A Fault Tolerant Control Method for Underwater Robot Propulsion System Failures

Abstract

The present invention discloses a fault-tolerant control method for an underwater robot propulsion system, which includes the steps of: (1) establishing a mathematical model of a six-degree-of-freedom underwater robot with four horizontal thrusters and two vertical thrusters; (2) obtaining the fault estimation of each thruster in real time through a sliding-mode observer, introducing a fault weight matrix, and establishing a fault prediction model based on the original thrust allocation matrix and the fault weight matrix; (3) using model predictive control to solve the optimization problem of thrust allocation. Taking the thrust vector of the thruster as the state quantity and the thrust change rate of the thruster as the control quantity, based on the objective function and constraint conditions, rolling optimization is carried out to solve the thrust of each thruster; (4) the calculation of the objective function in model predictive control is optimized using an improved multi-universe algorithm. The fault-tolerant control method of the present invention can solve the problem of over-saturation of faulty thrusters caused by traditional propulsion system fault-tolerant control, avoid the deepening of the fault degree, and improve the system stability.

Description

Technical Field

[0001] The present invention relates to a fault-tolerant control method for an underwater robot, and in particular to a fault-tolerant control method for a propulsion system failure of an underwater robot, belonging to the field of fault-tolerant control. Background Art

[0002] The complex marine environment makes the operation of underwater robots difficult and time-consuming underwater, and the unknown underwater environment may also cause failures of underwater robots. Failures may have varying degrees of impact on the operation performance of underwater robots. If an underwater robot "knows nothing" about the failure situation, or knows the failure situation but has no corresponding rescue measures, then the task may not be completed due to the failure, or the underwater robot may be lost, and even the safety of personnel may be endangered.

[0003] With the in-depth research on the multi-motor cooperative propulsion system of underwater robots, when the motor load and speed change, the coordination performance of multiple motors is guaranteed. However, when one motor fails, the original thrust cannot be achieved, and in severe cases, the multi-motor coordination characteristics will be damaged. Therefore, fault-tolerant control is required for the multi-motor cooperative propulsion system of underwater robots.

[0004] At present, the fault-tolerant control of underwater robot thrusters mostly adopts the method of directly compensating the thruster input to achieve the ideal thrust, without considering the importance of protecting the faulty thruster during the task execution of the underwater robot. Directly compensating the input of the faulty thruster will lead to a deeper degree of failure. If the faulty thruster changes from partial failure to complete failure during the task execution of the underwater robot, it will greatly increase the difficulty of the underwater robot's return journey. It is very necessary to use fault-tolerant control to protect the faulty thruster. Summary of the Invention

[0005] The purpose of the present invention is to provide a fault-tolerant control method for a propulsion system failure of an underwater robot. First, a mathematical model is established for the remotely operated underwater robot applied in the present invention. Aiming at the problems of traditional thrust allocation single-step optimization and protection of faulty thrusters, the present invention proposes to use the model predictive control method to solve the thrust allocation. When establishing the objective function of the prediction model, factors such as the thrust saturation characteristic, energy consumption, and thruster failure of the thruster are considered. At the same time, an improved multi-verse optimization algorithm (MVO) is used in the calculation of the objective function to improve the optimization ability and convergence, and the optimal thrust allocation can be better obtained.

[0006] The purpose of the present invention is realized through the following technical solutions:

[0007] To achieve the above object, the present invention provides a fault-tolerant control method for an underwater robot propulsion system. The underwater robot includes six thrusters, where thrusters No. 1, No. 2, No. 3, and No. 4 are horizontally distributed vector thrusters, and thrusters No. 5 and No. 6 are vertical thrusters. The method includes the following steps:

[0008] (1) Establish a six-degree-of-freedom mathematical model of the underwater robot with four horizontal thrusters and two vertical thrusters;

