2.2.3. Adaptive Planner

The adaptive planner is responsible for planning and optimizing the disassembly strategy. This process is known in the literature as disassembly sequence planning (DSP), and was described in detail in [25]. It finds application in both the planning and operating of disassembly lines, in addition to being used in the product development phase to ensure the guidelines of design for circularity (DfC). DSP starts with selecting the disassembly mode, followed by the modelling step, consisting of the two phases pre-processing and model building. Finally, the disassembly sequence can be optimized while considering a predefined objective function. Many publications have addressed only deriving disassembly sequences. Here, complete disassembly has been considered more frequently than incomplete disassembly [25]. However, the optimal circular economy strategies are supposed to be predefined in the literature and, therefore, are not seen as part of the disassembly planning [26]. An adaptive disassembly planner in an automated disassembly solution consists of an autonomous disassembly process planner that ensures the disassembly execution even when the required data are incomplete and a disassembly strategy optimizer to support decision-making by ensuring that data gaps are eliminated. In this way, different decisions can be made with the help of the optimizer, such as the optimal number of tool and direction changes as well as optimal circular economy strategies at the component level. In the context of battery disassembly, Ke et al. [12] proposed a disassembly sequence planning approach for EVBs based on a genetic algorithm using a frame-subgroup structure. The presented method has better convergence properties

compared to other genetic algorithm implementations. However, it requires the existence of a frame (one component) that has connection and precedence relationships with all other parts (subgroups). In addition, only disassembly sequences for a complete disassembly are considered. Incomplete disassembly and decisions about the optimal circular economy strategies at the component level were not taken into account. Choux et al. [13] proposed an autonomous disassembly task planner, which can generate disassembly sequences autonomously. Thereby, no information about the battery to be disassembled is required. However, the completeness of the automatically detected precedence relationships in [14] was not guaranteed. Thus, possible disassembly sequences that may represent an improved disassembly strategy cannot be considered by the adaptive planner. In addition, the proposed task planner cannot make decisions about the disassembly depth and cannot provide information about the circular economy strategies for the different components, while it has been shown in [14] that they are decisive factors for the disassembly planning for EVBs. These disadvantages can be overcome by extending the autonomous disassembly process planner by a disassembly strategy optimizer. This represents the main research focus of this paper.
