CN110794682A - Thrust distribution method for multi-propeller rotatable ship - Google Patents

Abstract

The present invention proposes a thrust distribution method for a rotatable multi-screw ship. First, a three-degree-of-freedom thrust distribution model of the ship is established, and then a multi-objective optimization objective regarding propeller energy consumption, mechanical wear, singularity and thrust error is established. Finally, the proposed problem is solved by an improved fast non-dominated genetic algorithm with elite retention strategy. The advantages are: the energy consumption and mechanical loss of the propulsion system can be effectively reduced, and the precision of ship control can be guaranteed.

Description

A thrust distribution method for a rotatable multi-propeller ship

technical field

The invention relates to the field of ship control, in particular to a thrust distribution method for a rotatable multi-propeller ship.

Background technique

The rotatable propeller can rotate around the axis, and can obtain the maximum thrust in any direction. It can make the ship turn on the spot, move laterally, and retreat rapidly and other special driving operations. The rotatable propeller is suitable for all kinds of engineering ships, such as tugboats, floating cranes, dredgers, ferries, working flat-bottomed boats, etc., and has broad market application prospects and military significance. In order to improve the maneuverability, efficiency and system reliability of these engineering ships, more than two propellers are usually configured. However, on the one hand, due to the interaction between the inertial force of the ship and the thrust of the propeller, the multi-propeller propulsion operating state presents asymmetric and irregular dynamic turbulence characteristics, and the superposition of the irregular characteristics of each propeller also brings about the ship heading control On the other hand, the propulsion load fluctuates violently under the action of sea conditions, which makes the power distribution among the propellers extremely unbalanced, which brings difficulties to the speed control of the ship. Therefore, how to precisely distribute the thrust and angle of each rotatable propeller becomes a key issue.

SUMMARY OF THE INVENTION

The present invention proposes a thrust distribution method for a rotatable multi-propeller ship. First, a three-degree-of-freedom thrust distribution model of the ship is established according to conditions, and then a multi-objective optimization of propeller energy consumption, wear, singularity and thrust error is established. The objective function and constraints are obtained. Finally, the proposed problem is solved by an improved fast non-dominated multi-objective optimization algorithm with elite retention strategy.

It mainly includes the following steps:

Step 1. Establish a three-degree-of-freedom thrust distribution model;

The motions in the heave, roll, and pitch directions are ignored, and the ship dynamics model of the three-degree-of-freedom motion in the heave, roll, and yaw directions can be summarized as follows:

Among them, η=[x, y, ψ] ^T is the actual position vector of the ship in the geodetic coordinate system X _E OY _E , and η _d is the expected position vector. v=[u, v, r] ^T is the actual velocity vector in the hull coordinate system XOY, and v _d is the desired velocity vector. R(ψ) is the rotation matrix, M is the inertia matrix of the hull, C(v) is the Coriolis centripetal force matrix, and D(v) is the damping matrix. τ is the actual resultant force and resultant moment generated by the ship propulsion system, and _τd is the expected resultant force and resultant moment.

For ships equipped with swivel multi-propellers, τ consists of the working state of the propeller and the thrust structure:

τ=B(α)f=B(α)[f ₁ , f ₂ , f ₃ , f ₄ ] ^T (3)

Among them, B(α)∈R ^3×4 thrust structure matrix, f is the thrust matrix composed of the thrust of each propeller, f _i is the thrust of the propeller i. The thrust structure matrices of the azimuth propeller and the ducted propeller are B _azi (α _i ) and B _tun , respectively:

B(α) ₌ [ _Bazi (α1),Bazi( _α2 ), _{Bazi(α3),Btun ] ( 5} )

i is the serial number of the propeller, i=1,2,3,4. α _i is the angle of the propeller i. l _i =[l _xi , l _yi ] ^T is the position coordinate of the propeller i installed on the hull under the hull coordinate system XOY, l _xi is the X coordinate of the propeller i, and l _yi is the Y coordinate of the propeller i.

Step 2. Establish thrust distribution constraints;

Among them, the maximum output forward and reverse thrusts of the propeller i are f _imax and f _imin , respectively, f _i0 is the thrust generated by the propeller i in the previous step, and f _i is the thrust at the current moment. α _imin and α _imax are the upper and lower limits of the propeller rotation angle, respectively, α _i0 is the angle of the propeller i in the previous step, and α _i is the angle of the propeller at the current moment. _Δfimin and _Δfimax are the lower and upper limits of the thrust change rate of the propeller i, respectively. Δα _imin and Δα _imax are the lower limit and the upper limit of the angle change rate of the propeller i, respectively.

