Ship control parameter tuning method based on hybrid multi-policy harvester optimization algorithm

Abstract

The application belongs to the technical field of robot motion control, and discloses a vehicle control parameter setting method based on a hybrid multi-strategy melolontha optimization algorithm. The method designs a hybrid multi-strategy melolontha optimization algorithm based on a traditional melolontha optimization algorithm, and proposes an optimization algorithm for population initialization and population position updating. A good point set theory is introduced to uniformly generate the population position in the initialization stage, so as to improve the diversity of the initialized population. After the position updating, the global search ability and the convergence speed of the algorithm are improved by adding the gravity reverse learning search and the Gaussian mutation in each dimension. The method is suitable for parameter setting optimization of an ADRC controller. The method not only can improve the diversity of the position updating mode of an amphibious vehicle, but also can improve the global search ability and the convergence speed of the algorithm while avoiding the algorithm from falling into a local optimal solution, so as to further improve the position and attitude control performance of the amphibious vehicle.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of robot motion control, and particularly relates to a vehicle control parameter setting method based on a hybrid multi-strategy melolontha optimization algorithm. BACKGROUND

[0002] With the increasing maturity of robot control technology, traditional robots that can only operate in a single medium have been difficult to meet the needs of today's society. Researchers have shown increasing interest in developing a water-air amphibious vehicle that can freely shuttle between underwater and air media. The water-air amphibious vehicle has the advantages of high-speed maneuvering, rapid deployment capability of unmanned aerial vehicles and rapid swimming, high concealment of underwater vehicles, and has great application potential in various working scenarios, and can be applied to activities such as monitoring ocean water quality, hydrological and meteorological measurement, and sea rescue.

[0003] Since the water-air amphibious vehicle needs to move in two different media, underwater and air, the traditional PID control method cannot meet the control accuracy of the water-air amphibious vehicle. Therefore, the stable motion control of the water-air amphibious vehicle in different media has always been the research focus of relevant researchers. In 2008, the Active Disturbance Rejection Control (ADRC) algorithm proposed by researcher Han Jingqing is to regard the control object as an integral series type, and use the Extended State Observer (ESO) to estimate and compensate the part different from the standard type in the dynamic object in real time, to suppress the influence of disturbance on the system output, so as to achieve better control effect than PID control under complex disturbance.

[0004] Although the ADRC controller is used for motion control of the water-air amphibious vehicle, it has good control effect, but due to the large number of parameters of the ADRC controller, the traditional parameter setting method with artificial experience is time-consuming and difficult to achieve optimal control effect.

[0005] Therefore, it is necessary to propose a water-air amphibious vehicle control parameter setting method that can solve the problem of long time consumption and inability to achieve optimal control effect of traditional water-air amphibious vehicle control parameter setting. SUMMARY

[0006] The purpose of the present application is to propose a vehicle control parameter setting method based on a hybrid multi-strategy melolontha optimization algorithm. The method is aimed at the problem that the position and attitude control parameters of the water-air amphibious vehicle are too many and show nonlinear characteristics, leading to difficult parameter adjustment. The control parameter setting is realized based on the hybrid multi-strategy melolontha optimization algorithm, thereby improving the control performance of the water-air amphibious vehicle.

[0007] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0008] The aircraft control parameter setting method based on the hybrid multi-strategy melolontha optimization algorithm comprises the following steps:

[0009] Step 1. Kinematics modeling of the amphibious aircraft is performed to obtain a kinematics model of the amphibious aircraft;

[0010] Step 2. Dynamics modeling of the amphibious aircraft is performed on the basis of the kinematics model of the amphibious aircraft to obtain a dynamics model of the amphibious aircraft containing unknown disturbances;

[0011] Step 3. An amphibious aircraft position and attitude control system is constructed on the basis of the kinematics model of the amphibious aircraft and the dynamics model of the amphibious aircraft containing unknown disturbances;

[0012] Step 4. Parameter setting of the amphibious aircraft position and attitude control system is performed based on the hybrid multi-strategy melolontha optimization algorithm, which is obtained by improving the melolontha optimization algorithm as follows: introducing the good point set theory for population initialization; introducing the spiral exploration strategy in the position update of the melolontha reproduction and stealing behavior; introducing the barycenter reverse learning strategy to search the global and Gaussian mutation of each dimension after the position update of the melolontha ball rolling, dancing, foraging, reproduction and stealing behavior.

[0013] In addition, on the basis of the above-mentioned aircraft control parameter setting method based on the hybrid multi-strategy melolontha optimization algorithm, the present application further proposes a computer device comprising a memory and one or more processors;

[0014] The memory stores executable codes, and the processor executes the executable codes to implement the steps of the above-mentioned aircraft control parameter setting method based on the hybrid multi-strategy melolontha optimization algorithm.

[0015] The present application has the following advantages:

[0016] As described above, the application refers to a kind of vehicle control parameter setting method based on hybrid multi-policy meloidogyne optimization algorithm, which is based on the design of hybrid multi-policy meloidogyne optimization algorithm based on traditional meloidogyne optimization algorithm, proposes the optimization algorithm of population initialization and population position update, the population position of initialization stage is generated uniformly by introducing the good point set theory to improve the diversity of initialization population, spiral exploration strategy is introduced in the position update of stealing behavior and breeding behavior, the diversity of position update mode is increased, and after position update, the global search ability and convergence speed of algorithm are improved by adding the center of gravity reverse learning search and Gaussian mutation of each dimension.The vehicle control parameter setting method based on hybrid multi-policy meloidogyne optimization algorithm of the application is suitable for parameter setting optimization of ADRC controller, can not only improve the diversity of position update mode of amphibious vehicle, but also improve the global search ability and convergence speed of algorithm while avoiding the algorithm falling into local optimal solution, thereby further improving the position and attitude control performance of amphibious vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flow chart of the vehicle control parameter setting method based on hybrid multi-policy meloidogyne optimization algorithm in the embodiment of the application.

[0018] Figure 2 The coordinate system diagram of amphibious vehicle in the embodiment of the application.

[0019] Figure 3 The structure diagram of position and attitude control system of amphibious vehicle in the embodiment of the application.

[0020] Figure 4 The comparison chart of control results of amphibious vehicle in surge direction by the method of the application and other optimization algorithms.

