytical signal by recording successive signal values of the analytical signal and applying a peak analysis methodology to the recorded successive signal values of the analytical signal within an interval to obtain a set of peak quantification results, the method comprising the steps of:
adding random noise to the recorded successive signal values of the analytical signal;
applying, in the computer, the peak analysis methodology to the random-noise-modified successive signal values of the analytical signal to obtain the set of peak quantification results;
repeating at least once the step of adding the random noise and the step of applying the peak analysis methodology, each random noise added to the recorded successive signal values of the analytical signal being different each time the step of adding is repeated;
applying, in the computer, a statistical evaluation methodology to obtained sets of peak quantification results to check peak quantification results for outliers; and
accepting, in the computer, peak quantification results from the recorded successive signal values of the analytical signal as valid results if there is no outlier and reporting the peak quantification results from the recorded successive signal values of the analytical signal if there is at least one outlier.
2. The method of one claim 1 , wherein the method is implemented during runtime of automated process analytics.
3. The method of claim 1 , wherein parameters of the peak analysis algorithm are adjusted dependent on a number and amount of outliers.
4. A method, implemented in a computer, for quantifying peaks in an analytical signal by recording successive signal values of the analytical signal and applying a peak analysis methodology to the recorded successive signal values of the analytical signal within an interval to obtain a set of peak quantification results, the method comprising the steps of:
shifting, in the computer, the recorded successive signal values of the analytical signal within the interval by at least one position, recorded successive signal values shifted out at one end of the interval being discarded and positions vacated at the other end of the interval being filled with one of original values before the shift and neighboring signal values shifted from outside in the interval;
applying, in the computer, the peak analysis methodology to the recorded successive signal values of the analytical signal in the interval to obtain the set of peak quantification results;
repeating at least once the step of shifting, by the computer, the recorded successive signal values of the analytical signal and the step of applying the peak analysis methodology, at least one of an amount and direction of shift against originally recorded successive signal values of the analytical signal being different each time the step of shifting is repeated;
applying a statistical evaluation methodology to the obtained sets of peak quantification results to check the sets of peak quantification results for outliers; and
accepting the sets of peak quantification results from the recorded successive signal values of the analytical signal as valid results if there is no outlier and reporting the peak quantification results from the recorded successive signal values of the signal if there is at least one outlier.
5. The method of one claim 4 , wherein the method is implemented during runtime of automated process analytics.
6. A method, implemented in a computer, for quantifying peaks in an analytical signal by recording successive signal values of the analytical signal and applying a peak analysis methodology to the signal values within an interval for obtaining a set of peak quantification results, the method comprising the steps of:
adding random noise to the recorded successive signal values of the analytical signal within the interval or shifting the recorded successive signal values of the analytical signal within the interval by at least one position, recorded successive signal values shifted out at one end of the interval being discarded and positions vacated at the other end of the interval being filled with originally recorded successive signal values before the shift or neighboring signal values shifted from outside in the interval;
applying, in the computer, the peak analysis methodology to the recorded successive signal values of the analytical signal values in the interval to obtain a set of peak quantification results;
repeating at least once the step of adding the random noise or shifting the recorded successive signal values of the analytical signal and the step of applying the peak analysis methodology to ensure that the steps of adding the random noise and shifting the recorded successive signal values of the analytical signal are each performed at least once, in cases in which the steps of adding random noise or shifting the signal values are performed at least twice, the random noise added to the recorded successive signal values of the analytical signal or at least one of an amount and direction of shift against the recorded successive signal values of the analytical signal being different each time the step of adding or shifting is repeated;
applying a statistical evaluation methodology to the obtained sets of peak quantification results to check the peak quantification results for outliers; and
accepting the peak quantification results from the recorded successive values of the analytical signal as valid results if there is no outlier and reporting the peak quantification results from the recorded successive signal values of the analytical signal if there is at least one outlier.
7. The method of one claim 6 , wherein the method is implemented during runtime of automated process analytics.