Step 3. Establish thrust distribution optimization objective;

Among them, P _w is the overall power consumption of the multi-propeller, c _i is the power coefficient of the propeller i; J _e is the thrust error penalty term, s is the error between the actual thrust and the ideal thrust, and Q is the weight matrix of the error penalty term. J _Ω is the angle change penalty item, and Ω is the weight matrix of the angle change penalty item. J _s is the singular structure penalty term, δ is the weight matrix of the singular structure penalty term, and ε is a constant.

Step 4. Use the improved fast non-dominated sorting genetic algorithm with elite retention strategy to solve the thrust distribution objective function, including the following sub-steps:

4-1) Select the angle and thrust of the propeller as decision variables, and design the chromosome Ch={f ₁ , f ₂ , f ₃ , f ₄ , α ₁ , α ₂ , α ₃ }; initialize the population to generate a population size of N Parent population P _t , the initial value of t is 0;

4-2) Perform polynomial mutation on the parent population P _t to generate the offspring population Q _t ;

4-3) fuse the parent population P _t with the child population Q _t to obtain a temporary fusion population R _t ;

4-4) Perform fast non-dominated sorting on R _t to obtain different Pareto frontiers F _j . N individuals in the population are divided into at most N Pareto frontiers, that is, j=1,2...N;

4-5) Arrange the individuals in each frontier F _j in descending order of crowding distance. The calculation method of crowding distance _Dk is as follows:

where |F _j | is the number of individuals in frontier F _j .

4-6) There are no individuals in the initial population of P _t+1 . The first NP _t+1 of F _j is selected and placed in P _t+1 . If F _j +P _t+1 <N, P _t+1 =P _t+1 ∪F _j , j=j+1, return to 4-4); otherwise, return to 4-5);

4-7) G _max is the maximum algebra of evolution, if t ≥ G _max , output the optimal solution set, and the optimization algorithm ends; otherwise, t=t+1, perform crossover and differential mutation operations on P _t to generate a population Q _t , and repeat steps 4-3) until the end. The way of differential mutation is as follows:

Q _t =βP _best +(1-β)P _t (9)

Among them, _β∈ [0,1], Pbest is the _best individual in frontier F1.

This method has the following effects and advantages:

The thrust obtained by the proposed thrust distribution method does not have frequent jumps, and can accurately track the expected force and moment of the ship. The thrust distribution method controls the angle of the propeller to not change repeatedly, and avoids the mechanical wear of the propulsion device; it can effectively reduce the energy consumption of the propulsion system, and can ensure the precision and real-time performance of the ship control.

Description of drawings

Figure 1 is a schematic diagram of the thrust distribution of a multi-propeller ship

Figure 2 is a schematic diagram of the distribution of multi-propellers

Fig. 3 Flow chart of improved fast non-dominated sorting genetic algorithm with elite retention strategy

Detailed ways

The present invention proposes a thrust distribution method for a rotatable multi-propeller ship. First, a three-degree-of-freedom thrust distribution model of the ship is established according to conditions, and then a multi-objective optimization of propeller energy consumption, wear, singularity and thrust error is established. The objective function and constraints are obtained. Finally, the proposed problem is solved by an improved fast non-dominated multi-objective optimization algorithm with elite retention strategy. Include the following steps:

Step 1. Establish a three-degree-of-freedom thrust distribution model;

The motions in the heave, roll, and pitch directions are ignored, and the ship dynamics model of the three-degree-of-freedom motion in the heave, roll, and yaw directions can be summarized as follows:

Among them, η=[x, y, ψ] ^T is the actual position vector of the ship in the geodetic coordinate system X _E OY _E , and η _d is the expected position vector. v=[u, v, r] ^T is the actual velocity vector in the hull coordinate system XOY, and v _d is the desired velocity vector. R(ψ) is the rotation matrix, M is the inertia matrix of the hull, C(v) is the Coriolis centripetal force matrix, and D(v) is the damping matrix. τ is the actual resultant force and resultant moment generated by the ship propulsion system, and _τd is the expected resultant force and resultant moment.

As shown in Figure 1, the purpose of the thrust distribution algorithm is to distribute the control force and torque calculated by the ship motion controller to each propeller, so that the actual resultant force and resultant moment τ generated by each propeller can satisfy τ _d .

