An underwater high-precision attitude control method of ROV

Abstract

The application discloses a kind of ROV underwater high-precision attitude control methods.Method includes: establishing dynamic model, attitude is decoupled to obtain control subsystem;Design sliding mode controller, desired attitude input, output desired thrust control variable control propeller group;Generate force and torque control ROV moves under external disturbance, output actual attitude input sliding mode controller;Establish RBF disturbance estimation model, input actual depth, bow angle and pitch angle, output estimated external disturbance and input sliding mode controller;Parameter self-correction is carried out to obtain parameter identification result feedback to sliding mode controller, control propeller group, realize the underwater attitude control of unmanned remote control submersible ROV.The method of the application adds model parameter self-correction and RBF neural network disturbance estimation process on the basis of sliding mode control, can effectively improve the attitude control precision and robustness, and can reduce chattering to some extent, improves the control performance of controller.

Description

Technical Field

[0001] The present invention relates to a posture control method, in particular to a high-precision underwater posture control method for an ROV based on parameter self-correction and RBF neural network disturbance estimation sliding mode control. Background Art

[0002] Underwater robots, as a crucial tool and equipment for marine resource development and environmental monitoring, have garnered widespread attention. Currently, they are used in a variety of applications, including underwater sampling, marine environmental monitoring, underwater search and rescue, and underwater construction operations. They have generated significant socioeconomic benefits and hold a broad range of potential applications. Compared to manned vehicles, unmanned remotely operated vehicles (ROVs) are more compact and can carry more sensors and operational equipment, enabling them to perform complex underwater operations in place of humans.

[0003] ROVs are subject to underwater resistance, buoyancy, and currents when moving underwater, all of which place higher demands on their underwater attitude control. Furthermore, with technological advancements, ROVs are no longer limited to low-speed underwater navigation; capabilities such as trajectory tracking and fixed-point observation are increasingly required. Therefore, high-precision attitude control for ROVs requires further research. Summary of the Invention

[0004] To address the problems presented in the prior art, the present invention provides a high-precision underwater attitude control method for ROVs. This method utilizes the nonlinear approximation capabilities of RBF neural networks to estimate external environmental disturbances and employs online parameter identification methods to identify system parameters in real time, improving model accuracy. Because external disturbances are estimated and model uncertainty is eliminated, a smaller switching term coefficient can be used compared to traditional sliding mode control, improving control robustness and reducing chattering.

[0005] The technical solution adopted in the present invention is:

[0006] The ROV underwater high-precision attitude control method of the present invention comprises the following steps:

[0007] Step 1: Establish a dynamic model of the underwater motion of an unmanned remotely operated vehicle (ROV) equipped with a thruster group, decouple the dynamic model at each posture of the unmanned remotely operated vehicle (ROV), and obtain the control subsystem at each posture.

[0008] Step 2: According to the control subsystem at each posture, a sliding mode controller is designed using a sliding mode control method based on the reaching law. The desired posture of the unmanned remotely operated vehicle (ROV) is input into the sliding mode controller, and the sliding mode controller outputs the desired thrust control quantity at each posture to control the thruster group.

[0009] Step 3: The thruster group generates force and torque control unmanned remote control submersible ROV moves in the state of being disturbed by external disturbance, and the unmanned remote control submersible ROV outputs actual attitude and inputs into the sliding mode controller in real time.

[0010] Step 4: An RBF disturbance estimation model is established, the actual depth, actual heading angle and actual pitch angle of the unmanned remote control submersible ROV are input into the RBF disturbance estimation model, the RBF disturbance estimation model outputs the estimated external disturbance of the unmanned remote control submersible ROV and inputs into the sliding mode controller in real time, so as to eliminate the influence of external disturbance on the control system, realize compensation for external disturbance, and use the disturbance estimation value instead of the actual disturbance.

[0011] Step 5: Recursive least square identification method is used to correct parameters of the control subsystem in each attitude to obtain parameter identification results, which are fed back to the sliding mode controller in real time to eliminate the influence of model parameter error on the control system, and the sliding mode controller outputs the expected thrust control amount in each attitude to control the thruster group, which controls the movement of the unmanned remote control submersible ROV, and realizes high-precision underwater attitude control of the unmanned remote control submersible ROV.

[0012] In the step 1, the thruster group comprises a plurality of thrusters; each attitude of the unmanned remote control submersible ROV comprises a depth, a heading angle and a pitch angle attitude. According to the different installation quantity and installation position of each thruster, each attitude is divided into different thruster control.

[0013] In the step 1, the established dynamic model of underwater movement of the unmanned remote control submersible ROV, i.e. the spatial six-degree-of-freedom motion equation of the ROV, is as follows:

[0014]

[0015] Wherein, M represents the inertia matrix; v and respectively represent the velocity vector and its differential of the unmanned remote control submersible ROV in the body coordinate system of the ROV; C(v) represents the Coriolis force and centripetal force matrix D(v) represents the hydrodynamic coefficient matrix; g(η) represents the force and torque vector generated by gravity and buoyancy; η represents the attitude vector of the unmanned remote control submersible ROV in the geodetic coordinate system; τ represents the force and torque vector generated by the thruster group.

