Search

Forecast errors can cause sub-optimal solutions in resource planning optimization, yet they are usually modeled simplistically by statistical models, causing unrealistic impacts on the optimal solutions. In this paper, realistic forecast errors are prescribed, and a corrective approach is proposed to mitigate the negative effects of day-ahead persistence forecast error by short-term forecasts from a state-of-the-art sky imager system. These forecasts preserve the spatiotemporal dependence structure of forecast errors avoiding statistical approximations. The performance of the proposed algorithm is tested on a receding horizon quadratic program developed for valley filling the midday net load depression through electric vehicle charging. Throughout one month of simulations the ability to flatten net load is assessed under practical forecast accuracy levels achievable from persistence, sky imager and perfect forecasts. Compared to using day-ahead persistence solar forecasts, the proposed corrective approach using sky imager forecasts delivers a 25% reduction in the standard deviation of the daily net load. It is demonstrated that correcting day-ahead forecasts in real time with more accurate short-term forecasts benefits the valley filling solution.

Low-resolution quantized time series signals present a challenge to big data machine learning (ML) prognostics in IoT industrial and transportation applications. The challenge for detecting anomalies in monitored sensor signals is compounded by the fact that many industries today use 8-bit sample-and-hold analog-to-digital (A/D) converters for almost all physical transducers throughout the system. This results in the signal values being severely quantized, which adversely affects the predictive power of prognostic algorithms and can elevate empirical false-alarm and missed-alarm probabilities. Quantized signals are dense and indecipherable to the human eye, and ML algorithms are challenged to detect the onset of degradation in monitored assets due to the loss of information in the digitization process. This paper presents an autonomous ML framework that detects and classifies quantized signals before instantiating two separate techniques (depending on the levels of quantization) to efficiently unquantize digitized signals, returning high-resolution signals possessing the same accuracy as signals sampled with higher-bit A/D chips. This new “unquantize” framework works in line with streaming sensor signals, upstream from the core ML anomaly detection algorithm, yielding substantially higher anomaly detection sensitivity, with much lower false-alarm and missed-alarm probabilities (FAPs/MAPs).

Scenario-based stochastic model predictive control traditionally considers the optimal strategy to be the expectation of the optimal strategies across all scenarios. However, while the stochastic problem involving uncertainties can be substantiated by a large number of scenarios, the expectation of the respective optimal control strategies derived from all scenarios as the optimal control strategy to the problem is challenging to justify. We therefore propose a different approach in which we artfully have the optimization program find the common optimal strategy across all scenarios for the first prediction step at each sample time, which, if it exists, yields the true optimal strategy with greater confidence. We demonstrate the efficacy of the proposed formulation through a case study of a research villa in Boras, Sweden, that is equipped with a battery and a photovoltaic system. We compute a covariance matrix that contains time-dependent information of the data and use it to generate autocorrelated scenarios from the probabilistic forecasts that serve as the uncertain input to the energy management system. We justify the credibility of the optimal solution derived from the proposed formulation with compelling reasoning and quantitative results such as improved self-consumption of photovoltaic power.

Photovoltaic (PV) systems are prone to irradiance variation caused by cloud shadows leading to fluctuations in generated power. Since these fluctuations can be harmful to the operation of power grids, there is a need to restrict the largest PV power ramp rates (RR). This article proposes a method to estimate the largest expected PV power RRs. The only inputs of the method are the minimum PV system dimension and the measurements of point irradiance and cloud shadow velocity. Since cloud shadows cause the largest power RRs for well-designed large-scale PV power plants, the relation between the largest RRs in irradiance and power during partial cloud shading events was studied based on irradiance measurements. The largest RRs in PV power are estimated from RRs in the average irradiance across the PV system. The proposed method was validated using measured data of 57 days from two PV systems. It showed superior performance compared to an existing method enveloping the RR in the measured power over 99.99% of the time. The method can be used in design and component sizing of PV power plants. publishedVersion

With ramp rate regulations for photovoltaic plants being discussed in many countries, the speed of clouds has gained significant importance lately. Besides, measuring cloud velocities and directions is of interest for validations of numerical weather predictions and solar nowcasting systems. Recently, the Cloud Shadow Speed Sensor (CSS) was developed and validated in San Diego for low cumulus clouds. In this publication, the CSS is studied under different weather and cloud conditions in the desert of Tabernas in southern Spain. Furthermore, a novel shadow camera based low-cost, low-maintenance approach to determine cloud shadow motion vectors is presented and used as a reference to benchmark the CSS. In comparison, the absolute velocities derived from the CSS and the shadow camera on 59 days for ± 5 min temporal medians show deviations of RMSD 2.1 m/s (28.0%), MAD 1.2 m/s (15.7%) and a bias of −0.2 m/s (2.8%). Deviations of the cloud shadow direction are RMSD 47.9° (26.6%), MAD 25.3° (14.0%) and bias 3.7° (2.0%). An adaption of the CSS software yields 91% more measurements on 59 days in comparison to the previously used algorithms at the expense of reduced accuracies, both for the measured velocities and for the measured directions. The CSS and the novel shadow camera based reference system enable long-time, low-maintenance ground measurements of cloud shadow speeds, which were previously not available. The distinct advantages and limitations of the two systems are discussed. In addition to the comparisons between the shadow camera system and the CSS on 59 days, the detection rates of the CSS are classified and measured on 223 days by analyzing CSS radiometer signals. Depending on the shading strength and shading durations, detection rates vary between 3.7% and 21.6%. Furthermore, the basic assumption as well as possible correction approaches of the linear cloud edge – curve fitting method are studied. The CSS was found to be a robust tool with great potential. However, optically thin clouds with diffuse edges pose a challenge and the detection rate leaves room for improvements. The newly developed shadow camera system provides more measurements which scatter less but needs certain geographical requirements. The shadow camera is found to be a feasible validation tool for cloud (shadow) motion vectors.

