Water Jet Cutting State Detection Based on Visual Information

本研究旨在阐明水射流的切割机理，建立一种通过可视化切割状态来检测水射流切割状态的方法。具体研究成果概括如下：(1)利用超高速摄像机对水射流切割过程进行了观察，获取了水射流切割状态，这对于阐明切割机理和建立切割状态识别方法是非常有用的。(2)发展并评价了一种利用超声波对非透明材料进行可视化的方法。提出了一种基于超声信息的状态识别方法。(3)提出了一种新的水射流切割机理模型，并对其进行了评价。该模型包括三种作用：(I)切割作用、(Ii)冲击作用和(Iii)腐蚀作用。提出了一种基于多维力传感器的水射流切割实时识别方法。在此基础上，提出了一种多维切割状态空间的概念，可以有效地将大量的传感器信息处理到一个紧凑的数据库中，从而可以检测到几种重要的切割状态，如：(A)工件分离不良；(B)水射流弯曲；(C)水射流对工件的侵彻；(D)工件上的阻力线。此外，基于该概念的切割系统可以自动确定切割条件，并在切割过程中对切割条件进行修改。(6)将传感器功能、切割状态识别方法、数据库概念和切割条件确定方法集成为一个基于传感器的自适应控制系统。利用该系统进行了切割实验，证明了该系统的有效性。

This research has been conducted to elucidate a cutting mechanism of water jet and establish a cutting state detection method of water jet by means of visualising cutting state. Concrete research results have been summarised as follows :(1) An observation of water jet cutting process and acquisition of water jet cutting state, which are very useful for a clarification of cutting mechanism and establishment of cutting state recognition method, have been done based on visualisation by means of a ultra high-speed video camera.(2) A visualisation method for not transparent materials by means of ultrasonic has been developed and evaluated. With this, a leading edge of cutting can be visualised to some extent, and a state recognition method based on ultrasonic information has been also proposed.(3) A new model of water jet cutting mechanism has been proposed and evaluated. This model consists of three actions ; (i) cutting action, (ii) impact action, and (iii) eroson action. Moreover, it has been clarified that abrasive in water jet plays different role in cutting for different materials.(4) A real-time recognition method of water jet cutting based on multi-axis force sensor has been developed. With this, several important cutting states, such as (a) bad separation of workpiece, (b) bend of water jet, (c) penetration of water jet through workpiece, (d) drag line on workpiece can be detected.(5) A new concept called as "A multi-dimensional cutting state space" has been proposed, which can efficiently deal with huge sensor information into a compact data base. Furthermore, a cutting system based on the concept can determine cutting conditions automatically and also modify cutting conditions during cutting.(6) The sensor function, cutting state recognition method, database concept, and cutting condition determination method are integrated as a sensor-based adaptive control system. Cutting experiments with this system have been conducted and shown its availability.