面向风湿免疫膝关节护理的多场耦合模型识别研究

Study on recognition of multi field coupling model for rheumatic immune knee care

针对单一物理场条件下传统识别模型对风湿免疫膝关节炎识别正确率的问题,提出一种力学模量场、血流场、温度场多物理场耦合的膝关节炎识别模型。模型以极限学习机(Extreme Learning Machine,ELM)为基础框架,通过采用粒子群优化算法(Particle Swarm Optimization,PSO)获取ELM模型输入层与隐含层之间的连接权值及隐含层神经元间的阈值,并将膝关节周围软组织肌肉和肌腱/韧带杨氏模量、血管内径大小、血流量、血流速度,以及膝关节周围组织肌肉、肌腱/韧带、血管表面温度的前25个主成分特征值作为改进ELM模型输入,实现了高效、精确的风湿免疫膝关节炎的识别。仿真结果表明,所提多物理场耦合的改进ELM模型,对风湿免疫膝关节炎的识别正确率较高,达到85%以上,相较于标准ELM模型和BP、ID3、LVQ模型,具有一定的优越性。

Aiming at the problem of the accuracy of traditional recognition model for rheumatic immune knee arthritis under the condition of a single physical field,a knee arthritis recognition model coupled with mechanical model field,blood flow field and tem-perature field is proposed.The model takes the extreme learning machine(ELM)as the basic framework,obtains the connection weight between the input layer and the hidden layer of the elm model and the threshold between the neurons of the hidden layer by u-sing particle swarm optimization(PSO),and combines the young's modulus of the soft tissue muscles and tendons/ligaments around the knee joint,the internal diameter of blood vessels,blood flow,blood flow velocity,and the tissue muscles around the knee joint The first 25 principal component eigenvalues of tendon/ligament and vascular surface temperature are used as the input of the im-proved elm model to realize the efficient and accurate recognition of rheumatic immune knee arthritis.The simulation results show that the proposed improved elm model coupled with multiple physical fields has a higher recognition accuracy of rheumatic immune knee arthritis,reaching more than 85%.Compared with the standard elm model and BP,ID3,LVQ model,it has certain advantages.