The planning of remanufacturing is significantly affected by the uncertainties associated with the availability of used components. This research addresses this challenge by leveraging Deep Learning (DL) algorithms to predict the Remaining Useful Life (RUL) of components, enabling optimal decision-making regarding the timing for sending parts for remanufacture. The study integrates DL methodologies with the Bees Algorithm (BA) to optimise DL learnable parameters. This optimisation is achieved by incorporating BA with the Adaptive Moment (ADAM) optimisation algorithm, facilitating updating weights and biases in the backpropagation process. This integration mitigates the occurrence of local optima, accelerating convergence towards global optima. The proposed method produces varying percentages of RUL prediction errors for the test datasets, culminating in an overall model score of 0.42, where the optimal value is 1.0. Furthermore, the developed model exhibits superior performance compared to four out of nine other research studies that utilised the IEEE PHM 2012 Data Challenge dataset, achieving a performance level of 44.44%.