Evolutionary algorithms find applicability in reinforcement learning of neural networks due to their independence from gradient-based methods. To achieve successful training of neural networks using evolutionary algorithms, careful considerations must be made to select appropriate algorithms due to the availability of various algorithmic variations. In Part1 and Part2, the author previously reported experimental evaluations on Genetic Algorithm and Evolution Strategy for reinforcement learning of neural networks, utilizing the Acrobot control task. This article constitutes Part3 of the series of comparative research. In this study, Differential Evolution is adopted as the third instance of major evolutionary algorithms. The experimental results show a statistically significant superiority of DE over both GA and ES (p < .01). In addition, DE exhibits its robustness to variations in hyperparameter configurations (the number of offsprings and generations). In the previous experiments, both ES and GA showed significant performance differences depending on the configurations, whereas in the experiment reported in this article, such differences are not detected for DE.

Evolutionary algorithms and swarm intelligence algorithms find applicability in reinforcement learning of neural networks due to their independence from gradient-based methods. To achieve successful training of neural networks using these algorithms, careful considerations must be made to select appropriate algorithms due to the availability of various algorithmic variations. In Part1, 2 and 3, the author previously reported experimental evaluations on Evolution Strategy, Genetic Algorithm, and Differential Evolution for reinforcement learning of neural networks, utilizing the Acrobot control task. This article constitutes Part4 of the series of comparative research. In this study, Particle Swarm Optimization is adopted as an instance of major swarm intelligence algorithms. The experimental result shows that PSO performed worse than all of DE, GA and ES. The difference between PSO and DE was statistically significant (p<.01). In addition, PSO exhibited lower capability in exploring solutions in high-dimensional search spaces than DE, GA, and ES did. A larger swarm size compensated for the weakness of PSO in global exploration, thus making itself more beneficial than a larger number of swarm search iterations.

Evolutionary algorithms find applicability in reinforcement learning of neural networks due to their independence from gradient-based methods. To achieve successful training of neural networks using evolutionary algorithms, careful considerations must be made to select appropriate algorithms due to the availability of various algorithmic variations. The author previously reported experimental evaluations on Evolution Strategy for reinforcement learning of neural networks, utilizing the Acrobot control task. In this study, Genetic Algorithm is adopted as another instance of major evolutionary algorithms. Experimental results demonstrate that there was no statistically significant difference between the experimental performances of GA and ES, but the priority of generations and offsprings was different; GA performed better with a greater number of generations while ES performed better with a greater number of offsprings. Eight hidden units were the best among four variations (4, 8, 16 or 32 units), which aligns with previous study using ES.

The removal of scale by high gradient magnetic separation (HGMS) has been investigated utilizing the magnetic properties of the scale. This study focuses on a thermal power plant with oxygen treatment (OT), which has been applied in many new power plants with large generation capacity in recent years. At high-temperature conditions where the HGMS system is expected to be installed, the viscosity of the fluid will be lower than that of at room temperature, which makes the separation ratio higher. In this study, the scale capture rate at room temperature (25°C) and 80°C was investigated by particle trajectory calculation and experiment, in order to design magnetic separation conditions suitable for the high-temperature environment of the chemical cleaning line of the thermal power plant adopting OT.

The reduction of carbon dioxide emissions becomes a global issue, the main source of carbon dioxide emissions in the Asian region is the energy conversion sector, especially coal-fired power plants. We are working to develop technologies that will at least limit the increase in carbon dioxide emissions from the thermal power plants as one way to reduce carbon dioxide emissions. Our research aims to reduce carbon dioxide emissions by removing iron oxide scale from the feedwater system of thermal power plants using a superconducting high-gradient magnetic separation (HGMS) system, thereby reducing the loss of power generation efficiency. In this paper, the background of thermal power plants in Asia is outlined, followed by a case study of the introduction of a chemical cleaning line at an actual thermal power plant in Japan, and the possibility of introducing it into the thermal power plants in China based on the results.

