ROV active disturbance rejection controller parameter optimization method based on improved NRBO

Abstract

The invention discloses an ROV active-disturbance-rejection controller parameter optimization method based on an improved NRBO, and the method comprises the steps: determining a to-be-optimized ROV active-disturbance-rejection controller parameter group and an optimization range of each parameter, and constructing a target function; initializing NRBO parameters, and generating an initial population; performing iterative optimization on the initial population based on an NRSR search rule and NTAO trap avoidance; in each iteration round, calculating an objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; finding an optimal individual in the current population according to the target function value, and taking the position of the optimal individual as a current optimal solution; wherein the optimal individual position is the corresponding ROV active disturbance rejection controller parameter group; and after a preset iteration convergence condition is satisfied, iteration is stopped, and an optimal ROV active disturbance rejection controller parameter set is obtained. According to the method, the control performance of the ROV active-disturbance-rejection controller is improved, and the anti-disturbance capability of the ROV active-disturbance-rejection controller is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of ROV non - linear control, and more specifically, to a method for optimizing the parameters of an active disturbance rejection controller of an ROV based on an improved NRBO. Background Technique

[0002] ROV (Remotely Operated Vehicle) is an underwater operation device connected to an on - water console through an umbilical cable and remotely controlled by an operator on the water surface. It has many characteristics such as high degrees of freedom, strong flexibility, strong adaptability, and mature technology. It is widely used in military and civilian fields and is often used to perform underwater operation tasks in high - pressure, low - temperature, dangerous areas or complex terrains. However, ROV is a strongly non - linear, under - actuated, and strongly coupled system, and the types of external disturbance forces are also relatively complex and diverse, such as external wind force, umbilical cable, etc., making it difficult to establish an accurate mathematical model. In other words, due to working in a complex and uncertain working environment, ROV requires a controller with high stability and strong anti - interference ability; and because the active disturbance rejection controller introduces too many parameters and it is difficult to adjust the parameters, often only the trial - and - error method can be used to manually adjust the parameters and observe the system response to gradually optimize, with low efficiency; the current intelligent swarm optimization algorithms have disadvantages such as poor global exploration ability, slow convergence speed, and easy to fall into local optimal values, resulting in poor control effects of the controller. Therefore, how to design a controller with good stability and strong anti - interference ability for this controlled system has become a hot issue in control engineering.

[0003] An active disturbance rejection controller (ADRC) includes a tracking differentiator (TD), a non - linear state error feedback (NLSEF), an extended state observer (ESO), and disturbance compensation. These components work together to improve the stability and anti - interference ability of the system. Next, these components will be described separately.

[0004] 1. Tracking Differentiator (TD):

[0005] In a PID controller, directly following a step signal will result in a large initial error and is prone to overshoot. The TD module can smooth the input signal and extract its differential signal, effectively solving the contradiction between fast tracking and avoiding overshoot, and providing a reasonable transition process for the system.

[0006] 2. Non - linear State Error Feedback (NLSEF):

[0007] Compared with the linear feedback used in traditional PID controllers, the NLSEF module uses a non - linear function to replace the traditional linear feedback, which not only speeds up the convergence rate but also enhances the anti - disturbance ability of the system.

[0008] 3. Extended State Observer (ESO):

[0009] In PID control, the integral link is mainly used to eliminate disturbances and can be regarded as a disturbance observer. However, its integral effect is slower and it will also cause overshoot. In the active disturbance rejection controller, ESO is used to estimate the system state in real - time and observe the total disturbance (including internal dynamics and external disturbances), thus eliminating the integral link. ESO is the core of the active disturbance rejection controller and realizes the dynamic compensation of the system state and disturbances.

[0010] 4. Disturbance compensation:

[0011] The disturbance compensation directly compensates the total disturbance estimated by ESO into the control quantity, enabling the controller to only handle other residual disturbances, which greatly simplifies the control task.

[0012] Currently, in the widely used mechanical system control, power system control, aircraft control, and ship engineering control, designing an ROV control system based on the active disturbance rejection controller can achieve a system with good stability and strong anti - disturbance ability.

[0013] However, in practical engineering applications, the active disturbance rejection controller has more adjustable parameters than the PID controller, the parameters are interrelated, and the calculation is more cumbersome; manual adjustment is more complex and difficult, which poses an obstacle to the application of the active disturbance rejection controller.

