A multi-parameter particle swarm identification system and method for unmanned ships

Abstract

The present invention relates to a multi-parameter particle swarm identification system and method for an unmanned ship. The system includes a state-measurable unmanned ship and a display and control terminal. The state-measurable unmanned ship includes an unmanned ship and an unmanned ship state monitoring device, which is used to control the navigation of the unmanned ship in real time, monitor the state information of the unmanned ship and upload it to the display and control terminal; the display and control terminal has a model building unit and an identification unit, the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, and inputs a speed instruction to obtain simulation information of the unmanned ship model, and compares the simulation information with the actual navigation state information of the unmanned ship to obtain forward speed error, lateral drift error and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identification parameters, and the identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model. The system and method are conducive to quickly and accurately identifying the parameters of the unmanned ship model.

Description

Technical Field

[0001] The present invention belongs to the field of unmanned ships, and in particular relates to a multi-parameter particle swarm identification system and method for unmanned ships. Background Art

[0002] With the development of commercial and military related to oceans and rivers, unmanned ships have become an important platform for performing tasks such as search and rescue, reconnaissance, monitoring, fire strikes, anti-mine and anti-submarine due to their advantages such as small size, high speed, low cost and no risk of casualties. Unmanned ship motion control is one of the key technologies to achieve autonomous, reliable and safe navigation. The mathematical model of unmanned ships, as the basis of motion control, contains very complex motion mechanisms. The model coefficients can be divided into inertia matrix coefficients, centripetal force matrix coefficients and hydrodynamic matrix coefficients according to the physical meaning of unmanned ship navigation. Therefore, it is very important to accurately identify the parameters of unmanned ships. At present, traditional parameter identification methods such as maximum likelihood, recursive least squares method, Kalman filter algorithm have the disadvantages of low identification accuracy and poor generalization ability. Although artificial intelligence algorithms such as recursive neural networks and support vector machines have high accuracy, the identification process is relatively complex and the identification speed is slow. In addition, most traditional identification methods must first collect the navigation data of unmanned ships in the identification process, and then return to the room for identification. This process is time-consuming and labor-intensive and prone to human errors. In summary, developing an identification method that can quickly and accurately obtain model parameters has important engineering significance. Summary of the invention

[0003] The object of the present invention is to provide a multi-parameter particle swarm identification system and method for an unmanned ship, which is conducive to quickly and accurately identifying the model parameters of the unmanned ship.

[0004] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a multi-parameter particle swarm identification system for unmanned ships, including a state-measurable unmanned ship and a display and control terminal, wherein the state-measurable unmanned ship includes an unmanned ship and an unmanned ship state monitoring device, which is used to control the navigation of the unmanned ship in real time, monitor the state information of the unmanned ship and upload it to the display and control terminal; the display and control terminal has a model building unit and an identification unit, the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, inputs the speed command into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned ship model, and compares the simulation information with the actual navigation state information of the unmanned ship to obtain forward speed error information, lateral drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identification parameters, and the identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model.

[0005] Furthermore, the unmanned ship is a twin-propeller unmanned ship, including a hull, a power module, a power module, a data communication module and a central processing unit module. The central processing unit module receives the navigation path and request data commands issued by the display and control terminal through the data communication module, controls the unmanned ship to navigate according to the set path, and controls the unmanned ship status monitoring device to obtain the status information of the unmanned ship; the unmanned ship status monitoring device includes a data acquisition module and a data processing module, and the data acquisition module includes a variety of sensing devices to collect the status information of the unmanned ship including speed, acceleration, propeller speed, positioning coordinates, heading angle and heading angle, and the acquired information is processed by the data processing module and sent to the central processing unit module, and then the status information of the unmanned ship is uploaded to the display and control terminal through the central processing unit module.