For ships equipped with rotatable multi-propellers, the configuration of the propellers is shown in Figure 2, and τ consists of the working state of the propellers and the thrust structure:

τ=B(α)f=B(α)[f ₁ , f ₂ , f ₃ , f ₄ ] ^T (3)

Among them, B(α)∈R ^3×4 thrust structure matrix, f is the thrust matrix composed of the thrust of each propeller, f _i is the thrust of the propeller i. Due to the fixed direction of the ducted propeller, its thrust structure matrix is only related to the position of the propeller. The thrust structure matrices of the azimuth propeller and the ducted propeller are B _azi (α _i ) and B _tun , respectively:

B(α) ₌ [ _Bazi (α1),Bazi( _α2 ), _{Bazi(α3),Btun ] ( 5} )

i is the serial number of the propeller, i=1,2,3,4. α _i is the angle of the propeller i. l _i =[l _xi , l _yi ] ^T is the position coordinate of the propeller i installed on the hull under the hull coordinate system XOY, l _xi is the X coordinate of the propeller i, and l _yi is the Y coordinate of the propeller i.

Step 2. Establish thrust distribution constraints;

Among them, the maximum output forward and reverse thrusts of the propeller i are f _imax and f _imin , respectively, f _i0 is the thrust generated by the propeller i in the previous step, and f _i is the thrust at the current moment. α _imin and α _imax are the upper and lower limits of the propeller rotation angle, respectively, α _i0 is the angle of the propeller i in the previous step, and α _i is the angle of the propeller at the current moment. Δf _imin and Δf _imax are the lower and upper limits of the thrust change rate of the propeller i, respectively, and Δα _imin and Δα _imax are the lower and upper limits of the angular change rate of the propeller i, respectively.

Step 3. Establish thrust distribution optimization objective;

Among them, P _w is the overall power consumption of the multi-propeller, c _i is the power coefficient of the propeller i; J _e is the thrust error penalty term, s is the error between the actual thrust and the ideal thrust, and Q is the weight matrix of the error penalty term. J _Ω is the angle change penalty item, and Ω is the weight matrix of the angle change penalty item. J _s is the singular structure penalty term, δ is the weight matrix of the singular structure penalty term, and ε is a constant. The goal of thrust distribution is to reduce the overall power consumption of the ship's propeller and avoid mechanical wear and thrust singularity under the premise of ensuring that the thrust error is sufficiently small.

Step 4. Use the improved fast non-dominated sorting genetic algorithm with elite retention strategy to solve the thrust distribution objective function, and the algorithm flow chart is shown in Figure 3. Step 4 includes the following sub-steps:

4-1) Select the angle and thrust of the propeller as decision variables, and design the chromosome Ch={f ₁ , f ₂ , f ₃ , f ₄ , α ₁ , α ₂ , α ₃ }; initialize the population to generate a population size of N Parent population P _t , the initial value of t is 0;

4-2) Perform polynomial mutation on the parent population P _t to generate the offspring population Q _t ;

4-3) fuse the parent population P _t with the child population Q _t to obtain a temporary fusion population R _t ;

4-4) Perform fast non-dominated sorting on R _t to obtain different Pareto frontiers F _j . N individuals in the population are divided into at most N Pareto frontiers, that is, j=1,2...N;

4-5) Arrange the individuals in each frontier F _j in descending order of crowding distance. The calculation method of crowding distance _Dk is as follows:

where |F _j | is the number of individuals in frontier F _j .

4-6) There are no individuals in the initial population of P _t+1 . The first NP _t+1 of F _j is selected and placed in P _t+1 . If F _j +P _t+1 <N, P _t+1 =P _t+1 ∪F _j , j=j+1, return to 4-4); otherwise, return to 4-5);

4-7) G _max is the maximum algebra of evolution, if t ≥ G _max , output the optimal solution set, and the optimization algorithm ends; otherwise, t=t+1, perform crossover and differential mutation operations on P _t to generate a population Q _t , and repeat steps 4-3) until the end. The way of differential mutation is as follows:

Q _t =βP _best +(1-β)P _t (9)

Among them, _β∈ [0,1], Pbest is the _best individual in frontier F1.

