CN105426341A - Parameter identification method and apparatus for complex object - Google Patents

Abstract

The present invention discloses a parameter identification method and apparatus for a complex object. The method comprises: initially setting a position of each particle; calculating a position of each particle in an iterative process and determining a particle which has a neighborhood historical optimal position by using a stimulus-response data set corresponding to each particle and according to the position of each particle; in an excellent-to-poor order of the neighborhood historical optimal positions, selecting multiple particles from particles that have never selected, and carrying out error evaluation on the neighborhood historical optimal positions of the currently selected multiple particles by using multiple different stimulus-response data sets; and determining whether a particle which meets a preset error requirement exists in the currently selected particles according to an evaluation result; and if yes, using a parameter value represented by the neighborhood historical optimal position of the particle, which meets the preset error requirement, as a parameter identification result; and otherwise returning to the evaluation operation if the number of particles that are selected reaches a preset number, or returning to the initial setting operation when the number of particles that are not selected does not reach the preset number.

Description

Parameter identification method and device for complex object

Technical Field

The present invention relates to a parameter identification technology, and more particularly, to a method and an apparatus for identifying parameters of a complex object.

Background

With the continuous and deep research and development of human beings in the fields of the earth, the outer space and the like, the use of complex objects is increased day by day, and for example, a mechanical arm with multi-arm joints arranged in a spacecraft such as a satellite is a complex object.

In practical application of a complex object, behavior prediction and parameter identification of the complex object are very important, for example, in an in-orbit operation process of a spacecraft, an unknown object is often grabbed by using a mechanical arm, in a process of controlling the mechanical arm to grab the unknown object, inertial parameters of the unknown object are accurately and timely identified, otherwise, the grabbing of the unknown object can cause unpredictable changes in mass distribution of the spacecraft, and certain difficulty is brought to subsequent planning of an operation path of the mechanical arm.

Because a complex object usually has the characteristics of multiple parameters, multiple states, multiple outputs, strong nonlinearity and the like, the existing parameter identification methods (such as a least square method, a maximum likelihood estimation method, a Newton method and the like) aiming at the linearity problem or the linearity-like problem are difficult to be applied to the parameter identification process of the complex object; and intelligent algorithms such as a neural network, a genetic algorithm or a particle swarm algorithm and the like are widely applied in the parameter identification process of the complex object.

The parameter identification using the intelligent algorithm generally optimizes the parameter to be identified in the identification model through an objective optimization function, so that the response error of the identification model and the actual object is minimized under the same excitation condition, and a parameter identification result is obtained. The objective optimization function is generally an objective optimization function based on the response error or the relative response error, such as a weighted euclidean distance for calculating the relative error or the absolute error of each channel response by using the objective optimization function.

The inventor discovers that in the process of implementing the invention: the existing intelligent algorithm is easy to have the problem of local minimum value, and the problem of local minimum value can have adverse effect on the parameter identification precision; in addition, with the increase of the complexity of the complex object, the number of the parameters to be identified is increased, which can greatly increase the calculation amount and the complexity in the parameter identification process; therefore, how to effectively control the amount of calculation and complexity in the parameter identification process and make the parameter identification have better identification accuracy is a significant concern in the parameter identification technology.

Disclosure of Invention

In view of the above, the present invention has been made to provide a method and apparatus for parameter recognition of a complex object that overcomes or at least partially solves the above problems.

According to an aspect of the present invention, there is provided a method for identifying parameters of a complex object, the method comprising: an initial setting step: different sets of parameter values are given to the parameter set to be identified so as to initially set the position of each particle; iteration step: calculating the position of each particle in the iterative process, and determining the particle with the optimal neighborhood history position according to the position of each particle by utilizing one excitation response data set corresponding to each particle; evaluation step: selecting a plurality of particles from the unselected particles with the neighborhood history optimal positions in the sequence from the optimal neighborhood history optimal positions to the inferior neighborhood history optimal positions, and respectively performing error evaluation on the neighborhood history optimal positions of the plurality of currently selected particles by taking different excitation response data sets as screening evaluation sequences; a judging step: and judging whether the particles meeting the preset error requirement exist in the currently selected particles according to the error evaluation result, if so, taking the parameter value represented by the neighborhood historical optimal position of the particles meeting the preset error requirement as a parameter identification result, otherwise, returning to the evaluation step when the unselected particles reach the preset number, and returning to the initial setting step when the unselected particles do not reach the preset number.

According to another aspect of the present invention, there is provided an apparatus for identifying parameters of a complex object, the apparatus comprising: the initial setting module is suitable for endowing different groups of parameter values for the parameter group to be identified so as to initially set the position of each particle; the iteration module is suitable for calculating the position of each particle in the iteration process and determining the particle with the optimal neighborhood history position according to the position of each particle by utilizing one excitation response data set corresponding to each particle; the evaluation module is suitable for sequentially selecting a plurality of particles from unselected particles with the neighborhood history optimal positions according to the sequence from the excellence to the inferiority of the neighborhood history optimal positions, and respectively performing error evaluation on the neighborhood history optimal positions of the plurality of currently selected particles by taking a plurality of different excitation response data sets as screening evaluation sequences; and the judging module is suitable for judging whether the particles meeting the preset error requirement exist in the currently selected particles according to the error evaluation result, if so, the parameter value represented by the neighborhood historical optimal position of the particles meeting the preset error requirement is used as a parameter identification result, otherwise, when the unselected particles reach the preset number, the evaluating module is triggered to continue to execute the selection and error evaluation operation, and when the unselected particles do not reach the preset number, the initial setting module is triggered to execute the initial setting operation.

The parameter identification method and the parameter identification device for the complex object provided by the invention at least have the following advantages and beneficial effects: the method comprises the steps of respectively allocating different excitation response data sets to each particle, determining the particle with the optimal neighborhood history position by using the excitation response data set corresponding to each particle (usually, one particle corresponds to the same excitation response data set in the iteration process of the previous time), and realizing the initial search and screening of all the particles under the condition that the calculated amount of parameter identification is as small as possible; the particles are continuously selected according to the sequence from superior to inferior of the neighborhood historical optimal position, and each selected particle is subjected to error evaluation by utilizing a plurality of different excitation response data sets, so that the global minimum value can be quickly and accurately determined, the phenomenon of large increase of calculated amount caused by using a plurality of different excitation response data sets is effectively controlled while the phenomenon of local minimum value is avoided, and meanwhile, the large increase of the parameter identification complexity cannot be caused due to the increase of the number of the parameters to be identified; therefore, the technical scheme provided by the invention has better parameter identification precision while effectively controlling the calculated amount and complexity in the parameter identification process.