[0006] Furthermore, the power module adopts a large-capacity lithium battery, and realizes voltage stabilization and current measurement through an ammeter; the power module adopts a dual-propeller brushless motor as a propeller, and realizes steering through differential speed; the data communication module adopts wireless communication equipment, and realizes real-time information interaction with the display and control terminal through the lightweight message transmission protocol Mavlink; the central processing unit module is the control core of the unmanned ship, which is mainly responsible for information processing and interaction of each module, and realizes the motion control of the unmanned ship.

[0007] Furthermore, the data acquisition module obtains the navigation speed and acceleration of the unmanned ship through an accelerometer, obtains the actual value of the propeller speed through a Hall sensor, obtains the latitude and longitude information of the unmanned ship, that is, the positioning coordinates, through a GPS module, obtains the heading angle information through an electronic compass, obtains the course angle information through a gyroscope, and transmits the obtained information to the data processing module; the data processing module processes the obtained information through a Kalman filter algorithm and then packages it and sends it to the central processing unit module.

[0008] Furthermore, the display and control terminal includes a computer and a parameter identification software module installed on the computer, and the parameter identification software module is mainly composed of a control unit, a display unit, a model building unit and an identification unit. The user controls the navigation path of the unmanned ship through the control unit, and displays the status information of the unmanned ship in real time through the display unit; the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, inputs the speed command into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned ship model, including simulated forward speed, simulated drift speed and simulated bow angular velocity, and compares the simulation information with the actual navigation status information of the unmanned ship to obtain forward speed error information, drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identified parameters; the identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model, thereby realizing the identification of the unmanned ship model parameters.

[0009] Furthermore, the model building unit establishes a Fossen three-degree-of-freedom unmanned ship model, which is expressed as follows:

[0010]

[0011]

[0012]

[0013]

[0014]

[0015] Among them, u, v, r are the longitudinal speed, transverse speed and bow turning speed of the unmanned ship respectively, m is the mass of the unmanned ship, I z is the moment of inertia about the z-axis, are the longitudinal and transverse additional masses, is the additional moment of inertia, X u , Y v , Y r 、N v 、N r is the linear hydrodynamic damping coefficient, X |u|u , Y |v|v , Y |r|v , Y |v|r , Y |r|r 、N |v|v 、N |r|v 、N |v|r 、N |r|r is the nonlinear hydrodynamic damping coefficient, c and d are the propeller thrust coefficients of the unmanned ship, and d p is the distance from the propeller to the center axis of the unmanned boat.

[0016] The present invention also provides an unmanned ship multi-parameter particle swarm identification method based on the above system, wherein the identification unit uses a multi-objective particle swarm algorithm to identify the unmanned ship model parameters, comprising the following steps:

[0017] Initialize the particle swarm and archive set: randomly generate M particles in the search space as the initial population, define the population as the unknown parameters that need to be identified in the model, calculate the objective function value of each particle, and obtain the initial Archive set according to the Pareto dominance principle;

[0018] Searching for guiding particles: From the initial Archive set, estimate the particle crowding according to the adaptive hypergrid method, and select guiding particles by using density information combined with the roulette method;

[0019] Update particle swarm and archive set: Update each particle according to the speed and position iteration formula, continue to calculate the objective function value of each particle, update the Archive archive set, and seek the optimal generation of guiding particles and individuals;

[0020] The particle swarm and the archive set are updated cyclically until the error between the simulation information and the measured navigation state information is minimized. Then, the relatively optimal solution is selected from the final archive set as the model parameter value according to the selection strategy.

[0021] Furthermore, the multi-objective particle swarm algorithm initializes the particle swarm in the dim-dimensional search space to generate M initial particles:

[0022] X=rand(M,dim)*(X max -X min )+X min

[0023] Where, X max and X min are the upper and lower bounds of the particle swarm search space, rand(M, dim) generates M dim-dimensional particles, and each element of the particle is a random decimal between [0, 1];

[0024] From the initialized particle swarm, find a set of non-inferior solutions according to the Pareto dominance principle;

[0025] Seeking guiding particles: Divide the objective function space into N1×N2×N3 networks, where the network module length of the i-th objective is n i for:

[0026] n i =(f i max -f i min ) / N,(i=1,2,3)