[0016] The body coordinate system of the ROV is a three-dimensional xyz coordinate system with the center of gravity of the unmanned remote control submersible ROV as the origin and the forward direction of the unmanned remote control submersible ROV as the x axis, which satisfies the right-hand rule.

[0017] In step 1, the kinetic model is decoupled at each attitude of the remotely operated vehicle (ROV), and other motions are not considered when single attitude control is considered, so the motion equation can be decoupled to single degree of freedom, thereby reducing the complexity of control and improving the control performance, and a single degree of freedom control equation is obtained after considering the real external disturbance, that is, a control subsystem at each attitude is obtained, as follows:

[0018] a) Depth control subsystem:

[0019]

[0020] wherein m represents the mass of the remotely operated vehicle (ROV); represents a first added mass coefficient; w and respectively represent the z-axis velocity and the differential of the remotely operated vehicle (ROV) in the ROV body coordinate system; Z w represents a first first-order hydrodynamic coefficient; Z w|w| represents a first second-order hydrodynamic coefficient; Z represents the force of the propeller acting on the remotely operated vehicle (ROV) at the depth attitude; d1 represents the external disturbance acting on the remotely operated vehicle (ROV) at the depth attitude, and the external disturbance specifically includes water flow, wave, ocean current and the like.

[0021] b) Heading angle control subsystem:

[0022]

[0023] wherein I zz represents the moment of inertia of the remotely operated vehicle (ROV) on the z-axis in the ROV body coordinate system; represents a second added mass coefficient; r and respectively represent the z-axis angular velocity and the differential of the remotely operated vehicle (ROV) in the ROV body coordinate system; N r represents a second first-order hydrodynamic coefficient; N r|r| a second second-order hydrodynamic coefficient; N represents the torque of the propeller acting on the remotely operated vehicle (ROV) at the heading angle attitude; d2 represents the external disturbance acting on the remotely operated vehicle (ROV) at the heading angle attitude.

[0024] c) Pitch angle control subsystem:

[0025]

[0026] wherein I yy represents the moment of inertia of the remotely operated vehicle (ROV) on the y-axis in the ROV body coordinate system; represents a third added mass coefficient; q and respectively represent the y-axis angular velocity and the differential of the remotely operated vehicle (ROV) in the ROV body coordinate system; Mq denotes the third first-order hydrodynamic coefficient; M q|q| denotes the third second-order hydrodynamic coefficient; M denotes the moment acting on the ROV at the pitch angle attitude; B denotes the buoyancy force received by the ROV underwater; h denotes the metacentric height of the ROV; θ denotes the pitch angle of the ROV; d3 denotes the external disturbance acting on the ROV at the pitch angle attitude.

[0027] The designed sliding mode controller in step 2 is specifically as follows:

[0028]

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036]

[0037] wherein ut1, ut2 and ut3 respectively denote the expected thrust control amount of the ROV at the depth, heading angle and pitch angle attitude, i.e. the depth control force, the heading angle control moment and the pitch angle control moment; m denotes the mass of the ROV; denotes the first additional mass coefficient; and respectively denote the first, second and third model parameter identification values in the depth system, and the initial values are taken as the modeling values; c1, c2 and c3 respectively denote the first sliding mode coefficients at the depth, heading angle and pitch angle attitudes, c1>0, c2>0, c3>0; e1 and respectively denote the attitude deviation and its differential of the ROV at the depth attitude; denotes the expected depth z of the ROV dsecond derivative of the depth; k1, k2 and k3 represent the second sliding mode coefficients in the depth, yaw angle and pitch angle postures respectively, k1 > 0, k2 > 0, k3 > 0; s1, s2 and s3 represent the first, second and third sliding mode functions respectively; ε1, ε2 and ε3 represent the third sliding mode coefficients in the depth, yaw angle and pitch angle postures respectively, ε1 > 0, ε2 > 0, ε3 > 0; sat() represents a saturation function; w represents the z-axis velocity of the ROV in the body coordinate system of the ROV; and represent the estimated external disturbances acting on the ROV in the depth, yaw angle and pitch angle postures respectively, the initial values are taken as 0; and represent the first, second and third exponential reaching laws respectively; sgn() represents a sign function; I zz represents the moment of inertia of the ROV in the z-axis in the body coordinate system of the ROV; represents the second added mass coefficient; and represent the first, second and third model parameter identification values in the yaw angle subsystem, the initial values are taken as the modeling values; e2 and represent the attitude deviation and its derivative of the ROV in the yaw angle posture; represents the desired yaw angle ψ d of the ROV; r represents the z-axis angular velocity of the ROV in the body coordinate system of the ROV; I yy represents the moment of inertia of the ROV in the y-axis in the body coordinate system of the ROV; represents the third added mass coefficient; and represent the first, second and third model parameter identification values in the pitch angle subsystem, the initial values are taken as the modeling values; e3 and represent the attitude deviation and its derivative of the ROV in the pitch angle posture; represents the desired pitch angle θ d of the ROV; q represents the y-axis angular velocity of the ROV in the body coordinate system of the ROV; B represents the buoyancy received by the ROV underwater; h represents the distance between the center of gravity and buoyancy of the ROV; θ represents the pitch angle of the ROV.