The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings of the present embodiment are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:

FIG. 1 is a flowchart illustrating a method for identifying parameters of a complex object according to a first embodiment of the present invention;

FIG. 2 is a flowchart illustrating a method for identifying parameters of a complex object according to a second embodiment of the present invention;

FIG. 3 is a schematic diagram of an identification result of a simulation experiment according to a second embodiment of the present invention;

FIG. 4 is a diagram illustrating a normalized error of the prior art in a simulation experiment according to the second embodiment of the present invention;

FIG. 5 is a schematic diagram of the identification result of the method of the present invention in the simulation experiment according to the second embodiment of the present invention;

FIG. 6 is a diagram illustrating normalized errors of the present invention in a simulation experiment according to the second embodiment of the present invention;

fig. 7 is a schematic structural diagram of a parameter identification apparatus for a complex object according to a third embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

Embodiment one, a method for identifying parameters of a complex object.

The present embodiment is a method for identifying parameters of a complex object, where the number of parameters to be identified is usually multiple, and the multiple parameters to be identified form a parameter group to be identified; of course, the parameter set to be identified in the present embodiment may also include only one parameter to be identified. The method of this embodiment is described below with reference to fig. 1.

In fig. 1, S100, an initial setting step (which may also be referred to as a particle initialization step): and giving different sets of parameter values to the parameter set to be identified so as to initially set the position of each particle.

Specifically, the parameter set to be recognized in this embodiment includes a plurality of parameters to be recognized (e.g., two, three, or four parameters), for example, the parameter set to be recognized includes the mass and the centroid position (three-dimensional coordinate value) of the target to be recognized. It should be noted that, in practical applications, the number of the parameters to be identified included in the parameter group to be identified may be more than four. The embodiment does not limit the specific content of the parameter to be identified included in the parameter group to be identified.

In this embodiment, parameter values may be given to each parameter to be identified in the parameter group to be identified for multiple times according to a certain value range, so that all particles correspond to different parameter values, that is, the positions of different particles are different.

In this embodiment, different sets of parameter values may be assigned to the parameter group to be identified by randomly generating parameter values, for example, randomly generating one parameter value within a predetermined numerical range of the mass, randomly generating three parameter values within a predetermined numerical range of the centroid position, and assigning the four parameter values to four parameters to be identified in the parameter group to be identified, thereby performing initialization setting on the position of one particle; after the above processes of random generation and assignment are repeatedly executed for many times, the positions of a plurality of particles are initialized and set respectively. The number of particles in this embodiment may be several hundred, and of course, the number of particles may be more. The number of particles used in the parameter identification process is not limited by the present embodiment.

At the same time as the initial setting of the position of each particle, an initial velocity should be set for each particle, i.e. the velocity at which each particle is initialized. The present embodiment may employ various existing methods to perform initialization setting on the position and the velocity of the particle, and the present embodiment does not limit the specific implementation manner of initializing the position and the velocity of the particle.

S110, iteration step: and calculating the position of each particle in the iterative process, and determining the particle with the optimal position of the neighborhood history by utilizing one excitation response data set corresponding to each particle according to the position of each particle.

Specifically, the present embodiment may utilize a particle swarm algorithm (e.g., an improved particle swarm algorithm) to implement the above iteration step. It is noted that other algorithms may be adopted to implement the iteration step in the present embodiment; in addition, no matter the particle swarm algorithm or other algorithms are adopted, the local minimum value and the global minimum value do not need to be distinguished in the embodiment, and only the neighborhood history optimal position set of the screened particles can contain the global minimum value.

The embodiment utilizes an improved particle swarm algorithm to realize the above iteration steps, so as to obtain a specific example of the particle with the neighborhood history optimal position as follows:

first, the position of each particle during the next iteration (i.e., the g +1 th generation) is calculated using the following equation (1).

x_n(g+1)＝x_n(g)+v_n(g +1) formula (1)

In the above formula (1), x_n(g +1) represents the position of the g +1 th generation particle n, x_n(g) Denotes the position of the g-th generation particle n, v_n(g +1) denotes the velocity of the g +1 th generation of particles N, and N denotes the respective particle number, N being 1, 2.

V in the above formula (1)_n(g +1) can be obtained by calculation from the following formula (2):

$v_n(g+1)={\mathrm{\ωv}}_n\left(g\right)+c_1\mathrm{\ξ}\[x_n^{\left(i\right)}\left(g\right)-x_n\left(g\right)\]+c_2\mathrm{\η}\[x_n^{\left(g\right)}\left(g\right)-x_n\left(g\right)\]$ formula (2)

In the above formula (2), v_n(g) Denotes the velocity of the g-th generation of particles n, ω denotes the inertial weight, c1 and c2 denote learning factors, ξ denotes random vectors within a predetermined interval,denotes an individual history optimal position of the particle n by the g-th generation, and i denotes an algebra corresponding to the individual history optimal position of the particle n, η denotes a random vector within a predetermined interval,represents the neighborhood history optimal position, x, in the neighborhood where the g-th generation of particle n is located_n(g) Denotes the position of the g-th generation of particles N, N denotes the respective particle number, and N is 1, 2.

Secondly, the positions of the particles in the iteration process are subjected to error evaluation by utilizing an excitation response data group corresponding to each particle, and neighborhood historical optimal position information of the neighborhood where the particles are located and individual historical optimal position information of each particle are recorded according to the result of the error evaluation.