[0027] In the formula, f i max 、f i min is the maximum and minimum value of the particle on the i-th objective function, and N is the number of members in the current iteration of the Archive set;

[0028] The position p of the non-inferior solution in the grid space i for:

[0029] p i =mod((f l,i -f i min ) / d i )+1

[0030] In the formula, fl,i is the ith objective function value corresponding to the lth non-inferior solution, mod is the rounding function, and the number of non-inferior solutions contained in each grid is obtained; the number of non-inferior solutions in a grid is used as the distribution density information of the particles, and the particles with the smallest distribution density are selected to guide the particles in combination with roulette;

[0031] Update particle swarm and archive set: Repeat the following formula to update the speed and position of each particle:

[0032]

[0033] Where V i (k+1) is the velocity of the ith particle at the k+1th iteration, w is the inertia factor, c1 and c2 are the individual learning factor and the overall learning factor, respectively, and rand() is a function that generates random numbers between 0 and 1. is the historical optimal position of the i-th particle at the kth iteration, X gbest (k) is the historical optimal position of all particles at the kth iteration, that is, the current guiding particle; the selection formula for selecting the relatively optimal solution in the final archive set is:

[0034]

[0035] Where N is the number of members in the final archive set, D is the sum of the distances from all solutions in the archive set to each objective function, and the particle that minimizes D is selected as the model parameter value.

[0036] Furthermore, the implementation method of the adaptive supergrid method is: for the adaptive grid of the k-dimensional target space, the target space is equally divided into N1×N2×…×N k Grid, where the network module length of the i-th target is n i =(f i max -f i min ) / N,(i=1,2,3),where f i max 、f i min is the maximum and minimum value of the particle on the i-th objective function, N is the number of members in the current iteration of the Archive set; the position p of the non-inferior solution in the grid space i It can be determined as p i =mod((f l,i -f i min ) / d i )+1, where f l,iis the ith objective function value corresponding to the lth non-inferior solution, mod is the rounding function, from which the number of non-inferior solutions contained in each grid can be calculated. The number of non-inferior solutions in a grid is used as the distribution density information of the particles, and the particles with the smallest distribution density are selected to guide the particles.

[0037] Furthermore, the objective function compares the output value of the unmanned ship model with the same input quantity with the actual output value of the unmanned ship measured by the unmanned ship status monitoring device to obtain a value that can evaluate the accuracy of the current model parameters. The smaller the value, the better the currently identified model parameters. Its form is as follows:

[0038]

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. The system and method are holistic and universal. Their integrity lies in providing a complete set of solutions from data measurement to parameter identification, and modularly encapsulating the more complicated identification algorithm, which can greatly shorten the time spent on parameter identification of unmanned ships; their universality is reflected in the fact that only by arranging relevant sensors on the unmanned ship used and conducting specified types of navigation tests, the terminal can obtain and process relevant data according to the communication module, thereby realizing model parameter identification. The system and method have no strict requirements on the hull shape and quality, and can be applied to unmanned ships of different models.

[0041] 2. The system and method adopt particle swarm optimization algorithm for parameter identification. Compared with other parameter identification algorithms, it has the advantages of fast search speed, fewer adjustment parameters, and less dependence on mathematical models. The particle swarm optimization algorithm has good stability and high precision, and has low requirements for the selection of initial parameters. It can well obtain parameters that can fully reflect the input and output characteristics of the unmanned ship model.

[0042] 3. The multi-objective particle swarm optimization algorithm adopted by the system and method, compared with the traditional single-objective particle swarm algorithm, establishes corresponding objective functions for each degree of freedom motion model and optimizes each objective function at the same time. It does not need to consider the weighted coefficients between each objective function, thus overcoming the subjectivity of the selection of model identification objective coefficients, thereby obtaining a set of Pareto optimal solution sets, and according to the relative distance strategy, selects a solution from the Pareto optimal solution set so that its corresponding objective functions are relatively optimal as the parameters of the multi-degree-of-freedom unmanned ship dynamics model; therefore, compared with the single-objective algorithm, this algorithm can solve the problem of difficult parameter identification of multi-degree-of-freedom coupling models.