[0038] The desired thrust control amount in each posture includes the desired thrust control amounts ut1, ut2 and ut3 of the ROV in the depth, yaw angle and pitch angle postures.

[0039] The control force and torque on each system calculated by the sliding mode controller are fed back to the thrust controller, and the ROV attitude adjustment process is completed by controlling the thruster group.

[0040] The attitude deviation e1 of the ROV in the depth attitude, the attitude deviation e2 of the ROV in the heading angle attitude, and the attitude deviation e3 of the ROV in the pitch angle attitude are as follows:

[0041] e1=z-z d

[0042] e2=ψ-ψ d

[0043] e3=θ-θ d

[0044] Wherein, z represents the actual depth of the ROV; z d represents the desired depth of the ROV; ψ d and θ d respectively represent the desired angle of the y-axis and the desired angle of the z-axis of the ROV in the ROV body coordinate system.

[0045] The desired attitude of the ROV includes the desired depth z d of the ROV, and the desired heading angle ψ d and the desired pitch angle θ d of the ROV in the ROV body coordinate system.

[0046] The total actual attitude of the ROV measured by the sensor group is ξ, and the total desired attitude is ξ d , and the sensor group includes at least two sensors, including angle sensors or electronic compasses, depth sensors, etc.

[0047] In step 4, the established RBF disturbance estimation model includes an input layer, a hidden layer, and an output layer, and in the specific implementation, the input layer node is 2, the hidden layer node is 7, and the output layer node is 1.

[0048] The estimated external disturbance of the ROV output by the RBF disturbance estimation model is as follows:

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] wherein, and represent the first, second and third network weights respectively; h() represents an activation function, h(x) = [h1(x) h2(x)... h n (x)] T , n is the number of hidden layer nodes; x1, x2 and x3 represent the actual depth matrix, the actual heading angle matrix and the actual pitch angle matrix of the remotely operated vehicle (ROV) respectively; and represent the first network weight the second network weight and the third network weight adjustment adaptive law; γ1, γ2 and γ3 represent the first, second and third adaptive learning rates of the RBF disturbance estimation model; ψ and represent the actual heading angle and its differential of the remotely operated vehicle (ROV) respectively; θ and represent the actual pitch angle and its differential of the remotely operated vehicle (ROV) respectively.

[0059] In the step 5, the recursive least square identification method is used to perform parameter self-correction on the control subsystem in each attitude to obtain parameter identification results, which are specifically as follows:

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] wherein, and respectively represent the parameter identification result of the first to-be-identified parameter vector Θ1 at time t and time t-1; and respectively represent the parameter identification result of the second to-be-identified parameter vector Θ2 at time t and time t-1; and respectively represent the parameter identification result of the third to-be-identified parameter vector Θ3 at time t and time t-1; K1(t), K2(t) and K3(t) respectively represent the gain matrix at time t; and respectively represent the depth, heading angle and pitch angle attitude information vector at time t; P1(t) and P1(t-1) respectively represent the covariance matrix at time t; α1, α2 and α3 respectively represent the system forgetting factors of the depth control subsystem, the heading angle subsystem and the pitch angle subsystem.

[0070] The recursive least square identification method is used to identify the system parameters, and the identification result is fed back to the sliding mode controller, so as to realize online identification and correction of the model parameters, correct the system parameters online, compensate for the underwater time-varying error and actual true value error, and thus eliminate the influence of the uncertainty term of the model.

[0071] The depth, heading angle and pitch angle attitude information vector at time t and are specifically as follows:

[0072]

[0073]

[0074]

[0075] The first to-be-identified parameter vector Θ1, the second to-be-identified parameter vector Θ2 and the third to-be-identified parameter vector Θ3 are specifically as follows:

[0076]

[0077]

[0078]

[0079] The present application has the following beneficial effects:

[0080] 1. The RBF neural network estimation process designed by the application can effectively estimate and compensate external environmental disturbances in ROV attitude control, and eliminate the influence of underwater interference.

[0081] 2. The parameter self-correction part designed by the application can effectively estimate and track the true value and change value of the ROV parameter, and eliminate the influence of the modeling error part.

[0082] 3. Compared with the traditional sliding mode control, the control model of the method is more perfect, high-precision attitude control can be realized under a smaller switching term coefficient, the control precision and robustness of the system are improved, and the chattering is reduced to a certain extent through a smaller switching term coefficient, and the control performance of the controller is improved. BRIEF DESCRIPTION OF DRAWINGS

[0083] Figure 1 is the ROV thruster group layout in the embodiment;

[0084] Figure 2 is the attitude control block diagram of the method of the application;

[0085] Figure 3 (a) of is the depth control result curve in the embodiment;

[0086] Figure 3 (b) of is the depth control error curve in the embodiment;

[0087] Figure 4 (a) of is the depth system disturbance estimation curve in the embodiment;

[0088] Figure 4 (b) of is the depth system parameter identification curve in the embodiment;

[0089] Figure 5 (a) of is the pitch angle control result curve in the embodiment;

[0090] Figure 5 (b) of is the pitch angle control error curve in the embodiment;

[0091] Figure 6 (a) of is the pitch angle system disturbance estimation curve in the embodiment;

[0092] Figure 6 (b) of is the pitch angle system parameter identification curve in the embodiment;

[0093] Figure 7 (a) of is the heading angle simulation result curve in the embodiment;

[0094] Figure 7 (b) of is the heading angle simulation error curve in the embodiment;

[0095] Figure 8 (a) is the heading angle system disturbance estimation graph in the embodiment;

[0096] Figure 8 (b) is the heading angle system parameter identification graph in the embodiment. DETAILED DESCRIPTION

[0097] The application will be further explained in detail below in connection with the accompanying drawings and specific embodiments, which are explained by way of example and not by way of limitation, i.e. the scope of protection of the application is not limited by the specific embodiments.