[0009] (2) Obtain the fault estimates γ1, γ2, γ3, γ4, γ5, γ6 of each thruster in real time through a sliding mode observer, where γ1, γ2, γ3, γ4 are the fault coefficients of the four horizontal thrusters, and γ5, γ6 are the fault coefficients of the two vertical thrusters. Introduce a fault weight matrix W. When there is no fault, establish a thrust prediction model of the underwater robot based on the thrust distribution vector matrix B(α). When a fault occurs, establish a fault prediction model based on the original thrust distribution matrix and the fault weight matrix W;

[0010] (3) Use model predictive control to solve the open-loop optimization problem of thrust distribution to obtain the optimal thrust solution. The optimization objectives considered for thrust redistribution include: minimum power consumption, minimum distribution error, and avoidance of singular terms in the thruster configuration matrix. The constraint conditions include equality constraints and thrust change constraints under fault conditions. Taking the thrust vector of the thrusters as the state quantity and the thrust change rate of the thrusters as the control quantity, perform rolling optimization based on the objective function and constraint conditions of the underwater robot thrust prediction model, and solve to obtain the thrust values of each thruster and distribute them to the corresponding thrusters;

[0011] (4) Use the multi-universe optimization algorithm to optimize the calculation of the objective function in model predictive control. Aiming at the problems of low search efficiency of the traditional multi-universe optimization algorithm and the difficulty in balancing the WEP (wormhole existence probability) and TDR (travel distance value) during the iteration process, improve the traditional MVO algorithm by using logarithmically increasing WEP and non-linearly converging TDR.

[0012] The object of the present invention can also be further achieved by the following technical measures:

[0013] Further, step (1) specifically includes:

[0014] Step (1.1): Construct the representation of the combined thrust and combined thrust moment of the four horizontal thrusters of the underwater robot in the body-fixed coordinate system:

[0015]

[0016] Among them, T1, T2, T3, and T4 are the thrusts of the underwater robot body received by thrusters No. 1, No. 2, No. 3, and No. 4, respectively, X h 、Y h 、Zh are the resultant thrusts of the underwater robot body by the horizontal thrusters along the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively, where x h , y h , z h are the position vectors of the horizontal thrusters about the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, and K h , M h , N h are the resultant thrust moments of the four horizontal thrusters on the underwater robot body about the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively, and α h is the angle between the thruster and the x-axis of the ship-fixed coordinate system;

[0017] Step (1.2): Construct the resultant thrust representation of the two vertical thrusters of the underwater robot in the ship-fixed coordinate system:

[0018]

[0019] where T5 and T6 are the thrusts of the 5th and 6th thrusters on the underwater robot body, respectively, and X v , Y v , Z v are the resultant thrusts of the vertical thrusters on the underwater robot body along the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively, and x v , y v are the position vectors of the vertical thrusters about the x-axis and y-axis of the ship-fixed coordinate system, and K v , M v , N v are the resultant thrust moments of the two vertical thrusters on the underwater robot body about the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively;

[0020] Step (1.3): From the above two equations, the resultant thrust of the six thrusters installed on the underwater robot on the underwater robot body to produce six-degree-of-freedom motion:

[0021]

[0022] where X T , Y T , Z T are the longitudinal thrust, lateral thrust, and vertical thrust produced by the six thrusters on the underwater robot body, respectively, and K T , M T , N T are the rolling thrust moment, pitching thrust moment, and yaw thrust moment produced by the six thrusters on the body, respectively;

[0023] The control vector of the six degrees of freedom acting on the underwater robot body is expressed by the following formula:

[0024] τ = B(α)T

[0025] where τ = [X T Y T Z T K T M T N T T is the six - degree - of - freedom thrust vector acting on the underwater robot body, T = [T1 T2 T3 T4 T5 T6] T is the thrust vector output by the thruster, and B(α) is the thruster vector arrangement matrix.