[0043] 4. The system and method adopt an adaptive supergrid method to obtain particle density information and select particles with lower density as guide particles, thereby improving the diversity of the particle swarm, avoiding the Pareto optimal solution from falling into the local optimum, and improving the accuracy of the Pareto optimal solution.

[0044] 5. The system and method adopt a redundant combination of multiple sensors. The central processor monitors the multi-channel sensor data in real time. Once a failure occurs during the navigation process, redundant switching will be executed immediately to improve the safety and stability during the navigation process. At the same time, the redundant combination of multiple sensors can provide more measurement data, and high-precision real-time measurement data can be obtained through the Kalman filter algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the system composition structure of an embodiment of the present invention.

[0046] Figure 2 It is a system working flow diagram of an embodiment of the present invention.

[0047] Figure 3 It is a parameter identification flow chart in an embodiment of the present invention.

[0048] In the figure: 1-display and control terminal, 2-data communication module, 3-electronic compass, 4-GPS, 5-hull, 6-accelerometer, 7-central processing unit module, 8-data processing module, 9-gyroscope, 10-electric speed controller, 11-power supply, 12-propeller, 13-Hall sensor, 14-brushless motor. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0050] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0051] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0052] like Figure 1As shown, this embodiment provides a multi-parameter particle swarm identification system for an unmanned ship, including a state-measurable unmanned ship and a display and control terminal 1, wherein the state-measurable unmanned ship includes an unmanned ship and an unmanned ship state monitoring device, which is used to control the navigation of the unmanned ship in real time, monitor the state information of the unmanned ship and upload it to the display and control terminal 1; the display and control terminal 1 has a model building unit and an identification unit, the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, inputs the speed command into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned ship model, and compares the simulation information with the actual navigation state information of the unmanned ship to obtain forward speed error information, lateral drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identification parameters, and the identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model.

[0053] In this embodiment, the unmanned ship is a twin-propeller unmanned ship, including a hull 5, a power module 11, a power module 14, a data communication module and a central processing unit module. The central processing unit module receives the navigation path and request data commands issued by the display and control terminal 1 through the data communication module, controls the unmanned ship to navigate according to the set path, and controls the unmanned ship status monitoring device to obtain the status information of the unmanned ship; the unmanned ship status monitoring device includes a data acquisition module and a data processing module, and the data acquisition module includes a variety of sensing devices to collect the status information of the unmanned ship including speed, acceleration, propeller speed, positioning coordinates, heading angle and heading angle, and the acquired information is processed by the data processing module and sent to the central processing unit module, and then the status information of the unmanned ship is uploaded to the display and control terminal through the central processing unit module.

[0054] The power module adopts a large-capacity lithium battery, and realizes voltage stabilization and current measurement through an ammeter; the power module adopts a dual-propeller brushless motor as a propeller, and realizes steering through differential speed; the data communication module adopts wireless communication equipment, and realizes real-time information interaction with the display and control terminal through the lightweight message transmission protocol Mavlink; the central processing unit module is the control core of the unmanned ship, which is mainly responsible for information processing and interaction of each module, and realizes the motion control of the unmanned ship.

[0055] The data acquisition module obtains the navigation speed and acceleration of the unmanned ship through the accelerometer 6, obtains the actual value of the rotation speed of the propeller 12 through the Hall sensor 13, obtains the latitude and longitude information of the unmanned ship, that is, the positioning coordinates, through the GPS module 4, obtains the heading angle information through the electronic compass 3, obtains the heading angle information through the gyroscope 9, and transmits the obtained information to the data processing module; the data processing module processes the obtained information through the Kalman filter algorithm and then packages it and sends it to the central processing unit module.

