System and method for UPS battery monitoring and data analysis

A computer implemented method for analyzing raw data relating to battery measurement values collected over a period of time, using an electronic processing/analysis system, to verify a quality of the raw data collected, wherein the raw collected data relates to battery measurement values of a battery string of a battery powered component, to improve an understanding of the raw collected data to better evaluate a health of the battery string, the method including:

• using a monitoring system to collect raw data point values from the battery string over a period of time;
• using the electronic processing/analysis system:
• to perform a data cleansing operation by identifying and removing specific ones of said raw data point values determined to be outliers residing outside of a predetermined range of data point values;
• to perform a data condensing operation by identifying and examining successive pairs of the raw data point values which were obtained adjacent one another in time, and whenever the two raw data point values of a given pair are identical, then discarding one said raw data point value of the pair being examined, to thus form a condensed collection of raw data point values;
• from the condensed collection of raw data point values, for each one of said raw data point values:
• using the hardware and software of the electronic processing/analysis system to identify a plurality of ones of the raw data point values that were obtained previous in time to a given one of the raw data point values under consideration;
• using the hardware and software of the electronic processing/analysis system to examine and assign a weight to each one of the plurality of ones of the raw data point values that were obtained previous in time to create weighted data point values;
• causing the hardware and software of the electronic processing/analysis system to use the weighted data point values to generate a predicted data point value which is associated with the raw data point value under consideration;
• using the hardware and software of the electronic processing/analysis system to perform a comparison between the predicted data point value and the raw data point value under consideration to determine an error relative to the raw data point value under consideration;
• further using the electronic processing/analysis system to evaluate the weighted data point values and the predicted data point values to improve an understanding of the raw data point values, to obtain an improved determination of a health of the battery string for the purpose of making a more accurate determination, by using the processing/analysis system, if a failure of the battery string is likely to occur, and to enable action to be taken to avoid compromising an integrity of the battery powered component;
• and
• if the error exceeds a predetermined threshold, then using the electronic processing/analysis system to determine that the raw data point value under consideration is an erroneous data point value.

A system for analyzing raw data collected over a period of time from a battery string of battery powered component, to verify a quality of the raw data collected and to eliminate from further consideration data point values determined to be erroneous data point values, to assist in determining a health of the battery string, wherein the raw data collected is battery measurement values taken from the battery string, the system including:

• an electronic processing system including hardware and software and configured to run an algorithm, and operable to collect raw data point values from the battery string over a period of time;
• a database in communication with the electronic processing system and configured to store the collected raw data point values in a memory;
• the hardware and software of the electronic processing system further being configured to:
• identify and remove, through electronic processing operations, specific ones of said raw data point values determined to be outliers residing outside of a predetermined range of data point values;
• for each one of said specific remaining one of the raw data point values, examine at least two other ones of the raw data point values obtained at points in time prior to a given one of the raw data point values being examined, and assigning different weights to each depending on which one is closer in time to the given one of the raw data point values being examined, to create weighted data point values;
• use the weighted data point values in connection with the algorithm to electronically determine a predicted data point value relative to the given one of the raw data point values being examined;
• use the predicted data point value and the given one of the raw data point values being examined to electronically assess the predicted data point value and the given one of the raw data point values to determine if the given one of the raw data point values being examined is an erroneous data point value;
• and
• causing the electronic processing system to use the weighted data point values and the predicted data point values to assess a health of the battery string for the purpose of making a more accurate determination if a failure of the battery string is likely to occur, and to enable action to be taken to avoid compromising an integrity of the battery powered component.