Predictive Health Management for Batteries in Connected Environments

Abstract

Batteries are prevalent energy storage devices, and their failures can cause huge losses such as the shutdown of entire systems. Therefore, the health management of batteries to increase their availability is highly desirable. To this end, prognostic health management (PHM) techniques can improve the availability and reliability of batteries.Thus, this study developed several novel methodologies for battery PHM in connected environments with batteries under constant monitoring. In the first work, we proposed a novel adaptive maintenance policy for batteries subject to a hard failure. Compared with traditional condition-based maintenance policies, the proposed predictive maintenance policy executes the maintenance decisions adaptively based on the model prognostic results. The prognostic model is continuously updated according to newly inspected data. The inspection schedule and preventive maintenance activities are scheduled online in a sequential manner. In addition, a more efficient optimization scheme with an adaptive sampling method is proposed for deriving the optimal maintenance parameters. The performance of the proposed method was evaluated using extensive simulations. The second work focuses on improving the serviceability of batteries for wireless sensor networks (WSN) deployed in remote and hard-to-reach places. We developed an active management strategy such that the batteries in a network will attain similar end-of-life times, in addition to lifetime extension. The fundamental idea is to adaptively adjust the node quality-of-service (QoS) to actively manage their degradation processes while ensuring a minimum level of network QoS. The framework first executes a prognostic algorithm that can predict the remaining useful life of a battery, given its assigned node-level QoS. Thereafter, an optimization algorithm is executed to cluster the battery end-of-life times by adjusting the node QoS levels. The proposed active intervention strategy converts the complex multi-unit maintenance problem for all batteries into a single-unit maintenance problem for the batteries as a group. This is because their failure instances are almost coincident with each other, and consequently, the maintenance activities for the entire battery network can be scheduled. This strategy extends the lifespan of the battery group by preventing early failures of the individual batteries. Based on the results from the second work, we further investigated and evaluated two opportunistic group maintenance strategies for battery replacement under the given primary maintenance policies, referring to the primary maintenance program performed for maintaining the underlying infrastructure (e.g., a bridge) monitored by the WSN. The simulation results revealed that both frameworks could remarkably simplify the battery replacement procedures, reduce the replacement frequency, and minimize maintenance costs.