[0014] For the above problems, Peng et al. (Motion control of autonomous underwater vehicle based on PIO-ADRC [J]. China Ocean Platform, 2024, 39(04): 47-53.) proposed a motion control method for AUV based on PIO (pigeon flock optimization algorithm), and compared and analyzed it with traditional ADRC and PIO-PID. It was concluded that PIO-ADRC showed the best performance in overshoot, rise time, and steady-state error when considering interference. Wang et al. (Depth control of ROV using the improved LADRC based on nutcracker optimization algorithm [J]. Ocean Engineering, 2024, 10, 309.) proposed a linear active disturbance rejection controller for ROV depth control based on the (NOA) nutcracker optimization algorithm, and compared and analyzed it with PID, LADRC, and cascaded LESO. It was concluded that it had high accuracy and anti-interference ability in terms of overshoot, settling time, fluctuation range, and steady-state error. Although the above current intelligent swarm optimization algorithms have certain advantages, they have disadvantages such as poor global exploration ability, slow convergence speed, and easy entry into local optimal values.

[0015] The Newton-Raphson-based optimizer (NRBO) is a new type of meta-heuristic algorithm (intelligent optimization algorithm), which has a significant effect on optimizing clustering algorithms.

[0016] Based on the NRBO algorithm, improving the control performance of the ROV active disturbance rejection controller, enhancing the anti-interference ability of the ROV active disturbance rejection controller, improving the control accuracy; simplifying the debugging process, reducing the manual debugging time, and reducing the dependence on experience; improving the stability of the system, enhancing the robustness, and avoiding parameter coupling problems are issues that need to be urgently solved by those skilled in the art. Summary of the Invention

[0017] In view of the above problems, the present invention provides a method for optimizing the parameters of an ROV active disturbance rejection controller based on improved NRBO to at least solve some of the technical problems mentioned in the above background technology.

[0018] To achieve the above object, the present invention adopts the following technical solutions:

[0019] The present invention provides a method for optimizing the parameters of an ROV active disturbance rejection controller based on improved NRBO, including the following steps:

[0020] Determine the parameter group of the ROV active disturbance rejection controller to be optimized and the optimization range of each parameter therein, and construct an objective function for optimizing the parameters of the ROV active disturbance rejection controller;

[0021] Initialize the NRBO parameters to generate an initial population; the dimension of the initial population corresponds to the number of parameters in the ROV active disturbance rejection controller parameter group.

[0022] Based on the NRSR search rule and NTAO trap avoidance, iteratively optimize the initial population to find the optimal ROV active disturbance rejection controller parameter group.

[0023] In each iteration round, calculate the objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; find the optimal individual in the current population based on the objective function value, and save the position of the optimal individual as the current optimal solution; among them, the position of the optimal individual is the corresponding ROV active disturbance rejection controller parameter group.

[0024] Stop the iteration after meeting the preset iteration convergence condition to obtain the optimal ROV active disturbance rejection controller parameter group.

[0025] Furthermore, it also includes:

[0026] Input the optimal ROV active disturbance rejection controller parameter group into the ROV active disturbance rejection controller for MATLAB / Simulink simulation to achieve the simulation control of the ROV.

[0027] Furthermore, the dimension of the initial population is 12-dimensional, corresponding to 12 parameters in the ROV active disturbance rejection controller parameter group: h0, r, δ, b0, α1, α2, α3, β 01 , β 02 , β 03 , β1, β2;

[0028] Among them, h0 represents the filtering factor; r represents the parameter determining the tracking speed; δ represents the interval length of the linear segment; b0 represents the disturbance compensation factor; β 01 、β 02 、β 03 are determined by the sampling step of the system; β1 represents the proportional coefficient; β2 represents the differential coefficient; α1, α2, α3 are all power exponents representing the nonlinear strength of the control function.

[0029] Furthermore, the objective function for optimizing the ROV active disturbance rejection controller parameters is expressed as:

[0030]

[0031] Among them, Q represents the fitness value of the individual position; t represents the rise time of the control system; T represents the sampling time; e i (t) represents the difference between the system output value and the desired input value at any moment within T; i = 1 represents the difference between the input signal and the controlled object position output; i = 2 represents the difference between the input signal differential and the controlled object position differential output.

[0032] Furthermore, the initial population is represented as:

[0033]

[0034] where n represents the nth generation population, and n = 1, 2, 3... N p ; j represents the jth dimension, and j = 1, 2, 3... dim; represents the individual position of the jth dimension of the nth generation population; rand represents a random number between (0, 1); lb and ub represent the lower and upper bounds of the individual position respectively; X n represents the population matrix of all dimensions.

