An Efficient Thrust Control Method Based on Optimized Energy Consumption for an Underwater Robot

Abstract

The present invention relates to the field of motion control of underwater robots, in particular to the technology of autonomous underwater robot motion control, realizing low-energy consumption thrust distribution control during the detection operation of autonomous underwater robots. Specifically, it is an efficient thrust control method for underwater robots based on optimized energy consumption, including the following steps: 1) Based on the thrust generated by the underwater robot, construct an objective function for the relationship between the thrusts at adjacent moments and the energy consumption; 2) Optimize and solve the objective function; 3) Loop steps 1) and 2), and use the thrust corresponding to the solution result of the objective function as the thrust corresponding to the optimal energy consumption. Based on this thrust, real-time allocation is performed for each thruster of the underwater robot. The present invention takes into account the total thrust constraint of the thrusters, optimizes the influence of the thrust difference between the thrusters at adjacent moments on the system energy consumption, and considers the influence of the actual operation ability of the thrusters on the thrust distribution algorithm.

Description

Technical Field

[0001] The present invention relates to the field of motion control of underwater robots, and more particularly to the technology of autonomous underwater robot motion control, realizing low-energy consumption thrust distribution control during the detection operation of autonomous underwater robots. Specifically, it is an efficient thrust control method for underwater robots based on optimized energy consumption. Background Art

[0002] With the development of metal, material, and computer technology levels, underwater robots are widely used in deep-sea scientific research, marine search and rescue, marine oil and gas development, lake and sea archaeology and other fields. Considering the operation efficiency and energy issues of underwater robots, integrating energy consumption optimization technology into the thrust control distribution of underwater robots will improve the operation efficiency under limited energy supply.

[0003] Traditional underwater robot thrust distribution methods rarely consider the problem of energy consumption optimization, have a large computational complexity, cannot be executed online, it is difficult to consider the impact of thrust distribution at adjacent moments on energy consumption, and it is impossible to achieve energy consumption optimization in a continuous time period in real-time online. Therefore, researching an efficient, online, and linear thrust control distribution method based on optimized energy consumption is a key technological breakthrough in the current thrust control distribution of underwater robots. Summary of the Invention

[0004] Based on the above technical background, in view of the deficiencies of the traditional underwater robot thrust distribution method based on energy consumption in the horizontal direction, the present invention proposes an efficient thrust control method for underwater robots based on optimized energy consumption.

[0005] The technical solution adopted by the present invention to achieve the above object is as follows:

[0006] An efficient thrust control method for underwater robots based on optimized energy consumption, comprising the following steps:

[0007] 1) Based on the thrust generated by the underwater robot, construct an objective function for the relationship between each thrust and energy consumption at adjacent moments;

[0008] 2) Optimize and solve the objective function;

[0009] 3) Loop steps 1) and 2), and take the thrust corresponding to the solution result of the objective function as the thrust corresponding to the optimal energy consumption. Based on this thrust, perform real-time distribution for each thruster of the underwater robot.

[0010] The objective function of step 1) is:

[0011]

[0012]

[0013]

[0014]

[0015] Among them, P is the objective function, k is the index number of time, μ is the influence factor of the thrust at adjacent moments, λ1 is the weight factor of the thrust error at a certain moment, λ2 is the weight factor of the thrust moment error at a certain moment, m and q are parameters used to adjust the approximation degree, and T x is the target thrust in the forward direction of the underwater robot at the k-th moment, and T MAX is the rated maximum thrust of the motor, and T i (k) is the thrust generated by a certain horizontal thruster at the k-th moment, and T i (k - 1) is the thrust generated by a certain horizontal thruster at the previous moment of k, and M z is the target turning moment of the horizontal thruster of the underwater robot at the k-th moment, and T x ’ is the difference between the target thrust and the actual thrust, and M z ’ is the difference between the target moment and the actual moment, θ is the angle between the thrusts T1, T2 and the vertical auxiliary line, is the angle between the thrusts T3, T4 and the vertical auxiliary line, l1 is the lever arm of the thrusts T1, T2, and l2 is the lever arm of the thrusts T3, T4.