We have been developing the scale removal system utilizing superconducting magnet that can remove iron scale from boiler feed-water in a thermal power plant. The scale removal prevents the plants from the reduction in power generation efficiency. Iron oxide scale consists of the ferromagnetic and the paramagnetic particles, and the optimal separation conditions largely differ depending on the magnetic properties of the particles. So, single separation condition may cause the filter blockage by over capture or inadequate capture. We proposed the two-stage magnetic separation according to the magnetic properties of the aggregates, where the ferromagnetic particles are captured in the 1st stage in low magnetic field and field gradient, and then the paramagnetic ones are captured in the 2nd stage in high magnetic field and field gradient. It was shown that two-stage magnetic separation system for the mixture of ferromagnetic and paramagnetic particles is possible by utilizing one superconducting solenoidal magnet.

Assuming that the high‐gradient magnetic separation (HGMS) system based on a superconducting magnet can reduce iron oxide scale in boiler feedwater and contribute to the efficient operation of thermal power plants, we design a new matrix structure, which should efficiently capture iron oxide particles in boiler feedwater. We then evaluate their practicability experimentally and confirm that this HGMS system with the newly developed matrix is applicable to thermal power plants with the desired efficiency. On the bases of the results, we designed practical large‐scale HGMS systems for boiler feedwater with a superconducting magnet and new matrices, which has long cleaning cycles and low‐pressure loss. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

In the boiler feed-water system of thermal power plants, iron oxide scale is generated due to the corrosion of piping, which decreases the effectiveness of the heat exchanger in the boiler and contributes to reduced power generation efficiency. Scale removal can prevent this phenomenon, which consequently results in decreased carbon dioxide emission. In a previous study, an iron scale removal system was developed using superconducting magnets, and high-gradient magnetic separation (HGMS) experiments were conducted using the height difference between the inflow and outflow tanks. As a result, particles were intensively captured by the filters at the inflow side. We succeeded in eliminating this capture by properly controlling the applied magnetic field. However, because the suspension also flowed due to the height difference in this experiment, the problems of partial blockage of the filter and the resulting decrease in the flow rate were not resolved even under the controlled magnetic field. In this study, HGMS experiments were conducted under a constant flow rate with a large pump in a large-scale system having a filter diameter of 300 mm to simulate a boiler feed-water system or chemical cleaning line. Furthermore, we investigated the feasibility of the large-scale HGMS system and elucidated the scale-up effect on the amount captured and the spatial distribution of the captured particles on the filters. Based on these results, a plant-scale HGMS system (800 mm filter diameter) was designed.

The development of magnetic separation systems for thermal power plants, including our recent studies, are reviewed.The magnetic separation systems reduce the scale in boiler feed water in thermal power plants and maintain energy conversion efficiency at a high level, reducing the amount of carbon dioxide discharged from thermal power plants.

We have been developing a scale removal system utilizing a superconducting magnet that can remove the iron oxide scale from the boiler feed-water in the thermal power plants. Scale removal can prevent a decrease in the power generation efficiency and contribute to reduce the carbon dioxide emissions. In our previous study, we conducted the large-scale experiments for practical use, but model particles were intensively captured by the inflow side filter. In this paper, we succeeded to homogenize the distribution of captured particles in the filter stacks by controlling the applied magnetic field strength. In addition, magnetic field strength dependency of filter blockage was examined in the lab-scale experiments to clarify a detailed mechanism of blockage.

Removal of iron oxide scale from feed-water in thermal power plant can improve power generation efficiency. We have proposed a novel scale removal system utilizing High Gradient Magnetic Separation (HGMS). This system can be applied to high temperature and pressure area. We have conducted the lab-scale model experiments using φ50 mm filters and it demonstrated high removal efficiency in HGMS, but scale-up of the system is required toward practical use. In this study, we conducted a large scale mock-up HGMS experiment. We used the superconducting solenoidal magnet with φ400 mm bore and demonstrated that our HGMS system can achieve sufficient scale removal capacity that is required to introduce into both off-line and on-line system.

