Power distribution method for stable and direct navigation of underwater chained robot

Abstract

The invention discloses a power distribution method for stable and direct navigation of an underwater chained robot. The method mainly comprises two steps. The method comprises the following steps: firstly, giving driving distribution characteristics and initial and target states of the underwater chained robot, and optimizing and deciding the driving force to be adopted by each unit in the next execution cycle based on a genetic algorithm; and 2, considering unit driving force input saturation conditions, establishing a penalty function, adding the penalty function into a genetic algorithm power optimization objective function, determining the magnitude of the driving force actually adopted by each unit in the next execution period, and restoring the spacing of each unit to an ideal state as soon as possible under the condition of considering constraints. Through effective combination of the genetic algorithm and the penalty function, rapid and stable direct navigation of the underwater chained robot under any disturbance or initial state is realized. The method has the advantages of being clear in principle, effective in method verification and the like, can be widely applied to power distribution tasks of stable and direct navigation of the power distribution type multi-body underwater robot, and can cope with interference of different ocean current environments.

Description

Technical Field

[0001] The invention relates to the field of underwater robot control, in particular to a power distribution method for an underwater chain-type robot to sail steadily and straight. Background Art

[0002] Autonomous underwater vehicles have the advantages of large range of activities, good maneuverability, and high intelligence, and have become an important tool for completing various underwater tasks. However, with the increasing complexity of tasks, the limited load carrying capacity of conventional AUVs has seriously restricted their role. In addition, large AUVs with strong load carrying capacity have shortcomings in concealment, transportation, storage, and deployment. Therefore, new concept underwater robots that take into account strong load carrying capacity, safety and concealment, and easy operation have received widespread attention in recent years.

[0003] The underwater chain robot is composed of multiple AUV units connected in series, and each unit can carry a load, which can effectively take into account the above-mentioned characteristic requirements for underwater robots. In addition, the underwater chain robot also has advantages in system fault tolerance, sensor sharing, data collaboration, etc. The driving propellers of the underwater chain robot are located on both sides of each AUV unit, and have a distributed power structural feature. In order to avoid rigid collisions between units, retractable springs are installed between each unit. However, system characteristics such as power distribution, coupling force, and relative displacement bring difficulties to the overall speed control and unit spacing control of the underwater chain robot, resulting in challenges in smooth and straight navigation.

[0004] In order to cope with different ocean current environment interference, it is of great significance to study the power distribution method for the underwater chain robot to sail smoothly and straight, especially in the presence of drive saturation constraints and uncertain ocean current interference. However, for underwater chain robots composed of more than three units, there is currently no method that can complete the power distribution of smooth straight navigation based on actual interference. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to provide a power distribution method for the smooth and straight navigation of an underwater chain robot, establish a dynamic model of straight navigation propulsion, and provide a power distribution decision method with drive saturation constraints and uncertain ocean current interference, so as to provide support for the smooth and straight navigation of the underwater chain robot.

[0006] The present invention is realized by the following technical scheme: a power distribution method for a stable and straight navigation of an underwater chain robot, the steps of which are:

[0007] Step 1: Given the driving distribution characteristics, initial state and target state of the underwater chain robot, the driving force size to be adopted by each unit in the next execution cycle is optimized based on the genetic algorithm;

[0008] The dynamic model of the underwater chain robot is established. Given the initial state and the target state, the power distribution result that minimizes the fitness function is selected based on the selection, crossover and mutation operations of the genetic algorithm. The fitness function is:

[0009]

[0010] in, and L b They are the unit spacing after one cycle of power distribution and the ideal unit spacing, and v o They are the overall speed and target overall speed after one cycle of power distribution execution, α is the weighting factor of spacing deviation, β is the weighting factor of overall speed deviation, and the optimization goal is to minimize the adaptation function J a The value of .

[0011] The dynamic model of the underwater chain robot is established. Given the initial state and the target state, the power distribution result that minimizes the fitness function is selected based on the selection, crossover and mutation operations of the genetic algorithm. The fitness function is:

[0012]

[0013] in, and L b They are the unit spacing after one cycle of power distribution and the ideal unit spacing, and v o They are the overall speed and target overall speed after one cycle of power distribution execution, α is the weighting factor of spacing deviation, β is the weighting factor of overall speed deviation, and the optimization goal is to minimize the adaptation function J a The value of .