[0056] In this embodiment, the display and control terminal includes a computer and a parameter identification software module installed on the computer. The parameter identification software module is mainly composed of a control unit, a display unit, a model building unit and an identification unit. The user controls the navigation path of the unmanned ship through the control unit, and displays the status information of the unmanned ship in real time through the display unit; the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, inputs the speed command into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned ship model, including simulated forward speed, simulated drift speed and simulated bow angular velocity, and compares the simulation information with the actual navigation status information of the unmanned ship to obtain forward speed error information, drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identified parameters; the identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model, thereby realizing the identification of the unmanned ship model parameters.

[0057] In this embodiment, the model building unit establishes a Fossen three-degree-of-freedom unmanned ship model, which is expressed as follows:

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] Among them, u, v, r are the longitudinal speed, transverse speed and bow turning speed of the unmanned ship respectively, m is the mass of the unmanned ship, I z is the moment of inertia about the z-axis, are the longitudinal and transverse additional masses, is the additional moment of inertia, X u , Y v , Y r 、N v 、N r is the linear hydrodynamic damping coefficient, X |u|u , Y |v|v , Y |r|v , Y |v|r , Y |r|r 、N |v|v 、N |r|v 、N |v|r 、N |r|r is the nonlinear hydrodynamic damping coefficient, c and d are the propeller thrust coefficients of the unmanned ship, and d pis the distance from the propeller to the center axis of the unmanned boat.

[0064] This embodiment also provides a multi-parameter particle swarm identification method for an unmanned ship based on the above system. The method is mainly implemented by an identification unit. The identification unit uses a multi-objective particle swarm algorithm to identify the parameters of the unmanned ship model. The specific method is:

[0065] 1) Randomly generate M particles in the search space as the initial population, and define the population as the unknown parameters that need to be identified in the model. The number of unknown parameters is the dimension dim of the particle population.

[0066] 2) Calculate the value of the objective function and obtain the individual extreme value of the particle.

[0067] 3) According to the Pareto dominance principle, the individual extreme values ​​of each particle are compared to obtain a set of non-inferior solutions for the particle swarm and store them in the Archive file set.

[0068] 4) From the Archive set, the particle crowding is estimated according to the adaptive hypergrid method, and the global optimal solution is selected by using the grid density information and the roulette method.

[0069] 5) Update the speed and position values ​​of each particle according to the speed and position iteration formula.

[0070] 6) Calculate the objective function value of the particle.

[0071] 7) Update individual extreme values.

[0072] 8) Update Archive file set: Adopt adaptive super grid method and use density information to maintain Archive file set so that the maximum value of the archive set remains unchanged.

[0073] 9) If the number of iterations does not reach the set value, return to 3) and continue execution.

[0074] 10) The iteration ends, and the Archive file set is the optimal Pareto solution. In order to obtain a compromise solution that is relatively better than the three objective functions, a selection strategy is set to select the relatively optimal solution from the optimal Pareto solution as the final solution.

[0075] The Pareto dominance principle is defined as follows:

[0076] If a solution s=(s1,s2,…,s n ) dominates another solution u=(u1,u2,…,u n ), denoted as s>u, if and only if:

[0077]

[0078]

[0079] In the formula, n represents the dimension of the solution, and m represents the number of objective functions to be solved. For a solution s, if there is no other solution u>s in the search space, then s becomes the Pareto optimal solution. The geometry of all Pareto optimal solutions is called the Pareto optimal set. The Pareto frontier is the set of optimal non-dominated solutions that satisfy the constraints and all objective functions.

[0080] The implementation method of the adaptive supergrid method is:

[0081] For the adaptive grid of k-dimensional target space, the target space is divided into N1×N2×…×N k Grid, where the network module length of the i-th target is n i =(f i max -f i min ) / N,(i=1,2,3),where f i max 、f i min is the maximum and minimum value of the particle on the i-th objective function, N is the number of members in the current iteration of the Archive set; the position p of the non-inferior solution in the grid space i It can be determined as p i =mod((f l,i -f i min ) / d i )+1, where f l,i is the ith objective function value corresponding to the lth non-inferior solution, mod is the rounding function, from which the number of non-inferior solutions contained in each grid can be calculated. The number of non-inferior solutions in a grid is used as the distribution density information of the particles, and the particles with the smallest distribution density are selected to guide the particles.

