Under combining the interpolation theory with non - linear capacity and parallel calculation structure of artificial neural network ( ann ) , a software implement technique for fine interpolation of the 1pt + 3tps hmt is researched . the roughing interpolation and fine interpolation algorithm of hmt are introduced in the first . the interpolation controller is constructed by bp neural network and the training with the lm algorithm to the controller is carried on 給出了混聯(lián)機床粗插補和精插補算法,以bp神經(jīng)網(wǎng)絡構建了插補控制器,用lm算法對控制器進行訓練,利用matlab中的神經(jīng)網(wǎng)絡工具箱進行了仿真,驗證了其可行性和有效性,為實現(xiàn)面向制造過程的高速實時插補方面,提供了一種新的插補控制器結構。