[0016] The said step 2) includes the following steps:

[0017] 2.1) Select a sample point g;

[0018] 2.2) Calculate the sample point g;

[0019] 2.3) Determine the sample point g;

[0020] 2.4) Loop steps 2.2) and 2.3) until the sample point g is exhausted;

[0021] 2.5) Loop steps 2.2) to 2.4), and calculate the minimum value of the objective function P of all sample point sets G.

[0022] The said step 2.1) is specifically:

[0023] Assume that the domain of the objective function P is D, G0(T1(k), T2(k), T3(k), T4(k)) is an interior point of D and there is a certain neighborhood of G0 Select several sample points g according to the normal distribution or the uniform distribution to form a sample point set G, and select any one of the sample points g.

[0024] The said step 2.2) includes the following steps:

[0025] 2.2.1) Take the first derivative of the thrusts in all directions \(T1(k)\), \(T2(k)\), \(T3(k)\), \(T4(k)\) within the sample point \(g\) respectively and set it to zero;

[0026] 2.2.2) Conduct a minimum approximation deduction for the thrusts in all directions \(T1(k)\), \(T2(k)\), \(T3(k)\), \(T4(k)\);

[0027] 2.2.3) Determine whether there is an extreme value of the objective function according to the deduction result. If there is, end the deduction. Otherwise, continue the deduction according to 2.2.2) until the sample point \(g\) ends.

[0028] The specific content of step 2.2.2) is as follows:

[0029]

[0030] Using the thrusts \(T1(n)\), \(T2(n)\), \(T3(n)\), \(T4(n)\) as inputs, calculate \(T1(n + 1)\) to \(T4(n + 1)\):

[0031]

[0032] Let Obtain the deduction result

[0033]

[0034] The specific content of step 2.2.3) is as follows:

[0035]

[0036] where \(fabs()\) is to take the absolute value and \(\epsilon\) is the threshold.

[0037] The specific content of step 2.3) is as follows:

[0038] If the objective function \(P\) k＝n+2 and \(P\) k＝n are both greater than \(P\) k＝n+1 , then there is a minimum value within the sample point \(g\), and the minimum value is \(P\) k＝n+1 ;

[0039] If the objective function \(P\) k＝n+2 and \(P\) k＝n are both less than \(P\) k＝n+1 , then there is a maximum value within the sample point \(g\), and the maximum value is \(P\) k＝n+1 , and this value is discarded.

[0040] where \(P\) k＝n is the value of the objective function \(P\) at time \(n\), \(P\) k＝n+1 is the value of the objective function \(P\) at time \(n + 1\), and \(P\) k＝n+2 is the value of the objective function \(P\) at time \(n + 2\).

[0041] The present invention has the following beneficial effects and advantages:

[0042] 1. Traditional thrust allocation only considers the energy consumption optimization at the current moment, ignoring the impact of thrust allocation at adjacent moments on energy consumption, and cannot calculate the optimal energy consumption within a period of time. The present invention establishes an algorithm for solving the optimal energy consumption in the thrust neighborhood at a sufficient number of relative moments, comprehensively considers the impact of thrust allocation at adjacent moments on energy consumption, and proposes a linear approximation optimization method.

[0043] 2. Compared with the disadvantages of complex calculation and inability to perform online calculation of traditional thrust allocation methods, the present invention has the ability of online calculation. It gradually optimizes through approximate deduction, avoiding the problem of complex matrix inversion operations, and reducing the complexity of calculation.

