This paper presents a novel drone system used for in-situ measurement of carbon dioxide (CO2), which is one of greenhouse gases. To obtain a vertical profile of atmospheric CO2 concentration, a Non-Dispersive Infrared (NDIR) analyzer was equipped on an off-the-shelf industrial drone. We designed and assembled an original mount which consists of carbon plates, hollow aluminum pipes, and resinous adapters created using a 3D printer. We obtained vertical CO2 profiles up to the 500m level through a flight test of five different dates after getting the flight permissions at an altitude of more than 150m above the ground and Beyond Visual Line of Sight (BVLOS) from the Ministry of Land, Infrastructure and Transport. It is shown that CO2 concentration generally increases with altitude, though the feature of CO2 profiles varies with observation dates. Based on the five profiles from September 2017 to January 2018, we revealed a seasonality of CO2 concentration over Akita. We suggest that this system is useful for in-situ CO2 measurement and enables to conduct frequent and easy observations.
Invention of atomic clock dramatically improved accuracy of time measurement, such as time scale, which is an application of atomic clock. However, even the best atomic clock accumulates error, or time deviation, which grows with time. For that reason, compensating for this deviation is a fundamental problem. In this paper, we propose an estimation method using Recursive Least-Squares Method with Vector-type Variable Forgetting Factor (VVFF-RLS) for multiple atomic clocks system. In multiple atomic clocks system, achieving reliable estimation of time deviations of atomic clocks is one of the most important tasks. An atomic clock, which counts the inter-atomic transition frequency to keep an accurate one second, sometimes suffers its abnormal behaviours of frequency due to environmental factors. The proposed method employs vector-type variable forgetting factor whose each component is allocated for the corresponding clock, and achieves a reliable estimation by sequentially determining the forgetting factor on the basis of the frequency stability. The usefulness of the proposed method is validated through numerical examples.
Considering cybersecurity for industrial control systems (ICS), the latency of a firewall could affect a timing restriction of a real-time control loops. To solve this issue, we propose the “real-time firewall”, the low and deterministic latency firewall for control networks. It employs on-the-fly rule matching method to minimize the latency of the firewall, which modifies the FCS (Frame Check Sequence) field of Ethernet frames to discard malicious frames. It also employs the Shift-and algorithm for signature pattern matching. We prototyped the real-time firewall using an FPGA and evaluated it, then confirmed that 1) it does not limit the throughput of 100BASE-TX wire speed, and 2) the latency ranges from 2.12µs to 2.2µs regardless of the frame size or the number of matching patterns to be inspected.
In this paper, we propose a day-ahead scheduling method for multi-period electricity markets using a machine learning approach based on neural networks. An aggregator, which has renewable energy generation devices, needs to schedule the energy production and consumption (prosumption) in a situation where the renewable power generation amount is not exactly predicted in day-ahead scheduling. If imbalance, defined as the difference between a day-ahead schedule and an actual prosumption profile, occurs, the aggregator is required to pay imbalance penalty costs. As a scheduling method to avoid paying imbalance penalty costs, we propose a scheduling model by machine learning based on the results of past transactions. In particular, the scheduling model is given as a neural network, which has an advantage in terms of computational costs compared to the kernel method. For developing a training algorithm, we show that the gradient of the profit function with respect to design parameters can be calculated from a solution to linear programming. Finally, we show the efficiency of the proposed method through a numerical example.
We consider rechargeable batteries used in a start-stop system in automobiles, which automatically shuts down and restarts the engine. When the start-stop system is activated, it is necessary to check in advance whether batteries can serve sufficient power for restarting the engine and to judge whether the engine should be stopped. To make the judgement, the unit step response in a specified time of an internal impedance of batteries is used as a criterion. In this report, we propose to utilize an impulse response model to calculate the unit step response in a specified time. In particular, we illustrate the µ-Markov model, which consists of the finite impulse response model and the auto-regressive with exogenous input model, is suitable for the purpose. The effectiveness of the proposed method is demonstrated in a numerical experiment.
A vibration suppression technique for a flexible slewing arm using tip accelerometer is presented where the first-order vibration mode as well as a transmission time delay are employed in an LQ compensator. This is implemented inside an I-PD servo major loop for controlling the angle of the hub. Simulation and experimental results demonstrate that making use of the time delay contributes to decrease unwanted vibrations.