[0035] Furthermore, the iteration of each individual position in the population is represented as:

[0036]

[0037] where represents the individual position of the nth generation population in the ITth iteration; represents the individual position of the nth generation population in the (IT + 1)th iteration, that is, the updated new individual position; NRSR represents the NRSR search rule; ρ n represents the direction guiding parameter, which is used to guide the individual population to the correct direction;

[0038] During the iteration process, the new trap avoidance operator NTAO is used to avoid local optima, which is represented as:

[0039]

[0040] where x NTAO1 represents the search space for expanding the individual position; x NTAO2 represents the search space for the worst solution to avoid the trap corresponding to the average position M of the current individual population; x NTAO3 represents the search space for adding a random individual position to the new trap avoidance formula to accelerate convergence; x NTAO represents the new individual position after avoiding local optimal solutions; C1 represents the probability of selecting an individual position in the search space x NTAO1 ; C2 represents the probability of selecting an individual position in the search space x NTAO2 ; C3 represents the probability of selecting an individual position in the search space x NTAO3 ; C1, C2, and C3 all follow a normal distribution; γ1, γ2, γ3, and γ4 all represent random numbers that follow the standard normal distribution, and γ1 ≠ γ2, γ3 ≠ γ4; γ5 and γ6 both represent random numbers in the range [0, 1]; It represents the individual position of the nth generation population in the (IT + 1)-th iteration; p and q are two arbitrary integers between 1 and N p and p≠q≠n, where p represents the pth generation population and q represents the qth generation population; ξ represents the adaptive coefficient; x b represents the optimal position of the individual x within the neighborhood [x - Δx, x]; x w represents the worst position of the individual x within the neighborhood [x, x + Δx]; Δx represents a finite increment that is infinitely close to x; It represents the individual position of the nth generation population in the (IT + 2)-th iteration.

[0041] Furthermore, the objective function value is calculated according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; specifically:

[0042] The ROV active disturbance rejection controller parameter group corresponding to each individual in the current population is input into the ROV active disturbance rejection controller to obtain the difference between the input signal and the output of the controlled object position, and the difference between the differential of the input signal and the output of the differential of the controlled object position;

[0043] The two obtained differences are input into the objective function to obtain the corresponding objective function value.

[0044] Furthermore, the preset iteration convergence condition includes: the change in the optimal fitness is less than the threshold, or the number of iterations reaches the preset number.

[0045] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an ROV active disturbance rejection controller parameter optimization method based on improved NRBO, which has the following beneficial effects:

[0046] The present invention solves the problem that the active disturbance rejection controller has too many parameters and is not easy to adjust, and avoids difficult methods such as empirical adjustment and dynamic adjustment. It can be more conveniently and quickly used for adjusting the pose of the active disturbance rejection controller of ROV and AUV.

[0047] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0049] Figure 1 It is a schematic structural diagram of an ROV attitude active disturbance rejection control system provided by an embodiment of the present invention.

[0050] Figure 2 Schematic diagram of the parameter optimization method for the ROV active disturbance rejection controller based on the improved NRBO provided by the embodiment of the present invention.

[0051] Figure 3 Schematic diagram of the simulation of the active disturbance rejection control system for depth (Z channel) provided by the embodiment of the present invention.

[0052] Figure 4 Schematic diagram of the comparison effect provided by the embodiment of the present invention. Detailed implementation manners

[0053] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0054] The embodiment of the present invention discloses a parameter optimization method for the ROV active disturbance rejection controller based on the improved NRBO. Refer to Figure 1 as shown, and it includes the following steps:

[0055] S1. Determine the parameter group of the ROV active disturbance rejection controller to be optimized and the optimization range of each parameter therein, and construct the objective function for the parameter optimization of the ROV active disturbance rejection controller;

[0056] S2. Initialize the NRBO parameters and generate the initial population; the dimension of the initial population corresponds to the number of parameters in the parameter group of the ROV active disturbance rejection controller;

[0057] S3. Based on the NRSR search rule and NTAO trap avoidance, perform iterative optimization on the initial population to find the parameter group of the ROV active disturbance rejection controller;

[0058] S4. In each iteration round, calculate the objective function value according to the parameter group of the ROV active disturbance rejection controller corresponding to each individual in the current population; find the optimal individual in the current population based on the objective function value, and save the position of the optimal individual as the current optimal solution; wherein, the position of the optimal individual is the corresponding parameter group of the ROV active disturbance rejection controller;

[0059] S5. Stop the iteration after meeting the preset iteration convergence condition to obtain the optimal parameter group of the ROV active disturbance rejection controller;

[0060] S6. Input the optimal parameter group of the ROV active disturbance rejection controller into the ROV active disturbance rejection controller for MATLAB / Simulink simulation to realize the simulation control of the ROV.