[0026] Furthermore, step (2) specifically includes:

[0027] Step (2.1): The mathematical model of the failure of the i - th thruster is expressed as:

[0028]

[0029] where γ i ∈(-1,0] is the failure factor of the i - th thruster, u i is the control voltage of the i - th thruster, is the control voltage of the i - th thruster after failure correction, γ i = 0 indicates that the i - th thruster is working normally without failure; when - 1 < γ i < 0, it indicates that the i - th thruster is partially failed but still working;

[0030] The state equation of the underwater robot with thruster failure is expressed as:

[0031] x p (k + 1)=A p x p (k)+B p u(k)+E(k)γ(k)+d(k)

[0032] where, is the system state variable, is the system input is the corresponding system matrix, d(k)=△A p x p (k)+△B p u p (k)+ω(k) represents the sum of system parameter uncertainties and external disturbances;

[0033] Define the new state variable z(k)=[x p (k) T γ(k) T T , and the observation model can be obtained:​​

[0034]

[0035] where and I m represents the identity matrix of order m, and I n represents the identity matrix of order n, 0 m×n represents the zero matrix of m×n, 0 m represents the zero matrix of m×m, 0 qm represents the zero matrix of q×m. If observable, the system state and the fault failure factor γ i can be estimated by the observer;

[0036] Step (2.2): Introduce the fault weight matrix:

[0037]

[0038] The modified thrust allocation matrix becomes τ = B(α)WT.

[0039] Furthermore, step (3) specifically includes:

[0040] Step (3.1): Objective function

[0041] Minimum power consumption:

[0042]

[0043] where is the power coefficient of the i-th thruster;

[0044] Minimum allocation error:

[0045] J s = s T Qs

[0046] where s = τ d - B(β)T is the allocation error;

[0047] Avoid singular terms of the thruster configuration matrix:

[0048]

[0049] where ρ and ε are adjustment parameters, aiming to make B(β)B(β) T ≠ 0, that is, B(β) has full row rank;

[0050] Step (3.2): Constraint conditions

[0051] Equality constraint:

[0052] B(β)T + s = τ d

[0053] Thrust change constraint:

[0054]

[0055] Wherein, T i,max is the maximum thrust that the i-th thruster can generate, T i,min is the minimum thrust, ΔT i,max is the maximum thrust change rate, ΔT i,min is the minimum thrust change rate, γ i is the failure factor of the i-th thruster;

[0056] Step (3.3): The objective function of the model predictive controller:

[0057] J = min(J P + J s + J sm )

[0058] Taking the thrust vector of the thruster as the state quantity and the thrust change rate of the thruster as the control quantity, based on the objective function and constraint conditions of the underwater robot thrust prediction model, perform rolling optimization, solve the thrust values of each thruster and allocate them to the corresponding thrusters.

[0059] Further, step (4) specifically includes:

[0060] Step (4.1): Initialize a multi-universe population U = [U1, U2, Ω, U n T , where n is the number of universes;

[0061] Step (4.2): Initialize the lower limit WEP min and upper limit WEP max of the probability of the existence of wormholes in the multi-universe space, the exploitation degree p, the current iteration number l, and the maximum iteration number L;

[0062] Step (4.3): Calculate the fitness value of the universe individuals and obtain the current optimal universe by comparison;

[0063] Step (4.4): Enter the main loop and update WEP (probability of the existence of wormholes) and TDR (travel distance value) according to the following formula;

[0064]

[0065] where WEP represents the probability of the existence of wormholes in the multi-universe space, and TDR represents the step length of the object moving towards the current optimal universe;

[0066] ​Step (4.5): In view of the low search efficiency of the traditional multi - universe algorithm and the difficulty in balancing WEP and TDR during the iteration process, the traditional MVO algorithm is improved by using logarithmically increasing WEP and non - linearly converging TDR, as shown in the following formula;

[0067]

[0068] Step (4.6): Implement the roulette wheel mechanism according to the following formula;

[0069]

[0070] where NI(U i ) represents the normalized expansion rate of the i - th universe, r1 is a random number in the range of [0, 1], and x k j represents the j - th object of the k - th universe selected by the roulette wheel mechanism.