[0098] As shown in Figure 2 the application ROV underwater high-precision attitude control method comprises the following steps:

[0099] Step 1: Establish a dynamic model of underwater motion of an unmanned remotely operated vehicle (ROV) installed with a propeller group, decouple the dynamic model at each attitude of the ROV, and obtain a control subsystem at each attitude.

[0100] In step 1, the propeller group includes a plurality of propellers; each attitude of the ROV includes a depth, a heading angle, and a pitch angle attitude. According to the number and position of the installation of each propeller, each attitude is divided into different propeller control.

[0101] In step 1, the dynamic model of underwater motion of the ROV, i.e. the spatial six-degree-of-freedom motion equation of the ROV, is established as follows:

[0102]

[0103] where M represents an inertia matrix; v and respectively represent the velocity vector and the differential of the ROV in the body coordinate system of the ROV; C(v) represents a Coriolis force and centripetal force matrix D(v) represents a hydrodynamic coefficient matrix; g(η) represents a force and moment vector generated by gravity and buoyancy; η represents an attitude vector of the ROV in the geodetic coordinate system; τ represents a force and moment vector generated by the propeller group.

[0104] The body coordinate system of the ROV is a three-dimensional xyz coordinate system with the center of gravity of the ROV as the origin and the forward direction of the ROV as the x-axis, which satisfies the right-hand rule.

[0105] In step 1, the kinetic model is decoupled at each attitude of the remotely operated vehicle (ROV), and no other motion is considered when single attitude control is considered, so the motion equation can be decoupled to single degree of freedom, thereby reducing the complexity of control and improving the control performance. The single degree of freedom control equation is obtained after considering the real external disturbance, that is, the control subsystem at each attitude is obtained, as follows:

[0106] a) Depth control subsystem:

[0107]

[0108] wherein m represents the mass of the remotely operated vehicle (ROV); represents the first added mass coefficient; w and respectively represent the z-axis velocity of the remotely operated vehicle (ROV) in the ROV body coordinate system and the differential thereof; Z w represents the first first-order hydrodynamic coefficient; Z w|w| represents the first second-order hydrodynamic coefficient; Z represents the force of the propeller acting on the remotely operated vehicle (ROV) at the depth attitude; d1 represents the external disturbance acting on the remotely operated vehicle (ROV) at the depth attitude, and the external disturbance specifically includes water flow, wave, ocean current and the like.

[0109] b) Heading angle control subsystem:

[0110]

[0111] wherein I zz represents the moment of inertia of the remotely operated vehicle (ROV) in the z-axis in the ROV body coordinate system; represents the second added mass coefficient; r and respectively represent the z-axis angular velocity of the remotely operated vehicle (ROV) in the ROV body coordinate system and the differential thereof; N r represents the second first-order hydrodynamic coefficient; N r|r| the second second-order hydrodynamic coefficient; N represents the torque of the propeller acting on the remotely operated vehicle (ROV) at the heading angle attitude; d2 represents the external disturbance acting on the remotely operated vehicle (ROV) at the heading angle attitude.

[0112] c) Pitch angle control subsystem:

[0113]

[0114] wherein I yy represents the moment of inertia of the remotely operated vehicle (ROV) in the y-axis in the ROV body coordinate system; represents the third added mass coefficient; q and respectively represent the y-axis angular velocity of the remotely operated vehicle (ROV) in the ROV body coordinate system and the differential thereof; Mq represents the third first-order hydrodynamic coefficient; M q|q| represents the third second-order hydrodynamic coefficient; M represents the moment acting on the remotely operated vehicle (ROV) at the pitch angle attitude; B represents the buoyancy received by the remotely operated vehicle (ROV) underwater; h represents the distance between the center of gravity and buoyancy of the remotely operated vehicle (ROV); θ represents the pitch angle of the remotely operated vehicle (ROV); and d3 represents the external disturbance acting on the remotely operated vehicle (ROV) at the pitch angle attitude.

[0115] Step 2: According to the control subsystem at each attitude, a sliding mode controller is designed using a sliding mode control method based on a reaching law, the desired attitude of the remotely operated vehicle (ROV) is input into the sliding mode controller, and the sliding mode controller outputs a desired thrust control amount at each attitude to control the thruster group.

