CN-121997485-A - Multi-algorithm cooperative external fixed support pose determination method, equipment and medium

CN121997485ACN 121997485 ACN121997485 ACN 121997485ACN-121997485-A

Abstract

The invention relates to a multi-algorithm collaborative external fixed support pose determining method, equipment and medium, which are used for acquiring algorithm starting information of a base layer and a collaborative layer in a processing system configured by a user, calling an started base algorithm in a base layer algorithm module to obtain a base pose, inputting a base pose solution into the started collaborative algorithm to carry out refinement processing if a base neural network approximation method is started and the collaborative algorithm in the collaborative layer algorithm module is started, and taking the refined pose solution as a final candidate pose solution, otherwise, directly taking the base pose solution as the final candidate pose solution of a corresponding path, respectively calculating theoretical branched chain length vectors corresponding to the final candidate pose solution through inverse kinematics, comparing the theoretical branched chain length vectors with actual branched chain length vectors of a current group, and selecting the final relative pose with the smallest matching error as a final relative pose result corresponding to the actual branched chain length vectors of the group, and outputting the final relative pose. Compared with the prior art, the method has the advantages of wide application range, high robustness, high efficiency and the like.

Inventors

ZHANG QING
JIA ZHIMIN
SUN YUANTAO
Min Guanyu

Assignees

同济大学

Dates

Publication Date: 20260508
Application Date: 20251231

Claims (10)

1. The multi-algorithm cooperative external fixed support pose determining method is characterized by running in a processing system preset with a plurality of pose solving algorithms, and comprises the following specific steps of: S1, acquiring a structural parameter set and an actually measured branched chain length of a target external fixed support, wherein the structural parameter set comprises a proximal ring parameter, a distal ring parameter, a branched chain parameter and installation position parameters of each branched chain on the proximal ring and the distal ring; S2, acquiring algorithm starting information of a base layer and a cooperative layer in the processing system configured by a user, calling a basic algorithm started in a basic layer algorithm module, and carrying out parallel calculation by taking the structure parameter set and the actual measurement branched chain length of the current group as common input; S3, if the basic neural network approximation method is started and the cooperative algorithm in the cooperative layer algorithm module is started, inputting a basic pose solution output by the basic neural network approximation method into the started cooperative algorithm for refinement, and taking the refined pose solution as a final candidate pose solution of the current algorithm; s4, calculating corresponding theoretical branched chain length vectors of all final candidate pose solutions through inverse kinematics, comparing the theoretical branched chain length vectors with actual measured branched chain length vectors of the current group, and selecting the final candidate pose solution with the smallest matching error as a final relative pose result corresponding to the actual measured branched chain length vectors of the current group to output.
2. The multi-algorithm collaborative external fixed support pose determination method according to claim 1 is characterized in that the collaborative algorithm comprises a collaborative numerical iteration algorithm and a collaborative intelligent optimization algorithm, wherein at most one algorithm is started in the collaborative algorithm, the basic numerical iteration algorithm and the collaborative numerical iteration algorithm are newton iteration methods for calculating a jacobian matrix by adopting a central difference method, and the basic intelligent optimization algorithm and the collaborative intelligent optimization algorithm are particle swarm optimization algorithm or genetic algorithm; when the activated cooperative algorithm is the cooperative numerical iterative algorithm, taking a basic pose solution output by the basic neural network approximation method as an iterative initial solution of the cooperative numerical iterative algorithm to carry out iterative calculation; when the activated collaborative algorithm is the collaborative intelligent optimization algorithm, a local search space is built by taking a basic pose solution output by the basic neural network approximation method as a center, and optimization search is performed in the local search space.
3. The multi-algorithm collaborative external fixed support pose determination method according to claim 1, wherein the basic neural network approximation method is implemented based on a pre-trained neural network model, the pre-trained neural network model is obtained by the following steps: Determining a sampling space of the pose parameter according to the near-end ring parameter, the far-end ring parameter and the branched-chain parameter in the structure parameter set; randomly generating a plurality of pose samples in the sampling space, and obtaining theoretical branched chain length vectors corresponding to each pose sample through inverse kinematics calculation to form a plurality of training sample pairs, wherein each training sample pair comprises a pose sample and the corresponding theoretical branched chain length vector; And training the initial neural network by utilizing the plurality of training samples to obtain the pre-trained neural network model.
4. A multi-algorithm collaborative external fixed support pose determination method according to claim 3 wherein said pre-trained neural network model comprises a back propagation neural network comprising a plurality of parallel sub-networks each for predicting a dimension in said final relative pose result and a radial basis function neural network employing gaussian radial basis functions as activation functions.
5. The multi-algorithm collaborative external fixed support pose determination method according to claim 1, wherein the actually measured branched chain lengths obtained in S1 are multiple groups, and continuous marks indicating whether there is a pose continuous change relationship among the multiple groups of data are correspondingly obtained; When the continuous flag indicates that there is a pose continuous variation relationship, when S2 is performed for the latter set of measured branch lengths: If the basic numerical iterative algorithm is started, the final relative pose result output by the basic numerical iterative algorithm in S4 aiming at the previous group of data is used as an iterative initial solution when solving the current group of data or used as a value of a corresponding dimension in the initial solution; And if the basic intelligent optimization algorithm is started, the final relative pose result output by the basic intelligent optimization algorithm in the S4 aiming at the previous group of data is used as the center of the search space when solving aiming at the current group of data.
6. The multi-algorithm collaborative external fixed support pose determination method according to claim 1, wherein the structural parameter set further comprises an installation sequence identifier of a proximal ring and a distal ring, and in the step S4, according to the installation sequence identifier, the correspondence between a proximal coordinate system and a distal coordinate system according to which inverse kinematics calculation is performed is determined.
7. The multi-algorithm collaborative external fixed support pose determination method according to claim 1, wherein installation position parameters of the branched chains on the proximal ring and the distal ring support asymmetric arrangement, the installation position parameters are specifically numbers of hole sites on the rings connected with the branched chains, the ring structure of the external fixed support is a full ring, a 2/3 ring or a U-shaped ring, and the joint types of the branched chains comprise a ball pair-a revolute pair-a ball pair and a revolute pair-a revolute pair.
8. The multi-algorithm collaborative external fixed support pose determination method according to claim 1, wherein in S4, the matching error is obtained by calculating a euclidean distance or an average absolute difference between the theoretical branched chain length vector and the measured branched chain length vector.
9. An electronic device comprising a memory and a processor, the memory having stored thereon a computer program, characterized in that the processor, when executing the program, implements the method according to any of claims 1-8.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when executed by a processor, implements the method according to any one of claims 1-8.

