Forward dynamic simulation is used to examine the causal relationships between muscle excitation patterns and human movement. The computed muscle control (CMC) algorithm computes a set of muscle excitations for a movement using proportional-derivative control. However, errors between experimental and simulated kinematics may cause rapid movements. Herein, we propose a novel algorithm, i.e., hybrid computed muscle control (HCMC), which uses a hybrid of numerical integration and optimization to compute muscle excitation patterns that accurately track kinematics, even for rapid movements. We compared the muscle excitation patterns and accuracies of the kinematics simulated by HCMC and CMC using synthetic and experimental data. Two simple musculoskeletal models were used. The synthetic data were generated for three repetitive movements from the rest position to the flexed position (the hip, knee, and ankle underwent 10°, 20°, and 10° plantar flexion, respectively) and back to the rest position for various times. Experimental data were obtained for a subject running at 220 steps/min. The maximum errors in all kinematics calculated using the HCMC algorithm were extremely lower than those calculated using CMC algorithm (HCMC: 0.04–0.07° [synthetic data] and 0.00–0.03° [experimental data]; CMC: 1.04–2.41° [synthetic data] and 0.48–2.50 [experimental data]). For rapid movements, muscle excitations estimated using HCMC occurred early and without delay than those estimated using CMC. The HCMC algorithm can provide muscle excitation patterns that accurately track kinematics and may be useful for perturbation studies using forward dynamic simulation of joints characterized by a low range of motion during rapid movements.

前向动态模拟用于检查肌肉刺激模式与人体运动之间的因果关系。计算的肌肉控制（CMC）算法使用比例微分控制为运动计算一组肌肉激励。但是，实验和模拟运动学之间的错误可能会导致快速运动。在这里，我们提出了一种新颖的算法，即混合计算肌肉控制（HCMC），该算法使用数值积分和优化的混合来计算即使是快速运动也能精确跟踪运动学的肌肉激励模式。我们使用合成和实验数据比较了HCMC和CMC模拟的肌肉兴奋模式和运动学的准确性。使用了两个简单的肌肉骨骼模型。从静止位置到屈曲位置（分别对髋部，膝盖和脚踝分别进行10°，20°和10°足底屈曲）并返回到静止位置的三个重复动作生成了综合数据，并返回了不同的时间。获得了以220步/分钟的速度运行的对象的实验数据。使用HCMC算法计算出的所有运动学的最大误差均大大低于使用CMC算法计算出的误差（HCMC：0.04–0.07°[综合数据]和0.00–0.03°[实验数据]； CMC：1.04–2.41°[综合数据] ]和0.48–2.50 [实验数据]）。对于快速运动，使用HCMC估计的肌肉兴奋比使用CMC估计的肌肉兴奋更早且没有延迟。