基于闭区间上的神经网络迭代计算模型与方法

Neural network computing model and method based on interval iterative algorithm

迭代算法被广泛应用于代数方程求解中,但传统的迭代方法通常对初值的选取要求很高,在迭代过程中不能给出迭代误差,计算得到的单一近似解也具有一定的局限性。把区间迭代算法与神经网络相结合,提出了一种基于闭区间上神经网络迭代计算模型与方法,该方法与传统的迭代方法相比,具有收敛速度快、误差小等特点.最后,通过数值算例表明,该方法是有效的、可行的,能够快速地获得任意代数方程的根。

Iterative algorithm is often used for algebraic equation. But traditional iterative algorithm is often very strict in selecting the first value and no providing iterative error in iterative process. Using iterative method to computing single approximate value has some limitations. Combining interval iterative algorithm with neural network, a neural network computing model and method based on interval iterative algorithm is proposed. Comparing with traditional iterative method, it convergence quickly and it reduced errors. The example showed the validity of the proposed model and method in finding the arbitrary roots of arbitrary algebraic equation.