Feature interaction is a newly proposed feature relevance relationship, but the unintentional removal of interactive features can result in poor classification performance for this relationship. However, traditional feature selection algorithms mainly focus on detecting relevant and redundant features while interactive features are usually ignored. To deal with this problem, feature relevance, feature redundancy and feature interaction are redefined based on information theory. Then a new feature selection algorithm named CMIFSI (Conditional Mutual Information based Feature Selection considering Interaction) is proposed in this paper, which makes use of conditional mutual information to estimate feature redundancy and interaction, respectively. To verify the effectiveness of our algorithm, empirical experiments are conducted to compare it with other several representative feature selection algorithms. The results on both synthetic and benchmark datasets indicate that our algorithm achieves better results than other methods in most cases. Further, it highlights the necessity of dealing with feature interaction.

Optimal control theory and reinforcement learning are gradually being used in the field of industrial control. In this article, a new optimal tracking control scheme for 160 MW boiler-turbine systems is proposed based on an online policy iteration integral reinforcement learning (IRL) method. Firstly, a self-learning state tracking control with a cost function is developed to deal with the optimal tracking control problems for the boiler-turbine nonlinear system. Then with a modified cost function, a policy iteration-based IRL method is introduced to obtain the optimal control law. Finally, the monotonicity and the convergence of the cost function is analyzed and the detailed implementation of the policy iteration-based IRL method is provided via neural networks. The simulation results show that the control of the boiler-turbine system can converge in a short time by using this online iterative method. Through a theoretical simulation case, it can be concluded that the proposed method is more advanced compared with the MPC method.

Field observations of coastal regions are important for studying physical and biological features. Observations of high-resolution coastal phenomena were obtained by using a tow-yo instrument and a turbulence profiler at Daya Bay in the South China Sea in October 2015. Details of coastal phenomena, including warm water from a nuclear plant discharge, as well as an upwelling, and front, were obtained. The upwelling, with a width of 2 km, resulted in saltier and more turbid water near the bottom, with low chlorophyll-a and dissolved oxygen contents being transported upward to the surface layer and changing the local water environment. The front, with the lateral salinity variations as large as 0.7 psu across 1 km, was active at the water intersection of the South China Sea and Daya Bay. Such events commonly form during weak stratification periods in autumn. Continuous measurements from VMP-250 profiler over circa 22 h revealed active fronts and an averaged dissipation rate of 8 × 10−8 W/kg and diffusivity of 5.8 × 10−5 m2/s (i.e., one order of magnitude larger than in the open ocean) in the thermocline. The front was accompanied by strong mixing, indicating that it had formed at the intersection of different water masses and played an important role in energy dissipation in Daya Bay, further affecting the distribution of ecological elements.

No abstract

This paper proposes an optimal trajectory planning model for automated on‐ramp merging. The minimum cost path based on the optimal control theory is selected as the optimal trajectory of the on‐ramp merging vehicle, which could avoid potential collisions and satisfy the kinematic constraints on the vehicle motion. Moreover, an optimal control strategy is presented to solve the trajectory planning problem of the facilitating vehicle. An analytical closed‐form solution is derived by using the Hamiltonian analysis and the path information of the merging vehicle. Particularly, the proposed planning process also considers the lateral movement of the on‐ramp merging, which is more efficient in the application. Owing to the location and time that the on‐ramp vehicle merges into the mainline can be obtained endogenously by its planning process, the model has high adaptation to various on‐ramp environments. Solutions to the above two optimal methods are implemented in a model predictive control framework to cope with possible external disturbances. Several numerical simulations and PreScan‐Simulink Co‐Simulation illustrate the effectiveness of the proposed model. Furthermore, this methodology is compared with a typical CACC strategy to demonstrate its potential to reduce fuel consumption, as well as improve passenger comfort.