[0021] Figure 5 The comparison chart of control results of amphibious vehicle in sway direction by the method of the application and other optimization algorithms.

[0022] Figure 6 The comparison chart of control results of amphibious vehicle in heave direction by the method of the application and other optimization algorithms.

[0023] Figure 7 The comparison chart of control results of amphibious vehicle in roll direction by the method of the application and other optimization algorithms.

[0024] Figure 8 The comparison chart of control results of amphibious vehicle in pitch direction by the method of the application and other optimization algorithms.

[0025] Figure 9The figure for comparing the control results of the method of the application and other optimization algorithms on the water-air amphibious vehicle in the yaw direction is shown. DETAILED DESCRIPTION

[0026] The application will be further described in detail below in combination with the drawings and specific embodiments:

[0027] Embodiment 1

[0028] In the field of water-air amphibious vehicle motion control technology, the position and attitude control of the water-air amphibious vehicle is an important content of the water-air amphibious vehicle motion control. The core of the six-degree-of-freedom position and attitude control of the water-air amphibious vehicle is that each dimension control needs to make the water-air amphibious vehicle more stable and fast to reach the ideal position or attitude. Therefore, aiming at the problem that the position and attitude control parameters of the water-air amphibious vehicle are more and show nonlinear characteristics, leading to difficult parameter adjustment, how to improve the control performance of the water-air amphibious vehicle through attitude control parameter setting is a technical problem to be solved by the application.

[0029] The general process of the method of the application includes the following parts: first, after the kinematics and dynamics modeling of the water-air amphibious vehicle, the water-air amphibious vehicle motion control system model is established, that is, the water-air amphibious vehicle position and attitude control system is constructed; then, aiming at the water-air amphibious vehicle in The control of the corresponding ADRC controller of the six degrees of freedom is performed on the water-air amphibious vehicle control parameter setting based on the mixed multi-strategy Scarab optimization algorithm for the parameters (β1, β2, β3) in each controller, wherein the mixed multi-strategy Scarab optimization algorithm adopted by the application introduces the strategy of spiral exploration in the position update of Scarab reproduction and theft, and the barycenter reverse learning is introduced after the position update of the five behaviors, and the Gaussian variation is introduced, wherein the five behaviors include Scarab rolling dung ball, dancing, foraging, reproduction and theft; the output parameters are used for adjusting the parameters (β1, β2, β3) in each ADRC controller. Finally, according to the output data of each ADRC controller after adjustment, the water-air amphibious vehicle is adjusted in the corresponding six degrees of freedom, and then the position and attitude are adjusted.

[0030] As shown in Figure 1 The water-air amphibious vehicle control parameter setting method based on the mixed multi-strategy Scarab optimization algorithm specifically includes the following steps:

[0031] Step 1. Kinematics modeling is performed on the water-air amphibious vehicle to obtain the kinematics model of the water-air amphibious vehicle.

[0032] The motion of the water-air amphibious vehicle belongs to typical six-degree-of-freedom motion. In order to analyze the six-degree-of-freedom motion of the water-air amphibious vehicle, a coordinate system needs to be defined first. The commonly used coordinate system in the motion of the water-air amphibious vehicle is a fixed coordinate system and mobile coordinate system As shown in Fig. 1, where Figure 2 is the origin of the fixed coordinate system, , , , is the coordinate axis of the fixed coordinate system; is the origin of the mobile coordinate system, , , is the coordinate axis of the mobile coordinate system.

[0033] For the convenience of analysis, it is generally assumed that the origin O of the mobile coordinate system coincides with the center of gravity of the water-air amphibious vehicle, the x-axis points to the bow, the y-axis is perpendicular to the x-axis and points to the starboard, and the z-axis is perpendicular to O-xy and points downward. The symbols for the motion and force of the water-air amphibious vehicle are shown in Table 1, where the heave corresponds to the heave direction, the roll corresponds to the pitch direction, the pitch corresponds to the pitch direction, and the yaw corresponds to the yaw direction.

[0034] Table 1 Symbols for the motion and force of the water-air amphibious vehicle

[0035]

[0036] In step 1 of this embodiment, the kinematic modeling of the water-air amphibious vehicle is performed according to the six-degree-of-freedom position and attitude, six-degree-of-freedom velocity and angular velocity vectors.

[0037] Definition of the six-degree-of-freedom position and attitude vector of the water-air amphibious vehicle is:

[0038] (1)

[0039] wherein represents the position vector of the water-air amphibious vehicle, , represents the position of the water-air amphibious vehicle in the surge direction, represents the position of the water-air amphibious vehicle in the sway direction, represents the position of the water-air amphibious vehicle in the heave direction, i.e. the heave direction; represents the attitude vector of the water-air amphibious vehicle, , represents the roll angle of the water-air amphibious vehicle, represents the pitch angle of the water-air amphibious vehicle, represents the yaw angle of the water-air amphibious vehicle.

[0040] Six-degree-of-freedom velocity and angular velocity vector of the water-air amphibious vehicle is:

[0041] (2)

[0042] where, is the velocity vector of the SWATH, , is the surge velocity of the SWATH, is the sway velocity of the SWATH, is the heave velocity of the SWATH; is the angular velocity vector of the SWATH, , is the angular velocity of the SWATH in the yaw direction, is the angular velocity of the SWATH in the pitch direction, is the angular velocity of the SWATH in the roll direction.

[0043] The velocity transformation equation of the SWATH in the three-axis linear motion is described by using the Euler angle description method as follows:

[0044] (3)

[0045] where, is the velocity coordinate transformation matrix.

[0046] The angular velocity transformation equation of the SWATH in the three-axis rotation is described as follows:

[0047] (4)

[0048] where, is the angular velocity coordinate transformation matrix.

[0049] By using the concept of rotation, Euler's theorem and the properties of skew-symmetric matrix, the transformation matrix is obtained as follows:

[0050] (5)

[0051] (6)

[0052] where, is the sine function, ; is the cosine function, ; is the tangent function, .

[0053] The total coordinate transformation matrix is defined as:

[0054] (7)

[0055] The kinematics model of the water-air amphibious vehicle is obtained from formula (3) and formula (4) as follows:

[0056] (8)

[0057] Step 2. Based on the kinematics model of the water-air amphibious vehicle, the dynamics modeling of the water-air amphibious vehicle is carried out to obtain the water-air amphibious vehicle dynamics model containing unknown disturbances.

