Method of adjusting at least one defective rotor of a rotorcraft

1. A method of adjusting at least one defective main or anti-torque rotor ( 6 , 10 ) of a particular rotorcraft, comprising the steps of:

using a neural network that represents relationships between firstly accelerations representative of a vibration generated on at least a portion of a reference rotorcraft and secondly rotor defects and adjustment parameters, wherein each adjustment parameter is with respect to adjusting an element that affects the generated vibration;

determining any defects of said rotor, wherein each one of said determined defects relates to a corresponding element that generated the vibration;

determining an adjustment value α for at least one of said adjustment parameters by minimizing the following relationship:

∑

c

⁢

⁢

λ

c

⁢

∑

a

⁢

⁢

λ

a

⁢

(

∑

h

=

1

B

-

2

⁢

⁢

(

λ

h

⁢



R

c

,

a

,

h

⁡

(

α

)

+

γ

c

,

a

,

h



2

)

+

λ

B

-

1

(



R

c

,

a

,

B

-

1

⁡

(

α

)

+

γ

c

,

a

,

B

-

1



2



γ

c

,

a

,

B



)

2

)

in which:

γ c , γ a , γ h , and γ B−1 are weighting coefficients depending respectively on the stage of flight c, an accelerometer a, harmonics h, and the number of blades B of the rotor minus one;

R c,a,h is an output cell of the neural network corresponding to the harmonic h, the accelerometer a, and the stage of flight c;

R c,a,B−1 is an output cell of the neural network corresponding to the harmonic B−1, the accelerometer a, and the stage of flight c;

γ c,a,h is an acceleration signal at the frequency h measured by the accelerometer a for the stage of flight c;

γ c,a,B is an acceleration signal at the frequency B measured by the accelerometer a for the stage of flight c; and

γ c,a,B−1 is an acceleration signal at the frequency (B−1) measured by the accelerometer a for the stage of flight c; and

based on the determined adjustment value, making an adjustment of at least one element to reduce the generated vibration.

2. A method according to claim 1 , wherein said neural network is obtained from at least a first series of measurements using said reference rotorcraft ( 1 ) considered as being a deformable body, in which defect-free main and anti-torque rotors ( 6 , 10 ) are adjusted to a reference adjustment at which the vibration level of at least said portion ( 3 , 8 ) of said reference rotorcraft ( 1 ) is at a minimum.

3. A method according to claim 2 , wherein said first series of measurements is taken during a particular operation of said reference rotorcraft by measuring the values of at least one acceleration, said measurements are taken at arbitrary locations on said portion ( 3 , 8 ) of the reference rotorcraft and are representative of the vibration generated at said portion ( 3 , 8 ) of the reference rotorcraft:

a) firstly with the defect-free main or anti-torque rotor ( 6 , 10 ) of the reference rotorcraft ( 1 ) adjusted to said reference adjustment;

b) secondly by introducing defects in said defect-free main or anti-torque rotor ( 6 , 10 ); and

c) then by varying the adjustment values of a plurality of adjustment parameters of said main or anti-torque rotor ( 6 , 10 ).

4. A method according to claim 3 , wherein said first series of measurements is performed during at least the following test flights:

a reference flight with the main or anti-torque rotor ( 6 , 10 ) adjusted to said reference adjustment;

flights with defects in said main or anti-torque rotor ( 6 , 10 );

a flight with a particular wrong adjustment of at least one balance weight ( 25 ) of a blade ( 7 );

a flight with a particular wrong adjustment of at least one rod ( 27 ) of said blade ( 7 ); and

a flight with a particular wrong adjustment of at least one compensator tab ( 28 ) provided on the trailing edge ( 29 ) of said blade ( 7 ).

5. A method according to claim 4 , wherein at least one of said test flights includes the following stages, during which measurements are taken:

a stage of hovering flight;

a stage of cruising flight at about 50 m/s;

a stage of flight at maximum continuous power; and

a test on the ground with the main and the anti-torque rotors ( 6 , 10 ) turning.

6. A method according to claim 3 , wherein, in order to detect said defects of said rotor ( 6 , 10 ), the following steps are performed:

d) taking a second set of measurements on said particular rotorcraft ( 1 ) by measuring the values of at least some of said accelerations at said portion ( 3 , 8 ) of the particular rotorcraft during particular operation of said particular rotorcraft; and

e) detecting from said second series of measurements of accelerations and from the neural network, said defects of said rotor ( 6 , 10 ).

7. A method according to claim 1 , wherein adjustment elements defining said adjustment parameters comprise at least:

a balance weight ( 25 ) for each of the blades ( 7 ) of the rotor ( 6 , 10 );

a rod ( 27 ) on each of the blades ( 7 ) of the rotor ( 6 , 10 ), with the exception of a blade that represents a reference blade; and

a compensator tab ( 27 ) on the trailing edge ( 29 ) of each of the blades ( 7 ) of the rotor if it is the main rotor ( 6 ),

each of the balance weight, the rod, and the compensator tab being adjustable to affect the generated vibration.