In this embodiment, one particle corresponds to one excitation response data set, and different particles correspond to different excitation response data sets; that is, the excitation response data set used by a particle during a past iteration is generally the same, and the excitation response data set used by a particle during a past iteration is different from the excitation response data set used by any other particle during a past iteration. It should be noted that, in this step, different sets of excitation response data corresponding to different particles are based on sets of excitation response data of different configurations of rigid bodies (e.g., robotic arms) in a multi-rigid-body system, for example, for a multi-rigid-body system including a robotic arm, one set of excitation response data is based on a set of excitation response data generated by rotating a robotic arm to the left, another set of excitation response data is based on a set of excitation response data generated by rotating a robotic arm to the right, and yet another set of excitation response data is based on a set of excitation response data generated by rotating a 1 st arm segment of the robotic arm. The configurations of the rigid bodies (e.g., robotic arms) corresponding to different sets of excitation response data should have large differences, i.e., the sets of excitation response data corresponding to different particles should have large differences.

In this embodiment, the following formula (3) may be used to perform error evaluation on the positions of the particles in the current iteration process:

formula (3)

In the above-mentioned formula (3),to use x_n(particle n) calculating the obtained complex object response result value,representing the complex object response observation for particle n,a calculation symbol representing the squared euclidean distance.

It is to be noted that in the above formula (3)Is obtained by calculation using the following formula (4):

formula (4)

In the above formula (4), f [. sup. ]]Representing a complex object computational dynamics model function, d (t)_n) Represents t_nThe complex object state matrix corresponding to the moment, J (t)_n) Represents t_nComplex object excitation matrix, x, corresponding to time of day_n(g) Denotes the position, t, of the g-th generation particle n_nThe complex object state and the excitation corresponding to the moment are the excitation response data set used by the particle n.

After the error evaluation result corresponding to the position of each particle in the current iteration process is obtained, the neighborhood history optimal position of the neighborhood where each particle is located and the individual history optimal position of each particle can be determined by comparing the error evaluation results of each particle. The neighborhood history optimal position information of the neighborhood where each particle is located and the individual history optimal position information of each particle recorded in this embodiment are not only applied to the next iteration process (for example, substituted into the above formula (2)), but also applied to the evaluation step of S120 described below.

Finally, according to the obtained error evaluation result of the current iteration, judging whether particles with errors meeting the preset requirements exist and whether the current iteration frequency reaches the preset frequency, and if the particles with errors meeting the preset requirements exist or the current iteration frequency reaches the preset frequency, performing the following evaluation step S120; otherwise, the next iteration process is carried out, namely the step of respectively calculating the positions of the particles in the current iteration process by using the formula (1) is returned.

In the above iterative process of the present embodiment, the error evaluation result of some particles in the past iterative process is likely to be always poor (i.e. the error is large), and such particles may be referred to as poor performing particles; while some particles may always have better error evaluation results (i.e., smaller errors) during the past iterations, such particles may be referred to as better performing particles.

For the particles with poor performance, the embodiment should shift the particles to the vicinity of the particles with good performance in the above iteration process; one specific example is:

a certain number of iterations per interval (e.g. per interval G)_reGeneration), the worst K among all particles can be evaluated_reN particles x_l(g) The transformation (i.e. changing the current position and current velocity of the particle) is performed according to the following equation (5) to achieve the transfer of the particle, G_reMay be a preset constant, K_reRepresents the restart particle ratio, and K_re<1, N represents the total number of particles.

$x_l\left(g\right)=x_r^{\left(g\right)}\left(g\right)+\mathrm{\&xi;};$ Formula (5)

In the above-mentioned formula (5),indicating the neighborhood history optimal position of the particle r randomly selected from the g-th generation and having an error rating meeting the predetermined requirement, and ξ indicating a random vector within a predetermined interval.

It should be noted that the neighborhood in the present embodiment is a region composed of a plurality of particles that are close to each other, for example, the neighborhood is N whose euclidean distance from a particle is the smallest_nb(g) AnParticle constitution of N in_nb(g) Can be expressed by the following formula (6);

$N_{nb}\left(g\right)=\frac{g}{G}{K}_{nb}N$ formula (6)

In the above equation (6), G represents the current iteration number, G represents the total iteration number, K_nbRepresents the neighborhood particle fraction (i.e., the fraction of particles in the neighborhood over the total number of particles), and K_nb<1, N represents the total number of particles.

From the above equation (6), the number N of particles constituting the neighborhood_nb(g) Will gradually increase as the number of iterations increases.

In addition, the embodiment may sort the neighborhood history optimal positions of the particles before jumping to S120, and in general, the embodiment may sort the neighborhood history optimal positions of the particles according to the neighborhood history optimal position information of the currently recorded particles in an order from good to bad (i.e., from small to large in error) of the neighborhood history optimal positions.

S120, evaluation step: and sequentially selecting a plurality of particles from the unselected particles with the neighborhood history optimal positions according to the sequence from the excellence to the inferiority of the neighborhood history optimal positions, and respectively performing error evaluation on the neighborhood history optimal positions of the currently selected particles by taking different excitation response data sets as screening evaluation sequences.

Specifically, in the case where the neighborhood history optimal positions of the particles have been sorted in S110:

first, from the sorted sequence unselectedSelecting a predetermined number of particles from the selected particles (e.g. selecting N)_fIndividual particles);

secondly, taking a plurality of different excitation response data groups as screening evaluation sequences respectively, and performing error evaluation on the neighborhood history optimal positions of the currently selected particles respectively by using the following formula (7);

$E_n=\frac{1}{N_r}\underset{u=1}{\overset{N_r}{\&Sigma;}}\left|\right|f\&lsqb;d\left(t_u\right),J{\left(t_u\right)}_u;x^{\left(g\right)}\left(G\right)\&rsqb;-{\&lsqb;{\tilde{p}}_1\left(t_u\right),...,{\tilde{p}}_z\left(t_u\right)\&rsqb;}^T|{|}_2$ formula (7)

In the above formula (7), E_nDenotes the result of error evaluation, N_rA group number representing a plurality of excitation response data groups, | | Y | | | non-calculation₂The calculation symbol, f [. multidot. ] representing the squared Euclidean distance]Representing a complex object computational dynamics model function, d (t)_u) Represents t_uThe complex object state matrix corresponding to the moment, J (t)_u) Represents t_uA corresponding complex object excitation matrix is generated,represents the optimal position of the neighborhood history of the currently selected u-th particle by the g-th generation,represents t_uAnd responding the observed value by the complex object corresponding to the moment.

