US 11,694,124 B2
Artificial intelligence (AI) based predictions and recommendations for equipment
Rajarajan Thangavel Ramalingam, Bangalore (IN); Vladimir Valeryevich Ryabovol, Windham, NH (US); Auri Priyadharshini Munivelu, Bangalore (IN); Ramanathan Lakshmanan, Bangalore (IN); Ravi Kanth Vinnakota, Bangalore (IN); Sunil Kumara D S, Hassan District (IN); Basavaraj Chidanandappa, Bangalore (IN); and Venkata Rama Krishna Perumalla, Chebrole (IN)
Assigned to ACCENTURE GLOBAL SOLUTIONS LIMITED, Dublin (IE)
Filed by ACCENTURE GLOBAL SOLUTIONS LIMITED, Dublin (IE)
Filed on Jun. 14, 2019, as Appl. No. 16/441,941.
Prior Publication US 2020/0394533 A1, Dec. 17, 2020
Int. Cl. G06N 20/00 (2019.01); G06N 20/20 (2019.01); G06Q 50/04 (2012.01); G06Q 30/0201 (2023.01); G06F 18/23 (2023.01); G06F 18/2113 (2023.01); G06F 18/21 (2023.01)
CPC G06N 20/20 (2019.01) [G06F 18/2113 (2023.01); G06F 18/2185 (2023.01); G06F 18/23 (2023.01); G06N 20/00 (2019.01); G06Q 30/0206 (2013.01); G06Q 50/04 (2013.01)] 19 Claims
OG exemplary drawing
 
1. An Artificial Intelligence (AI) based equipment attribute prediction system comprising:
at least one processor;
a non-transitory processor readable medium storing machine-readable instructions that cause the at least one processor to:
access historical data related to at least one attribute of at least one customized equipment wherein the historical data includes data values tracking the equipment attribute and features of the equipment attribute over different time periods;
extract the features of the equipment attribute from the historical data using text processing techniques;
generate a plurality of feature combinations that include each possible combination of the features of the equipment attribute from the historical data, the plurality of feature combinations are generated based on a data type of the features and operators that can be applied to the data types;
train a plurality of AI models implementing different methodologies for generating predictions for the equipment attribute;
obtain a plurality of attribute predictions for the equipment attribute from the plurality of AI models, by applying each of the possible feature combinations to each of the plurality of AI models;
select a top N AI models and corresponding possible feature combinations from the plurality of AI models wherein N is a natural number, wherein the top N AI models are selected using model selection criteria that include an error estimation of the plurality of attribute predictions and multi-sample testing for consistency;
identify a best scoring AI model of the top N AI models and a particular feature combination from the possible feature combinations for the equipment attribute using a scoring scheme;
load into the non-transitory, processor-readable medium a single instance of the best scoring AI model in response to a user request for prediction for the equipment attribute for a single piece of the at least one customized equipment;
obtain from the single instance of the best scoring AI model and the particular feature combination, an attribute prediction for the equipment attribute in accordance with a user request wherein the user request includes new data values for the features included in the possible feature combination corresponding to the best scoring AI model; and
derive at least one recommendation by analyzing coefficients of an equation representing the best scoring AI model based on the possible feature combination corresponding to the best scoring AI model and the attribute prediction;
calculate variance between an actual value of the equipment attribute and the attribute prediction; and
further train the best scoring AI model on the actual value based at least on the variance.