[0116] In step 2, the designed sliding mode controller is as follows:

[0117]

[0118]

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126] wherein ut1, ut2 and ut3 respectively represent the desired thrust control amount of the remotely operated vehicle (ROV) at the depth, heading angle and pitch angle attitude, i.e. the depth control force, heading angle control moment and pitch angle control moment; and m represents the mass of the remotely operated vehicle (ROV). represents the first additional mass coefficient; and respectively represent the first, second and third model parameter identification values in the depth system, and the initial values are taken as the modeling values; c1, c2 and c3 respectively represent the first sliding mode coefficients at the depth, heading angle and pitch angle attitudes, c1>0, c2>0, c3>0; e1 and respectively represent the attitude deviation and its differential of the remotely operated vehicle (ROV) at the depth attitude. z represents the desired depth of the remotely operated vehicle (ROV) d represents the second derivative of the depth; k1, k2, and k3 represent the second sliding mode coefficients in the depth, yaw, and pitch attitude, respectively, k1 > 0, k2 > 0, k3 > 0; s1, s2, and s3 represent the first, second, and third sliding mode functions, respectively; ε1, ε2, and ε3 represent the third sliding mode coefficients in the depth, yaw, and pitch attitude, respectively, ε1 > 0, ε2 > 0, ε3 > 0; sat() represents the saturation function; w represents the z-axis velocity of the remotely operated vehicle (ROV) in the body coordinate system of the ROV; and represent the estimated external disturbances acting on the remotely operated vehicle (ROV) in the depth, yaw, and pitch attitude, respectively, with initial values taken as 0; and represent the first, second, and third exponential reaching law, respectively; sgn() represents the sign function; I zz represents the moment of inertia of the remotely operated vehicle (ROV) in the z-axis in the body coordinate system of the ROV; represents the second added mass coefficient; and represent the first, second, and third model parameter identification values in the yaw attitude subsystem, with initial values taken as the modeling values; e2 and represent the attitude deviation and its derivative of the remotely operated vehicle (ROV) in the yaw attitude, respectively; z represents the desired yaw angle of the remotely operated vehicle (ROV) d represents the second derivative of the yaw; r represents the z-axis angular velocity of the remotely operated vehicle (ROV) in the body coordinate system of the ROV; I yy represents the moment of inertia of the remotely operated vehicle (ROV) in the y-axis in the body coordinate system of the ROV; represents the third added mass coefficient; and represent the first, second, and third model parameter identification values in the pitch attitude subsystem, with initial values taken as the modeling values; e3 and represent the attitude deviation and its derivative of the remotely operated vehicle (ROV) in the pitch attitude, respectively; z represents the desired pitch angle of the remotely operated vehicle (ROV) d represents the second derivative of the pitch; q represents the y-axis angular velocity of the remotely operated vehicle (ROV) in the body coordinate system of the ROV; B represents the buoyancy received by the remotely operated vehicle (ROV) underwater; h represents the distance between the center of gravity and buoyancy of the remotely operated vehicle (ROV); θ represents the pitch angle of the remotely operated vehicle (ROV).

[0127] The desired thrust control amounts in each attitude include desired thrust control amounts ut1, ut2 and ut3 of the ROV in depth, heading angle and pitch angle attitudes.

[0128] After the control forces and moments on each system are calculated by the sliding mode controller, they are fed back to the thrust controller, and the ROV attitude adjustment process is completed by controlling the thruster group. In the sliding mode controller, the saturation function is used to replace the sign function to reduce chattering.

[0129] The attitude deviation e1 of the ROV in the depth attitude, the attitude deviation e2 in the heading angle attitude and the attitude deviation e3 in the pitch angle attitude are as follows:

[0130] e1 = z - z d

[0131] e2 = ψ - ψ d

[0132] e3 = θ - θ d

[0133] Wherein, z represents the actual depth of the ROV; z d represents the desired depth of the ROV; ψ d and θ d respectively represent the desired angle of the z-axis and the desired angle of the y-axis of the ROV in the ROV body coordinate system.

[0134] The desired attitude of the ROV includes the desired depth z d of the ROV and the desired heading angle ψ d and the desired pitch angle θ d of the z-axis and the y-axis of the ROV in the ROV body coordinate system.

[0135] The total actual attitude of the ROV measured by the sensor group is ξ, and the total desired attitude is ξ d , and the sensor group includes at least two sensors, including angle sensors or electronic compasses, depth sensors and the like.

[0136] Step 3: The thruster group generates forces and moments under the control of the desired thrust control amount to control the movement of the ROV under external disturbance. The ROV outputs the actual attitude and inputs it into the sliding mode controller in real time.

[0137] Step 4: Establish an RBF disturbance estimation model and input the actual depth, actual heading angle and actual pitch angle of the unmanned remotely operated vehicle (ROV) into the RBF disturbance estimation model. The RBF disturbance estimation model outputs the estimated external disturbance of the unmanned remotely operated vehicle (ROV) and inputs it into the sliding mode controller in real time to eliminate the influence of the external disturbance on the control system, realize compensation for the external disturbance, and use the disturbance estimation value instead of the actual disturbance.

[0138] In step 4, the established RBF disturbance estimation model includes an input layer, a hidden layer, and an output layer. In the specific implementation, the input layer node is 2, the hidden layer node is 7, and the output layer node is 1.