[0071] Step (4.7): Calculate the updated optimal universe according to the following formula. If it is better than the current optimal universe, replace it; otherwise, retain the current optimal universe;

[0072]

[0073] where X j represents the j - th object of the current optimal universe, lb j and ub j represent the lower and upper bounds of x respectively, and r2, r4 are random numbers in the range of [0, 1].

[0074] Step (4.8): Judge the termination criterion. If the maximum number of iterations or the minimum precision requirement is reached, exit the main loop and output the optimal universe and the objective function value; otherwise, return to Step (4.3).

[0075] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0076] 1. The thrust values assigned to each thruster of the underwater robot by the thrust allocation method of the underwater robot proposed by the present invention can accurately track the expected thrust of the underwater robot;

[0077] 2. Compared with the traditional single - step optimization method, the method of multi - step optimization solution of thrust allocation based on model predictive control proposed by the present invention has a larger feasible region for thrust solution and can avoid the problem of falling into local optimality;

[0078] 3. By modifying the thrust prediction model of the underwater robot, the present invention can avoid the problem of thrust saturation of the faulty thruster in the case of thruster failure, accurately track the expected thrust of the underwater robot, and ensure the stability and safety of path tracking;

[0079] 4. By introducing the multiverse algorithm into the solution of the optimization problem in model predictive control, the improved model predictive control algorithm used in the present invention has stronger local and global optimization capabilities and convergence in a large search space, and it is easier to obtain the global optimal solution of the thruster allocation of the underwater robot. Description of the Drawings

[0080] Figure 1 is the coordinate system diagram of the underwater robot with respect to the ship;

[0081] Figure 2 is the top view of the thruster distribution of the underwater robot;

[0082] Figure 3 is the flow chart of the model predictive control thruster allocation of the present invention;

[0083] Figure 4 is the flow chart of the multiverse algorithm optimizing the model predictive control of the present invention. Detailed Embodiment

[0084] The present invention will be further described below with reference to the drawings and specific embodiments.

[0085] In the embodiment of the present invention, the underwater robot includes 6 thrusters, among which the No. 1, No. 2, No. 3, and No. 4 are horizontally distributed vector thrusters, and the No. 5 and No. 6 are vertical thrusters. The coordinate system of the underwater robot with respect to the ship is as Figure 1 shown, and the layout positions of each thruster are as Figure 2 shown. The present invention represents the thruster allocation matrix based on this underwater robot, which specifically includes the following steps:

[0086] Step (1.1): Construct the representation of the combined thrust and combined thrust moment of the four horizontal thrusters of the underwater robot in the coordinate system with respect to the ship:

[0087]

[0088] Among them, T1, T2, T3, and T4 are the thrusts of the No. 1, No. 2, No. 3, and No. 4 thrusters received by the underwater robot body respectively, X h , Y h , Z h are the combined thrusts of the horizontal thrusters received by the underwater robot body along the x-axis, y-axis, and z-axis of the coordinate system with respect to the ship respectively, K h , M h , N h are the combined thrust moments of the four horizontal thrusters on the underwater robot body along the x-axis, y-axis, and z-axis of the coordinate system with respect to the ship respectively, and α h is the angle between the thruster and the x-axis of the coordinate system with respect to the ship;

[0089] Step (1.2): Construct the representation of the combined thrust of the two vertical thrusters of the underwater robot in the ship-fixed coordinate system:

[0090]

[0091] where T5 and T6 are the thrusts exerted on the underwater robot body by thruster No. 5 and No. 6 respectively, and X v , Y v , Z v are the combined thrusts of the vertical thrusters on the underwater robot body along the x-axis, y-axis, and z-axis of the ship-fixed coordinate system respectively, and x v , y v are the position vectors of the vertical thrusters about the x-axis and y-axis of the ship-fixed coordinate system, and K v , M v , N v are the combined thrust moments of the two vertical thrusters on the underwater robot body about the x-axis, y-axis, and z-axis of the ship-fixed coordinate system respectively;

