Applications of the Bees Algorithm: Nature-Inspired Optimisation of Manufacturing Processes

Introduction

In today’s highly competitive and globalised market, automation is a common solution to reduce labour costs and improve the efficiency and consistency of manufacturing processes (Frohm et al. 2006). A crucial step in the manufacturing of products is the assembly of the individual components and sub-components into the finished product. Assembly operations in today’s manufacturing processes account for 20%-50% of the whole production time, and more than 40% of the whole production costs (Boothroyd, 1994). One of the main challenges in the automation of assembly operations is the optimisation of the sequence of placement of the components, in order to minimise the total manufacturing time. Efficient assembly operations contribute to increasing plant throughput and reducing machine hours.

Assembly sequence optimisation is typically an NP-complete combinatorial problem ((Rashid et al. 2012; Han et al. 2021), where the computational time-complexity using any known solver grows in non-polynomial fashion with the number of elements (i.e. components to assemble). For any non-trivial assembly problem, a brute force evaluation of all possible assembly sequences is not feasible, and heuristic search methods must be used.

Of even greater complexity is the problem of disassembling End-of-Life (EoL) products. Disassembly is a critical operation in closed-loop supply chain since it is a prerequisite for recycling, remanufacturing and reusing of EoL products and components (Guo et al. 2020). It is nowadays gaining increasing importance due to the need of making more efficient use of limited natural resources, and reduce environmental pollution. In many ways, disassembly is more challenging than assembly to automate, due to variability in the condition of the components (Vongbunyong et al. 2015). Unlike the assembly of new or remanufactured products, which is deterministic because the components to be assembled are of known geometries, dimensions and states, disassembly is stochastic as has to contend with used products of uncertain shapes, sizes and conditions. For example, components might be worn-out, corroded, or deformed, they might have been intentionally modified or replaced with equivalents of a different maker, or they might be altogether missing. Due to these uncertainties, the sequence of disassembly operations can not be pre-set and fixed as it is customary for assembly, and has to be adapted to the state of each individual product.

In this chapter, two case studies involving the assembly and disassembly of products will be examined, and tackled using the popular Bees Algorithm metaheuristics (Pham et al. 2006a; Pham and Castellani 2009). The Bees Algorithm takes inspiration from the collective problem-solving abilities of biological bees (apis mellifera) to provide solution to complex search problems. It mimics the foraging behaviour of bee colonies to scout the space of candidate solutions, and perform parallel local searches in the most promising regions. The Bees Algorithm has found application in several manufacturing fields involving the solution of NP-hard optimisation problems such as preliminary design (Pham et al. 2007a), job scheduling for a machine (Pham et al. 2007b), quality control (Pham et al. 2006b), control of robotic manipulators (Fahmy et al. 2012), manufacturing cell formation (Pham et al. 2007c), and assembly line balancing (Baykasoglu et al. 2009).