[0139] The estimated external disturbance of the ROV output by the RBF disturbance estimation model is as follows:

[0140]

[0141]

[0142]

[0143]

[0144]

[0145]

[0146]

[0147]

[0148]

[0149] in, and denote the first, second, and third network weights respectively; h() denotes the activation function, h(x)=[h1(x) h2(x) ... h n (x)] T , n is the number of hidden layer nodes; x1, x2 and x3 represent the actual depth matrix, actual heading angle matrix and actual pitch angle matrix of the unmanned remotely operated underwater vehicle ROV respectively; and Represents the first network weight Second network weight and the third network weight The adaptive law of regulation; γ1, γ2 and γ3 represent the first, second and third adaptive learning rates of the RBF perturbation estimation model; ψ and are the actual heading angle of the ROV and its differential; θ and They represent the actual pitch angle of the ROV and its differential respectively.

[0150] Step 5: Use the recursive least squares identification method to perform parameter self-correction on the control subsystem under each posture to obtain the parameter identification results, and feed the parameter identification results back to the sliding mode controller in real time to eliminate the influence of the model parameter error on the control system. The sliding mode controller outputs the expected thrust control amount under each posture in real time to control the thruster group. The thruster group controls the movement of the unmanned remotely operated vehicle (ROV) to achieve high-precision underwater posture control of the unmanned remotely operated vehicle (ROV).

[0151] In step 5, the recursive least squares identification method is used to perform parameter self-correction on the control subsystem under each posture to obtain the parameter identification results, which are as follows:

[0152]

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161] in, and Respectively represent the parameter identification results of the first parameter vector Θ1 to be identified at time t and time t-1; and Respectively represent the parameter identification results of the second parameter vector Θ2 to be identified at time t and time t-1; and Respectively represent the parameter identification results of the third parameter vector Θ3 to be identified at time t and time t-1; K1(t), K2(t) and K3(t) respectively represent the gain matrix at time t; and They represent the depth, heading angle and pitch angle attitude information vectors at time t respectively; P1(t) and P1(t-1) represent the covariance matrices at time t respectively; α1, α2 and α3 represent the system forgetting factors of the depth control subsystem, heading angle subsystem and pitch angle subsystem respectively.

[0162] Recursive least square identification method is used to identify system parameters, and the identified result Feedback to the sliding mode controller, realizing online identification and correction of model parameters, online correction of system parameters, making up the underwater time-varying error and actual true value error, so as to eliminate the influence of the uncertainty term of the model.

[0163] Depth, heading angle and pitch angle attitude information vector at time t And Specifically as follows:

[0164]

[0165]

[0166]

[0167] The first to be identified parameter vector Θ1, the second to be identified parameter vector Θ2 and the third to be identified parameter vector Θ3 are specifically as follows:

[0168]

[0169]

[0170]

[0171] The specific implementation of the application is as follows:

[0172] As Figure 1 shown, it is a ROV propeller layout in the application example, and six propellers are arranged on the ROV body and symmetrically distributed along the central axis of the ROV, including the front propeller T1 and the propeller T2 arranged vertically, the middle propeller T3 and the propeller T4, and the tail propeller T5 and the propeller T6 arranged horizontally. Among them, the depth is adjusted by the front propeller 1, the front propeller 2, the middle propeller 3 and the middle propeller 4, the thrust direction of each propeller is the same, the size is inversely proportional to the horizontal distance from the center of gravity, a resultant force is formed on the ROV to control the depth. The pitch angle is adjusted by the front propeller 1, the front propeller 2, the middle propeller 3 and the middle propeller 4, and the front propeller 1 and the front propeller 2 constitute group 1, and the middle propeller 3 and the middle propeller 4 constitute group 2, the forces exerted on the ROV by group 1 and group 2 are equal in size and opposite in direction, forming a resultant couple to realize the pitch angle control. The heading angle is adjusted by the tail propeller 5 and the tail propeller 6, and the forces exerted on the ROV by each propeller are equal in size and opposite in direction, forming a resultant couple to realize the heading angle control.

[0173] When controlling the ROV in the embodiment, the parameters used by the controller are shown in Table 1 below.

[0174] Table 1

[0175]

[0176] Set the expected target position to time-varying. The target position is set as follows:

[0177]

[0178] The ROV model parameters in the embodiment are shown in Table 2 below.

[0179] Table 2

[0180]

[0181] Figures 3 to 8 The depth, pitch angle and heading angle control examples are position and error curves in actual water environment, RBF neural network disturbance estimation curves and parameter identification result curves. Figure 3 、 Figure 5 、 Figure 7 The target position tracking curves in Figure 2 show that, under the control method of the present invention, the system's depth, pitch angle, and heading angle can stably track the given target position, demonstrating the system output's ability to track changing target positions under the control method of the present invention. The error curves for the three systems show that the system output has a certain error at the beginning of tracking, but the error quickly converges to near zero and remains very stable in the later stages, demonstrating that the control method of the present invention has high control accuracy for the system's attitude output.

[0182] from Figure 4 、 Figure 6 、 Figure 8 From the disturbance estimation and model parameter identification curves, it can be seen that the external environment disturbance estimation and model parameter identification results designed by the present invention can quickly converge to near the actual value, and the results are relatively stable thereafter, reflecting that the present invention has good estimation performance for external disturbances and model parameters. In summary, compared with traditional sliding mode control, the control method of the present invention does not require a large switching term coefficient to obtain better control accuracy and robustness, and has a better control effect on ROV attitude adjustment.