[0092] Step (1.3): From the above two sets of equations, the combined thrust of the six thrusters installed on the underwater robot to produce six-degree-of-freedom motion of the underwater robot body:

[0093]

[0094] where X T , Y T , Z T are the longitudinal thrust, lateral thrust, and vertical thrust produced by the six thrusters on the underwater robot body respectively, and K T , M T , N T are the rolling thrust moment, pitching thrust moment, and yaw thrust moment produced by the six thrusters on the body respectively;

[0095] The six-degree-of-freedom control vector acting on the underwater robot body can be expressed by the following formula:

[0096] τ = B(α)T

[0097] where τ = [X T Y T Z T K T M T N T T is the six-degree-of-freedom thrust vector acting on the underwater robot body, T = [T1 T2 T3 T4 T5 T6] T is the thrust vector output by the thrusters, and B(α) is the thruster vector arrangement matrix;

[0098] ​The present invention mathematically expresses the failure fault of the underwater robot thruster and introduces the thrust distribution equation. The specific steps are as follows:

[0099] Step (2.1): The mathematical model of the failure fault of the i-th thruster can be expressed as:

[0100]

[0101] where γ i ∈(-1, 0] is the failure factor of the i-th thruster, u i is the control voltage of the i-th thruster, is the control voltage of the i-th thruster after failure correction, γ i = 0 indicates that the i-th thruster is working normally without faults; when -1 < γ i < 0, it indicates that the i-th thruster is partially failed but still working;

[0102] The state equation of the underwater robot with a thruster fault can be expressed as:

[0103] x p (k + 1) = A p x p (k) + B p u(k) + E(k)γ(k) + d(k)

[0104] where, is the system state variable, is the system input is the corresponding system matrix, d(k) = △A p x p (k) + △B p u p (k) + ω(k) represents the sum of system parameter uncertainties and external disturbances;

[0105] Define the new state variable z(k) = [x p (k) T γ(k) T T , and the observation model can be obtained:

[0106]

[0107] where and I m represents the m-order identity matrix, I n represents the n-order identity matrix, 0 m×n represents the m×n zero matrix, 0 m represents the m×m zero matrix, 0 qm represents the q×m zero matrix, if​ Observable, system state and fault failure factor γ i Can be estimated by the observer;

[0108] Step (2.2): Introduce the fault weight matrix:

[0109]

[0110] The modified thrust allocation matrix becomes τ = B(α)WT.

[0111] As Figure 3 shown, the present invention provides a method for solving the thrust allocation matrix using a model predictive controller in the case of a thruster fault of an underwater robot, comprising the following steps:

[0112] Step (3.1): Objective function

[0113] Minimum power consumption:

[0114]

[0115] Among them, Is the power coefficient of the i-th thruster;

[0116] Minimum allocation error:

[0117] J s = s T Qs

[0118] Among them, s = τ d - B(β)T is the allocation error;

[0119] Avoid singular terms of the thruster configuration matrix:

[0120]

[0121] Among them, ρ and ε are adjustment parameters, and the purpose is to make B(β)B(β) T ≠ 0, that is, B(β) is row full rank;

[0122] Step (3.2): Constraint conditions

[0123] Equality constraint:

[0124] B(β)T + s = τ d

[0125] Thrust change constraint:

[0126]

[0127] Among them, T i,max Is the maximum thrust that the i-th thruster can generate, T i,minis the minimum thrust, ΔT i,max is the maximum thrust change rate, ΔT i,min is the minimum thrust change rate, γ i is the failure factor of the i-th thruster.