After the error evaluation is performed on the neighborhood history optimal positions of the currently selected particles, the following step S130 is performed.

It should be particularly noted that, in this step, the excitation response data sets used for performing error evaluation on the neighborhood history optimal positions of different particles should be the same; that is, the present embodiment performs error calculation on the neighborhood history optimal positions of different particles using the same screening evaluation sequence. In addition, the plurality of different excitation response data sets in the present embodiment are excitation response data sets based on different configurations of rigid bodies (e.g., robotic arms) in a multi-rigid-body system, such as one in which a first excitation response data set is an excitation response data set generated based on left rotation of a robotic arm, a second excitation response data set is an excitation response data set generated based on right rotation of a robotic arm, a third excitation response data set is an excitation response data set generated based on 1 st arm joint rotation of a robotic arm, and so on. The configurations of the rigid bodies (such as mechanical arms) corresponding to different excitation response data sets should have large differences, namely, the screening evaluation sequences should have large differences.

S130, a judging step: and judging whether the particles meeting the preset error requirement exist in the currently selected particles according to the error evaluation result, if so, taking the parameter value represented by the neighborhood historical optimal position of the particles meeting the preset error requirement as a parameter identification result, otherwise, returning to the step S120 when the unselected particles can reach the preset number, and returning to the step S100 when the unselected particles can not reach the preset number.

Specifically, in a plurality of E calculated by the above formula (7)_nIn, if min (E)_n) If the preset error requirement is met, min (E) can be adjusted_n) Neighborhood history optimal position of corresponding particleThe represented parameter value is used as a parameter identification result; if min (E)_n) The preset error requirement is not met, and the quantity of unselected particles in the neighborhood history optimal position sequence of the sequenced particles is not less than N_fReturning to S120 to continue to select the subsequent N from the neighborhood history optimal position sequence of the ordered particles_fA plurality of particles; if min (E)_n) The preset error requirement is not met, and the quantity of unselected particles in the neighborhood history optimal position sequence of the sequenced particles is less than N_fAnd returning to the step S100 to change the position and the speed of each particle which are initially set, and performing iteration by using the reset position and speed of each particle.

As can be seen from the above description, in the embodiment, a plurality of different excitation response data sets are used to perform error evaluation on the neighborhood history optimal position of the particle, and for the particle trapped in a local minimum value rather than a global minimum value, the local minimum value always shows a larger error for some excitation response data sets with larger differences, so that a parameter identification result is not generated according to the local minimum value; thereby fully verifying the naive physical connotation of 'true and false'.

Embodiment two, a parameter identification method of a complex object.

In this embodiment, a parameter identification process of this embodiment is described by taking a robot arm installed in a spacecraft as an example of a complex object.

It should be noted that the parameter identification method provided in the present embodiment is also applicable to cases where the complex object is represented in other forms.

The mechanical arm arranged in the spacecraft usually has a plurality of arm sections, and a star arm coupling system is usually formed between the spacecraft and the mechanical arm, for example, the mechanical arm comprises q-1 arm sections (such as arm sections comprising connecting rods and joints), and the q-1 arm sections are respectively connected (such as hinged) in sequence through joints (total q-1 joints, which can also be called rotary joints) on the q-1 arm sections and are finally fixed on the spacecraft in a continuous mode, so that the star arm coupling system with the q-1 arm sections is formed.

Under the condition that the target to be identified (namely an unknown target) is grabbed by the mechanical arm, the tail end arm section and the target to be identified can be regarded as being in rigid connection, so that the spacecraft, the mechanical arm and the target to be identified can be regarded as a multi-rigid system. Under the condition of neglecting the influence factors such as earth oblateness, atmospheric damping, sunlight pressure, earth magnetic field and the like, the multi-rigid-body system meets the angular momentum conservation principle of the multi-rigid-body system. In addition, the on-orbit flight state of the star-arm coupling system under the action of trackless control thrust and attitude-free control moment can be set, namely the position and attitude of the main star are not controlled by a booster or other external forces.

In the application scenario, the parameters to be identified in this embodiment may be inertial parameters of the target to be identified, that is, the mass, the centroid position, and the moment of inertia of the target to be identified; however, since the inertia moment of the object to be recognized can be expressed as a function with respect to the mass and the centroid position of the object to be recognized, the parameter to be recognized in the present embodiment can be converted into the mass and the centroid position of the object to be recognized, and the centroid position of the object to be recognized is generally a three-dimensional coordinate of the centroid, so that the parameter to be recognized in the application scenario is converted into four scalars, that is, the parameter group to be recognized in the present embodiment includes four scalars.

The method for identifying parameters of a complex object in this embodiment mainly includes two parts, one part is an iterative computation part, and the other part is a composite evaluation part, and the method in this embodiment is described in detail below with reference to fig. 2.

In FIG. 2, S10, position x where all particles are initialized_i(0) And velocity v_i(0) The positions and velocities of all particles are initialized as described above using equations (1) and (2).

S11, judging whether the current iteration number is G or not_reAnd judging whether the number of the particles with poor performance is not less than K_reN, if the current iteration number is G_reIs not less than K, and the number of particles currently exhibiting poor performance is not less than K_reN, then to S12; and if the current iteration number is not G_reIs an integer multiple of K or the number of particles currently exhibiting poor performance is less than K_reN, then to S13.

Whether the particle is poor in performance in this embodiment may be determined according to whether the error of the particle reaches a certain error value, for example, in the first iteration process, the performance of each particle may be set to be good, and in the subsequent iteration process, the poor-performing particle may be identified by using the error of each particle calculated by the following formula (8).

And S12, randomly distributing the particles with poor performance to the periphery of the particles with better performance, and updating the neighborhood historical optimal position of the particles and the individual historical optimal position of the particles.

S13, evaluating the behavior of each particle using the corresponding one of the sets of excitation response data for each particle, i.e., calculating the error of each particle using the corresponding one of the sets of excitation response data for each particle, to S14.