[0061] The above-mentioned marks S1 - S6 are only for convenience of explanation and do not limit the specific execution order of the steps. Next, each of the above steps will be specifically described.

[0062] In the above step S1, determine the ROV active disturbance rejection controller parameter group to be optimized and the optimization range of each parameter therein, and construct the objective function for optimizing the ROV active disturbance rejection controller parameters; specifically:

[0063] In the embodiment of the present invention, taking the ROV depth control (Z channel) as an example, the ROV active disturbance rejection controller parameter group to be optimized includes h0, r, δ, b0, α1, α2, α3, β 01 , β 02 , β 03 , β1, β2, these 12 parameters;

[0064] Among them, h0 represents the filtering factor; r represents the parameter determining the tracking speed; δ represents the interval length of the linear segment; b0 represents the disturbance compensation factor; β 01 , β 02 , β 03 is determined by the sampling step of the system; β1 represents the proportional coefficient; β2 represents the differential coefficient; α1, α2, α3 are all power exponents, representing the nonlinear strength of the control function;

[0065] The optimization range of each of the above parameters, that is, the upper bound and the lower bound of each parameter, is specifically shown in Table 1:

[0066] Table 1: Optimization Range of ROV Active Disturbance Rejection Controller Parameters

[0067] Parameter Value range Parameter Value range <![CDATA[h0]]> [0.0006,0.1] r [0.1,5] δ [0.1,1] <![CDATA[b0]]> [0.5,5] <![CDATA[α1]]> [0.1,10] <![CDATA[α2]]> [0.1,10] <![CDATA[α3]]> [0.1,10] <![CDATA[β 01 > [50,200] <![CDATA[β 02 > [1000,3000] <![CDATA[β 03 > [2000,4000] <![CDATA[β1]]> [0,500] <![CDATA[β2]]> [0,1000]

[0068] After that, based on the ROV active disturbance rejection controller parameter group to be optimized, construct the objective function for optimizing the ROV active disturbance rejection controller parameters, expressed as:

[0069]

[0070] Among them, Q represents the fitness value of the individual position; t represents the rise time of the control system; T represents the sampling time; e i (t) represents the difference between the system output value and the desired input value at any time within T; i = 1 represents the difference between the input signal and the output of the controlled object position; i = 2 represents the difference between the differential of the input signal and the differential output of the controlled object position.

[0071] In the above step S2, initialize the NRBO parameters and generate the initial population; the dimension of the initial population corresponds to the number of parameters in the ROV active disturbance rejection controller parameter group;

[0072] The initial population is expressed as:

[0073]

[0074] where n represents the nth generation population, and n = 1, 2, 3…N p ; j represents the jth dimension, and j = 1, 2, 3…dim; represents the individual position of the jth dimension of the nth generation population; rand represents a random number between (0, 1); lb and ub represent the lower and upper bounds of the individual position respectively; X n represents the population matrix of all dimensions.

[0075] For example, in the embodiments of the present invention, let the population size be 20 and the population dimension be 12 (h0, r, δ, b0, α1, α2, α3, β 01 , β 02 , β 03 , β1, β2), the maximum number of iterations is 50, the adaptive coefficient ξ is dynamically adjusted, Δx is based on the random perturbation of the current optimal solution and the individual position, and the initial population is randomly generated. The mathematical expression is:

[0076]

[0077] where n = 1, 2, 3…20 represents the nth population, and j = 1, 2, 3…12 represents the jth dimension.

[0078] In the above step S3, based on the NRSR search rule and NTAO trap avoidance, the initial population is iteratively optimized to find the optimal ROV active disturbance rejection controller parameter group;

[0079] The NRSR search rule is the core key strategy of NRBO and is derived based on Taylor expansion. The second-order Taylor expansion of the optimization objective function f(x) has the mathematical expression:

[0080]

[0081] And the first-order and second-order derivatives are approximated by the difference method, with the mathematical expression:

[0082]

[0083] where f′(x), f″(x) represent the first-order and second-order derivatives of the objective function f(x); x represents the population individual; Δx represents a finite increment that is infinitely close to x.

[0084] The Newton-Raphson iteration formula is introduced, with the mathematical expression:

[0085]

[0086] Among them, x n+1 represents the updated individual position in the (n + 1)-th generation population.