[0014] The selection, crossover and mutation operations based on genetic algorithms are as follows:

[0015] (1) Selection: Using the random traversal sampling selection operator, the power allocation individuals with higher fitness are selected with a higher probability from the randomly generated power population to form a new power population;

[0016] (2) Crossover: Using the crossover operator of single-point crossover, a crossover point is randomly selected from the binary-coded power group obtained by the selection operation, and the binary codes are exchanged to obtain a new generation of power groups;

[0017] (3) Mutation: A bit-flipping mutation operator is used to randomly select mutation bits from the binary-coded dynamic population obtained by the crossover operation, and then perform a binary bit inversion operation to obtain a new generation of population.

[0018] The process of establishing the underwater chain robot dynamics model is as follows:

[0019] Under ideal conditions, the forces acting on each unit include driving force, resistance, and reaction force of adjacent units. The driving force is generated by the propellers on both sides of the AUV unit, the resistance is generated by the fluid resistance, and the reaction force is generated by the spring deformation. Define n as the number of underwater chain robot units, L a is the length of each unit, u i , l i and r i They are the total driving force of unit i, the driving force of the left propeller and the driving force of the right propeller, respectively. When flying straight, l i =r i and u i = l i +r i , w i is the fluid resistance of unit i, f i(i+1) is the force exerted by unit i on unit (i+1), f (i+1)i is the force exerted by unit (i+1) on unit i.

[0020] Ideal conditions include:

[0021] The shape, center of gravity, center of buoyancy, and driving position of each unit in the underwater chain robot are the same, and the gravity is equal to the buoyancy in the initial state; the underwater chain robot maintains horizontal and straight navigation during movement; the gravity and buoyancy of the connecting mechanism between units are negligible; the driving force effects on both sides of each unit are synchronous and equal in size; the spacing between the units of the underwater chain robot is equal when the springs between the units are not deformed; the driving force direction of each unit is forward, and the performance is the same.

[0022] According to Newton's second law, the dynamic equations of each unit are as follows:

[0023]

[0024] Among them, m i 、v i and Represent the mass, velocity and acceleration of the unit respectively. If the spring coefficient between units is h i(i+1) , then the above formula can be transformed into:

[0025]

[0026] Among them, x i(i+1) and L b are the current spacing of each unit and the ideal spacing of each unit, (x (i-1)i -L b )≤0 and (x (i-1)i -L b)>0 corresponds to the compression and extension states of the spring respectively. The power distribution of the underwater chain robot for smooth straight navigation is realized based on the power optimization distribution of the genetic algorithm and the power distribution constraint with the penalty function, which can cope with random disturbances and arbitrary initial states.

[0027] Step 2: Establish a penalty function and add it to the objective function of the genetic algorithm power optimization to determine the actual driving force used by each unit in the next execution cycle.

[0028] If the saturation constraint force of each unit is T max , then the driving force u allocated to each unit i Should meet: i ≤T max Since genetic algorithms are difficult to handle constraints in optimization problems, in order to convert constraints into unconstrained expressions, a method that can i >T max A function that penalizes the power allocation result, namely the penalty function, and adds it to the fitness function of the genetic algorithm;

[0029] The penalty function also adopts a quadratic form. Considering the minimization optimization mode of the fitness function, the specific terms are expressed as

[0030]

[0031] Among them: J b is the penalty function, γ is the penalty factor, for the above penalty function: when u i The value is not greater than T max When u i Value greater than T max When the sub-item value is u i -T max , resulting in penalties, the fitness function takes a larger value,

[0032] Fitness function J with penalty function added a 'The specific form is as follows:

[0033]

[0034] The beneficial effects of the present invention are:

[0035] 1. Based on the established direct-flight propulsion dynamics model, the power distribution result is determined by genetic algorithm, so that the underwater chain robot can quickly move towards a stable and direct flight.

[0036] 2. Establish a penalty function and add it to the genetic algorithm objective function to optimize the power distribution result so that the underwater chain robot can still tend to sail smoothly and straight when there are driving force constraints.

[0037] 3. The present invention can handle the power distribution problem of underwater chain robots under the conditions of driving force constraints and uncertain ocean current interference, and achieve high reliability in achieving smooth and direct navigation.