[0183] The above are specific examples of the present invention, but they cannot limit the scope of protection of the present invention. Any simple changes made within the technical solution of the present invention are within the scope of protection of the present invention.

Claims

1. A high-precision underwater attitude control method for ROV, characterized by: The method comprises the following steps: Step 1: Establish a dynamic model of the underwater motion of an ROV equipped with a thruster group, decouple the dynamic model at each posture of the ROV, and obtain the control subsystem at each posture; Step 2: Based on the control subsystems at each attitude, a sliding mode controller is designed using a sliding mode control method based on the reaching law. The desired attitude of the ROV is input into the sliding mode controller, which outputs the desired thrust control variable at each attitude to control the thruster group. Step 3: The thruster group generates force and torque under the control of the desired thrust control amount to control the ROV to move under the state of external disturbance. The ROV outputs the actual attitude and inputs it into the sliding mode controller in real time. Step 4: Establish an RBF disturbance estimation model and input the actual depth, actual heading angle, and actual pitch angle of the ROV into the RBF disturbance estimation model. The RBF disturbance estimation model outputs the estimated external disturbance of the ROV and inputs it into the sliding mode controller in real time. Step 5: Use the recursive least squares identification method to perform parameter self-correction on the control subsystem under each posture to obtain the parameter identification results, and feed the parameter identification results back to the sliding mode controller in real time. The sliding mode controller outputs the desired thrust control amount under each posture in real time to control the thruster group. The thruster group controls the movement of the unmanned remotely operated vehicle (ROV) to achieve underwater posture control of the unmanned remotely operated vehicle (ROV).

2. A ROV underwater high-precision attitude control method according to claim 1, characterized in that: In step 1, the thruster group includes a plurality of thrusters; and the postures of the unmanned remotely operated underwater vehicle (ROV) include depth, heading angle, and pitch angle.

3. The ROV underwater high-precision attitude control method according to claim 1, characterized in that: In step 1, the dynamic model of the underwater motion of the unmanned remotely operated vehicle (ROV) is established as follows: Where M represents the inertia matrix; v and They represent the velocity vector and its differential of the ROV in the ROV body coordinate system; C(v) represents the Coriolis force and centripetal force matrix; D(v) represents the hydrodynamic coefficient matrix; g(η) represents the force and torque vector generated by gravity and buoyancy; η represents the attitude vector of the ROV in the geodetic coordinate system; τ represents the force and torque vector generated by the thruster group; The ROV body coordinate system is specifically a three-dimensional xyz coordinate system with the center of gravity of the unmanned remotely operated vehicle ROV as the origin and the forward direction of the unmanned remotely operated vehicle ROV as the x-axis.

4. A method for high-precision underwater attitude control of an ROV according to claim 3, characterized in that: In step 1, the dynamic model is decoupled at each posture of the ROV to obtain the control subsystem at each posture, as follows: a) Depth control subsystem: Where m represents the mass of the unmanned remotely operated vehicle ROV; represents the first additional mass coefficient; w and They represent the z-axis velocity and its differential of the ROV in the ROV body coordinate system; Z w represents the first-order hydrodynamic coefficient; Z w|w| Represents the first and second order hydrodynamic coefficients; Z represents the force of the thruster acting on the ROV in the deep attitude; d1 represents the external disturbance acting on the ROV in the deep attitude; b) Heading angle control subsystem: Among them, I zz It represents the inertia moment of the ROV on the z-axis in the ROV body coordinate system; represents the second additional mass coefficient; r and They represent the axis angular velocity and its differential of the ROV in the ROV body coordinate system; N r represents the second first-order hydrodynamic coefficient; N r|r| The second-order hydrodynamic coefficient; N represents the torque of the propeller acting on the ROV in the heading angle attitude; d2 represents the external disturbance acting on the ROV in the heading angle attitude; c) Pitch angle control subsystem: Among them, I yy It represents the inertia moment of the ROV on the y-axis in the ROV body coordinate system; represents the third additional mass coefficient; q and They represent the y-axis angular velocity and its differential of the ROV in the ROV body coordinate system; M q represents the third first-order hydrodynamic coefficient; M q|q| The third second-order hydrodynamic coefficient; M represents the torque acting on the unmanned remotely operated vehicle ROV in the pitch angle posture; B represents the buoyancy of the unmanned remotely operated vehicle ROV underwater; h represents the distance from the center of gravity of the unmanned remotely operated vehicle ROV; θ represents the pitch angle of the unmanned remotely operated vehicle ROV; d3 represents the external disturbance acting on the unmanned remotely operated vehicle ROV in the pitch angle posture.

