A company in the logistics field has a supply chain system that includes different robots for managing materials and orders in its warehouse.
The failure rate of some of the robots used in the order management chain is higher than desirable. Every time a failure occurs, the supply chain is interrupted, generating a great economic impact, which also influences delivery times and the customer experience.
To try to detect the causes that are producing an unexpected number of breakdowns, an intelligent system is developed that, through the analysis of process data, environmental conditions and use, is capable of determining the conditions that make more it favours the appearance of system failures, allowing the establishment of policies to reduce the number of unexpected failures.
METHODOLOGY & results
Architecture: Cloud architecture based on AWS, including AWS IoT and using RedShift as a data warehouse.
Developing language: Java, Python
ML techniques: Supervised learning algorithms (Decision Trees, Random Forest), unsupervised learning algorithms (HCA)
Results: Knowledge of the conditions that favour the appearance of failures allowed their control, reducing them by 30%.