| CPC G05B 23/0259 (2013.01) [G05B 13/0265 (2013.01); G05B 13/04 (2013.01); G05B 13/041 (2013.01); G05B 13/048 (2013.01); G05B 17/02 (2013.01); G06F 30/20 (2020.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01); G01N 33/18 (2013.01); G06F 2111/10 (2020.01); G06Q 50/06 (2013.01)] | 17 Claims |
|
1. A computer implemented method for detecting anomalies in a water distribution system composed of a network comprising a plurality of nodes and a plurality of arcs and a plurality of pieces of equipment, the pieces of equipment comprising at least valves and pumps, an entity referring to a set of at least one of the plurality of nodes and/or at least one of the plurality of arcs and/or at least one of the plurality of pieces of equipment, said method comprising:
parametrizing a hydraulic model of the water distribution system with a set of values of control variables;
using sensors of at least water velocity and pressure on the network to acquire observations of a subset of state variables at first time references; then
classifying at least one entity in a set of classes comprising at least one class representative of an abnormal state based on at least said observations; then
identifying at least one target entity where to change the values of control variables, said identifying depending at least on one entity classified in said at least one class representative of an abnormal state; then
changing the set of values of control variables of only said at least one target entity identified, using a stepwise adjustment of the control variables and a break criterion based on residue values of the state variables, wherein changing the set of values of control variables results in changing water velocity or pressure on the at least one target entity in the network; and then
performing a classification in a class representative of an abnormal state of at least one entity of the network according to the set of control variables issued from said changing step,
wherein changing the set of values of control variables using a stepwise adjustment of the control variables and a break criterion based on residue values of the state variables comprises:
A) changing the set of values of the control variables;
B) using the hydraulic model to calculate predicted values of a set of state variables comprising at least water velocity and pressure at the nodes at the time references;
C) computing residue values of the set of state variables as a difference between predicted values and observations at the time references;
D) if said differences satisfy a break criterion, going to step F);
E) if not, changing the set of values of the control variables and going back to step B);
F) if said differences do not satisfy a refinement criterion, selecting a subset of the network where to calculate predicted values, going back to step B),
wherein identifying at least one target entity where to change the values of control variables based on at least said observations comprises:
using the hydraulic model to calculate predicted values of a set of state variables comprising at least a water velocity and pressure at the nodes, said predicted values being associated to second time references;
computing residue values of the subset of the state variables as a difference between predicted values and observations at the second time references;
performing a statistical analysis of residue values at an entity of the network at a selection of the time references;
classifying the entities of the network based on rules applied to the output of the statistical analysis.
|