【摘要】 针对植入式膀胱尿液容量监测系统存在损伤膀胱组织、传感器易脱落及尿液容量预测模型精度低等问题，设计了一种基于多点位分布式力检测的植入式膀胱尿液容量监测系统，并提出了一种基于SVM-RBFNN的变姿态膀胱尿液容量预测模型。该系统通过体内装置多个微型力传感器和无线通讯模块实现膀胱压力信号的采集和传输；体外装置对压力信号进行接收、处理、显示和存储，并通过无线供能模块向体内装置供电。该模型通过结合支持向量机(SVM)和径向基函数神经网络(RBFNN)实现变姿态下的尿液容量预测。通过搭建模拟实验平台，对该系统和预测模型进行测试。测试结果表明：监测系统可稳定采集膀胱压力信号，传感器不易脱落。预测模型姿态识别正确率达93.75%。尿液容量预测平均精度达96%以上，最大误差绝对值和预测精度均优于BP神经网络。可为设计新型的膀胱尿液容量监测系统提供技术指导。更多还原

【Abstract】 Aiming at the problems of bladder tissue damage, sensor’s easily falling off and the low accuracy of urine volume prediction model in the implantable bladder urine volume monitoring system, an implantable bladder urine volume monitoring system based on multi-point distributed force detection is designed, and a variable posture bladder urine volume prediction model based on SVM-RBFNN is proposed. The system realizes the collection and transmission of bladder pressure signal through multiple micro force sensors and wireless communication modules installed in the body; The external device receives, processes, displays and stores the pressure signal, and supplies power to the internal device through the wireless power supply module. The model is based on support vector machine(SVM) and radial basis function neural network(RBFNN), and enables the system to predict urine volume under changing posture. The system and prediction model are verified by the established simulated experiment platform. The test results show that the monitoring system can stably collect bladder pressure signals, and the sensor does not easily fall off. The accuracy of attitude recognition of the prediction model is 93.75%. The average accuracy of urine volume prediction is higher than 96%, and the absolute value of maximum error and prediction accuracy are better than those of BP neural network. It can provide technical guidance for the design of a new bladder urine volume monitoring system.更多还原