The invention discloses an intelligent working system for unmanned engineering machinery and a working method and relates to the field of unmanned and intelligent control. The problems that accordingto existing engineering machinery, in an application process, manpower resources are wasted, working efficiency is low, working accuracy is relatively low, physical and psychological health of a driver are seriously impaired, and intelligent operation cannot be realized due to high workload and a severe working environment are solved. According to the system provided by the invention, a 3D terrain, a vehicle state, working task progress, a working medium and a practical scene are taken as input, vehicle energy consumption and working efficiency are taken as optimization goals, and a working mode of a working device is adjusted and improved continuously based on a deep learning method. Deep learning is taken as an important method for the artificial intelligence. According to the unmanned engineering machinery, for different complex working media, different working scenes and different working modes, a working task and the working mode can be dynamically and rationally changed accordingto data collected by the unmanned engineering machinery itself, the working efficiency and working precision are improved, and the working energy consumption is reduced.