Granted Patent B2

US 12,339,195 · App. 17/991,009 · Granted Jun 24, 2025

Inventors: Simone Baliva (Saluzzo, IT); Marco Cambiano (Cumiana, IT)

Assignee: AKTIEBOLAGET SKF

G01M17/02B60B27/0068G06N3/02B60B2380/50

Abstract

Claims (62)

1. A system for accurate real-time detection of forces and moments acting on a wheel of a vehicle, comprising:

an outer ring of a wheel hub unit;

a plurality of strain sensors carried integral by the outer ring, the plurality of strain sensors configured to generate first electrical signals (Ds 1 ) as a function of respective mechanical stresses(S) acting on the outer ring;

a first processing unit configured to,

receive, upon deformation of the plurality of sensors, a plurality of said first electrical signals and a second electrical signal (Ds 2 ) proportional to an internal temperature of the wheel hub unit, and

generate a pseudo-sinusoidal signal comprising frequency and instantaneous amplitude proportional to the mechanical stresses;

a second processing unit configured to receive and process a plurality of third electrical signals (Dv 1 ,n) corresponding to selected parameters of the vehicle; and

a third processing unit configured to,

receive the plurality of third electrical signals and the pseudo-sinusoidal signal,

calculate, in real time, on the basis of the received plurality of third electrical signals and the pseudo-sinusoidal signal, forces and moments applied to the vehicle wheel,

extrapolating the forces and moments on the basis of a series of data previously fed to said third processing unit and corresponding to forces and moments previously experimentally detected on the vehicle wheel following the execution with the vehicle of a series of predetermined maneuvers.

2. The detection system of claim 1 , wherein:

each one of the plurality of strain sensors comprises at least four pairs of strain gauges,

each pair of the strain gauges is arranged in correspondence with a respective elastic strain zone of the outer ring,

the respective elastic strain zone includes a predetermined circumferential and radial extent; and

each pair of the strain gauges are piezoelectric.

3. The detection system of claim 1 , wherein:

the third processing unit is configured to process the plurality of third electrical signals and the pseudo-sinusoidal signal based on a calibration map, and

the calibration map is previously empirically determined based on a non-linear regression function and a database based on a data set corresponding to forces and moments experimentally detected on the vehicle wheel.

4. The detection system of claim 1 , wherein:

the third processing unit is configured to process the plurality of third electrical signals and the pseudo-sinusoidal signal through a shallow neural network, and

the shallow neural network is previously trained based on a data set corresponding to forces and moments experimentally detected on the vehicle wheel.

5. The detection system of claim 4 , wherein:

the shallow neural network is of a non-linear regression approach type and comprises at least a first layer and a second layer, and

the first layer is formed of at least six neurons and the second layer is formed of at least three neurons.

6. The detection system of claim 4 , wherein the shallow neural network is configured to receive and process signals corresponding to a set of predetermined maneuvers, the signals comprising at least one of:

signals resulting from at least three ABS braking events with at least 1 g of longitudinal deceleration in a straight line;

signals resulting from at least three 0.7 g braking events in a straight line;

signals resulting from at least three 0.5 g braking events in a straight line;

signals resulting from at least two double line lane changes with 1 g or more of lateral acceleration;

signals consequent to the execution of a plurality of complete vehicle turns with constant speed and increasing steering radius until the loss of adherence, both clockwise and counterclockwise; and

signals resulting from performing at least five minutes of mixed normal driving and drifting during turns.

7. The detection system of claim 4 , wherein the shallow neural network is configured to receive and process signals corresponding to the selected vehicle state parameters, the signals comprising at least one of:

vehicle lateral acceleration signals;

vehicle longitudinal acceleration signals;

vehicle steering angle signals;

vehicle speed signals; and

brake pressure signals.

8. The detection system of claim 1 , wherein the first processing unit comprises a first electronic board carried by the wheel hub unit.

9. The detection system of claim 1 , wherein,

the second and third processing units are selected from the group consisting of: a same second electronic board suitably programmed, a second and a third electronic board suitably programmed, and one or more software programs loaded on a vehicle control unit or ECU; and

the third processing unit is configured to write to a register of the vehicle ECU at a sampling interval of the order of ten milliseconds, a set of data comprising respective forces (Fx and Fz) and a moment (Mx) acting on the vehicle wheel,

wherein x is an axis parallel to a direction of vehicle motion, y is an axis transverse to the direction of vehicle motion and z is a substantially vertical axis perpendicular to the x- and y-axes.

10. A method for accurately detecting, in real time, forces and moments acting on a wheel of a vehicle carried by a wheel hub unit, comprising:

a first step in which a plurality of first electrical signals and second electrical signals are processed to generate a pseudo-sinusoidal signal comprising an instantaneous frequency and amplitude proportional to mechanical stresses;

a second step of detecting and processing a plurality of third electrical signals (Dv 1 ,n) proportional to selected vehicle state parameters; and

a third step in which each of the plurality of third electrical signals and the pseudo-sinusoidal signal are processed to provide in real time the forces and moments (Fx,y;Mx) applied to the vehicle wheel, the third step being performed on the basis of a series of data comprising values of forces and moments experimentally detected on the vehicle wheel following execution of a series of predetermined maneuvers by the vehicle,

wherein the wheel hub unit comprises an outer ring provided with a plurality of strain sensors integral therewith, each of the plurality of strain sensors being configured to generate first electrical signals (Ds 1 ) as a function of respective mechanical stresses(S) acting locally on the outer ring, and a temperature sensor configured to generate a second electrical signal (Ds 2 ) proportional to an internal temperature of the wheel hub unit.

11. The method of claim 10 , wherein the third step is performed by a shallow neural network trained by data set.

12. The method of claim 10 , wherein the set of predetermined maneuvers comprises:

at least three ABS braking events with at least 1 g longitudinal deceleration in a straight line;

at least three 0.7 g braking events in a straight line;

at least three 0.5 g braking events in a straight line;

at least two double lane changes with 1 g or more of lateral acceleration;

performing a plurality of complete vehicle turns with constant speed and increasing turning radius until loss of grip, both clockwise and counterclockwise; and

at least five minutes of mixed normal driving and of drifting during turns.

13. The method of claim 11 , wherein the selected vehicle state parameters comprise at least one of:

lateral acceleration of the vehicle;

longitudinal acceleration of the vehicle;

steering angle of the vehicle;

vehicle speed; and

brake pressure.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (13)