[0087] Combined with the above formula, the mathematical expression for position update of NRSR is obtained:

[0088]

[0089] Δx = rand(1, dim) × |x b - x n |

[0090] Among them, x b represents the optimal position of individual x within the neighborhood [x - Δx, x] (i.e., the upper bound); x w represents the worst position of individual x within the neighborhood [x, x + Δx] (i.e., the lower bound); Δx represents a finite increment that is infinitely close to x; rand(1, dim) represents a random number with dimension decision variables.

[0091] Then, the individual population is guided to the correct direction through the ρ parameter, expressed as:

[0092]

[0093] Among them, a and b represent random numbers between (0, 1), r1 and r2 represent different integers randomly selected from the population and r1 ≠ r2, represents the positions of other individuals in the population.

[0094] Finally, the mathematical expression for individual position iteration is:

[0095]

[0096] Among them, represents the individual position of the n-th generation population in the IT-th iteration; represents the individual position of the n-th generation population in the (IT + 1)-th iteration, that is, the updated new individual position; ρ n represents the direction guiding parameter, which is used to guide the individual population to the correct direction;

[0097] Since the performance parameter (DF) in the original trap avoidance operator (TAO) of the NRBO optimization algorithm is a fixed constant value, lacking diversity and having weak adaptability to dynamic environments, and introducing too many parameters leads to high computational complexity. Now, inspired by the CLJAYA algorithm and the GWO algorithm, a new trap avoidance operator (NTAO) is designed to avoid local optima, and the mathematical expression for position update:

[0098]

[0099] Among them, x NTAO1 represents the search space for expanding the individual position; x NTAO2 represents the search space where the worst solution is guided by the average position M of the current individual population to avoid traps; x NTAO3 represents the search space for adding the random individual position into the new formula for avoiding traps to accelerate convergence; x NTAO represents the new individual position after avoiding local optimal solutions; C1 represents the probability of selecting an individual position in the search space x NTAO1 ; C2 represents the probability of selecting an individual position in the search space x NTAO2 ; C3 represents the probability of selecting an individual position in the search space x NTAO3 ; C1, C2, and C3 all follow a normal distribution; γ1, γ2, γ3, and γ4 all represent random numbers following the standard normal distribution, and γ1≠γ2, γ3≠γ4, which are used to increase the population richness and avoid falling into local errors; γ5 and γ6 both represent random numbers in the range of [0,1], which are used to increase the population richness and avoid falling into local errors; represents the individual position of the nth generation population in the (IT + 1)th iteration; p and q are two arbitrary integers between 1 and N p and p≠q≠n, where p represents the pth generation population and q represents the qth generation population; ξ represents the adaptive coefficient; x b represents the optimal position of individual x in the neighborhood of [x - Δx, x]; x w represents the worst position of individual x in the neighborhood of [x, x + Δx]; Δx represents a finite increment infinitely close to x; represents the individual position of the nth generation population in the (IT + 2)th iteration.

[0100] Among them, M represents the average position of the current individual population, and the mathematical expression is:

[0101]

[0102] The adaptive coefficient ξ is used to enhance the algorithm, and the mathematical expression is:

[0103]

[0104] Among them, IT represents the current iteration; Max IT represents the maximum number of iterations.

[0105] Specifically, taking the filtering factor h0 as an example, find the position of the filtering factor h0 in the neighborhood of [h0 - Δx, h0] to make it reach the best position in the neighborhood, denoted as h 0b ; find the position of the filtering factor h0 in the neighborhood of [h0, h0 + Δx] to make it reach the worst position in the neighborhood, denoted as h 0w; Utilize the optimal solution h 0b and the worst solution h 0w to adjust the current position, combine with Δx for gradient direction correction, and then introduce random coefficients a, b and random individual positions to enhance the global search ability. The iterative mathematical expression of parameter h0 is:

[0106]

[0107] respectively represent the individual positions of the nth generation population in the ITth iteration and the IT+1th iteration; ρ n represents the direction guiding parameter, which is used to guide the individual population to the correct direction; NRSR represents the Newton-Raphson search rule;

[0108] NTAO trap avoidance operator. Due to the randomness of parameters γ1, γ2, γ3, γ4, γ5, γ6, the individual population becomes more diverse and escapes from the local optimal solution. Then, the search step size is controlled by the adaptive coefficient ξ, which is adaptively adjusted with the number of iterations. It is responsible for global exploration in the initial stage and focuses on local search in the later stage, which helps to avoid local optimality. The mathematical expression of the NTAO trap avoidance operator is:

[0109]

[0110] where h 0NTAO1 represents the search space for expanding h0; h 0NTAO2 represents the worst solution h 0b has the search space for avoiding traps guided by the average position M of the current individual population; h 0NTAO3 represents the search space for adding the random h0 to the new trap avoidance formula to accelerate convergence; h 0NTAO represents the new individual position after avoiding local optimal solutions.