[0128] Step (3.3): Using the thrust vector of the thrusters as the state variables and the thrust change rate of the thrusters as the control variables, perform rolling optimization based on the objective function and constraints of the underwater robot thrust prediction model, solve the thrust values of each thruster, and allocate them to the corresponding thrusters;

[0129] Objective function of the model predictive controller:

[0130] J = min(J P + J s + J sm )

[0131] As Figure 4 shown, the method for solving model predictive control thrust allocation proposed by the present invention based on an improved multi-universe optimization algorithm includes the following steps:

[0132] Step (4.1): Initialize a multi-universe population U = [U1, U2, …, U n ) T , where n is the number of universes;

[0133] Step (4.2): Initialize the lower limit WEP min and upper limit WEP max of the probability of the existence of wormholes in the multi-universe space, the exploitation degree p, the current iteration number l, and the maximum iteration number L;

[0134] Step (4.3): Calculate the fitness value of the universe individuals and obtain the current optimal universe by comparison;

[0135] Step (4.4): Enter the main loop and update WEP (probability of the existence of wormholes) and TDR (travel distance value) according to the following formula;

[0136]

[0137] where WEP represents the probability of the existence of wormholes in the multi-universe space, and TDR represents the step size for the object to move towards the current optimal universe;

[0138] Step (4.5): Aiming at the problems of low search efficiency of the traditional multi-universe algorithm and the difficulty in balancing WEP and TDR during the iteration process, improve the traditional MVO algorithm, and use logarithmically increasing WEP and non-linearly converging TDR, as shown in the following formula;

[0139]

[0140] Step (4.6): Execute the roulette mechanism according to the following formula;

[0141]

[0142] where NI(U i ) represents the normalized inflation rate of the i-th universe (i.e., with a length of 1), r1 is a random number within the range of [0, 1], represents the j-th object of the k-th universe selected by the roulette mechanism;

[0143] Step (4.7): Calculate the updated optimal universe according to the following formula. If it is better than the current optimal universe, replace it; otherwise, retain the current optimal universe;

[0144]

[0145] where X j represents the j-th object of the current optimal universe, lb j and ub j respectively refer to the lower and upper limits of x, and r2, r4 are random numbers within the range of [0, 1];

[0146] Step (4.8): Judge the termination criterion. If the maximum number of iterations or the minimum precision requirement is reached, exit the main loop and output the optimal universe and the objective function value; otherwise, return to Step (4.3).

[0147] In addition to the above embodiments, the present invention may also have other implementation manners. All technical solutions formed by equivalent replacement or equivalent transformation fall within the protection scope required by the present invention.

Claims

1. A fault-tolerant control method for an underwater robot propulsion system, characterized in that, The method includes the following steps: (1) Establish a six-degree-of-freedom underwater robot mathematical model with four horizontal thrusters and two vertical thrusters, where thrusters No. 1, 2, 3, and 4 are horizontally distributed vector thrusters, and thrusters No. 5 and 6 are vertical thrusters; (2) Obtain the fault estimates γ1, γ2, γ3, γ4, γ5, γ6 of each thruster in real time through a sliding mode observer. Among them, γ1, γ2, γ3, γ4 are the fault coefficients of the four horizontal thrusters, and γ5, γ6 are the fault coefficients of the two vertical thrusters. Introduce a fault weight matrix W. When there is no fault, establish an underwater robot thrust prediction model based on the thrust allocation vector matrix B(α). When a fault occurs, establish a fault prediction model based on the original thrust allocation matrix and the fault weight matrix W; (3) Use model predictive control to solve the open-loop optimization problem of thrust allocation to obtain the optimal thrust solution. The optimization objectives considered for thrust reallocation include: minimum power consumption, minimum allocation error, and avoidance of singular terms in the thruster configuration matrix. The constraint conditions include equality constraints and thrust change constraints under fault conditions. Taking the thrust vector of the thrusters as the state quantity and the thrust change rate of the thrusters as the control quantity, perform rolling optimization based on the objective function and constraint conditions of the underwater robot thrust prediction model, and solve to obtain the thrust values of each thruster and allocate them to the corresponding thrusters; (4) Use the multi-universe optimization algorithm to optimize the calculation of the objective function in model predictive control, improve the traditional MVO algorithm, and use the logarithmically increasing WEP and non-linearly converging TDR. WEP represents the probability of the existence of a wormhole in the multi-universe space, and TDR represents the step size for an object to move towards the current optimal universe.