[0058] Based on the kinematics modeling, the dynamics modeling of the water-air amphibious vehicle is carried out to obtain the water-air amphibious vehicle dynamics model containing unknown disturbances as shown in formula (9):

[0059] (9)

[0060] wherein M represents the inertia matrix including the added mass, C(v) represents the Coriolis force and centripetal force matrix including the added mass, D(v) represents the hydrodynamic damping matrix, represents the force / torque vector generated by the gravity and buoyancy, represents the control force / torque vector generated by the propulsion system, represents the unknown disturbance force / torque vector, which is usually generated by the sea current, surge, time-varying of dynamic model parameters, etc.

[0061] For the convenience of analysis, it is generally assumed that the water-air amphibious vehicle is a rigid body with constant mass, the center of buoyancy is located directly above the center of gravity, has three approximate symmetry planes, and does not consider the changes of temperature, pressure, etc. in water. Based on the above assumptions, the parameter matrix of the simplified water-air amphibious vehicle dynamics model is obtained as follows:

[0062] (10)

[0063] (11)

[0064] (12)

[0065] (13)

[0066] wherein m represents the mass of the water-air amphibious vehicle, represents the derivative of the force along the x-axis, i.e. the forward force to the forward acceleration , represents the derivative of the force along the y-axis, i.e. the lateral force to the lateral acceleration , represents the derivative of the force along the z-axis, i.e. the vertical force to the vertical acceleration The derivative of The moment of inertia about the x-axis is represented by . This represents the moment of inertia about the y-axis. This represents the moment of inertia about the z-axis. The torque about the x-axis, i.e., the pitching moment versus the pitching acceleration. The derivative of The torque about the y-axis, i.e., the pitching moment with respect to the pitching acceleration. The derivative of The torque about the z-axis, i.e., the yaw torque versus the yaw acceleration. The derivative of .

[0067] This represents the rate of change of the forward force with respect to the forward velocity u. This represents the rate of change of lateral force with respect to lateral velocity v. This represents the rate of change of vertical force with respect to vertical velocity w. This represents the rate of change of the forward force with respect to the square of the forward velocity, u|u|. This represents the rate of change of the lateral force with respect to the square of the lateral velocity v|v|. This represents the rate of change of vertical force with respect to the square of vertical velocity w|w|. This represents the rate of change of the roll moment with respect to the roll angular velocity p. This represents the rate of change of the pitching moment with respect to the pitching angular velocity q. This represents the rate of change of the yaw moment with respect to the yaw angular velocity r. This represents the rate of change of the roll moment with respect to the square of the roll angular velocity p|p|. This represents the rate of change of the pitching moment with respect to the square of the pitching angular velocity q|q|. Let r be the rate of change of the yaw moment with respect to the square of the yaw angular velocity r|r|.

[0068] W represents the weight of the amphibious vehicle, W=mg, where g is the gravitational constant, and B represents buoyancy. , The density of water, Indicates the volume of water displaced. Indicates the position of the center of buoyancy. This indicates the x-coordinate of the center of buoyancy. This indicates the y-coordinate of the center of buoyancy. This indicates the coordinate of the center of buoyancy on the z-axis.

[0069] Define the thrust control coefficient matrix A(k) at time k as:

[0070] (14)

[0071] Then the control force / torque vector generated by the propulsion system at time k is expressed as:

[0072] (15)

[0073] wherein, is the pseudo-inverse matrix of the thrust control coefficient matrix at the k-1 moment h is a six-degree-of-freedom control output vector.

[0074] Step 3. On the basis of the kinematic model of the amphibious vehicle and the dynamic model of the amphibious vehicle containing unknown disturbances, a position and attitude control system of the amphibious vehicle is constructed.

[0075] The structural block diagram of the position and attitude control system of the amphibious vehicle is shown in Figure 3 The position and attitude control system of the amphibious vehicle includes an AUV model and six ADRC controllers, and the AUV model is established on the basis of the kinematic model of the amphibious vehicle in step 1 and the dynamic model of the amphibious vehicle containing unknown disturbances in step 2. The ADRC controller contains a tracking differentiator (TD), a nonlinear state error feedback (NLSEF) and an extended state observer (ESO).

[0076] In step 3 of the embodiment, a motion control system model of the amphibious vehicle is established according to the expected position and attitude angle of the amphibious vehicle and the actual position and attitude angle of the amphibious vehicle.

[0077] On the basis of the kinematic and dynamic models, a position and attitude control system of the amphibious vehicle is constructed, and the expected position and attitude angle and the actual position and attitude angle of the amphibious vehicle are input into the ADRC controller, wherein represents the expected position of the amphibious vehicle in the surge direction, represents the expected position of the amphibious vehicle in the sway direction, represents the expected position of the amphibious vehicle in the heave direction, represents the expected roll angle of the amphibious vehicle, represents the expected pitch angle of the amphibious vehicle, represents the expected yaw angle of the amphibious vehicle, and the amphibious vehicle adjusts its position and attitude according to the received output , wherein, a control output quantity representing a position and an attitude angle in the vector h, a control output quantity representing a position in the surge direction, a control output quantity representing a position in the sway direction, a control output quantity representing a position in the heave direction, a control output quantity representing a position in the roll direction, a control output quantity representing a position in the pitch direction, a control output quantity representing a position in the yaw direction.

[0078] Taking the control of the vehicle yaw angle ψ as an example:

[0079] The tracking differentiator TD is used as a filter and a differential function, which functions to filter the input signal and reasonably extract the differential signal of the input signal. The TD has the following advantages in the transition process: first, it solves the contradiction between the fast response and the overshoot of the system; second, it reduces the initial error of the system, expands the selection range of the error feedback gain and the error differential feedback gain, and facilitates the tuning of the system. The TD is expressed as:

[0080] .

[0081] wherein, represents the tracking error at time k, represents the state value of the smooth tracking signal generated inside the TD at time k, represents the target parameter value that the amphibious vehicle is expected to reach at time k, represents the state update value of the smooth tracking signal generated inside the TD at time k+1, is a sampling period, represents the state value of the rate of change of the tracking signal generated inside the TD at time k, represents the state update value of the rate of change of the tracking signal generated inside the TD at time k+1, is a fastest control synthesis function. is a parameter that determines the speed of tracking, The larger the value is, the shorter the transition process is, and the faster the response speed is. According to the design experience of the control system, generally, = 10 and = 0.01.

