As safety and reliability critical components, lithium-ion batteries always require real-time\ndiagnosis and prognosis. This often involves a large amount of computation, which makes diagnosis\nand prognosis difficult to implement, especially in embedded or mobile applications. To address this\nissue, this paper proposes a run-time Reconfigurable Computing (RC) system on Field Programmable\nGate Array (FPGA) for Relevance Vector Machine (RVM) to realize real-time Remaining Useful Life\n(RUL) estimation. The system leverages state-of-the-art run-time dynamic partial reconfiguration\ntechnology and customized computing circuits to balance the hardware occupation and computing\nefficiency. Optimal hardware resource consumption is achieved by partitioning the RVM algorithm\naccording to a multi-objective optimization. Moreover, pipelined and parallel computation circuits\nfor kernel function and matrix inverse are proposed on FPGA to further accelerate the computation.\nExperimental results with two different battery data sets show that, without sacrificing the RUL\nprediction performance, the embedded RC platform significantly reduces the computation time\nand the requirement of hardware resources. This demonstrates that complex prognostic tasks can\nbe implemented and deployed on the proposed system, and it can be extended to the embedded\ncomputation of other machine learning algorithms.
Loading....