装配线平衡是生产系统中的一类重要优化问题。研究先进的优化算法对于解决装配线平衡问题具有十分重要的意义。基于模型的数学规划方法和数据驱动的智能优化算法是求解装配线平衡问题的两类常用手段。为了追求更好的优化性能，进一步提高生产效率，将二者融合提出混合智能算法已成为相关领域的前沿方向。本学位论文分别针对机器人装配线平衡问题和分布式装配线平衡问题，提出数据与模型共融的智能优化算法，为装配线平衡问题的高效求解提供算法支撑。在综述装配线平衡问题以及混合智能优化的研究现状的基础上，本论文通过深入研究主要取得了如下成果：1. 针对机器人装配线平衡问题，设计了问题相关的概率模型，提出了利用分枝定界的采样方法，设计了基于关键工作站的局部搜索方法，进而提出了一种结合分枝定界的分布估计算法，并对算法的计算复杂度进行了分析。2． 针对绿色机器人装配线平衡问题，提出了多目标混合整数规划模型，设计了利用分枝定界的双种群采样方法，提出了Pareto最优机器人分配方法，进而提出了一种结合分枝定界的多目标分布估计算法，并分析了算法的计算复杂度。3. 针对分布式装配线平衡问题，提出了一种问题分解策略，将问题分解成为一个上层问题和两个下层问题。针对上层问题，设计了一种差分进化算法和种群的协同初始化方法。在差分进化的评价环节，采用数学规划方法求解下层的运输问题，并通过分析问题性质提出了一种查找表方法求解下层的装配线平衡问题。进而，提出了将差分进化与数学规划相结合的一种基于分解的数模共融智能优化算法。4. 针对考虑设施选择的分布式装配线平衡问题，提出了一种问题分解策略，将问题分解成为一个上层问题和三个下层问题。针对上层问题，设计了一种遗传算法和启发式种群初始化方法。在遗传算法的评价环节，分别采用数学规划方法、基于知识的查表法和启发式方法求解下层问题。进而，提出了将遗传算法与数学规划、启发式方法相结合的一种基于分解的数模共融智能优化算法。通过试验设计方法探讨了参数对算法性能的影响，并采用大量标准算例通过性能对比和统计分析验证了各章所提算法的有效性和高效性。

Assembly line balancing is an important optimization problem in production systems. It is of great significance to study advanced optimization algorithms for assembly line balancing problems (ALBP). Model-based mathematical programming and data-driven intelligent optimization are two types of widely-used techniques to solve ALBP. To achieve better performances and to improve production efficiency, it has been research frontier to develop hybrid intelligent algorithms fusing two kinds of techniques. Aiming at the robotic ALBP (RALBP) and the distributed ALBP (DALBP), this dissertation proposes the model-data fusion hybrid intelligent optimization algorithms, which provides algorithmic support for solving ALBP.After the literature review of assembly line balancing problem and hybrid intelligent optimization algorithms, this dissertation achieves the following main results via deep research.1. For RALBP, a probability model is designed, and a bound-guided sampling method is proposed, and a local search method is designed based on critical workstation. Then, a BGEDA is proposed and its computational complexity is analyzed.2. For energy-efficient RALBP, a multi-objective mixed-integer programming model is developed, and a bi-population bound-guided sampling method is designed, and a Pareto-optimal robot allocation method is proposed. Then, a BGEDA is presented and its computational complexity is analyzed.3. For DALBP, a decomposition strategy is proposed to decompose the problem into an upper-level problem and two lower-level problems. For the upper-level problem, a differential evolution (DE) algorithm is developed and a cooperative initialization mechanism is designed. In the evaluation part of DE, mathematical approaches are adopted to solve the lower-level transportation problem and a lookup table method is developed to solve the lower-level assembly line balancing problem based on the analysis of problem properties. Then, a DMAT is proposed by fusing DE and model-based approach.4. For DALBP with facility selection, a decomposition strategy is proposed to decompose the problem into an upper-level problem and three lower-level problems. For the upper-level problem, a genetic algorithm (GA) is developed and a heuristic is designed to generate the initial population. In the evaluation part of GA, mathematical approaches, lookup table methods, and knowledge-based heuristic methods are developed to solve the lower-level problems. Then, a DMAT is proposed by fusing GA, model-based approaches, and knowledge-based heuristics.The effect of parameters on the performances of the algorithms is investigated by using the design-of-experiment method. The effectiveness and efficiency of the proposed algorithms are demonstrated by performance comparisons and statistical analysis using extensive benchmarking instances.