[0082] .

[0083] .

[0084] .

[0085] wherein, represents a basic speed change amount, This indicates the threshold for dividing the error region. This indicates the overall prediction error. This represents the term used in the calculation of nonlinear feedback intensity.

[0086] sign is a sign function, expressed as follows:

[0087] .

[0088] The Extended State Observer (ESO) is the core component of the ADRC controller. It takes all types of disturbances from inside and outside the aircraft system as the total system disturbance and estimates it in real time. Simultaneously, the ESO compensates for this total disturbance and transforms the nonlinear and uncertain object with unknown disturbances into a linear object of an integral series. The ESO is based solely on the input and output information of the controlled object, without relying on the specific mathematical model generating the disturbance, thus exhibiting strong robustness. The ESO expression is as follows:

[0089] .

[0090] in, This represents the estimated value output by the controlled object. This represents an estimated rate of change in the output of the controlled object. This represents the estimated total disturbance. β1, β2, and β3 are parameters to be tuned. β1 affects the estimation of the state variables, β2 affects the estimation of the disturbance to the entire system, and β3 can speed up the system's response. δ is the length of the linear segment of the fal function, δ = 0.1. b0 is the control gain, b0 = 1. ψ (k) is for controlling the total amount.

[0091] It is a non-linear function, and its expression is as follows:

[0092] .

[0093] in, That is , α1, α2, α3, α4, and α5 are variable parameters between 0 and 1, and α1, α2, α3, α4, and α5 are generally taken as 0.5.

[0094] Nonlinear error feedback (NLSEF) uses a nonlinear combination of tracking error signals to obtain the control quantity h. ψ0 (k) is used to achieve better control, and its expression is as follows:

[0095] .

[0096] in, Indicates positional deviation. The speed deviation is represented by k1 and k2, which are the controller gains: k1=500, k2=5, and α4=α5=0.5. The control quantity h... ψ0 (k) Same as the extended observer for the total disturbance The compensation amount together constitutes the total control quantity h of the system. ψ (k) is:

[0097] .

[0098] Step 4. Parameter tuning of the position and attitude control system of the amphibious vehicle is performed based on the hybrid multi-strategy dung beetle optimization algorithm. The hybrid multi-strategy dung beetle optimization algorithm is obtained by improving the dung beetle optimization algorithm as follows: introducing the optimal point set theory for population initialization, introducing a spiral exploration strategy in the position update of dung beetle reproduction and theft behavior, and introducing a centroid back learning strategy to search the global and dimension-wise Gaussian mutation after the position update of dung beetle rolling dung balls, dancing, foraging, reproduction and theft behavior.

[0099] In this embodiment, step 4 involves tuning the model parameters of the amphibious vehicle's motion control system, ultimately outputting the optimal position and attitude control parameters to ensure the amphibious vehicle's motion control achieves the desired performance. Step 4 includes population initialization, position updates for five behaviors, global centroid back-learning search, and dimensional Gaussian mutation.

[0100] Amphibious vehicles in Controlling six degrees of freedom requires six corresponding ADRC controllers. The parameters β1, β2, and β3 in each controller need to be tuned, for a total of 18 parameters, represented as β = [β...]. 1x ,β 2x ,β 3x ,β 1y ,β 2y ,β 3y ,β 1z ,β 2z ,β 3z ,β 1φ ,β 2φ ,β 3φ ,β 1θ ,β 2θ ,β 3θ ,β 1ψ ,β 2ψ ,β 3ψ ], where β 1x β 2x β 3x The parameter β represents the ADRC controller parameter in the sway direction. 1y β 2y β 3yParameters of the ADRC controller in the sway direction, β 1z 2z 3z Parameters of the ADRC controller in the heave direction, β 1φ 2φ 3φ Parameters of the ADRC controller in the roll direction, β 1θ 2θ 3θ Parameters of the ADRC controller in the pitch direction, β 1ψ 2ψ 3ψ Parameters of the ADRC controller in the yaw direction.

[0101] The conventional parameter setting method is artificial experience setting, which is time-consuming and cannot achieve optimal control effect. Therefore, the method of the present application designs a vehicle control parameter setting method based on a hybrid multi-strategy melolontha optimization algorithm, and the parameter setting process is as follows:

[0102] The vehicle control parameter setting method based on the hybrid multi-strategy melolontha optimization algorithm is used to obtain the optimal solution of the ADRC controller parameters (β1, β2, β3) of the amphibious vehicle in six degrees of freedom, and then the optimal solution of the position and attitude control parameters is obtained. The parameter setting method in this embodiment specifically includes the following steps:

[0103] Step 4.1. Population initialization.

[0104] Input the maximum number of iterations , the number of populations N, the dimension D of the solution, the upper limit of the solution , and the lower limit of the solution .

[0105] According to the initialization behavior switching rule, the population is divided into four groups, which are marked as rolling dung beetle, breeding dung beetle, foraging dung beetle and stealing dung beetle.

[0106] In this embodiment, the population initialization is uniformly generated in the initialization stage by introducing the theory of good point set to improve the diversity of the initialized population. The construction of the good point set is as follows:

[0107] (16)

[0108] Wherein, D represents the dimension of the optimization problem, i.e. the dimension of the solution, to represent the 1st to Dth good points in the good point set, , , ​​​​​​​​To meet the minimum prime number.

[0109] The expression of initializing an individual position by using the set of good points is:

[0110] (17)

[0111] wherein, denotes the jth individual position, and denote the lower limit and upper limit of the jth dimension of the optimization problem, respectively.

[0112] Step 4.2. Position update.

[0113] Secondly, the position of the population is updated according to the inspiration of the five behaviors of dung beetle rolling dung ball, dancing, foraging, reproduction and stealing, wherein the spiral exploration strategy is introduced in the position update of the reproduction and stealing of the dung beetle, and the specific process is as follows:

[0114] Step 4.2.1. Position update of rolling dung ball behavior.

[0115] In the position update process, the individual marked as the rolling dung ball dung beetle applies the position update algorithm of the rolling dung ball behavior.

