Abstraction of Abstraction - an Outline of General Scheduling Platform for Space Missions

Scheduling has always been a bottleneck in space mission control in China. Issues at ideology and feasibility level are essential for plotting a brand new scheduling platform. Through introspection on traditional scheduling methods, theoretical and technical solutions are proposed to support advanced scheduling algorithms and products. Academic issues including theory and methodology are quoted to make scheduling intelligent, flexible and expandable. To design the platform, the first concern is to make it all-purpose, supporting totally different missions and spacecrafts, even unknown ones in the future. Two abstracting methods at higher level, Attribute Transposing Method and Weak Attribute Method, are discussed to generalize similar logic models into one. The Attribute Transposing Method uses a key-value pair list descripting similar objects. The keys and the length of the list varies and different object types can be defined. The Weak Attribute Method ignores the exact meaning of attributes, abstracting similar objects as indistinct ones. Once the weak attributes, the number of which is finite and presumably adequate, are given with certain data, an abstracted object is converted into specified one. With these two methods, no matter how the spacecraft models and missions evolve, the logic model stays stable, while still capable containing different logic objects and being processed by spacecraft-model-irrelevant scheduling algorithms. Based on general modeling technology, a three-layer architecture including general layer, spacecraft model layer and UI layer is given. With this architecture, different mission and spacecraft models can be described by pure configurations, and even solving procedures can be appointed and costumed. Rules, designed and provided in libraries, are core of the scheduling platform, with which solving algorithms can be carried, scheduling environment can be configured and even spacecraft model relevant functions can be automated. Rule mechanism together with this three-layer architecture design is essential to make the platform promising for challenging scheduling applications. Then, sample logical models, required functions and features, such as quasi-natural language descriptions and two-way visualization, are designed to outline the features of the platform. A use case and corresponding solving procedure within the platform is given illuminating how convenient, like drag & drop arrangement, and intelligent, like self-adapting resource dispatching, mission scheduling and control could be in the future. Validated by a prototype, scheduling theories, methods, preliminary design and technologies that easy and intelligent scheduling needs proved feasible. The platform will be realized in several years and bring revolutionary improvement to China's space mission scheduling in the foreseeable future.