Claims (1)

1. a thrust distribution method for rotatable multi-screw ship, is characterized in that: Step 1. Establish a three-degree-of-freedom thrust distribution model; The motions in the heave, roll, and pitch directions are ignored, and the ship dynamics model of the three-degree-of-freedom motion in the heave, roll, and yaw directions can be summarized as follows:

Among them, η=[x,y,ψ] ^T is the actual position vector of the ship in the geodetic coordinate system X _E OY _E ,

is the first derivative of η, η _d is the desired position vector; v=[u, v, r] ^T is the actual velocity vector in the hull coordinate system XOY,

is the first derivative of v, v _d is the desired velocity vector; R(ψ) is the rotation matrix, M is the inertia matrix of the hull, C(v) represents the Coriolis centripetal force matrix, D(v) is the damping matrix; τ is the ship The actual resultant force and resultant moment generated by the propulsion system, τ _d is the expected resultant force and resultant moment; For ships equipped with swivel multi-propellers, τ consists of the working state of the propeller and the thrust structure: τ=B(α)f=B(α)[f ₁ , f ₂ , f ₃ , f ₄ ] ^T (3) Among them, B(α)∈R ^3×4 thrust structure matrix, f is the thrust matrix composed of the thrust of each propeller, f _i is the thrust of the propeller i; the thrust structure matrix of the azimuth propeller and the ducted propeller are B _azi ( α _i ) and B _tun :

B(α) ₌ [ _Bazi (α1),Bazi( _α2 ), _{Bazi(α3),Btun ] ( 5} ) i is the serial number of the propeller, i=1, 2, 3, 4; α _i is the angle of the propeller i; l _i = [l _xi , l _yi ] ^T is the position coordinate of the propeller i installed on the hull under the hull coordinate system XOY , l _xi is the X coordinate of the propeller i, and l _yi is the Y coordinate of the propeller i; Step 2. Establish thrust distribution constraints;

Among them, the maximum output forward and reverse thrusts of propeller i are f _imax and f _imin , respectively, f _i0 is the thrust generated by propeller i in the previous step, f _i is the thrust at the current moment; α _imin and α _imax are the rotation angle of the propeller, respectively. Upper and lower limits, α _i0 is the angle of the propeller i in the previous step, α _i is the angle of the propeller at the current moment; Δf _imin and Δf _imax are the lower and upper limits of the thrust change rate of the propeller i, respectively, Δα _imin and Δα _imax are the propeller i The lower and upper limits of the angular rate of change; Step 3. Establish thrust distribution optimization objective;

Among them, P _w is the overall power consumption of the multi-propeller, ci is the power coefficient of the propeller _i ; J _e is the thrust error penalty term, s is the error between the actual thrust and the ideal thrust, and Q is the weight matrix of the error penalty term; J _Ω is the angle change penalty term, Ω is the weight matrix of the angle change penalty term; J _s is the singular structure penalty term, δ is the weight matrix of the singular structure penalty term, and ε is a constant; Step 4. Use the improved fast non-dominated sorting genetic algorithm with elite retention strategy to solve the thrust distribution objective function, including the following sub-steps: 4-1) Select the angle and thrust of each propeller as decision variables to design chromosomes; initialize the population to generate a parent population P _t with a population size of N, and the initial value of t is 0; 4-2) Perform polynomial mutation on the parent population P _t to generate the offspring population Q _t ; 4-3) fuse the parent population P _t with the child population Q _t to obtain a temporary fusion population R _t ; 4-4) Perform fast non-dominated sorting on R _t to obtain different Pareto frontiers F _j , and the maximum value of j is N; 4-5) Arrange the individuals in each front F _j in descending order of crowding distance; the calculation method of crowding distance _Dk is as follows:

where |F _j | is the number of individuals in frontier F _j ; 4-6) There is no individual in the initial population of P _t+1 ; select the former NP _t+1 of F _j and put it into P _t+1 ; if F _j +P _t+1 <N, P _t+1 =P _{t +1} ∪F _j , i=i+1, return to execute 4-4); otherwise, return to execute 4-5); 4-7) G _max is the maximum algebra of evolution, if t ≥ G _max , output the optimal solution set P _t , and the optimization algorithm ends; otherwise, t=t+1, perform crossover and differential mutation operations on P _t to generate a population Q _t , and cyclically execute steps 4-3) until the end; the method of differential mutation is as follows: Q _t =βP _best +(1-β)P _t (9) Among them, _β∈ [0,1], Pbest is the _best individual in frontier F1.

Images