The fast-growing wind power industry faces the challenge of reducing operation and maintenance (O&M) costs for wind power plants. Predictive maintenance is essential to improve wind turbine reliability and prolong operation time, thereby reducing the O&M cost for wind power plants. This study presents a solution for predictive maintenance of wind turbine generators. The proposed solution can: (1) predict the remaining useful life (RUL) of wind turbine generators before a fault occurs and (2) diagnose the state of the wind turbine generator when the fault occurs. Moreover, the proposed solution implies low-deployment costs because it relies solely on the information collected from the widely available supervisory control and data acquisition (SCADA) system. Extra sensing hardware is needless. The proposed solution has been deployed and evaluated in two real-world wind power plants located in China. The experimental study demonstrates that the RUL of the generators can be predicted 18 days ahead with about an 80% prediction accuracy. When faults occur, the specific type of generator fault can be diagnosed with an accuracy of 94%.

Operation anomalies are common phenomena in large-scale solar farms. Effective anomaly detection and classification is essential for improving operation reliability and electricity generation. However, this is a challenging task due to the high complexity and wide variety of frequently occurring anomalies. Furthermore, existing pre-installed supervisory control and data acquisition systems (SCADA) can only provide a limited amount of information regarding the healthy condition of solar farms, making accurate anomaly detection and classification difficult. This paper presents a data-driven anomaly detection and classification solution, which can accurately detect and classify diverse photovoltaic system anomalies. The proposed solution does not require additional equipment or non-SCADA data collection. More specifically, the proposed work consists of two methods: (1) a hierarchical context-aware anomaly detection method using unsupervised learning, and (2) a multimodal anomaly classification method. The proposed solution has been deployed in two large-scale solar farms (39.36 MWp and 21.62 MWp). Multi-month operation demonstrates the effectiveness, robustness, as well as cost- and computation-efficiency of the proposed solution.

Developing chatbots as personal companions has long been a goal of artificial intelligence researchers. Recent advances in Large Language Models (LLMs) have delivered a practical solution for endowing chatbots with anthropomorphic language capabilities. However, it takes more than LLMs to enable chatbots that can act as companions. Humans use their understanding of individual personalities to drive conversations. Chatbots also require this capability to enable human-like companionship. They should act based on personalized, real-time, and time-evolving knowledge of their users. We define such essential knowledge as the common ground between chatbots and their users, and we propose to build a common-ground-aware dialogue system from an LLM-based module, named OS-1, to enable chatbot companionship. Hosted by eyewear, OS-1 can sense the visual and audio signals the user receives and extract real-time contextual semantics. Those semantics are categorized and recorded to formulate historical contexts from which the user's profile is distilled and evolves over time, i.e., OS-1 gradually learns about its user. OS-1 combines knowledge from real-time semantics, historical contexts, and user-specific profiles to produce a common-ground-aware prompt input into the LLM module. The LLM's output is converted to audio, spoken to the wearer when appropriate. We conduct laboratory and in-field studies to assess OS-1's ability to build common ground between the chatbot and its user. The technical feasibility and capabilities of the system are also evaluated. Our results show that by utilizing personal context, OS-1 progressively develops a better understanding of its users. This enhances user satisfaction and potentially leads to various personal service scenarios, such as emotional support and assistance.

Data-driven approaches have gained increasing interests in fault detection of photovoltaic systems due to the availability of sensor data. However, the noise introduced by environmental variations and measurement variabilities pose significant challenges on effective fault detection. Furthermore, the change in electrical signal magnitude of a faulty photovoltaic component is usually small, making it difficult to distinguish an anomaly from normal ones. As such, incipient faults are nearly undetectable when they cause less loss of electricity generation. This article proposes a collaborative fault detection solution based on collaborative filtering techniques. Specifically, the proposed solution first predicts photovoltaic strings' current values according to similar strings using historical data. Faults are then detected by long-term differences between the predicted and actual values. A key advantage of the proposed solution is its ability to capture similarities among different photovoltaic strings under noisy and spatial-temporally variant conditions, which significantly enhances fault detection performance. The proposed solution has been deployed in two large-scale solar farms (39.36 MWp and 51.04 MWp). The results show that the proposed solution is superior to existing data-driven solutions in terms of efficiency, effectiveness, and robustness.

Deep-learning-based video processing has yielded transformative results in recent years. However, the video analytics pipeline is energy intensive due to high data rates and reliance on complex inference algorithms, which limits its adoption in energy-constrained applications. Motivated by the observation of high and variable spatial redundancy and temporal dynamics in video data streams, we design and evaluate an adaptive-resolution optimization framework to minimize the energy use of multi-task video analytics pipelines. Instead of heuristically tuning the input data resolution of individual tasks, our framework utilizes deep reinforcement learning t dynamically govern the input resolution and computation of the complete video analytics pipeline. By monitoring the impact of variable resolution on the quality of high-dimensional video analytics features, hence the accuracy of video analytics results, the proposed end-to-end optimization framework learns the best non-myopic policy for dynamically controlling the resolution of input video streams to achieve globally optimized energy efficiency. Governed by reinforcement learning, optical flow is incorporated into the framework to minimize unnecessary temporal-redundancy induced spatial feature re-computation as well as preserving the performance. The proposed framework is applied to video instance segmentation, one of the most challenging machine vision tasks, and achieves the best energy efficiency on the YouTube-VIS dataset among all baseline methods.