Deploying big-data Machine Learning (ML) services in a cloud environment presents a challenge to the cloud vendor with respect to the cloud container configuration sizing for any given customer use case. OracleLabs has developed an automated framework that uses nested-loop Monte Carlo simulation to autonomously scale any size customer ML use cases across the range of cloud CPU-GPU "Shapes" (configurations of CPUs and/or GPUs in Cloud containers available to end customers). Moreover, the OracleLabs and NVIDIA authors have collaborated on a ML benchmark study which analyzes the compute cost and GPU acceleration of any ML prognostic algorithm and assesses the reduction of compute cost in a cloud container comprising conventional CPUs and NVIDIA GPUs.

This paper presents a novel tandem human-machine cognition approach for human-in-the-loop control of complex business-critical and mission-critical systems and processes that are monitored by Internet-of-Things (IoT) sensor networks and where it is of utmost importance to mitigate and avoid cognitive overload situations for the human operators. We present an advanced pattern recognition system, called the Multivariate State Estimation Technique-2, which possesses functional requirements designed to minimize the possibility of cognitive overload for human operators. These functional requirements include: (1) ultralow false alarm probabilities for all monitored transducers, components, machines, subsystems, and processes; (2) fastest mathematically possible decisions regarding the incipience or onset of anomalies in noisy process metrics; and (3) the ability to unambiguously differentiate between sensor degradation events and degradation in the systems/processes under surveillance. The prognostic machine learning innovation presented herein does not replace the role of the human in operation of complex engineering systems, but augments that role in a manner that minimizes cognitive overload by very rapidly processing, interpreting, and displaying final diagnostic and prognostic information to the human operator in a prioritized format that is readily perceived and comprehended.

Advanced machine learning (ML) prognostics are leading to increasing Return-on-Investment (ROI) for dense-sensor Internet-of-Things (IoT) applications across multiple industries including Utilities, Oil-and-Gas, Manufacturing, Transportation, and for business-critical assets in enterprise and cloud data centers. For all of these IoT prognostic applications, a nontrivial challenge for data scientists is acquiring enough time series data from executing assets with which to evaluate, tune, optimize, and validate important prognostic functional requirements that include false-alarm and missed-alarm probabilities (FAPs, MAPs), time-to-detect (TTD) metrics for early-warning of incipient issues in monitored components and systems, and overhead compute cost (CC) for real-time stream ML prognostics. In this paper we present a new data synthesis methodology called the Telemetry Parameter Synthesis System (TPSS) that can take any limited chunk of real sensor telemetry from monitored assets, decompose the sensor signals into deterministic and stochastic components, and then generate millions of hours of high-fidelity synthesized telemetry signals that possess exactly the same serial correlation structure and statistical idiosyncrasies (resolution, variance, skewness, kurtosis, auto-correlation content, and spikiness) as the real telemetry signals from the IoT monitored critical assets. The synthesized signals bring significant value-add for ML data science researchers for evaluation and tuning of candidate ML algorithmics and for offline validation of important prognostic functional requirements including sensitivity, false alarm avoidance, and overhead compute cost. The TPSS has become an indispensable tool in Oracle's ongoing development of innovative diagnostic/prognostic algorithms for dense-sensor predictive maintenance applications in multiple industries.

A significant challenge for Machine Learning (ML) prognostic analyses of large-scale time series databases is variable clock skew between/among multiple data acquisition (DAQ) systems across assets in a fleet of monitored assets, and even inside individual assets, where the sheer numbers of sensors being deployed are so large that multiple individual DAQs, each with their own internal clocks, can create significant clock-mismatch issues. For Big Data prognostic anomaly detection, we have discovered and amply demonstrated that variable clock skew issues in the timestamps for time series telemetry signatures cause poor performance for ML prognostics, resulting in high false-alarm and missed-alarm probabilities (FAPs and MAPs). This paper describes a new Analytical Resampling Process (ARP) that embodies novel techniques in the time domain and frequency domain for interpolative online normalization and optimal phase coherence so that all system telemetry time series outputs are available in a uniform format and aligned with a common sampling frequency. More importantly, the "optimality" of the proposed technique gives end users the ability to select between "ultimate accuracy" or "lowest overhead compute cost", for automated coherence synchronization of collections of time series signatures, whether from a few sensors, or hundreds of thousands of sensors, and regardless of the sampling rates and signal-to-noise (S/N) ratios for those sensors.

Five strains of Skeletonema costatum-like species were isolated from different coastal regions of the China Sea, where red tides occurred. The internal transcribed spacer regions (ITS-1, 5.8S rDNA and ITS-2) and the partial large subunit rDNA (D1-D2 hyper-variable domain) regions of these strains were sequenced and aligned with sequences obtained from other Skeletonema strains. The five strains showed considerable sequence variation and also length variation among their ITS regions, whereas their partial LSU regions exhibited less variation. Their 5.8S rDNA were identical. The LSU and ITS trees provided different topologies and phylogenetic inferences suggested that the five isolates belonged to at least three different species. The strain from Yellow Sea might be S. tropicum; the strains from East Sea and Xiamen were grouped in the same clade in both trees; the other two strains clustered in the LSU tree, but belonged to different clades in the ITS tree. Results reveal considerable diversity in the genus Skeletonema from the Chinese coast, especially strains causing red tides.