[0044] 3. The present invention considers the total thrust constraint of the thrusters, optimizes the impact of the thrust difference between thrusters at adjacent moments on system energy consumption, and considers the impact of the actual operation ability of the thrusters on the thrust allocation algorithm. Description of the Drawings

[0045] Figure 1 is a schematic diagram of the thruster distribution on the water surface of the underwater robot;

[0046] Figure 2 is a schematic diagram of solving the minimum value of the objective function within the sampling region. Detailed Embodiment

[0047] The following further describes the present invention in detail with reference to the drawings and embodiments.

[0048] The present invention aims to establish a dependency relationship describing the relationship between the thruster thrust and energy consumption at any adjacent moment, and then linearly approximate and deduce the dependency relationship to find the minimum value, so as to obtain the optimal thrust allocated at this stage. The technical solution adopted by the present invention to achieve the above purpose is: an efficient thrust control method for an underwater robot based on optimized energy consumption. It includes the following steps:

[0049] 1. Construct an objective function for the relationship between each thrust and energy consumption at adjacent moments;

[0050] 2. Design an online optimization solution process for the objective function;

[0051] (1) Select a sample point g;

[0052] (2) Calculate the sample point g;

[0053] 1) Respectively take the first derivative of the thrusts T1(k), T2(k), T3(k), T4(k) in each direction within the sample point g and set it to zero;

[0054] 2) Perform minimum approximation deduction on the thrusts in all directions T1(k), T2(k), T3(k), and T4(k);

[0055] 3) Determine that the objective function has an extreme value;

[0056] (3) Calculate the determination condition of the sample point g;

[0057] (4) Repeat steps (2) and (3) until the sample point g ends;

[0058] (5) Loop and call steps (2) to (4) to calculate the minimum value of the objective function P of all sample point sets G;

[0059] 3. Solve for the thrust corresponding to the optimal energy consumption.

[0060] 4. Loop and call steps 1 to 3 to calculate the optimal energy consumption of the underwater robot during movement online, allocate the thrusts of each thruster in real time, and achieve efficient thrust control allocation.

[0061] As Figure 1 shown, T1, T2, T3, and T4 are the thrusts generated by the thrusters at the front left, front right, rear left, and rear right positions of the underwater robot respectively. O is the hydrodynamic center of the underwater robot. Horizontal and vertical auxiliary lines are drawn respectively with the hydrodynamic center O as the reference. The angles between the thrusts T1, T2 and the vertical auxiliary line are θ, and the angles between the thrusts T3, T4 and the vertical auxiliary line are The moment arms of the thrusts T1, T2 are l1, and the moment arms of the thrusts T3, T4 are l2. l1 and l2 may not be equal.

[0062] As Figure 2 shown, the specific process of the present invention is as follows:

[0063] 1. Construct an objective function for the relationship between the thrusts and energy consumption at adjacent moments

[0064]

[0065]

[0066]

[0067]

[0068] where P is the objective function, k is the index number of time, μ is the influence factor of the thrust at adjacent moments, λ1 is the weight factor of the thrust error at a certain moment, λ2 is the weight factor of the thrust moment error at a certain moment. By adjusting λ1 and λ2, the thrust and moment at the current moment are affected. μ, λ1, and λ2 are engineering experience coefficients and are all constants is a parameter used to adjust the approximation degree. The larger q is, the better the approximation effect.

[0069] T x is the target thrust in the forward direction of the underwater robot at time k, T MAX is the rated maximum thrust of the motor, which is a calibrated known quantity, T i (k) is the thrust generated by a certain horizontal thruster at time k, T i (k - 1) is the thrust generated by a certain horizontal thruster at the previous moment of k, which is a known quantity obtained. M z is the target steering torque of the horizontal thrusters of the underwater robot at time k. T x ’ is the difference between the target thrust and the actual thrust, M z ’ realizes the difference between the target torque and the actual torque. T x ’、M z ’ are intermediate variables for solving the objective function.