[0111] In the above step S4, in each iteration round, calculate the objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; find the optimal individual in the current population based on the objective function value, and save the position of the optimal individual as the current optimal solution; among them, the position of the optimal individual is the corresponding ROV active disturbance rejection controller parameter group;

[0112] Among them, calculate the objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; specifically: input the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population into the ROV active disturbance rejection controller to obtain: the difference between the input signal and the output of the controlled object position, and the difference between the differential of the input signal and the output of the differential of the controlled object position; input the two obtained differences into the objective function to obtain the corresponding objective function value.

[0113] In the embodiment of the present invention, the structure diagram of the ROV active disturbance rejection control system can be seen Figure 2 as shown. In the ROV active disturbance rejection control system, the control links include attitude control and control quantity change, which are respectively used to control the position of the ROV: forward and backward, translation, depth (x, y, z) and attitude: roll angle, pitch angle, yaw angle

[0114] Taking the ROV depth control (Z channel) as an example in the embodiment of the present invention, the Z-channel active disturbance rejection controller designed includes a tracking differentiator (TD), a non-linear error feedback (NLSEF), an extended state observer (ESO) and disturbance compensation. When the input desired position signal is z d , the tracking differentiator (TD) arranges a transition process for the desired input signal. The input end inputs z d . One output end is the position tracking signal, and the other output end is the position tracking differential signal; the non-linear error feedback (NLSEF) adopts a non-linear combined control law in the form of PD. One input end is the error signal between the position tracking signal and the internal state variable of the Z-channel system, and the other input end is the error signal between the position tracking differential signal and the differential of the internal state variable of the Z-channel system, and the output generates the preliminary control quantity of the Z channel; the extended state observer (ESO) estimates the state variables and total disturbances of the system. One input end is the position signal of the difference between the preliminary control quantity and the total disturbance, and the other input end is the feedback signal of the ROV pose control quantity. The first and second output ends are the ESO real-time observation system state variables, and the third output end is the ESO real-time observation system total disturbance quantity; the difference is made between the preliminary control quantity of the Z channel and the real-time observation total disturbance observation value of the ESO, and the output end goes to the control cabin of the ROV. The thrust distribution output end generates a control quantity for the thruster, and the thruster output end generates thrust to control the depth of the ROV for underwater operations.

[0115] Specifically:

[0116] (1) Tracking differentiator (TD):

[0117] The tracking differentiator (TD) arranges a transition process for the desired input signal to avoid overshoot and oscillation phenomena caused by the desired input signal. The mathematical expression is:

[0118]

[0119] Among them, v0, v1, and v2 respectively represent the input signal, the tracking signal of the input signal, and the tracking differential signal of the input signal; T represents the sampling time; k represents the kth moment; fst(·) represents the fastest control synthesis function; r represents the parameter determining the tracking speed; h0 represents the filtering factor, which affects the smoothness of the differential signal; f1 is an intermediate parameter for easy explanation;

[0120] (2) Nonlinear Error Feedback (NLSEF):

[0121] The Nonlinear Error Feedback (NLSEF) adopts a nonlinear combined control law in the PD form, and its mathematical expression is:

[0122]

[0123] where, 0 < α1 < 1 < α2; k p = β1 represents the proportional coefficient; k d = β2 represents the differential coefficient; δ represents the interval length of the linear segment; e1 represents the difference between the input signal and the position output of the controlled object; e2 represents the difference between the differential of the input signal and the differential output of the position of the controlled object; u0 represents the output of the nonlinear error feedback control law; fal(·) represents the saturation function; α1, α2 represent the power exponents, representing the nonlinear strength of the control function; where, e includes e1 and e2; α includes α1 and α2; f2 is an intermediate parameter for easy explanation;

[0124] (3) Extended State Observer (ESO):

[0125] The Extended State Observer (ESO) estimates the state variables and total disturbances of the system in real time, and its mathematical expression:

[0126]

[0127] where, z1 and z2 respectively represent the state observations of v1 and v2; z3 represents the extended state observation; β 01 , β 02 , β 03 respectively represent the feedback parameters of the state observations z1, z2, and z3, which are determined by the step size of the system; b0 represents the disturbance compensation factor; u represents the disturbance compensation control quantity; χ is the error value, representing the difference between the tracking signal observation and the actual tracking signal quantity of the input signal; α3 is the power exponent, representing the nonlinear strength of the control function.