[0038] 4. The present invention can provide theoretical support for the research on power distribution for the smooth and straight navigation of underwater chain robots. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of an underwater chain robot composed of five underwater robot units;

[0040] Figure 2 is a flow chart of the method of the present invention;

[0041] Figure 3 This is a schematic diagram of the force situation of the underwater chain robot;

[0042] Figure 4 This is a graph showing the change in power distribution, unit spacing and overall speed of a 5-unit underwater chain robot heading for steady straight navigation under the action of the genetic algorithm.

[0043] Figure 5 This is a diagram showing the change in power distribution and unit spacing when a 5-unit underwater chain robot is heading for steady straight navigation under the interference of ocean currents.

[0044] Figure 6 This is a graph showing the change in power distribution and unit spacing when a 5-unit underwater chain robot tends to sail smoothly and straight ahead when a penalty function is added. DETAILED DESCRIPTION

[0045] The present invention will be further described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0047] In order to achieve the smooth and straight navigation of the underwater chain robot, it is necessary to first establish the dynamic model of the system. Under ideal conditions, the forces acting on each unit include driving force, resistance, and reaction force of adjacent units. The driving force is generated by the propellers on both sides of the AUV unit, the resistance is generated by the fluid resistance, and the reaction force is generated by the spring deformation. Figure 3is a schematic diagram of the force situation of the underwater chain robot. n is the number of underwater chain robot units. L a Indicates the length of each unit. i , l i and r i They represent the total driving force of unit i, the driving force of the left propeller, and the driving force of the right propeller, respectively. When flying straight, l i =r i andu i = l i +r i .w i Represents the fluid resistance of unit i, which varies from unit to unit. i(i+1) represents the force exerted by unit i on unit (i+1), f (i+1)i Represents the force exerted by unit (i+1) on unit i.

[0048] According to Newton's second law, the dynamic equations of each unit are as follows:

[0049]

[0050] Among them, m i 、v i and represent the mass, velocity and acceleration of the unit respectively. If the spring constant between units is h i(i+1) , then the above formula can be changed to:

[0051]

[0052] Among them, x i(i+1) and L b are the current spacing of each unit and the ideal spacing of each unit respectively. (x (i-1)i -L b )≤0 and (x (i-1)i -L b )>0 corresponds to the compression and tension states of the spring respectively.

[0053] The power distribution of the underwater chain robot for steady straight navigation can be achieved based on the power optimization distribution of the genetic algorithm and the power distribution constraint with the penalty function, which can cope with random disturbances and arbitrary initial states.

[0054] The principle of power optimization allocation based on genetic algorithm is explained below.

[0055] Genetic algorithms screen individuals through selection, crossover and mutation in order to retain individuals with high fitness values ​​and form new groups. The new group inherits the information of the previous generation, but is better than the previous generation. In this way, the fitness of individuals in the group continues to improve until certain conditions are met. The fitness function is the basis for the selection of genetic algorithms, and its expression directly affects the convergence speed and the quality of the solution. Since quadratic forms have good mathematical properties, simple solution methods and wide practical applications, quadratic forms are used here to establish fitness functions. In addition, the underwater chain robot sails at a uniform speed at the target speed and the ideal unit spacing is an important guarantee for smooth and direct navigation. When establishing the fitness function, the specific terms should be associated with the overall speed difference and the unit spacing difference. Therefore, the fitness function is defined as follows:

[0056]

[0057] in, and L b They are the unit spacing after one cycle of power distribution and the ideal unit spacing, and v o are the overall speed and target overall speed after one cycle of power distribution execution, α is the weighting factor of spacing deviation, and β is the weighting factor of overall speed deviation. The optimization goal is to minimize the adaptation function J a The value of .

[0058] When using genetic algorithms, the selection, crossover and mutation process of the power group is as follows:

[0059] (1) Selection: Using the random traversal sampling selection operator, the power allocation individuals with higher fitness are selected from the randomly generated power population with a higher probability to form a new power population;

[0060] (2) Crossover: Using the crossover operator of single-point crossover, a crossover point is randomly selected from the binary-coded power group obtained by the selection operation, and the binary codes are exchanged to obtain a new generation of power groups;

[0061] (3) Mutation: A bit-flipping mutation operator is used to randomly select mutation bits from the binary-coded dynamic population obtained by the crossover operation, and then a binary bit inversion operation is performed to obtain a new generation of population.

[0062] If there are unit driving force constraints in the underwater chain robot system, a penalty function is established and added to the objective function of the genetic algorithm power optimization to achieve power optimization allocation considering the constraints. The principle is explained below.