Removal of iron oxide scale from feed-water in thermal power plant can improve power generation efficiency. We have proposed a novel scale removal system utilizing High Gradient Magnetic Separation (HGMS). This system can be applied to high temperature and pressure area. We have conducted the lab-scale model experiments using phi 50 mm filters and it demonstrated high removal efficiency in HGMS, but scale-up of the system is required toward practical use. In this study, we conducted a large scale mock-up HGMS experiment. We used the superconducting solenoidal magnet with phi 400 mm bore and demonstrated that our HGMS system can achieve sufficient scale removal capacity that is required to introduce into both off-line and on-line system.

Precise protein structure determination provides significant information on life science research, although high-quality crystals are not easily obtained. We developed a system for producing high-quality protein crystals with high throughput. Using this system, gravity-controlled crystallization are made possible by a magnetic microgravity environment. In addition, in-situ and real-time observation and time-lapse imaging of crystal growth are feasible for over 200 solution samples independently. In this paper, we also report results of crystallization experiments for two protein samples. Crystals grown in the system exhibited magnetic orientation and showed higher and more homogeneous quality compared with the control crystals. The structural analysis reveals that making use of the magnetic microgravity during the crystallization process helps us to build a well-refined protein structure model, which has no significant structural differences with a control structure. Therefore, the system contributes to improvement in efficiency of structural analysis for "difficult" proteins, such as membrane proteins and supermolecular complexes.

We are developing a superconducting magnetic separation system to remove scale or iron oxide from boiler feedwater in thermal power plants. The reduction of scale improves energy conversion efficiency of thermal power plants and reduces discharged CO 2 . We have studied suitable installation locations and operation conditions and examined the separation ability of the system located near the boiler where the water is at high temperatures and pressures. We have concluded that the magnetic separation system is effective to remove scale of boiler feedwater and that the use of a superconducting magnet is essential for the system.

To improve thermal power plant efficiency, we proposed a water treatment system with a high gradient magnetic separation (HGMS) system using a superconducting magnet, which is applicable in high-temperature and high-pressure conditions. This is a method to remove the scale from feed-water utilizing magnetic force. One of the issues for practical use of the system is how to extend the continuous operation period. In this paper, we succeeded in solving the problem by eliminating the deviation of captured scale quantity by each filter. In fact, in the HGMS experiment using the solenoidal superconducting magnet, it was shown that a decrease in separation rate and an increase in pressure loss were prevented, and the total quantity of captured scale increased by proper filter design. The design method of the magnetic filter was proposed and will be suitable for long-term continuous scale removal in the feed-water system of the thermal power plant.

In-situ observations of particles deposition process on a ferromagnetic filter in high gradient magnetic separation were carried out under high magnetic fields to obtain information for the optimization of separation condition. The spike-like deposition structure was observed on the upper stream of the magnetic filter, different from the conventional deposition image obtained for paramagnetic particles. The length of the spike structure tends to be long with lower flow velocity and lower applied magnetic field. It was also observed that the chain structure or the bundle of such chaines were formed on the way to the filter under the condition of the low applied magnetic field and low flow rates. Results obtained here indicate that the effect of deposited particles on the spatial distribution of the magnetic field and the hydrodynamics, they are often ignored in the simulation so far, should be considered appropriately.

A Superconducting High Gradient Magnetic Separation (HGMS) system is proposed for treatment of feed-water in thermal power plant [1]. This is a method to remove the iron scale from feed-water utilizing magnetic force. One of the issues for practical use of HGMS system is to extend continuous operation period. In this study, we designed the magnetic filters by particle trajectory simulation and HGMS experiments in order to solve this problem. As a result, the quantity of magnetite captured by each filter was equalized and filter blockage was prevented. A design method of the magnetic filter was proposed which is suitable for the long-term continuous scale removal in the feed-water system of the thermal power plant.