US 12,265,199 B2
Microelectromechanical system and corresponding method for weather pattern recognition
Stefano Paolo Rivolta, Desio (IT); Lorenzo Bracco, Chivasso (IT); Roberto Mura, Milan (IT); and Federico Rizzardini, Settimo Milanese (IT)
Assigned to STMICROELECTRONICS S.r.l., Agrate Brianza (IT)
Filed by STMICROELECTRONICS S.r.l., Agrate Brianza (IT)
Filed on Nov. 16, 2020, as Appl. No. 17/099,582.
Claims priority of application No. 102019000022188 (IT), filed on Nov. 26, 2019.
Prior Publication US 2021/0158190 A1, May 27, 2021
Int. Cl. G06F 18/214 (2023.01); G01W 1/02 (2006.01); G06F 18/243 (2023.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G16Y 20/10 (2020.01)
CPC G01W 1/02 (2013.01) [G06F 18/2148 (2023.01); G06F 18/24323 (2023.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G16Y 20/10 (2020.01)] 18 Claims
OG exemplary drawing
 
1. A microelectromechanical system, comprising:
at least one microelectromechanical movement sensor that is configured to, in operation, generate a movement signal in response to a weather condition of an environment in which the at least one microelectromechanical movement sensor is positioned; and
a processing circuitry that includes:
a recognition circuitry coupled to the at least one microelectromechanical movement sensor, and the recognition circuitry configured to:
receive the movement signal generated by the at least one microelectromechanical movement sensor at an input of the recognition circuitry;
filter the movement signal received at the input of the recognition circuitry coupled to the at least one microelectromechanical movement sensor, the movement signal is filtered into a filtered movement signal that includes one or more respective components;
extract a plurality of features from the filtered movement signal filtered from the movement signal received at the input of the recognition circuitry; and
determine a plurality of weather patterns including a weather pattern of the weather condition by processing the plurality of features, wherein processing the plurality of features includes:
implementing at least one machine-learning algorithm trained to recognize the weather condition, wherein implementing the at least one machine-learning algorithm includes implementing a classification algorithm to classify the plurality of features into classes each representative of a respective weather pattern of the plurality of weather patterns, via assignment criteria defined by the classification algorithm, wherein the classification algorithm uses a decision tree, and using the decision tree includes:
 comparing a first feature of the plurality of features to a first value at a first node of the decision tree:
 when the first feature is greater than the first value, initiating performing a first branch of the decision tree by comparing a second feature to a second value at a second node along the first branch; and
 when the first feature is less than or equal to the first value, initiating performing a second branch of the decision tree by comparing a third feature to a third value at a third node along the second branch.