Specifically, the error of each particle can be calculated by the following formula (8):

formula (8)

In the above-mentioned formula (8),indicates the use of x_n(i.e. particle n) calculates the angular velocity value of the derived spacecraft attitude angle α,an angular velocity observation representing the spacecraft attitude angle α for particle n,indicates the use of x_nThe angular velocity values of the obtained attitude angles β of the spacecraft are calculated,an angular velocity observation representing the spacecraft attitude angle β for particle n,indicates the use of x_nCalculating the angular velocity value of the attitude angle gamma of the spacecraft,representing an angular velocity observation representing the spacecraft attitude angle gamma for the particle n,a calculation symbol representing the squared euclidean distance.

It should be noted that, in the above formula (8)Is obtained by calculation using the following formula (9):

formula (9)

In the above formula (9) [. ]]^TA transposed matrix representing f]Represents a complex object computational dynamics model function,represents t_nThe excitation corresponding to the moment (i.e. the angular velocity matrix of each arm segment), an Indicates the angular velocity of the rotation angle of the corresponding arm section,represents t_nThe state of the mechanical arm (i.e. the angle matrix of each arm section) corresponding to the moment phi_M＝[φ₁,φ₂,...,φ_q]^T，φ_iIs the corner of the arm section i,represents t_nAttitude angle state phi of spacecraft corresponding to moment_S≡[α,β,γ]^Tα, gamma are the attitude angle values of the spacecraft in 3 axes, x, respectively_n(g) Denotes the position, t, of the g-th generation particle n_nThe state of the mechanical arm, the state of the attitude angle of the spacecraft and the excitation corresponding to the moment are excitation response data sets used by the particles n.

S14, updating the neighborhood history optimal position of each particle and the individual history optimal position of each particle according to the calculation result of S13, and going to S15.

S15, judging whether particles with errors meeting the preset error requirement exist or not and whether the preset iteration number is reached or not, if the particles with errors meeting the preset error requirement exist or the preset iteration number is reached, going to S20, and if the particles with errors meeting the preset error requirement exist or the preset iteration number is reached, going to S16.

S16, performing the next iteration, such as updating the velocity of each particle, and using the updated velocity to calculate the position of the next generation of the particle, etc., to S11.

S20, sorting the neighborhood history optimal positions of the particles according to the sequence from the superior to the inferior of the neighborhood history optimal positions, and taking the top N_fThe neighborhood history optimal position of the individual particle goes to S21.

S21, selecting N by using a plurality of excitation response data sets based on different configurations_fThe individual particles are subjected to error calculation, respectively, to S22.

The present embodiment can perform error calculation using the following equation (10):

$E_n=\frac{1}{N_r}\underset{u=1}{\overset{N_r}{\&Sigma;}}\left|\right|f\&lsqb;{\widetilde{\stackrel{\&CenterDot;}{\mathrm{\&phi;}}}}_M\left(t_u\right),{\widetilde{\mathrm{\&phi;}}}_M\left(t_u\right),{\widetilde{\mathrm{\&phi;}}}_S\left(t_u\right);x_u^{\left(g\right)}\left(G\right)\&rsqb;-{\widetilde{\stackrel{\&CenterDot;}{\mathrm{\&phi;}}}}_S\left(t_u\right)|{|}_2$ formula (10)

In the above formula (10), E_nIndicates the error calculation result, N_rA group number representing a plurality of excitation response data groups, | | Y | | | non-calculation₂Representing a calculation symbol representing the squared Euclidean distance, f]Represents a complex object computational dynamics model function,represents t_uThe excitation corresponding to the moment (i.e. the angular velocity matrix of each arm segment), for the purpose of the corresponding angular velocity,represents t_uThe mechanical arm state (each joint angle matrix) phi corresponding to the moment_M＝[φ₁,φ₂,...,φ_q]^T，φ_iIs the corner of the arm section i,represents t_uAttitude angle state phi of spacecraft corresponding to moment_S≡[α,β,γ]^TAnd α, gamma are respectively the 3-axis attitude angle values of the spacecraft,represents the optimal position of the neighborhood history of the currently selected u-th particle by the g-th generation,to represent^t _uAnd (4) observing the attitude angular velocity state of the spacecraft corresponding to the moment.

S22, judging N obtained by calculation_fA E_nMin (E) in (1)_n) Whether the predetermined requirement is met, if the predetermined requirement is met, go to S25, otherwise, go to S23.

S23, judging whether the quantity of unselected particles in the sequenced sequence reaches N_fIf N is reached_fOne, then to S24, otherwise, to S10.

S24, sequentially selecting the subsequent N from the sequenced sequence_fThe neighborhood history optimal position of the individual particle and to S22.

S25, mixing min (E)_n) And taking the parameter value represented by the neighborhood history optimal position of the corresponding particle as a final identification result, and outputting the parameter identification result.

As can be seen from the above description, the present embodiment adopts a design scheme of performing particle screening hierarchically, and evaluates the performance of the particles by using a single excitation response data set in the iterative computation process, so that the iterative computation process does not require distinguishing a local minimum value from a global minimum value, but can traverse a minimum value point equivalent to the global minimum value performance; in the composite evaluation process, a plurality of excitation response data sets are utilized to calculate the errors of the particles screened out in the iterative calculation process, and the particles screened out in the iterative calculation process are very limited relative to the parameter space to be identified, so that although the error calculation is carried out by utilizing the excitation response data sets, the increment of the calculated amount is very limited, and the composite evaluation process can quickly screen out the global minimum value because the iterative calculation process traverses the minimum value point equivalent to the global minimum value; therefore, the embodiment effectively controls the calculated amount and complexity in the parameter identification process and improves the accuracy of parameter identification.

The following is a practical simulation experiment for the method of the present embodiment and the existing identification method:

in the simulation experiment, a modified particle swarm algorithm is used, and the definition of relevant parameters in the modified particle swarm algorithm is as follows: the total number N of particles is 500, the total number of iterations G is 50, and the neighborhood particle ratio (i.e., the ratio of particles in the neighborhood to the total number of particles) K_nb10% of the total number of iterations G_reIs 10, restart particle ratio K_reThe content was 20%.