2. The fault-tolerant control method for the propulsion system failure of an underwater robot, where the underwater robot includes six thrusters, among which thrusters No. 1, No. 2, No. 3, and No. 4 are horizontally distributed vector thrusters, and thrusters No. 5 and No. 6 are vertical thrusters, characterized in that, In the above step (1), the establishment of the six-degree-of-freedom underwater robot mathematical model includes the following steps: Step (1.1): Construct the combined thrust and combined thrust moment expressions of the four horizontal thrusters of the underwater robot in the ship-fixed coordinate system: Among them, T1, T2, T3, and T4 are the thrusts of the underwater robot body received by thrusters 1, 2, 3, and 4 respectively; X h , Y h , Z h are the resultant thrusts of the horizontal thrusters on the x-axis, y-axis, and z-axis of the ship-fixed coordinate system received by the underwater robot body respectively. x h , y h , z h are the position vectors of the horizontal thrusters around the x-axis, y-axis, and z-axis of the ship-fixed coordinate system. K h , M h , N h are the resultant thrust moments of the four horizontal thrusters on the x-axis, y-axis, and z-axis of the ship-fixed coordinate system on the underwater robot body respectively. α h is the angle between the thruster and the x-axis of the ship-fixed coordinate system; Step (1.2): Construct the combined thrust expression of the two vertical thrusters of the underwater robot in the ship-fixed coordinate system: Among them, T5 and T6 are the thrusts of the 5th and 6th thrusters on the underwater robot body, respectively, X v , Y v , Z v are the resultant thrusts of the vertical thrusters on the underwater robot body along the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively. x v , y v are the position vectors of the vertical thrusters around the x-axis and y-axis of the ship-fixed coordinate system. K v , M v , N v are the resultant torque moments of the two vertical thrusters on the underwater robot body around the x-axis, y-axis, and z-axis of the ship-fixed coordinate system, respectively; Step (1.3): From the above two equations, the combined thrust of the six thrusters installed on the underwater robot generates six-degree-of-freedom motion of the underwater robot body: Among them, X T , Y T , Z T are respectively the longitudinal thrust, lateral thrust and vertical thrust generated by six thrusters on the underwater robot body, and K T , M T , N T are respectively the rolling thrust moment, pitching thrust moment and yaw thrust moment generated by six thrusters on the body; The six-degree-of-freedom control vector acting on the underwater robot body is expressed by the following formula: τ = B(α)T where τ = [X T Y T Z T K T M T N T T is the six-degree-of-freedom thrust vector acting on the underwater robot body, T = [T1 T2 T3 T4 T5 T6] T is the thrust vector output by the thruster; B(α) is the thruster vector arrangement matrix.​ 3. The fault-tolerant control method for the underwater robot propulsion system according to claim 1, characterized in that In the above step (2), obtaining the fault estimates of each thruster and introducing a fault weight matrix to establish a fault prediction model includes the following steps: Step (2.1): The mathematical model of the failure of the i-th thruster is expressed as: where γ i ∈ (-1, 0] is the failure factor of the i-th thruster, u i is the control voltage of the i-th thruster, is the control voltage of the i-th thruster after failure correction, γ i = 0 indicates that the i-th thruster is working properly without faults; when -1 < γ i < 0, it indicates that the i-th thruster is partially failed but still working; The state equation of the underwater robot with thruster faults is expressed as: x p (k + 1)= A p x p (k)+ B p u(k)+ E(k)γ(k)+ d(k) Among them, is the system state variable, is the system input is the corresponding system matrix, d(k) = △A p x p (k) + △B p u p (k) + ω(k) represents the sum of system parameter uncertainties and external disturbances; Define a new state variable \(z(k)=[x p (k) T \gamma(k) T \ T , and the observation model can be obtained as follows: Among them and I m represents the identity matrix of order m, and I n represents the identity matrix of order n, 0 m×n represents the zero matrix of m×n, 0 m represents the zero matrix of m×m, 0 qm represents the zero matrix of q×m. If is observable, the system state and the fault failure factor γ i can be estimated by the observer; Step (2.2): Introduce a fault weight matrix: The modified thrust allocation matrix becomes τ = B(α)WT.