[0082] The objective function compares the output value of the unmanned ship model with the same input quantity with the actual output value of the unmanned ship measured by the unmanned ship status monitoring device to obtain a value that can evaluate the accuracy of the current model parameters. The smaller the value, the better the currently identified model parameters. Its form is as follows:

[0083]

[0084] like Figure 2As shown, when the system is working, by setting different left and right propeller 12 speeds, the unmanned boat hull 5 is controlled to achieve different actions. First, the desired left and right propeller 12 speeds are set on the display and control terminal 1, and the command will be transmitted to the unmanned boat through the data transmission module 2. After receiving the relevant command, the unmanned boat starts self-testing, and the self-test content includes the attitude of the unmanned boat, the power supply and the sensor status, etc. If the self-test fails, the status monitoring is performed cyclically. After the self-test passes, the unmanned boat enters the navigation mode and performs parameter initialization, mainly including the initialization of sensor-related parameters and the initialization of the state parameters of the unmanned boat. After completing the parameter initialization, the values ​​of the left and right Hall sensors 13 are read, and the real-time speeds of the left and right propellers 12 are obtained by calculation. At the same time, the sensor data such as the electronic compass 3, GPS4, accelerometer 6 and gyroscope 9 are read, and the data is filtered and processed by the data processing module 8 and sent to the central processing unit module 7. The control of the propeller 12 speed is mainly realized by the PID algorithm, so the next step is to call the existing PID library function to calculate, and the PWM value is obtained. The electric adjustment 10 controls the motor 14 according to the obtained PWM value. At the same time, the data obtained by the communication module determines whether a stop command is received. If there is no stop command, the sensor data is recorded cyclically and the speed of the propeller 12 is controlled. After receiving the stop command, the unmanned boat transmits all experimental data to the display and control terminal 1 for subsequent parameter identification.

[0085] In this embodiment, the workflow of the multi-objective particle swarm algorithm is as follows: Figure 3 As shown, first, initialize the particle swarm, calculate the values ​​of each objective function of the particle swarm, and calculate the individual optimal X pbest , and then according to the Pareto principle, from X pbest The Pareto optimal solution is selected and stored in an external archive. Then, an iterative loop is entered: the grid is automatically divided according to the number of solutions in the external archive according to the adaptive hypergrid method; the number of particles in the same grid is calculated and used as the individual density value of the particles, and particles with the same individual density value are recorded as the same individual density level. At the same time, the proportion of each individual density level to the total sum of individual density levels is calculated, and then the individual density level is selected by roulette, and a particle is randomly selected from the selected individual density level as the guide particle X. gbest ; Then, update the particle X(t+1) through the iterative formula and compare it with X pbest Compare whether Pareto is dominant, if so, update X pbest , otherwise it remains unchanged, and then the Pareto dominance principle is used to select the Pareto optimal solution and update the external file; then, the loop end condition is judged, if the condition is met, the external file is output, otherwise, the loop continues; finally, a set of relatively optimal compromise solutions is selected from the final external file as the model parameter value.