[0116] λ is a random number between 0 and 1, when λ < 0.9, it means that the dung beetle performs the rolling dung ball behavior without obstacles, and the position update is as follows:

[0117] (18)

[0118] wherein, t is the current iteration number, denotes the position of the ith dung beetle in the tth iteration, denotes the position of the ith dung beetle in the t+1th iteration, denotes the position of the ith dung beetle in the t-1th iteration, η denotes a natural coefficient, and η takes the value of -1 or 1; is a disturbance coefficient, ; is a random number between 0 and 1, used to simulate the intensity change of light:

[0119] .

[0120] wherein, denotes the global worst position of the rolling dung ball dung beetle; the parameters generated by the position update of the rolling dung ball behavior are substituted into the target function ITAE, and the value of the target function I is obtained by integrating the product of the system time and the absolute error, when the result of I is the maximum, the corresponding position is the global worst position of the rolling dung ball dung beetle .

[0121] The ITAE expression is:

[0122] .

[0123] wherein, represents the stable time of the system, represents the error function.

[0124] Step 4.2.2. Dance behavior position update.

[0125] When , it indicates that the dung beetle encounters an obstacle, at which time the dung beetle needs to update the position through the dance behavior:

[0126] (19)

[0127] wherein, is the deflection angle, , when , , , the position is not updated.

[0128] Step 4.2.3. Reproduction behavior position update.

[0129] In the position update process, the individual marked as the reproduction dung beetle applies the reproduction behavior position update algorithm. The expression for simulating the female dung beetle's egg-laying area through the boundary selection strategy by the reproduction behavior of the dung beetle is:

[0130] (20)

[0131] wherein, and represent the lower limit and the upper limit of the egg-laying area; , represents the maximum number of iterations; and represent the lower limit and the upper limit of the to-be-adjusted parameter, i.e., the lower limit and the upper limit of the solution; represents the global optimal position of the reproduction dung beetle.

[0132] Substitute the parameter generated by the reproduction behavior position update into the objective function ITAE, and obtain the objective function value I by integrating the product of the system time and the absolute error, and when the result of I is the minimum, the corresponding position is the global optimal position of the reproduction dung beetle .

[0133] The method of the present application introduces a spiral exploration strategy in the reproduction behavior position update, and has a probability of 0.5 to update the position by using the spiral exploration strategy, thereby increasing the diversity of the position update mode. The spiral exploration factor is defined as follows:

[0134] (21)

[0135] where v1 and v2 are random numbers between (0, 1), and are independent random vectors of D, D is the dimension of the optimization problem, i.e., the number of parameters to be adjusted.

[0136] is a random number between (0, 1), when the position update of the breeding behavior is as follows:

[0137] (22)

[0138] when the spiral exploration strategy is introduced, and the position update of the breeding behavior is as follows:

[0139] (23)

[0140] where and are independent random vectors of D, represents the position of the brood ball of the ith dung beetle in the tth iteration, represents the position of the brood ball of the ith dung beetle in the t+1th iteration. Step 4.2.4. Position update of foraging behavior.

[0141] In the process of position update, individuals marked as foraging dung beetles apply the foraging behavior position update algorithm.

[0142] The optimal foraging area generated by the foraging behavior of the dung beetle is represented as:

[0143]

[0144] (24) where

[0145] and are the lower and upper limits of the foraging area, respectively, represents the global optimal position of the foraging dung beetle. The parameters generated by the foraging behavior position update are substituted into the objective function ITAE, and the value of the objective function I is obtained by integrating the product of the system time and the absolute error. When the result of I is the smallest, the corresponding position is the global optimal position of the foraging dung beetle .

[0146] Then the position update of the foraging behavior of the dung beetle is as follows:

[0147] (25) ​​

[0148] where, is a random number obeying normal distribution, is a random vector of (0, 1).

[0149] Step 4.2.5. Stealing behavior position update.

[0150] In the position update process, the individuals marked as stealing ants apply the stealing behavior position update algorithm.

[0151] The stealing behavior of stealing ants also introduces a half-probability spiral exploration strategy in the position update, to increase the diversity of the position update method, and the spiral exploration factor is defined as follows:

[0152] (26)

[0153] is a random number between (0, 1), when the stealing behavior generated position update is as follows:

[0154] (27)

[0155] where, g is a random vector of (0, 1) obeying normal distribution, S is a constant.

[0156] When the spiral exploration strategy is introduced, and the stealing behavior generated position update is as follows:

[0157] (28)

[0158] Step 4.3. Barycenter reverse learning strategy searches for global.

[0159] After all groups have completed the position update of the five behaviors of rolling dung balls, dancing, foraging, breeding and stealing, the barycenter reverse learning strategy is further introduced. The principle of this strategy is to generate a reverse solution with the population barycenter as the symmetry point, compare the objective function values of the current solution and the reverse solution, and then select the better solution to enter the next iteration.

[0160] The value of the population barycenter in the jth dimension is represented as:

[0161] (29)

[0162] where, N is the population size, is the position of the ith ant in the jth dimension, .

[0163] Then the reverse solution of the ith ant in the jth dimension is:

[0164] (30)

[0165] The method of the application in the parameter setting process based on the hybrid multi-policy melolontha algorithm, the target function used is the integral of time multiplied by absolute error ITAE, and the expression is:

[0166] (31)

[0167] Wherein, The stability time of the system is represented, The error function is represented.

[0168] The current solution, that is, The reverse solution generated by the barycenter reverse learning strategy , is substituted into the target function ITAE, and I is obtained by integrating the system time multiplied by the absolute error, the result of I is the minimum, then the corresponding solution is a better solution, then the better solution is selected into the next iteration, so as to improve the global search ability.

[0169] Step 4.4. Gaussian mutation of the optimal solution is performed dimension by dimension.

[0170] After the reverse learning strategy, the optimal solution obtained by the target function ITAE, that is, the optimal solution before Gaussian mutation, is subjected to Gaussian mutation dimension by dimension, and the value of the jth dimension of the optimal solution after mutation is:

[0171] (32)

[0172] Wherein, μ is a random variable obeying Gaussian distribution, That is, the optimal solution after Gaussian mutation.

[0173] By Gaussian mutation of the optimal solution, the convergence speed of the algorithm can be improved.

