• Energy Research

Abstract As a representation of smart and green city development, bike-sharing system is one of the hottest topic in the fields of transportation, public health, urban planning, and so on. With the development of Mobility as a Service (MaaS), emerging technologies such as mobile data mining give some new solutions for optimizing bike-sharing system and predicting the emission reduction. Here, we propose a bike-sharing layout optimization and emission reduction potential analysis structure under the concept of MaaS. A human travel mode detection method and a geometry-based probability model are proposed to support the particle swarm optimization process. We implement a comparison study to analyze the computational efficiency. Taking Setagaya ward, Tokyo as the study case with about 3 million GPS trajectories, the result shows that with the increase of station number from 30 to 90, the adoption of bike-sharing system can reduce about 3.1-3.8 thousand tonnes of CO2 emission.

Abstract As a representation of smart and green city development, bike-sharing system is one of the hottest topic in the fields of transportation, public health, urban planning, and so on. With the development of Mobility as a Service (MaaS), emerging technologies such as mobile data mining give some new solutions for optimizing bike-sharing system and predicting the emission reduction. Here, we propose a bike-sharing layout optimization and emission reduction potential analysis structure under the concept of MaaS. A human travel mode detection method and a geometry-based probability model are proposed to support the particle swarm optimization process. We implement a comparison study to analyze the computational efficiency. Taking Setagaya ward, Tokyo as the study case with about 3 million GPS trajectories, the result shows that with the increase of station number from 30 to 90, the adoption of bike-sharing system can reduce about 3.1-3.8 thousand tonnes of CO2 emission.

Abstract Considering the tremendous economic losses and human injury caused by pipeline leaks, it is critical to detect and locate the pipeline leakage in time. This work proposes a generative adversarial networks (GANs) framework for leak detection and localization from the perspective of data science instead of physical meaning. The GANs are designed by two powerful neural networks: generative (G) network and discriminative (D) network. Real experiments are performed to verify the effectiveness of the proposed GANs framework, confirming that it can be applied to pipeline leakages for the estimations of the location, coefficient, and the starting time. To qualify the performance of the approach, sensitivity analysis for the structure of the GANs framework is evaluated. Finally, the proposed generative model is validated by two pipeline leakages. The errors of these two examples are 3.9% and 3.5%, respectively, indicating that the proposed method is better than the improved PSO and ANN.

Abstract Considering the tremendous economic losses and human injury caused by pipeline leaks, it is critical to detect and locate the pipeline leakage in time. This work proposes a generative adversarial networks (GANs) framework for leak detection and localization from the perspective of data science instead of physical meaning. The GANs are designed by two powerful neural networks: generative (G) network and discriminative (D) network. Real experiments are performed to verify the effectiveness of the proposed GANs framework, confirming that it can be applied to pipeline leakages for the estimations of the location, coefficient, and the starting time. To qualify the performance of the approach, sensitivity analysis for the structure of the GANs framework is evaluated. Finally, the proposed generative model is validated by two pipeline leakages. The errors of these two examples are 3.9% and 3.5%, respectively, indicating that the proposed method is better than the improved PSO and ANN.

Abstract The pressure changes dramatically during the shutdown process of the multi-product pipeline. When the pipeline pressure comes to decrease, it is often mistaken as pipeline leakage or other abnormal condition which increases the burden of the operator on-site. At present, the method of pipeline shutdown pressure analysis is mainly based on numerical simulation which can not monitor shutdown pressure in real-time. In this work, the time-series approximate ability of long short-term memory (LSTM) is taken advantage of to construct a shutdown pressure prediction model. To overcome the drawback of this deep learning algorithm that is trained only by ample data, the scientific principle and theory are integrated into LSTM. Subsequently, the theory-guided long short-term memory (TG-LSTM) is proposed for pipeline shutdown pressure prediction. The proposed model is trained with available data and simultaneously guided by the theory (physical principle and engineering theory) of the underlying problem. In the training process, the data mismatch, as well as monotonicity constraints, and boundary constraints are coupled into loss function. After acquiring the parameters of the neural network, a TG-LSTM model is established which not only fits the data, but also follows the physical principle and the engineering theory. The proposed model is verified by three real-world multi-product pipelines. The results indicate that TG-LSTM achieves better accuracy than other prediction models, with MAPE being 0.246%, 0.186%, and 0.143%, respectively. Finally, the sensitivity analysis of different hyper-parameter is conducted to illustrate the robustness of TG-LSTM in pipeline shutdown pressure prediction.

Abstract The pressure changes dramatically during the shutdown process of the multi-product pipeline. When the pipeline pressure comes to decrease, it is often mistaken as pipeline leakage or other abnormal condition which increases the burden of the operator on-site. At present, the method of pipeline shutdown pressure analysis is mainly based on numerical simulation which can not monitor shutdown pressure in real-time. In this work, the time-series approximate ability of long short-term memory (LSTM) is taken advantage of to construct a shutdown pressure prediction model. To overcome the drawback of this deep learning algorithm that is trained only by ample data, the scientific principle and theory are integrated into LSTM. Subsequently, the theory-guided long short-term memory (TG-LSTM) is proposed for pipeline shutdown pressure prediction. The proposed model is trained with available data and simultaneously guided by the theory (physical principle and engineering theory) of the underlying problem. In the training process, the data mismatch, as well as monotonicity constraints, and boundary constraints are coupled into loss function. After acquiring the parameters of the neural network, a TG-LSTM model is established which not only fits the data, but also follows the physical principle and the engineering theory. The proposed model is verified by three real-world multi-product pipelines. The results indicate that TG-LSTM achieves better accuracy than other prediction models, with MAPE being 0.246%, 0.186%, and 0.143%, respectively. Finally, the sensitivity analysis of different hyper-parameter is conducted to illustrate the robustness of TG-LSTM in pipeline shutdown pressure prediction.