The timescale of the energy-release process of an energy storage system has put forward higher requirements with the increasing proportion of new energy power generation in the power grid. In this paper, a new type of compressed-air energy storage system with an ejector and combustor is proposed in order to realize short-timescale and long-timescale energy-release processes under the non-supplementary combustion condition and ejector supplementary combustion condition, respectively. A simulation model of the new system is established in APROS software. The results of this study show that the new system can realize continuous power output when energy storage and energy release operate simultaneously, and especially when the ejector coefficient is 0.8 and burner thermal power is 10 MW, the power-generation time is 12.45 h and the total generated power is 140,052 kW∙h, which are 15.6 and 17.5 times greater those of the short-timescale condition, respectively. In summary, the compressed-air energy storage system with an ejector and combustor that is proposed in this paper can flexibly meet the demands of multiple timescales’ power generation.
Channel fading and impulse noise in low-voltage station areas severely hinder data acquisition, and effective estimation of dual-mode channel data and state is the critical prerequisite for adaptive transmission. To address the impact of impulse noise on the performance of Orthogonal Frequency Division Multiplexing (OFDM) power line communication systems, this paper proposes a wireless-assisted joint estimation algorithm of channel fading and impulse noise in dual-mode OFDM systems. Firstly, a model of the dual-mode OFDM system combining power line and wireless links is established, and the initial channel estimation is obtained based on Linear Minimum Mean Square Error (LMMSE), the wireless link is used as an aid to provide a relatively reliable initial symbol estimation for the power line link. Then, a two-stage approach for initial impulse noise estimation is proposed: firstly, the possible impulse noise is located based on constructed residuals to form a support set, followed by refinement of the impulse amplitude using Single Measurement Vector Sparse Bayesian Learning (SMV-SBL). Finally, a decision-feedback iterative update is performed to iteratively optimize the symbol detection, channel estimation, and impulse noise estimation processes. Combined with adaptive dual-mode fusion, this approach improves estimation accuracy. Simulation results demonstrate that the proposed algorithm achieves significant improvements in estimation accuracy and bit error rate.