In the simulation experiment, the number of times of the identification experiment is 115, when the parameter identification is performed by using the conventional parameter identification method without using the method of the present embodiment, the identification result is shown in fig. 3, and the normalized error of the parameter identification is shown in fig. 4; when the method of the embodiment is used for parameter identification, N is set_f20 and N_rIn the case of 30, the composite evaluation may be performed N only once_fSelecting the optimal neighborhood history position of each particle to obtain the final parameter identification result, wherein the centroid position b_tAnd mass m_tThe identification result of (2) is shown in FIG. 5, and the normalization of the parameter identificationThe error is shown in fig. 6.

The error ratios of the parameters obtained in the simulation are shown in table 1 below.

TABLE 1

The data in table 1 can be clearly known by comparing, under the same simulation experiment condition, compared with the existing parameter identification technology, the embodiment improves the identification precision by combining the improved particle swarm algorithm and the composite evaluation strategy, and is close to an order of magnitude, so that the parameter identification precision is greatly improved, the average identification error is less than 8%, and meanwhile, more than 90% of the identification error is less than 5%, which fully proves that the method of the embodiment effectively solves the problem of local minimum values.

In the 115 embodiments, different N is used_rIn the case of the composite evaluation, the normalized statistical error average value of each parameter is shown in table 2 below.

TABLE 2

The data in table 2 shows that as the number of excitation response data sets used in the composite evaluation process increases, the recognition accuracy increases, but the rate at which the recognition accuracy increases slows.

As can be seen from the above simulation experiment results, in this embodiment, the calculation cost of the composite evaluation process is about 1.2 generation calculation cost in the iterative calculation process, and the increased calculation amount is about 2.4%, which is worth of the parameter identification effect.

Embodiment three, a parameter identification device for a complex object. The structure of the device is shown in fig. 7.

In fig. 7, the apparatus for identifying parameters of a complex object of the present embodiment mainly includes: an initial setup module 700, an iteration module 710, an evaluation module 720, and a decision module 730.

The initial setting module 700 is mainly adapted to assign different sets of parameter values to the parameter set to be identified, so as to initially set the positions of the particles.

Specifically, the parameter set to be recognized in this embodiment includes a plurality of parameters to be recognized (e.g., two, three, or four parameters), for example, the parameter set to be recognized includes the mass and the centroid position (three-dimensional coordinate value) of the target to be recognized. It should be noted that, in practical applications, the number of the parameters to be identified included in the parameter group to be identified may be more than four. The embodiment does not limit the specific content of the parameter to be identified included in the parameter group to be identified.

The initial setting module 700 may assign a parameter value to each parameter to be identified in the parameter set to be identified for multiple times according to a certain value range, so that all the particles correspond to different parameter values, that is, the positions of different particles are different.

The initial setting module 700 may assign different sets of parameter values to the parameter group to be identified by randomly generating the parameter values, for example, the initial setting module 700 randomly generates one parameter value within a predetermined numerical range of the quality and randomly generates three parameter values within a predetermined numerical range of the centroid position, and the initial setting module 700 assigns the four parameter values to four parameters to be identified in the parameter group to be identified, so that the initial setting module 700 performs the initial setting on the position of one particle; after the above processes of random generation and assignment are repeatedly executed by the initial setting module 700 for many times, the initial setting module 700 performs initial setting on the positions of the plurality of particles, respectively. The number of particles in this embodiment is typically several hundred, and the number of particles may be larger. The number of particles used in the parameter identification process is not limited by the present embodiment.

While the initial setting module 700 performs the initial setting on the position of each particle, an initial velocity should be set for each particle, that is, the initial setting module 700 initializes the velocity of each particle. The initial setting module 700 may perform the initial setting on the position and the speed of the particle by using various existing methods, and this embodiment does not limit the specific implementation manner of the initial setting module 700 for initializing the position and the speed of the particle.

The iteration module 710 is primarily adapted to calculate the position of each particle during the iteration process, and determine the particle having the neighborhood history optimal position using a respective excitation response data set corresponding to each particle according to the position of each particle.

Specifically, the iteration module 710 may utilize a particle swarm algorithm (e.g., an improved particle swarm algorithm) to implement the above iteration steps. It is noted that the iteration module 710 may also use other algorithms to implement the iteration step; no matter the particle swarm algorithm or other algorithms are adopted, the iteration module 710 does not need to distinguish the local minimum value from the global minimum value, and the iteration module 710 only needs to enable the neighborhood history optimal position set of the screened particles to contain the global minimum value.

The iteration module 710 utilizes a modified particle swarm algorithm to implement the above iteration steps to obtain a specific example of the particle with the neighborhood history optimal position, as described in the above first embodiment for the formulas (1) to (4) and the related description in the second embodiment, and the description is not repeated here.

In the iterative process, the error evaluation result of some particles in the past iterative process is probably always poor (namely the error is large), and such particles can be called as poor-performing particles; while some particles may always have better error evaluation results (i.e., smaller errors) during the past iterations, such particles may be referred to as better performing particles.

For a particle with poor performance, the iteration module 710 should shift the particle to the vicinity of a particle with good performance in the iteration process, and specific examples are as described in the above embodiment one with respect to the formula (5) and the related description in embodiment two, and will not be repeated here.

In addition, the iteration module 710 may rank the neighborhood history optimal positions of the particles, and in general, the iteration module 710 may rank the neighborhood history optimal positions of the particles according to the neighborhood history optimal position information of the currently recorded particles in an order from good to bad (i.e., from small to large in error) of the neighborhood history optimal positions.

The evaluation module 720 is mainly adapted to sequentially select a plurality of particles from the unselected particles having the neighborhood history optimal position according to the sequence from the superior to the inferior of the neighborhood history optimal position, and perform error evaluation on the neighborhood history optimal positions of the currently selected plurality of particles respectively by using different excitation response data sets as a screening evaluation sequence.