[0063] If the saturation constraint force of each unit is T max , then the driving force u allocated to each unit i Should meet: i ≤Tmax Since genetic algorithms are difficult to handle constraints in optimization problems, in order to convert constraints into unconstrained expressions, it is necessary to establish an i >T max A function that penalizes the power allocation result (i.e., penalty function) is used and added to the fitness function of the genetic algorithm.

[0064] The penalty function also adopts a quadratic form. Considering the minimization optimization mode of the fitness function, the specific terms can be expressed as:

[0065]

[0066] γ is the penalty factor. For the above penalty function: when u i The value is not greater than T max When u i Value greater than T max When the sub-item value is u i -T max , resulting in penalty, the fitness function takes a larger value.

[0067] Fitness function J with penalty function added a 'The specific form is as follows:

[0068]

[0069] In order to verify the effectiveness of the proposed power distribution method, a simulation study was carried out in the MATLAB environment using a 5-unit underwater chain robot as an example. The values ​​of the parameters in the simulation are as follows:

[0070] (1) Initial state: driving force of each unit u1=u2=u3=u4=u5=2N; initial velocity of each unit v1=v2=v3=v4=v5=1knot; overall velocity v p =1knot; unit spacing

[0071] x 12 =x 23 =x 34 =x 45 =0.13m.

[0072] (2) Target state: the distance between each unit x 12 =x 23 =x 34 =x 45 =L b =0.12m; overall speed v o =1.5knot.

[0073] (3) Platform parameters: number of units n = 5; mass of each unit m1 = m2 = m3 = m4 = m5 = 5 kg; unit length L a =0.9m; unit resistance w1=2.3N, w2=2.5N, w3=2.5N, w4=2.4N, w5=2.6N; spring coefficient is h 12 =h 23 =h 34 =h 45 =25; each unit drives saturation constraint force T max =3.5N.

[0074] (4) Method parameters: Sample size N = 500; Number of selections with replacement N' = 15; Crossover probability p i =0.5; mutation probability p v =0.02; sampling period T s =0.75s; spacing deviation weighting factor α=130; overall speed deviation weighting factor β=110; penalty factor γ=10000.

[0075] (5) Interference form: When considering the ocean current interference, a sinusoidal power output interference is applied to unit 5. The interference force f c The specific form of the change over time t is as follows:

[0076]

[0077] Figure 4 The figure shows the power distribution change, unit spacing change and overall speed change results of the 5-unit underwater chain robot tending to steady straight navigation under the action of the genetic algorithm. It can be seen from the figure that under the action of the genetic algorithm, the driving force of each unit tends to be stable, the unit spacing tends to the ideal 0.12m from 0.13m, and the overall speed tends to the target 1.5knot from 1knot.

[0078] Figure 5 The power distribution change and unit spacing change results of the 5-unit underwater chain robot tending to steady straight navigation under the disturbance of ocean current. It can be seen from the figure that unit 5 is disturbed by the sinusoidal disturbance of ocean current, and its driving force is constantly changing with time t to offset the influence of the disturbance. Under the power optimization distribution based on genetic algorithm, the spacing between each unit is from 0.13m to the ideal 0.12m, which proves that the proposed method can effectively deal with uncertain ocean current disturbance.

[0079] Figure 6The figure shows the change in power distribution and unit spacing of the 5-unit underwater chain robot towards steady straight navigation when the penalty function is added. It can be seen from the figure that except for the sinusoidal fluctuation of unit 5, the driving forces of the other units tend to be stable, and the driving forces do not exceed the 3.5N constraint. In addition, due to the sinusoidal wave interference of unit 5 and the added penalty function, the process of the unit spacing from 0.13m to the ideal 0.12m oscillates significantly. As time goes by, the unit spacing will tend to the ideal value.

[0080] The above simulation results verify the feasibility and effectiveness of the proposed method, which can provide theoretical support for the optimal power distribution of underwater chain robots.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, any modifications, improvements and equivalent substitutions made without departing from the principles of the present invention are all included in the protection scope of the present invention.

Claims

1. A power distribution method for the steady and straight navigation of an underwater chain robot. It is characterized in that The steps are: Step 1: Given the driving distribution characteristics, initial state and target state of the underwater chain robot, the driving force size to be adopted by each unit in the next execution cycle is optimized based on the genetic algorithm; Step 2: Establish a penalty function and add it to the objective function of the genetic algorithm power optimization to determine the actual driving force used by each unit in the next execution cycle.