[0070] 2. Design the online optimization solution process of the objective function

[0071] 2.1 Select the sample point g

[0072] Assume that the domain of the objective function P is D, and G0(T1(k), T2(k), T3(k), T4(k)) is an interior point of D and there is a certain neighborhood of G0 Select a set G of several sample points according to the normal distribution or the uniform distribution, and select any one of the sample points g as an example to describe the algorithm.

[0073] 2.2 Calculate the sample point g

[0074] 2.2.1 Respectively, take the first derivative of each directional thrust T1(k), T2(k), T3(k), T4(k) in the sample point g and set it to zero.

[0075] Assume

[0076]

[0077] Respectively, take the first derivative of and set it to zero.

[0078] That is

[0079] 2.2.2 Conduct a minimum value approximation deduction for each directional thrust T1(k), T2(k), T3(k), T4(k).

[0080] When k = 0, the initial values T1(0), T2(0), T3(0), T4(0)

[0081]

[0082]

[0083] Deducing with T1(0), T2(0), T3(0), and T4(0) as inputs gives:

[0084]

[0085] Similarly, the following can be deduced:

[0086]

[0087] Based on the previous step, with T1(n), T2(n), T3(n), and T4(n) as inputs, the steps to calculate T1(n + 1) to T4(n + 1) are as follows:

[0088]

[0089]

[0090]

[0091] 2.2.3 Determine that the objective function has an extreme value

[0092] Until ε is a very small value, then it is considered that the objective function P has an extreme value, and the deduction of the sample point g ends.

[0093] 2.3 Calculate the determination conditions for the sample point g

[0094] If the objective function P k＝n+2 and P k＝n are both greater than P k＝n+1 , then it indicates that there is a minimum value within the sample point g, that is, P k＝n+1 ;

[0095] If the objective function P k＝n+2 and P k＝n are both less than P k＝n+1 , then it indicates that there is a maximum value within the sample point g, that is, P k＝n+1 . Abandon this value and continue to the next step. P k＝n+2 is the value of the objective function P at time n + 2, and P k＝n and P k＝n+1 are the same.

[0096] 2.4 Repeat steps 2.2 and 2.3 until the sample point g ends.

[0097] Continue to deduce the remaining domain of the sample point g according to the methods of 2.2 and 2.3 until the sample point g ends, and determine whether there are other minimum values.

[0098] 2.5 Loop and call steps 2.2 to 2.4 to calculate the minimum value of the objective function P for all sample point sets G.

[0099] There are N mutually distinct sample points g (N is a certain number) in the set G of several sample points of the objective function P within the domain. The steps to calculate the minimum value of the objective function P for all sample point sets G are as follows:

[0100] Obtain the minimum value of the objective function P in a sample point g according to steps 2.2 to 2.4; loop and call steps 2.2 to 2.4 to calculate all the minimum values within N sample points; compare all the minimum values and select the minimum value, which is the minimum value of the objective function P for all sample point sets G. The specific algorithm is as follows:

[0101] While(1)

[0102] if (there are N mutually distinct sample points g in the set G of several sample points of the objective function P within the domain)

[0103] for i = 1:N

[0104] Obtain the minimum value of the objective function P in a sample point g according to steps 2.2 to 2.4;

[0105] Loop and call steps 2.2 to 2.4 to calculate all the minimum values within N sample points;

[0106] Compare all the minimum values and select the minimum value.

[0107] End

[0108] 3. Solve for the thrust corresponding to the optimal energy consumption

[0109] As can be seen from the above, when the objective function P obtains the minimum value for all sample point sets G, the current T1(k), T2(k), T3(k), and T4(k) are the minimum thrust values for each thruster to reach the optimal energy consumption.

[0110] By looping and executing steps 1 to 3, the optimal energy consumption of the underwater robot during online calculation of motion can be obtained, the thrust of each thruster can be allocated in real time, and efficient thrust control allocation can be achieved.