4. The fault tolerance control method for the underwater robot propulsion system according to claim 1, characterized in that, In the above step (3), using model predictive control to solve the thrust reallocation problem includes the following steps: Step (3.1): Objective function Minimum power consumption: Among them, is the power coefficient of the i-th thruster; Minimum allocation error: J s = s T Qs where s = τ d -B(β)T is the allocation error; Avoidance of singular terms in the thruster configuration matrix: where ρ and ε are adjustment parameters, aiming to make B(β)B(β) T ≠ 0, that is, the rows of B(β) are full rank; Step (3.2): Constraint conditions Equality constraints: B(β)T + s = τ d Thrust change constraints: Among them, T i,max is the maximum thrust that the i-th thruster can generate, T i,min is the minimum thrust, ΔT i,max is the maximum thrust change rate, ΔT i,min is the minimum thrust change rate, γ i is the failure factor of the i-th thruster; Step (3.3): The objective function of the model predictive controller: J = min(J P + J s + J sm ) Taking the thrust vector of the thruster as the state variable and the thrust change rate of the thruster as the control variable, based on the objective function and constraints of the underwater robot thrust prediction model, perform rolling optimization to solve the thrust values of each thruster and allocate them to the corresponding thrusters.

5. The fault tolerance control method of the underwater robot multi-motor propulsion system according to claim 1, characterized in that In step (4) above, using the improved multi - universe algorithm to optimize and obtain the objective function of the rolling optimization in model predictive control includes the following steps: Step (4.1): Initialize a multiverse population U = [U1, U2, …, U n T , where n is the number of universes;​ Step (4.2): Initialize the lower limit WEP of the probability of the existence of a wormhole in the multiverse space min and the upper limit WEP max , the exploitation degree p, the current iteration number l, and the maximum iteration number L; Step (4.3): Calculate the fitness value of the universe individuals and obtain the current optimal universe by comparison. Step (4.4): Enter the main loop and update WEP and TDR according to the following formula: where WEP represents the probability of the existence of a wormhole in the multi - universe space, and TDR represents the step size for an object to move towards the current optimal universe. Step (4.5): Aiming at the problems of low search efficiency of the traditional multi - universe algorithm and the difficulty in balancing WEP and TDR during the iteration process, improve the traditional MVO algorithm by using logarithmically increasing WEP and non - linearly convergent TDR, as shown in the following formula: Step (4.6): Execute the roulette wheel mechanism according to the following formula: where NI(U i ) represents the normalized expansion rate of the i-th universe, r1 is a random number in the range of [0, 1], represents the j-th object of the k-th universe selected by the roulette mechanism; Step (4.7): Calculate the updated optimal universe according to the following formula. If it is better than the current optimal universe, replace it; otherwise, retain the current optimal universe. Among them, X j represents the j-th object of the current optimal universe, lb j and ub j respectively refer to the lower and upper limits of x, and r2, r4 are random numbers within the range of [0, 1]; Step (4.8): Judge the termination criterion. If the maximum number of iterations or the minimum accuracy requirement is reached, exit the main loop and output the optimal universe and the objective function value; otherwise, return to step (4.3).

Images