Description

Multi-algorithm cooperative external fixed support pose determination method, equipment and medium Technical Field The invention relates to the technical field of skeletal deformity correction, in particular to a multi-algorithm cooperative external fixed support pose determining method, equipment and medium. Background Medical external fixation stents play an important role in the field of bone surgery. In response to the various symptoms, the physician needs to select the appropriate stent based on clinical experience, and thus design a wide variety of anchor configurations. Six-axis external fixation stents, one of the typical advanced ring-shaped external fixation devices, have been widely used by orthopedic surgeons worldwide. The six-axis external fixing support is an improved Stewart mechanism frame and consists of two rings which are connected by six adjustable telescopic screws at universal joints. This allows the external fixation frame to simultaneously correct complex multi-dimensional deformities like an Ilizarov stent. The treatment algorithm is the planning of the treatment process based on the external fixed support, the solving algorithm of the support pose is the core of the treatment algorithm, and doctors can accurately control the correction process under the assistance of a computer. The theoretical screw length under each correction state can be calculated in advance through a treatment algorithm, and the treatment plan is adjusted according to actual conditions in the treatment process to predict the treatment effect. Because of the characteristics of the parallel mechanism, the inverse kinematics algorithm is used for calculating the length of the theoretical screw rod simply and efficiently, but the forward kinematics algorithm which assists in designing the treatment process and predicting the actual treatment effect of the stent is less researched. The conventional forward kinematics algorithm has various limitations that when the external fixed support in the form of a typical Taylor support and the like is used for processing special abnormal shapes, solving failure is easy to occur due to the fact that the difference of the positions and the lengths of support links is large, personalized treatment requirements need personalized supports, such as asymmetric branched chain installation modes or different branched chain support forms, the conventional forward kinematics algorithm is difficult to meet complex and various support configurations and application scenes, the application range is limited, the recovery process and the final effect of a patient are seriously influenced by the low-efficiency and low-precision forward kinematics algorithm, the potential high risk is brought to treatment due to the fact that the result precision of the algorithm is not high, the solving speed is not timely, even in certain cases, the convergence result cannot be given, meanwhile, the algorithm has certain requirements on the performance of local medical equipment and even a network server, the algorithm with low efficiency increases equipment load, and the instantaneity of calculation response is influenced. Therefore, how to provide an external fixed support positive kinematics solving method which is efficient, stable, strong in adaptability and capable of reliably running in different clinical scenes is a technical problem to be solved. Disclosure of Invention The invention aims to overcome the defects of the prior art and provide a multi-algorithm collaborative external fixed support pose determining method, equipment and medium, and by establishing a basic layer parallel prediction or solving and collaborative layer refined double-layer algorithm architecture, the rapid prediction and high-precision optimized directional collaboration is realized, so that the problems of failure, non-convergence and insufficient precision caused by poor initial values in the traditional external fixed support positive kinematics solving are solved while the computing efficiency is ensured, and the robustness, speed and accuracy of the pose determining process are improved. The aim of the invention can be achieved by the following technical scheme: According to one aspect of the invention, there is provided a multi-algorithm cooperative external fixed support pose determining method, which is operated in a processing system preset with a plurality of pose solving algorithms, and the specific steps include: S1, acquiring a structural parameter set and an actually measured branched chain length of a target external fixed support, wherein the structural parameter set comprises a proximal ring parameter, a distal ring parameter, a branched chain parameter and installation position parameters of each branched chain on the proximal ring and the distal ring; S2, acquiring algorithm starting information of a base layer and a cooperative layer in the processing system configured by a user, calling a basic algor