Claims

1. An efficient thrust control method for an underwater robot based on optimized energy consumption, characterized in that, It includes the following steps: 1) Based on the thrust generated by the underwater robot, construct an objective function for the relationship between each thrust and energy consumption at adjacent moments; 2) Optimize and solve the objective function; 3) Loop steps 1) and 2), and use the thrust corresponding to the solution result of the objective function as the thrust corresponding to the optimal energy consumption. Based on this thrust, perform real-time allocation for each actuator of the underwater robot; The said step 2) includes the following steps: 2.1) Select a sample point g; 2.2) Calculate the sample point g; 2.3) Judge the sample point g; 2.4) Loop steps 2.2) and 2.3) until the sample point g ends; 2.5) Loop steps 2.2) to 2.4), and calculate the minimum value of the objective function P of all sample point sets G; The said step 2.1) is specifically: Suppose the domain of the objective function P is D, G0(T1(k), T2(k), T3(k), T4(k)) is an interior point of D and there exists a neighborhood of G0 Select a number of sample points g according to the normal distribution or the uniform distribution to form a sample point set G, and select any one of the sample points g; The said step 2.2) includes the following steps: 2.2.1) Respectively take the first derivative of each of the thrusts T1(k), T2(k), T3(k), T4(k) in the sample point g and set it to zero; 2.2.2) Perform a minimum value approximation deduction on the thrusts T1(k), T2(k), T3(k), T4(k); 2.2.3) Determine whether there is an extreme value in the objective function according to the deduction result. If so, end the deduction. Otherwise, continue the deduction according to 2.2.2) until the sample point g ends.

2. The efficient thrust control method for an underwater robot based on optimized energy consumption according to claim 1, characterized in that, The objective function of the said step 1) is: Among them, P is the objective function, k is the index number of time, μ is the influence factor of the thrust at adjacent moments, λ1 is the weight factor of the thrust error at a certain moment, λ2 is the weight factor of the thrust moment error at a certain moment, m and q are parameters for adjusting the approximation degree, T x is the target thrust in the forward direction of the underwater robot at time k, T MAX is the rated maximum thrust of the motor, T i (k) is the thrust generated by a certain horizontal thruster at time k, T i (k - 1) is the thrust generated by a certain horizontal thruster at the previous moment of k, M z is the target steering moment of the horizontal thruster of the underwater robot at time k, T x ' is the difference between the target thrust and the actual thrust, M z ' realizes the difference between the target moment and the actual moment, θ is the angle between the thrusts T1, T2 and the vertical auxiliary line, is the angle between the thrusts T3, T4 and the vertical auxiliary line, l1 is the lever arm of the thrusts T1, T2, and l2 is the lever arm of the thrusts T3, T4.

3. The efficient thrust control method for an underwater robot based on optimized energy consumption according to claim 1, characterized in that, The said step 2.2.2) is specifically: Taking the thrusts T1(n), T2(n), T3(n), T4(n) as inputs, calculate T1(n + 1) to T4(n + 1): Let obtain the deduction result 4. The efficient thrust control method for an underwater robot based on optimized energy consumption according to claim 3, characterized in that, The said step 2.2.3) is specifically: and and and Where fabs() is to take the absolute value and ε is a threshold value.

5. The efficient thrust control method for an underwater robot based on optimized energy consumption according to claim 1, characterized in that, The said step 2.3) is specifically: If the objective function P k＝n+2 and P k＝n are both greater than P k＝n+1 , then there is a minimum value within the sample point g, and the minimum value is P k＝n+1 ; If the objective function P k＝n+2 and P k＝n are both less than P k＝n+1 , then there is a maximum value within the sample point g, and the maximum value is P k＝n+1 , and this value is discarded. Among them, P k＝n is the value of the objective function P at the n-th moment, P k＝n+1 is the value of the objective function P at the (n + 1)-th moment, P k＝n+2 is the value of the objective function P at the (n + 2)-th moment.

Images