5. The method for high-precision underwater attitude control of an ROV according to claim 3, characterized in that: In step 2, the sliding mode controller designed is as follows: Wherein, ut1, ut2 and ut3 represent the desired thrust control quantities of the ROV at depth, heading angle and pitch angle, i.e., depth control force, heading angle control torque and pitch angle control torque, respectively; m represents the mass of the ROV; represents the first additional mass coefficient; and Represent the first, second and third model parameter identification values ​​in the depth system respectively; c1, c2 and c3 represent the first sliding mode coefficients at depth, heading angle and pitch angle respectively, c1>0, c2>0, c3>0; e1 and They represent the attitude deviation and its differential of the ROV in the deep attitude respectively; represents the expected depth z of the ROV d The second-order differential of ; k1, k2 and k3 represent the second sliding mode coefficients at depth, heading angle and pitch angle attitude, respectively, k1>0, k2>0, k3>0; s1, s2 and s3 represent the first, second and third sliding mode functions, respectively; ε1, ε2 and ε3 represent the third sliding mode coefficients at depth, heading angle and pitch angle attitude, respectively, ε1>0, ε2>0, ε3>0; sat() represents the saturation function; w represents the z-axis velocity of the unmanned remotely operated vehicle ROV in the ROV body coordinate system; and are the estimated external disturbances acting on the ROV at depth, heading angle, and pitch angle attitude, respectively; and Represent the first, second and third exponential reaching laws respectively; sgn() represents the sign function; I zz It represents the inertia moment of the ROV on the z-axis in the ROV body coordinate system; represents the second additional mass coefficient; and denote the first, second and third model parameter identification values ​​in the heading angle subsystem respectively; e2 and They represent the attitude deviation and its differential of the ROV in the heading angle attitude respectively; represents the desired heading angle ψ of the ROV d The second-order differential of r represents the z-axis angular velocity of the ROV in the ROV body coordinate system; I yy It represents the inertia moment of the ROV on the y-axis in the ROV body coordinate system; represents the third additional mass coefficient; and denote the first, second and third model parameter identification values ​​of the pitch angle subsystem respectively; e3 and They represent the attitude deviation and its differential of the ROV in pitch angle attitude respectively; represents the desired pitch angle θ of the ROV d The second-order differential of ; q represents the y-axis angular velocity of the unmanned remotely operated submersible ROV in the ROV body coordinate system; B represents the buoyancy of the unmanned remotely operated submersible ROV under water; h represents the gravity center distance of the unmanned remotely operated submersible ROV; θ represents the pitch angle of the unmanned remotely operated submersible ROV; The expected thrust control amount under each attitude includes the expected thrust control amount ut1, ut2 and ut3 of the unmanned remotely operated vehicle ROV under the attitude of depth, heading angle and pitch angle.

6. The method for high-precision underwater attitude control of an ROV according to claim 5, characterized in that: The attitude deviation e1 of the ROV in the depth attitude, the attitude deviation e2 in the heading angle attitude, and the attitude deviation e3 in the pitch angle attitude are specifically as follows: e1=zz d e2=ψ-ψ d e3=θ-θ d Where z represents the actual depth of the ROV; z d represents the expected depth of the unmanned remotely operated vehicle ROV; ψ d and θ d They represent the z-axis expected angle and y-axis expected angle of the unmanned remotely operated vehicle ROV in the ROV body coordinate system respectively; The desired attitude of the ROV includes the desired depth z of the ROV. d And the z-axis expected heading angle ψ of the unmanned remotely operated underwater vehicle ROV in the ROV body coordinate system d and the desired pitch angle θ on the y-axis d .

7. The method for high-precision underwater attitude control of an ROV according to claim 5, characterized in that: In step 4, the established RBF disturbance estimation model includes an input layer, a hidden layer and an output layer; The estimated external disturbance of the ROV output by the RBF disturbance estimation model is as follows: in, and denote the first, second and third network weights respectively; h() denotes the activation function; x1, x2 and x3 denote the actual depth matrix, actual heading angle matrix and actual pitch angle matrix of the unmanned remotely operated underwater vehicle ROV respectively; and Represents the first network weight Second network weight and the third network weight The adaptive law of regulation; γ1, γ2 and γ3 represent the first, second and third adaptive learning rates of the RBF perturbation estimation model; ψ and are the actual heading angle of the ROV and its differential; θ and They represent the actual pitch angle of the ROV and its differential respectively.

8. The method for high-precision underwater attitude control of an ROV according to claim 4, characterized in that: In step 5, the recursive least squares identification method is used to perform parameter self-correction on the control subsystem under each posture to obtain parameter identification results, which are as follows: in, and Respectively represent the parameter identification results of the first parameter vector Θ1 to be identified at time t and time t-1; and Respectively represent the parameter identification results of the second parameter vector Θ2 to be identified at time t and time t-1; and Respectively represent the parameter identification results of the third parameter vector Θ3 to be identified at time t and time t-1; K1(t), K2(t) and K3(t) respectively represent the gain matrix at time t; and They represent the depth, heading angle and pitch angle attitude information vectors at time t respectively; P1(t) and P1(t-1) represent the covariance matrices at time t respectively; α1, α2 and α3 represent the system forgetting factors of the depth control subsystem, heading angle subsystem and pitch angle subsystem respectively.

9. The method for high-precision underwater attitude control of an ROV according to claim 8, characterized in that: The depth, heading angle and pitch angle attitude information vector at time t and The details are as follows:

10. The method for high-precision underwater attitude control of an ROV according to claim 8, characterized in that: The first parameter vector Θ1 to be identified, the second parameter vector Θ2 to be identified, and the third parameter vector Θ3 to be identified are specifically as follows:

Images