Specifically, in the case where the iteration module 710 has ordered the neighborhood history optimal positions of the particles:

first, the evaluation module 720 selects a predetermined number of particles from the unselected particles in the sorted sequence (e.g., selects N)_fIndividual particles); next, the evaluation module 720 uses the multiple different excitation response data sets as screening evaluation sequences, and performs error evaluation on the neighborhood history optimal positions of the currently selected particles by using the above formula (7).

It should be particularly noted that the excitation response data sets used by the evaluation module 720 to perform error evaluation on the neighborhood history optimal positions of different particles should be the same; that is, the evaluation module 720 performs error calculation on the neighborhood history optimal positions of different particles using the same screening evaluation sequence. In addition, the plurality of different sets of excitation response data used by the evaluation module 720 are sets of excitation response data based on different configurations of rigid bodies (e.g., robotic arms) in a multi-rigid-body system, such as a set of excitation response data in which a first set of excitation response data is generated based on left rotation of a robotic arm, a second set of excitation response data is generated based on right rotation of a robotic arm, a third set of excitation response data is generated based on 1 st arm rotation of a robotic arm, and so on. The configurations of the rigid bodies (such as mechanical arms) corresponding to different excitation response data sets should have large differences, namely, the screening evaluation sequences should have large differences.

The determining module 730 is mainly adapted to determine whether there are particles meeting the preset error requirement in the currently selected particles according to the result of the error evaluation, if so, take the parameter value represented by the neighborhood history optimal position of the particles meeting the preset error requirement as the parameter identification result, otherwise, when the unselected particles reach the predetermined number, trigger the evaluating module to continue to execute the operations of selection and error evaluation, and when the unselected particles do not reach the predetermined number, trigger the initial setting module to execute the initial setting operation.

Specifically, a plurality of E calculated by the above equation (7) in the evaluation module 720_nIn, if min (E)_n) If the predetermined error requirement is satisfied, the determination module 730 may compare min (E)_n) Neighborhood history optimal position of corresponding particleThe represented parameter value is used as a parameter identification result; if min (E)_n) The preset error requirement is not met, and the quantity of unselected particles in the neighborhood history optimal position sequence of the sequenced particles is not less than N_fThen the determining module 730 triggers the evaluating module 720 so that the evaluating module 720 continues to select the subsequent N from the neighborhood history optimal position sequence of the sorted particles_fA plurality of particles; if min (E)_n) The number of unselected particles in the neighborhood history optimal position sequence of the sorted particles which do not meet the preset error requirementIn a quantity of less than N_fThen the decision module 730 triggers the initial setting module 700 to change the initial setting module 700 to the initial set position and velocity of each particle, and the iteration module 710 iterates using the reset position and velocity of each particle.

The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.

In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

Furthermore, those skilled in the art will appreciate that although embodiments described herein include some features included in other embodiments, not other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.

The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functions in a parameter recognition apparatus for complex objects according to embodiments of the present invention. The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from a website on the internet, provided on a carrier signal, or provided in any other form.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps or the like not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.

Claims (10)

1. A method for identifying parameters of a complex object, the method comprising the steps of:

an initial setting step: different sets of parameter values are given to the parameter set to be identified so as to initially set the position of each particle;

iteration step: calculating the position of each particle in the iterative process, and determining the particle with the optimal neighborhood history position according to the position of each particle by utilizing one excitation response data set corresponding to each particle;

evaluation step: selecting a plurality of particles from the unselected particles with the neighborhood history optimal positions in the sequence from the optimal neighborhood history optimal positions to the inferior neighborhood history optimal positions, and respectively performing error evaluation on the neighborhood history optimal positions of the plurality of currently selected particles by taking different excitation response data sets as screening evaluation sequences;

a judging step: and judging whether the particles meeting the preset error requirement exist in the currently selected particles according to the error evaluation result, if so, taking the parameter value represented by the neighborhood historical optimal position of the particles meeting the preset error requirement as a parameter identification result, otherwise, returning to the evaluation step when the unselected particles reach the preset number, and returning to the initial setting step when the unselected particles do not reach the preset number.

2. The method of claim 1, wherein in an application environment of a multi-rigid-body system consisting of a spacecraft, a robotic arm having a plurality of arms mounted in the spacecraft, and an object to be identified grasped by the robotic arm, the set of parameters to be identified comprises: the mass of the target to be identified and the centroid position of the target to be identified.

3. The method according to claim 1, characterized in that said iteration step comprises in particular:

respectively calculating the position of each particle in the iteration process by using the following formula;

respectively carrying out error evaluation on the position of each particle in the current iteration process by utilizing one excitation response data group corresponding to each particle, and recording neighborhood historical optimal position information of the neighborhood where the particle is located and individual historical optimal position information of each particle according to the error evaluation result, wherein different particles correspond to different excitation response data groups;

when the particles with errors meeting the preset requirements exist in the particles or the current iteration times reach the preset times, the evaluation step is carried out, otherwise, the next iteration process is carried out;

x_n(g+1)＝x_n(g)+v_n(g +1), and $v_n(g+1)={\mathrm{\&omega;v}}_n\left(g\right)+c_1\mathrm{\&xi;}\&lsqb;x_n^{\left(i\right)}\left(g\right)-x_n\left(g\right)\&rsqb;+c_2\mathrm{\&eta;}\&lsqb;x_n^{\left(g\right)}\left(g\right)-x_n\left(g\right)\&rsqb;;$

wherein x is_n(g +1) represents the position of the g +1 th generation particle n, x_n(g) Denotes the position of the g-th generation particle n, v_n(g +1) denotes the velocity of the g +1 th generation of particles N, N denotes the respective particle number, and N is 1,2_n(g) Denotes the velocity of the g-th generation of particles n, ω denotes the inertial weight, c1 and c2 denote learning factors, ξ denotes random vectors within a predetermined interval,indicates the cut-off to the g generationUntil the individual historical optimal position of the particle n, η represents a random vector within a predetermined interval,representing the historical optimal position of the neighborhood in the neighborhood where the g-th generation of particle n is located.