[0128] (4) Disturbance compensation with ESO:

[0129] The disturbance compensation directly compensates the total disturbance estimated by the ESO into the control quantity. At this time, the main control quantity of the system needs to handle the uncompensated residual disturbance, and its mathematical expression is:

[0130]

[0131] where, u0 represents the output of the nonlinear error feedback control law, and u represents the disturbance compensation control quantity.

[0132] In the above step S5, after meeting the preset iteration convergence condition, stop the iteration to obtain the optimal ROV active disturbance rejection controller parameter group;

[0133] The preset iteration convergence condition includes: the change in the optimal fitness in consecutive several iterations is less than a threshold value or the number of iterations reaches a preset number (for example, 50 times);

[0134] The results of 50 iterations are shown in Table 2:

[0135] Table 2

[0136] Parameter Optimal value Parameter Value range <![CDATA[h0]]> 0.0049 r 4.9998 δ 0.8262 <![CDATA[b0]]> 1.0017 <![CDATA[α1]]> 0.9126 <![CDATA[α2]]> 2.6532 <![CDATA[α3]]> 0.1000 <![CDATA[β 01 > 59.6794 <![CDATA[β 02 > 2999.4758 <![CDATA[β 03 > 3976.8700 <![CDATA[β1]]> 490.2402 <![CDATA[β2]]> 887.8246

[0137] In the above step S6, the optimal ROV active disturbance rejection controller parameter group is input into the ROV active disturbance rejection controller for MATLAB / Simulink simulation to achieve the simulation control of the ROV;

[0138] In the embodiment of the present invention, the simulation time is 10 s, the simulation step size is 0.001, the solver adopts discrete (no continuous state), the desired step signal is set with a step time of 2 s and a termination time of 4 s. The schematic diagram of the simulation experiment results is as Figure 3 shown. The simulation of the objective function based on the improved NRBO for the depth (Z channel) designed by the present invention and the active disturbance rejection controller system of the ROV.

[0139] The simulation experiment is compared with three other active disturbance rejection control methods for Z channel control. The comparison results of the control performance are shown in Table 3, and the schematic diagram of the experimental comparison is as Figure 4 shown.

[0140] Table 3

[0141] Method Overshoot / (%) Rise time / (s) Steady-state error Traditional ADRC 0.400 5.984 0.0011 PSO-ADRC 2.395 4.045 0.0092 NRBO-ADRC 0.600 3.816 0.0051 Improved NRBO-ADRC 0.025 3.774 0.0001

[0142] From Table 3 and Figure 4 it can be seen that the simulation results show that there is almost no overshoot in the improved NRBO-ADRC of the present invention, the rise time is significantly faster than that of the traditional ADRC and PSO-ADRC, slightly faster than that of the NRBO-ADRC, and the steady-state error is also smaller than that of the traditional ADRC, PSO-ADRC and NRBO-ADRC. In general, the control performance of the active disturbance rejection controller based on the improved NRBO of the present invention is superior to that of the traditional ADRC, PSO-ADRC and NRBO-ADRC.

[0143] In summary, the method for optimizing the parameters of the ROV active disturbance rejection controller based on the improved NRBO provided by the embodiment of the present invention improves the control performance, enhances the anti-disturbance ability, and improves the control accuracy; simplifies the debugging process, reduces the manual debugging time, and reduces the dependence on experience; improves the stability of the system, enhances the robustness, and avoids the problem of parameter coupling. It solves the problem of more accurate depth control during the deep-sea operation of the ROV, and the dynamic indicators exceed the optimization of the controller parameters by other intelligent swarm optimization algorithms.

[0144] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other.

[0145] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An optimization method for the parameters of an ROV auto-disturbance rejection controller based on improved NRBO, characterized in that, It includes the following steps: Determine the ROV active disturbance rejection controller parameter group to be optimized and the optimization range of each parameter therein, and construct the objective function for optimizing the ROV active disturbance rejection controller parameters; Initialize the NRBO parameters to generate an initial population; the dimension of the initial population corresponds to the number of parameters in the ROV active disturbance rejection controller parameter group; Based on the NRSR search rule and NTAO trap avoidance, iteratively optimize the initial population to find the ROV active disturbance rejection controller parameter group; In each iteration round, calculate the objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; Find the optimal individual in the current population according to the objective function value, and save the position of the optimal individual as the current optimal solution; among them, the position of the optimal individual is the corresponding ROV active disturbance rejection controller parameter group; Stop the iteration after meeting the preset iteration convergence condition to obtain the optimal ROV active disturbance rejection controller parameter group.