2. A power distribution method for a stable and straight underwater chain robot according to claim 1, Features: In the step 1, the specific method is: The dynamic model of the underwater chain robot is established. Given the initial state and the target state, the power distribution result that minimizes the fitness function is selected based on the selection, crossover and mutation operations of the genetic algorithm. The fitness function is: in, and L b They are the unit spacing after one cycle of power distribution and the ideal unit spacing, and v o They are the overall speed and target overall speed after one cycle of power distribution execution, α is the weighting factor of spacing deviation, β is the weighting factor of overall speed deviation, and the optimization goal is to minimize the adaptation function J a The value of .

3. A power distribution method for a stable and straight underwater chain robot according to claim 2, Features: The process of establishing the underwater chain robot dynamics model is as follows: Under ideal conditions, the forces acting on each unit include driving force, resistance, and reaction force of adjacent units. The driving force is generated by the propellers on both sides of the AUV unit, the resistance is generated by the fluid resistance, and the reaction force is generated by the spring deformation. Define n as the number of underwater chain robot units, L a is the length of each unit, u i , l i and r i They are the total driving force of unit i, the driving force of the left propeller and the driving force of the right propeller, respectively. When flying straight, l i =r i andu i = l i +r i , w i is the fluid resistance of unit i, f i(i+1) is the force exerted by unit i on unit (i+1), f (i+1)i is the force exerted by unit (i+1) on unit i. According to Newton's second law, the dynamic equations of each unit are as follows: Among them, m i 、v i and Represent the mass, velocity and acceleration of the unit respectively. If the spring coefficient between units is h i(i+1) , then the above formula can be transformed into: Among them, x i(i+1) and L b are the current spacing of each unit and the ideal spacing of each unit, (x (i-1)i -L b )≤0 and (x (i-1)i -L b )>0 corresponds to the compression and extension states of the spring respectively. The power distribution of the underwater chain robot for smooth straight navigation is realized based on the power optimization distribution of the genetic algorithm and the power distribution constraint with the penalty function, which can cope with random disturbances and arbitrary initial states.

4. A power distribution method for a stable and straight underwater chain robot according to claim 3, Features: The ideal conditions specifically include: The shape, center of gravity, center of buoyancy, and driving position of each unit in the underwater chain robot are the same, and the gravity is equal to the buoyancy in the initial state; the underwater chain robot maintains horizontal and straight navigation during movement; the gravity and buoyancy of the connecting mechanism between units are negligible; the driving force effects on both sides of each unit are synchronous and equal in size; the spacing between the units of the underwater chain robot is equal when the springs between the units are not deformed; the driving force direction of each unit is forward, and the performance is the same.

5. A power distribution method for a stable and straight navigation of an underwater chain robot according to claim 2, Features: The selection, crossover and mutation operations based on the genetic algorithm are specifically as follows: (1) Selection: Using the random traversal sampling selection operator, the power allocation individuals with higher fitness are selected from the randomly generated power population with a higher probability to form a new power population; (2) Crossover: Using the crossover operator of single-point crossover, a crossover point is randomly selected from the binary-coded power group obtained by the selection operation, and the binary codes are exchanged to obtain a new generation of power groups; (3) Mutation: A bit-flipping mutation operator is used to randomly select mutation bits from the binary-coded dynamic population obtained by the crossover operation, and then perform a binary bit inversion operation to obtain a new generation of population.

6. A power distribution method for a stable and straight underwater chain robot according to claim 1, Features: The specific method of step 2 is: If the saturation constraint force of each unit is T max , then the driving force u allocated to each unit i Should meet: i ≤T max Since genetic algorithms are difficult to handle constraints in optimization problems, in order to convert constraints into unconstrained expressions, a method that can i >T max A function that penalizes the power allocation result, namely the penalty function, and adds it to the fitness function of the genetic algorithm; The penalty function also adopts a quadratic form. Considering the minimization optimization mode of the fitness function, the specific terms are expressed as Among them: J b is the penalty function, γ is the penalty factor, for the above penalty function: when u i The value is not greater than T max When u i Value greater than T max When the sub-item value is u i -T max , resulting in penalties, the fitness function takes a larger value, Fitness function J with penalty function added a 'The specific form is as follows:

Images