4. The method of claim 3, wherein the performing the error evaluation on the positions of the particles in the current iteration by using the excitation response data set corresponding to each particle comprises:

respectively carrying out error evaluation on the positions of the particles in the current iteration process by utilizing an excitation response data set corresponding to each particle based on the following formula;

$e\left(x_n\right)=\left|\right|\&lsqb;\frac{\stackrel{^^}{p_1}\left(x_n\right)-{\tilde{p}}_1\left(x_n\right)}{\stackrel{^^}{p_1}\left(x_n\right)},...,\frac{p_z\left(x_n\right)-{\tilde{p}}_z\left(x_n\right)}{p_z\left(x_n\right)}\&rsqb;|{|}_2^2;$

wherein,to use x_nCalculating the obtained complex object response result value x_nWhich represents the number of particles n,representing the complex object response observation for particle n,a calculation symbol representing the squared euclidean distance;

and saidIs obtained by the following formula:

${\&lsqb;\stackrel{^^}{p_1}\left(x_n\right),...,p_z\left(x_n\right)\&rsqb;}^T=f\left(d\right(t_n),J(t_n);x_n(g\left)\right)$

wherein, f [. X [ ]]Representing a complex object computational dynamics model function, d (t)_n) Represents t_nThe complex object state matrix corresponding to the moment, J (t)_n) Represents t_nComplex object excitation matrix, x, corresponding to time of day_n(g) Denotes the position, t, of the g-th generation particle n_nThe complex object state and the excitation corresponding to the moment are the excitation response data set used by the particle n.

5. The method of claim 3, wherein the iterating step further comprises:

each interval G_reInstead, the worst K was evaluated for error among all particles_reN particles x_l(g) Carrying out conversion according to the following formula;

$x_l\left(g\right)=x_r^{\left(g\right)}\left(g\right)+\mathrm{\&xi;};$

wherein, K_reRepresents the restart particle ratio, and K_re<1, N represents the total number of particles,representing the neighborhood history optimal position of the particle r that meets the predetermined requirement from the randomly chosen error evaluation in the g-th generation, ξ representing a random vector within a predetermined interval.

6. The method of claim 1, wherein the neighborhood is defined by N being the smallest euclidean distance from a particle_nb(g) A plurality of particles;

wherein,g represents the current iteration number, G represents the total iteration number, K_nbRepresents the proportion of particles in the neighborhood to the total number of particles, and K_nb<1, N represents the total number of particles.

7. The method of claim 1, wherein the error evaluation of the neighborhood history optimal positions of the currently selected plurality of particles using each of the different excitation response data sets as a screening evaluation sequence comprises:

respectively carrying out error evaluation on the neighborhood history optimal positions of the currently selected particles by using the following formula;

$E_n=\frac{1}{N_r}\underset{u=1}{\overset{N}{\&Sigma;}}\left|\right|f\&lsqb;d\left(t_u\right),J\left(t_u\right);x_u^{\left(g\right)}\left(G\right)\&rsqb;-{\&lsqb;{\tilde{p}}_1\left(t_u\right),...,{\tilde{p}}_z\left(t_u\right)\&rsqb;}^T|{|}_2$

wherein E is_nDenotes the result of error evaluation, N_rRepresenting multiple stimulus response dataNumber of groups, | × | non-conducting phosphor₂The calculation symbol, f [. multidot. ] representing the squared Euclidean distance]Representing a complex object computational dynamics model function, d (t)_u) Represents t_uThe complex object state matrix corresponding to the moment, J (t)_u) Represents t_uA corresponding complex object excitation matrix is generated,represents the optimal position of the neighborhood history of the currently selected u-th particle by the g-th generation,represents t_uAnd responding the observed value by the complex object corresponding to the moment.

8. An apparatus for recognizing parameters of a complex object grasped by a space robot, the apparatus comprising:

the initial setting module is suitable for endowing different groups of parameter values for the parameter group to be identified so as to initially set the position of each particle;

the iteration module is suitable for calculating the position of each particle in the iteration process and determining the particle with the optimal neighborhood history position according to the position of each particle by utilizing one excitation response data set corresponding to each particle;

the evaluation module is suitable for sequentially selecting a plurality of particles from unselected particles with the neighborhood history optimal positions according to the sequence from the excellence to the inferiority of the neighborhood history optimal positions, and respectively performing error evaluation on the neighborhood history optimal positions of the plurality of currently selected particles by taking a plurality of different excitation response data sets as screening evaluation sequences;

and the judging module is suitable for judging whether the particles meeting the preset error requirement exist in the currently selected particles according to the error evaluation result, if so, the parameter value represented by the neighborhood historical optimal position of the particles meeting the preset error requirement is used as a parameter identification result, otherwise, when the unselected particles reach the preset number, the evaluating module is triggered to continue to execute the selection and error evaluation operation, and when the unselected particles do not reach the preset number, the initial setting module is triggered to execute the initial setting operation.

9. The apparatus of claim 8, wherein in an application environment of a multi-rigid-body system consisting of a spacecraft, a robotic arm having a plurality of arms installed in the spacecraft, and an object to be identified grasped by the robotic arm, the set of parameters to be identified comprises: the mass of the target to be identified and the centroid position of the target to be identified.

10. The apparatus of claim 8, wherein the evaluation module is specifically adapted to:

respectively carrying out error evaluation on the neighborhood history optimal positions of the currently selected particles by using the following formula;

$E_n=\frac{1}{N_r}\underset{u=1}{\overset{N_r}{\&Sigma;}}\left|\right|f\&lsqb;d\left(t_u\right),J\left(t_u\right);x_u^{\left(g\right)}\left(G\right)\&rsqb;-{\&lsqb;{\tilde{p}}_1\left(t_u\right),...,{\tilde{p}}_z\left(t_u\right)\&rsqb;}^T|{|}_2$

wherein E is_nDenotes the result of error evaluation, N_rA group number representing a plurality of excitation response data groups, | | Y | | | non-calculation₂The calculation symbol, f [. multidot. ] representing the squared Euclidean distance]Representing a complex object computational dynamics model function, d (t)_u) Represents t_uThe complex object state matrix corresponding to the moment, J (t)_u) Represents t_uA corresponding complex object excitation matrix is generated,represents the optimal position of the neighborhood history of the currently selected u-th particle by the g-th generation,represents t_uAnd responding the observed value by the complex object corresponding to the moment.