2. The parameter optimization method of the ROV auto-disturbance rejection controller based on the improved NRBO according to claim 1, characterized in that, It also includes: Input the optimal ROV active disturbance rejection controller parameter group into the ROV active disturbance rejection controller for MATLAB / Simulink simulation to realize the simulation control of the ROV.

3. An optimization method for the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 1, characterized in that The dimension of the initial population is 12-dimensional, corresponding to the 12 parameters in the ROV active disturbance rejection controller parameter group: h0, r, δ, b0, α1, α2, α3, β 01 , β 02 , β 03 , β1, β2; Among them, h0 represents the filtering factor; r represents the parameter determining the tracking speed; δ represents the interval length of the linear segment; b0 represents the disturbance compensation factor; β 01 , β 02 , β 03 is determined by the sampling step of the system; β1 represents the proportional coefficient; β2 represents the differential coefficient; α1, α2, and α3 are all power exponents representing the nonlinear strength of the control function.

4. A method for optimizing the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 1, characterized in that The objective function for optimizing the ROV active disturbance rejection controller parameters is expressed as: Among them, Q represents the fitness value of the individual position; t represents the rise time of the control system; T represents the sampling time; e i (t) represents the difference between the system output value and the desired input value at any moment within T; i = 1 represents the difference between the input signal and the output of the controlled object position; i = 2 represents the difference between the differential of the input signal and the differential output of the controlled object position.

5. A method for optimizing the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 1, characterized in that The initial population is expressed as: Among them, n represents the nth generation of population, and n = 1, 2, 3... N p ; j represents the jth dimension, and j = 1, 2, 3... dim; represents the individual position of the jth dimension of the nth generation of population; rand represents a random number between (0, 1); lb and ub respectively represent the lower bound and upper bound of the individual position; X n represents the population matrix of all dimensions.

6. An optimization method for the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 5, characterized in that, The iteration of the position of each individual in the population is expressed as: Among them, represents the individual position of the n-th generation population in the IT-th iteration; represents the individual position of the n-th generation population in the (IT + 1)-th iteration, that is, the updated new individual position; NRSR represents the NRSR search rule; ρ n represents the direction guiding parameter, which is used to guide the individual population to the correct direction; During the iteration process, the new trap avoidance operator NTAO is used to avoid local optima, which is expressed as: Among them, x NTAO1 represents the search space for expanding the individual position; x NTAO2 represents the search space where the worst solution is guided by the average position M of the current individual population to avoid traps; x NTAO3 represents the search space for adding a random individual position to the new formula for avoiding traps to accelerate convergence; x NTAO represents the new individual position after avoiding local optima; C1 represents the probability of selecting an individual position in the search space x NTAO1 ; C2 represents the probability of selecting an individual position in the search space x NTAO2 ; C3 represents the probability of selecting an individual position in the search space x NTAO3 ; C1, C2, and C3 all follow a normal distribution; γ1, γ2, γ3, and γ4 all represent random numbers that follow a standard normal distribution, and γ1≠γ2, γ3≠γ4; γ5 and γ6 both represent random numbers in the range [0, 1]; represents the individual position of the nth generation population in the (IT + 1)th iteration; p and q are two arbitrary integers between 1 and N p and p≠q≠n, where p represents the pth generation population and q represents the qth generation population; ξ represents the adaptive coefficient; x b represents the optimal position of the individual x in the neighborhood [x - Δx, x]; x w represents the worst position of the individual x in the neighborhood [x, x + Δx]; Δx represents a finite increment that is infinitely close to x; represents the individual position of the nth generation population in the (IT + 2)th iteration.

7. An optimization method for the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 1, characterized in that Calculate the objective function value according to the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population; Specifically: Input the ROV active disturbance rejection controller parameter group corresponding to each individual in the current population into the ROV active disturbance rejection controller to obtain: the difference between the input signal and the position output of the controlled object, and the difference between the differential of the input signal and the differential of the position output of the controlled object; Input the two obtained differences into the objective function to obtain the corresponding objective function value.

8. A method for optimizing the parameters of an ROV auto-disturbance rejection controller based on improved NRBO according to claim 1, characterized in that The preset iteration convergence condition includes: the change in the optimal fitness is less than the threshold, or the number of iterations reaches the preset number.

Images