[0174] The values before and after Gaussian mutation, that is, the values of the optimal solution before Gaussian mutation and the optimal solution after Gaussian mutation, are substituted into the target function ITAE again, and the value I obtained by the target function is used to judge whether the position after Gaussian mutation is better than the position before Gaussian mutation, if the position after Gaussian mutation is better than the position before Gaussian mutation, then the position after Gaussian mutation is used for position updating.

[0175] The optimal selection of six ADRC controller parameters (β1, β2, β3) is guided by ITAE, when the iteration number reaches the maximum iteration number input in the initialization, the optimal solution judged by ITAE is output, and then the output data after the controller processing is used to adjust the position and attitude of the amphibious vehicle.

[0176] This invention designs a vehicle control parameter tuning method based on a hybrid multi-strategy dung beetle optimization algorithm for the parameter tuning optimization of ADRC controllers. This improves the diversity of position update methods for amphibious vehicles, avoids the algorithm getting stuck in local optima, and enhances the global search capability and convergence speed of the algorithm, thereby further improving the position and attitude control performance of amphibious vehicles.

[0177] To verify the effectiveness of the method proposed in this invention, a kinematic and dynamic model of the amphibious vehicle was established using the Simulink module of MATLAB software in this embodiment. The model parameters are shown in Table 2.

[0178] Table 2. Dynamics and Kinematic Model Parameters of Amphibious Vehicles

[0179]

[0180] Assuming the initial position and attitude angles of the amphibious vehicle are all 0, the desired position is x. d =2、y d =3、z d =2, the desired attitude angle is φ d =π / 3、θ d =π / 3、ψ d =π / 3. In this embodiment, the overall optimization capability based on the hybrid multi-strategy dung beetle optimization algorithm was simulated, and the convergence curve and tracking effect of the amphibious vehicle position and attitude control parameter tuning were compared with the traditional dung beetle optimization algorithm, gray wolf optimization algorithm and particle swarm optimization algorithm. The population size N=40, and other relevant parameters are shown in Table 3.

[0181] Table 3. Parameter settings for the four algorithms

[0182]

[0183] Where c1 represents the tendency factor controlling the particle to move towards its own historical best position, c2 represents the tendency factor controlling the particle to move towards the group's historical best position, and w max w represents the maximum inertia weight. min denoted by , a2 represents the minimum inertia weight, a2 represents the attenuation coefficient, and S0 is a Bernoulli random variable.

[0184] Table 4. Comparison of overshoot and steady-state time results of the method of this invention with other optimization algorithms for the control of amphibious vehicles.

[0185]

[0186] Depend on Figures 4 to 9It can be known that the ITAE effect of the ADRC controller based on the grey wolf optimization algorithm is the worst, the effects of the particle swarm optimization algorithm and the traditional blt optimization algorithm are similar, and the effect of the mixed multi-strategy blt optimization algorithm is the best.

[0187] From Figures 4 to 9 Compared with the artificial parameter adjustment, the overall control performance of the ADRC adjusted by the grey wolf optimization algorithm is not obvious, and the overall control performance of the ADRC adjusted by the particle swarm optimization algorithm, the blt optimization algorithm and the mixed multi-strategy blt optimization algorithm is obviously improved. In the control of the bow shake angle, although the overshoot of the mixed multi-strategy blt optimization algorithm is higher than that of the artificial parameter adjustment and the grey wolf optimization algorithm, the steady state time is obviously reduced. Overall, except that the steady state time of the ADRC adjusted by the mixed multi-strategy blt optimization algorithm is higher in the control of the lateral shake angle, the overall control performance of the ADRC adjusted by the mixed multi-strategy blt optimization algorithm is the best.

[0188] Embodiment 2

[0189] This embodiment 2 describes a computer device, which comprises a memory and one or more processors.

[0190] The executable code is stored in the memory, and when the processor executes the executable code, the steps of the aircraft control parameter adjustment method based on the mixed multi-strategy blt optimization algorithm in the above-mentioned embodiment 1 are implemented.

[0191] The computer device in this embodiment is any device or apparatus with data processing capability, which will not be described here.

[0192] Of course, the above description is only for the preferred embodiments of the present application, and the present application is not limited to the above-mentioned embodiments. It should be noted that any skilled person in the art can make all equivalent substitutions and obvious modifications under the teaching of the present application, which all fall within the scope of the present application, and should be protected by the present application.

Claims

1. A method for tuning the control parameters of a vehicle based on a hybrid multi-policy harvester optimization algorithm, characterized in that, The method comprises the following steps: Step 1. Kinematic modeling of the water-air amphibious vehicle is performed to obtain a kinematic model of the water-air amphibious vehicle; Step 2. On the basis of the kinematic model of the water-air amphibious vehicle, dynamic modeling of the water-air amphibious vehicle is performed to obtain a water-air amphibious vehicle dynamic model containing unknown disturbances; Step 3. On the basis of the kinematic model of the water-air amphibious vehicle and the water-air amphibious vehicle dynamic model containing unknown disturbances, a water-air amphibious vehicle position and attitude control system is constructed; Step 4. Parameter setting of the water-air amphibious vehicle position and attitude control system is performed based on a hybrid multi-strategy scarab optimization algorithm, which is obtained by improving a scarab optimization algorithm as follows: introducing a good point set theory for population initialization; introducing a spiral exploration strategy in the position update of scarab reproduction and stealing behavior; introducing a barycenter reverse learning strategy to search the global and Gaussian mutation in each dimension after the position update of the scarab ball-rolling, dancing, foraging, reproduction and stealing behaviors; In the step 4, the process of introducing the good point set theory for population initialization is specifically as follows: By introducing the theory of well-set, the population positions in initialization phase are generated uniformly The well-set is constructed as follows: (16) wherein D denotes the dimension of the optimization problem, i.e. the dimension of the solution, to denote the 1st to Dth best points in the set of best points, respectively, , , to satisfy the smallest prime number; An expression for initializing individual positions by using the good point set is as follows: (17) wherein, denotes the jth individual position, and denotes the lower and upper bounds of the jth dimension of the optimization problem, respectively.