[0086] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A multi-parameter particle swarm identification system for unmanned ships, characterized in that: It comprises a state-measurable unmanned ship and a display and control terminal, wherein the state-measurable unmanned ship comprises an unmanned ship and an unmanned ship state monitoring device, which is used for real-time control of the navigation of the unmanned ship, monitoring the state information of the unmanned ship and uploading it to the display and control terminal; the display and control terminal is provided with a model building unit and an identification unit, wherein the model building unit establishes a three-degree-of-freedom motion model of the unmanned ship, inputs a rotation speed instruction into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned ship model, and compares the simulation information with the actual navigation state information of the unmanned ship to obtain forward speed error information, lateral drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identification parameters, and the identification unit adopts a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the value of the objective function, that is, to obtain the optimal parameter value of the unmanned ship model; The model building unit establishes a Fossen three-degree-of-freedom unmanned ship model, and its expression is: Among them, u, v, r are the longitudinal speed, transverse speed and bow turning speed of the unmanned ship respectively, m is the mass of the unmanned ship, I z is the moment of inertia about the z-axis, are the longitudinal and transverse additional masses, is the additional moment of inertia, X u , Y v , Y r 、N v 、N r is the linear hydrodynamic damping coefficient, X |u|u , Y |v|v , Y |r|v , Y |v|r , Y |r|r 、N |v|v 、N |r|v 、N |v|r 、N |r|r is the nonlinear hydrodynamic damping coefficient, c and d are the propeller thrust coefficients of the unmanned ship, and d p is the distance from the propeller to the center axis of the unmanned boat; The identification unit uses a multi-objective particle swarm algorithm to identify the parameters of the unmanned ship model, including the following steps: Initialize the particle swarm and archive set: randomly generate M particles in the search space as the initial population, define the population as the unknown parameters that need to be identified in the model, calculate the objective function value of each particle, and obtain the initial Archive set according to the Pareto dominance principle; Searching for guiding particles: From the initial Archive set, estimate the particle crowding according to the adaptive hypergrid method, and select guiding particles by using density information combined with the roulette method; Update particle swarm and archive set: Update each particle according to the speed and position iteration formula, continue to calculate the objective function value of each particle, update the Archive archive set, and seek the optimal generation of guiding particles and individuals; The particle swarm and the archive set are updated cyclically until the error between the simulation information and the measured navigation state information is minimized. Then, the relatively optimal solution is selected from the final archive set as the model parameter value according to the selection strategy.

2. The unmanned ship multi-parameter particle swarm identification system according to claim 1 is characterized in that: The unmanned ship is a twin-propeller unmanned ship, including a hull, a power module, a power module, a data communication module and a central processing unit module. The central processing unit module receives the navigation path and request data commands issued by the display and control terminal through the data communication module, controls the unmanned ship to navigate according to the set path, and controls the unmanned ship status monitoring device to obtain the status information of the unmanned ship; the unmanned ship status monitoring device includes a data acquisition module and a data processing module. The data acquisition module includes a variety of sensing devices to collect the status information of the unmanned ship including speed, acceleration, propeller speed, positioning coordinates, heading angle and heading angle. The acquired information is processed by the data processing module and sent to the central processing unit module, and then the status information of the unmanned ship is uploaded to the display and control terminal through the central processing unit module.

3. The unmanned ship multi-parameter particle swarm identification system according to claim 2 is characterized in that: The power module adopts a large-capacity lithium battery, and realizes voltage stabilization and current measurement through an ammeter; the power module adopts a dual-propeller brushless motor as a propeller, and realizes steering through differential speed; the data communication module adopts wireless communication equipment, and realizes real-time information interaction with the display and control terminal through the lightweight message transmission protocol Mavlink; the central processing unit module is the control core of the unmanned ship, which is mainly responsible for information processing and interaction of each module, and realizes the motion control of the unmanned ship.

4. The unmanned ship multi-parameter particle swarm identification system according to claim 2 is characterized in that: The data acquisition module obtains the navigation speed and acceleration of the unmanned ship through an accelerometer, obtains the actual value of the propeller speed through a Hall sensor, obtains the latitude and longitude information of the unmanned ship, that is, the positioning coordinates, through a GPS module, obtains the heading angle information through an electronic compass, obtains the course angle information through a gyroscope, and transmits the obtained information to the data processing module; the data processing module processes the obtained information through a Kalman filter algorithm and then packages it and sends it to the central processing unit module.

