玩法设计是游戏开发过程中至关重要的环节，游戏设计师进行玩法设计是一个快速迭代的过程。快速原型设计方法是传统的玩法设计方法，该方法存在设计周期长、缺乏量化分析方法等问题。已有研究证明计算机能够自动生成机制简单的小规模游戏，游戏生成的关键技术是游戏描述语言及其配套的系统。本研究旨在基于玩法原型设计平台《创意游戏机》，探索结合玩法验证系统的快速原型设计方法，提升玩法设计的效率。具体而言，探索通过游戏描述语言定义玩法原型，结合自动游戏生成技术提高玩法探索的效率，通过玩法质量的量化分析方法定位影响玩法质量的因素，并尝试预测玩法质量。本研究的主要工作在于：构建玩法定义、玩法生成及玩法评价理论，设计并开发相应的平台系统。第一，本研究结合游戏设计理论和经典游戏描述语言，设计专用于本研究的游戏描述语言，并实现对应的玩法解析功能。该语言能够描述10 种类型共计 574 个玩法。第二，本研究设计并实现玩法的自动生成系统，通过父代玩法选择，规则交叉、变异和修正，关卡的合成及修正和玩法的修正，完成玩法的自动生成过程。该方法生成的玩法基本可玩，且整体质量达到人类设计水平。第三，本研究提出评价玩法质量的方法及影响玩法质量的玩法构成要素和玩家体验要素，提出量化玩家体验的计算方法，实现配套的测试系统及数据处理系统。该方法能够定位部分影响玩法质量的潜在因素并初步预测玩法质量。第四， 基于本研究提出的玩法定义、玩法生成及玩法评价理论，开发玩法验证平台。该平台支持解析与运行玩法、配置参数并一键生成玩法和自动记录玩法构成及玩家测试数据等功能。本研究提出的玩法验证平台具备改进传统玩法设计方法的潜力，对比传统玩法设计方法，可以缩短玩法设计的周期，节约游戏开发的成本。第一，本研究结合游戏设计理论改进游戏描述语言，提升游戏描述语言的可读性，拓展玩法描述的边界。第二，结合游戏设计理论拓展游戏生成技术，提升玩法设计过程中设计师进行玩法探索的效率。第三，提出玩法质量的量化分析方法，证明机器学习算法具备预测玩法质量的潜力，促进游戏生成领域相关的研究。

Gameplay design is an indispensable facet of the game development process. It is an iterative procedure for game designers to conduct gameplay design. Traditional gameplay design methods, such as rapid prototyping, are plagued by lengthy design cycles and a dearth of quantitative analysis techniques. Prior research has demonstrated that computers can autonomously produce small-scale games with simple mechanics. The crux of game generation technology is the game description language and its corresponding system.This study aims to explore a rapid prototyping method, in conjunction with a gameplay verification system, to enhance the efficiency of gameplay design using the gameplay prototyping platform ＂Creative Game Machine＂. Specifically, this thesis endeavor to define gameplay prototypes through a game description language, fuse automatic game generation technology to augment the efficiency of gameplay exploration, and employ quantitative analysis methods to identify the factors that influence gameplay quality and endeavor to prognosticate gameplay quality.The principal objectives of this study encompass the construction of gameplay definition, gameplay generation, and gameplay evaluation theories, along with designing and developing corresponding platform systems. First, amalgamating game design theory and classical game description language, this thesis devise a game description language specifically for this study and implement associated gameplay analysis functions. This language can describe a total of 574 gameplay types spanning 10 categories. Second, this thesis design and implement an automatic gameplay generation system. By selecting parent gameplay, rule crossover, mutation and correction, level synthesis and correction, and gameplay correction, this thesis complete the process of automatic gameplay generation. The gameplay generated by this method is essentially playable, and the overall quality is commensurate with human design. Third, this thesis propound a methodology to evaluate gameplay quality and identify the gameplay constituent elements and player experience elements that impact gameplay quality. This thesis propose a quantitative method for computing player experience, implement matching testing and data processing systems. This method can unearth potential factors that impact gameplay quality and preliminarily predict gameplay quality. Fourth, based on the gameplay definition, gameplay generation, and gameplay evaluation theories posited in this study, this thesis develop a gameplay verification platform. The platform supports parsing and running gameplay, configuring parameters, generating gameplay with one click, and automatically recording gameplay composition and player test data.The gameplay verification platform advanced in this study has the potential to ameliorate traditional gameplay design methods. In comparison to traditional gameplay design methods, it can truncate the gameplay design cycle and economize game development expenses. First, this thesis enhance the game description language by amalgamating game design theory to boost the comprehensibility of the game description language and broaden the horizons of gameplay depiction. Second, this thesis expand game generation technology predicated on game design theory to enhance the efficacy of game designers in exploring gameplay during the gameplay design process. Third, this thesis posit a quantitative analysis technique for gameplay quality, evincing that machine learning algorithms have the potential to prognosticate gameplay quality and invigorate related research in the game generation arena.