2. The hybrid multi-policy Blattella germanica optimization algorithm based vehicle control parameter tuning method according to claim 1, wherein The step 1 is specifically as follows: Defining six degree of freedom position and attitude vectors for water-air amphibious vehicles are: (1) wherein represents a position vector of the water-air amphibious vehicle, , represents a position of the water-air amphibious vehicle in the surge direction, represents a position of the water-air amphibious vehicle in the sway direction, represents a position of the water-air amphibious vehicle in the heave direction; represents an attitude vector of the water-air amphibious vehicle, , represents a roll angle of the water-air amphibious vehicle, represents a pitch angle of the water-air amphibious vehicle, represents a yaw angle of the water-air amphibious vehicle; Defining six degrees of freedom velocity and angular velocity vectors for a water-air amphibious vehicle are: (2) wherein denotes the velocity vector of the water-air amphibious vehicle, , denotes the velocity of the water-air amphibious vehicle in the surge direction, i.e. the forward velocity, denotes the velocity of the water-air amphibious vehicle in the sway direction, i.e. the lateral velocity, denotes the velocity of the water-air amphibious vehicle in the heave direction, i.e. the vertical velocity; denotes the angular velocity vector of the water-air amphibious vehicle, , denotes the angular velocity of the water-air amphibious vehicle in the roll direction, i.e. the roll angular velocity, denotes the angular velocity of the water-air amphibious vehicle in the pitch direction, i.e. the pitch angular velocity, denotes the angular velocity of the water-air amphibious vehicle in the yaw direction, i.e. the yaw angular velocity; A velocity conversion equation for describing the water-air amphibious vehicle along the three-axis linear motion by using the Euler angle description method is as follows: (3) wherein is the velocity coordinate transformation matrix; An angular velocity conversion equation for describing the water-air amphibious vehicle rotation around the three-axis is as follows: (4) wherein is the angular velocity coordinate transformation matrix; By using the concept of rotation, Euler's theorem and the properties of skew-symmetric matrices, a conversion matrix is obtained as follows: (5) (6) wherein denotes a sine function, denotes a cosine function, denotes a tangent function; defining a total coordinate transformation matrix is: (7) A kinematic model of the water-air amphibious vehicle is obtained from the formula (3) and the formula (4) as follows: (8)。 3. The hybrid multi-policy Blattella germanica optimization algorithm based vehicle control parameter tuning method according to claim 2, characterized in that, The step 2 is specifically as follows: On the basis of the kinematic model of the water-air amphibious vehicle, dynamic modeling of the water-air amphibious vehicle is performed to obtain a water-air amphibious vehicle dynamic model containing unknown disturbances as shown in the formula (9): (9) where M is the inertia matrix including added mass, C(v) is the Coriolis and centripetal force matrix including added mass, D(v) is the hydrodynamic damping matrix, Fg is the force due to gravity and buoyancy, Fc is the control force due to the propulsion system, Fi is the unknown disturbance force; A parameter matrix of the simplified water-air amphibious vehicle dynamic model is as follows: (10) (11) (12) (13) Where m represents the mass of the amphibious vehicle. This represents the force along the x-axis, i.e., the forward force and the forward acceleration. The derivative of This represents the force along the y-axis, i.e., the lateral force and the lateral acceleration. The derivative of This represents the force along the z-axis, i.e., the vertical force and the vertical acceleration. The derivative of The moment of inertia about the x-axis is represented by . This represents the moment of inertia about the y-axis. This represents the moment of inertia about the z-axis. The torque about the x-axis, i.e., the pitching moment versus the pitching acceleration. The derivative of The torque about the y-axis, i.e., the pitching moment with respect to the pitching acceleration. The derivative of The torque about the z-axis, i.e., the yaw torque versus the yaw acceleration. The derivative; represents the rate of change of the forward force with respect to the forward velocity u, represents the rate of change of the lateral force with respect to the lateral velocity v, represents the rate of change of the vertical force with respect to the vertical velocity w, represents the rate of change of the forward force with respect to the forward velocity squared u | u |, represents the rate of change of the lateral force with respect to the lateral velocity squared v | v |, represents the rate of change of the vertical force with respect to the vertical velocity squared w | w |, represents the rate of change of the roll moment with respect to the roll angular velocity p, represents the rate of change of the pitch moment with respect to the pitch angular velocity q, represents the rate of change of the yaw moment with respect to the yaw angular velocity r, represents the rate of change of the roll moment with respect to the roll angular velocity squared p | p |, represents the rate of change of the pitch moment with respect to the pitch angular velocity squared q | q |, represents the rate of change of the yaw moment with respect to the yaw angular velocity squared r | r |. W denotes the weight of the amphibious vehicle, W = mg, g being the gravity coefficient, B denotes the buoyancy, , is the density of water, denotes the volume of water displaced, denotes the position of the center of buoyancy, denotes the coordinate of the position of the center of buoyancy on the x-axis, denotes the coordinate of the position of the center of buoyancy on the y-axis, denotes the coordinate of the position of the center of buoyancy on the z-axis; A thrust control coefficient matrix A(k) of the k-th moment is defined as follows: (14) the control force generated by the propulsion system at the kth time instant is expressed as: (15) wherein, is the thrust control coefficient matrix at the k-1 time is the pseudo-inverse matrix of is the six-degree-of-freedom control output vector.

4. The hybrid multi-policy Blum optimization algorithm based vehicle control parameter tuning method according to claim 3, wherein, In the step 3, The water-air amphibious vehicle position and attitude control system comprises an AUV model and six ADRC controllers; wherein the AUV model is established based on the kinematic model of the water-air amphibious vehicle in the step 1 and the water-air amphibious vehicle dynamic model containing unknown disturbances in the step 2; the ADRC controller comprises a tracking differentiator, a nonlinear error feedback and an extended state observer; by comparing the desired position and attitude angle of the amphibious vehicle with the actual position and attitude angle of the amphibious vehicle and the actual position and attitude angle of the amphibious vehicle inputting the ADRC controller to obtain a control output , the amphibious vehicle adjusts its position and attitude according to the received control output ​ wherein denotes a desired position of the water-air amphibious vehicle in the surge direction, denotes a desired position of the water-air amphibious vehicle in the sway direction, denotes a desired position of the water-air amphibious vehicle in the heave direction, denotes a desired roll angle of the water-air amphibious vehicle, denotes a desired pitch angle of the water-air amphibious vehicle, denotes a desired yaw angle of the water-air amphibious vehicle; denotes a control output quantity in the surge direction, denotes a control output quantity in the sway direction, denotes a control output quantity in the heave direction, denotes a control output quantity in the roll direction, denotes a control output quantity in the pitch direction, denotes a control output quantity in the yaw direction.