5. The unmanned ship multi-parameter particle swarm identification system according to claim 1 is characterized in that: The display and control terminal includes a computer and a parameter identification software module installed on the computer. The parameter identification software module is mainly composed of a control unit, a display unit, a model building unit and an identification unit. The user controls the navigation path of the unmanned boat through the control unit, and displays the status information of the unmanned boat in real time through the display unit; the model building unit establishes a three-degree-of-freedom motion model of the unmanned boat, inputs the speed instruction into the three-degree-of-freedom motion model to obtain simulation information of the three-degree-of-freedom unmanned boat model, including simulated forward speed, simulated lateral drift speed and simulated bow angular velocity, and compares the simulation information with the actual navigation state information of the unmanned boat to obtain forward speed error information, lateral drift error information and bow angular velocity error information, and then establishes an objective function to evaluate the quality of the identification parameters; The identification unit uses a multi-objective particle swarm algorithm to find a set of unmanned ship model parameters to minimize the objective function value, that is, to obtain the optimal parameter value of the unmanned ship model, thereby realizing the identification of the unmanned ship model parameters.

6. The unmanned ship multi-parameter particle swarm identification system according to claim 1 is characterized in that: The multi-objective particle swarm algorithm initializes the particle swarm in the dim-dimensional search space and generates M initial particles: X=rand(M,dim)*(X max -X min )+X min Where, X max and X min are the upper and lower bounds of the particle swarm search space, rand(M, dim) generates M dim-dimensional particles, and each element of the particle is a random decimal between [0, 1]; From the initialized particle swarm, find a set of non-inferior solutions according to the Pareto dominance principle; Seeking guiding particles: Divide the objective function space into N1×N2×N3 networks, where the network module length of the i-th objective is n i for: n i =(f i max -f i min ) / N,(i=1,2,3) In the formula, f i max 、f i min is the maximum and minimum value of the particle on the i-th objective function, and N is the number of members in the current iteration of the Archive set; The position p of the non-inferior solution in the grid space i for: p i =mod((f l,i -f i min ) / d i )+1 In the formula, f l,i is the ith objective function value corresponding to the lth non-inferior solution, mod is the rounding function, and the number of non-inferior solutions contained in each grid is obtained; the number of non-inferior solutions in a grid is used as the distribution density information of the particles, and the particles with the smallest distribution density are selected to guide the particles in combination with roulette; Update particle swarm and archive set: Repeat the following formula to update the speed and position of each particle: Where V i (k+1) is the velocity of the ith particle at the k+1th iteration, w is the inertia factor, c1 and c2 are the individual learning factor and the overall learning factor, respectively, and rand() is a function that generates random numbers between 0 and 1. is the historical optimal position of the i-th particle at the kth iteration, X gbest (k) is the historical optimal position of all particles at the kth iteration, that is, the current guiding particle; the selection formula for selecting the relatively optimal solution in the final archive set is: Where N is the number of members in the final archive set, D is the sum of the distances from all solutions in the archive set to each objective function, and the particle that minimizes D is selected as the model parameter value.

7. The unmanned ship multi-parameter particle swarm identification system according to claim 1 is characterized in that: The implementation method of the adaptive supergrid method is as follows: for the adaptive grid of the k-dimensional target space, the target space is equally divided into N1×N2×…×N k Grid, where the network module length of the i-th target is n i =(f i max -f i min ) / N,(i=1,2,3),where f i max 、f i min is the maximum and minimum value of the particle on the i-th objective function, N is the number of members in the current iteration of the Archive set; the position p of the non-inferior solution in the grid space i It can be determined as p i =mod((f l,i -f i min ) / d i )+1, where f l,i is the ith objective function value corresponding to the lth non-inferior solution, mod is the rounding function, from which the number of non-inferior solutions contained in each grid can be calculated. The number of non-inferior solutions in a grid is used as the distribution density information of the particles, and the particles with the smallest distribution density are selected to guide the particles.

8. The unmanned ship multi-parameter particle swarm identification system according to claim 1 is characterized in that: The objective function compares the output value of the unmanned ship model with the same input quantity with the actual output value of the unmanned ship measured by the unmanned ship status monitoring device to obtain a value that can evaluate the accuracy of the current model parameters. The smaller the value, the better the currently identified model parameters. Its form is as follows:

Images