5. The hybrid multi-policy Blattella germanica optimization algorithm based vehicle control parameter tuning method according to claim 4, wherein In the step 4, Water amphibious vehicle in The control of six degrees of freedom needs six corresponding ADRC controllers, and the parameters β1, β2, β3 in each controller need to be tuned, a total of 18 parameters need to be tuned. The parameters to be tuned are expressed as β=[β 1x , β 2x , β 3x , β 1y , β 2y , β 3y , β 1z , β 2z , β 3z , β 1φ , β 2φ , β 3φ , β 1θ , β 2θ , β 3θ , β 1ψ , β 2ψ , β 3ψ ] wherein β 1x , β 2x , β 3x denote parameters of the ADRC controller in the surge direction, β 1y , β 2y , β 3y denote parameters of the ADRC controller in the sway direction, β 1z , β 2z , β 3z denote parameters of the ADRC controller in the heave direction, β 1φ , β 2φ , β 3φ denote parameters of the ADRC controller in the roll direction, β 1θ , β 2θ , β 3θ denote parameters of the ADRC controller in the pitch direction, β 1ψ , β 2ψ , β 3ψ denote parameters of the ADRC controller in the yaw direction.

6. The hybrid multi-policy Blattella germanica optimization algorithm based vehicle control parameter tuning method according to claim 5, wherein The process of the position update in the step 4 is specifically as follows: The population position is updated according to the five behaviors of the scarab ball-rolling, dancing, foraging, reproduction and stealing; The process of the ball-rolling behavior position update is specifically as follows: In the position update process, the ball-rolling behavior position update algorithm is applied to the individual marked as the ball-rolling scarab; λ is a random number between 0 and 1, when λ<0.9, it means that the scarab performs the ball-rolling behavior without obstacles, and the position update is as follows: (18) where t is the current iteration number, denotes the position of the i-th dung beetle in the t-th iteration, denotes the position of the i-th dung beetle in the t+1-th iteration, denotes the position of the i-th dung beetle in the t-1-th iteration, and η denotes a natural coefficient; is a perturbation coefficient; is a random number between (0, 1), for simulating the intensity variation of light: ； wherein, represents the global worst position of the tumblebug; The process of the dancing behavior position update is specifically as follows: When means that the dung beetle encounters an obstacle, at which point the dung beetle needs to update its position by performing a dance: (19) wherein is the deflection angle, when , , the position is not updated; The process of the reproduction behavior position update is specifically as follows: In the position update process, the reproduction behavior position update algorithm is applied to the individual marked as the reproduction scarab; An expression for simulating the female scarab egg-laying region by the boundary selection strategy of the scarab reproduction behavior is as follows: (20) wherein, and denotes the lower and upper limits of the egg laying area; , denotes the maximum number of iterations; and denotes the lower and upper limits of the parameter to be adjusted; denotes the globally optimal position of the breeding dung beetle; In the reproduction behavior update position, the spiral exploration strategy is introduced, and the spiral exploration factor is defined as follows: (21) where v1 and v2 are random numbers between (0, 1), and are independent random vectors; is a random number between (0, 1), when The position update resulting from the reproductive behavior is as follows: (22) When the spiral exploration strategy is introduced, the position update generated by the breeding behavior is as follows: (23) wherein, and is an independent random vector, represents the position of the brood ball of the ith dung beetle in the tth iteration, represents the position of the brood ball of the ith dung beetle in the t+1th iteration; The process of foraging behavior position update is as follows: In the position update process, the foraging behavior position update algorithm is applied to the individual marked as the foraging scarab; The optimal foraging area generated by the foraging behavior of the scarab is represented as: (24) wherein, and are the lower and upper limits of the foraging zone, respectively, denotes the global optimal position of the foraging scarab. The position update of the scarab foraging behavior is as follows: (25) wherein, is a random number that is normally distributed, is a random vector of (0,1). The process of stealing behavior position update is as follows: In the position update process, the stealing behavior position update algorithm is applied to the individual marked as the stealing scarab; In the position update of the stealing behavior of the stealing scarab, the spiral exploration strategy is introduced, and the spiral exploration factor is defined as follows: (26) When the theft occurs, the location update is as follows: (27) where g is normally distributed a random vector, S is a constant; When the spiral exploration strategy is introduced, the position update of the stealing behavior is as follows: (28)。 7. The hybrid multi-policy Blum optimization algorithm based vehicle control parameter tuning method according to claim 6, wherein, In step 4, the process of the barycenter reverse learning strategy searching the global is as follows: The value of the population barycenter in the jth dimension is represented as: (29) wherein N is the population number, is the position of the ith dung beetle in the jth dimension, ; The inverse solution of the ith dung beetle in the jth dimension is then is: (30) The target function ITAE is the integral of the system time multiplied by the absolute error, and the expression of the target function value I is as follows: (31) wherein denotes the settling time of the system, denotes the error function; The current solution is The inverse solution generated by the centroid inverse learning strategy Substituting into the objective function shown in formula (31), the objective function value I is obtained by multiplying the system time by the integral of the absolute error. The solution with the smallest I is selected to enter the next iteration to improve the global search capability.

8. The hybrid multi-policy Blum optimization algorithm based vehicle control parameter tuning method according to claim 7, wherein, In step 4, the process of the Gaussian mutation is as follows: The optimal solution before Gaussian variation obtained by the objective function shown in formula (31) is subjected to Gaussian variation dimension by dimension, and the value of the jth dimension of the optimal solution after variation is is: (32) wherein μ is a random variable following a Gaussian distribution; is the optimal solution after Gaussian variation; The values of the optimal solution before the Gaussian mutation and the optimal solution after the Gaussian mutation are substituted into the target function ITAE, and whether the position after the Gaussian mutation is better than the position before the Gaussian mutation is judged by the target function value I.

9. A computer device comprising a memory and one or more processors, the memory having stored therein executable code, the computer device characterized in that, When the processor executes the executable code, the steps of the aircraft control parameter setting method based on the hybrid multi-policy scarab optimization algorithm are realized.

Images