Energy flow distribution prediction method and system for regional integrated energy system

Abstract

The invention discloses an energy flow distribution prediction method and a system for a regional integrated energy system. The energy flow distribution prediction method comprises the following steps: collecting multi-energy flow network data; constructing a multi-energy flow network data representation model; mining multi-source spatial and temporal features related to energy flow distribution;and outputting an energy flow distribution prediction result. The system comprises a multi-energy flow network data representation module, an energy flow distribution spatial feature extraction module, an energy flow distribution timing feature extraction module, an energy flow distribution external factor feature extraction module, an energy flow distribution feature fusion module and an energy flow distribution prediction output module. According to the energy flow distribution prediction method and the system for the regional integrated energy system, multi-energy flow network data is converted and analyzed based on a network representation learning method and a deep learning method, spatial and temporal features and external factor features related to energy flow distribution are mined, and a new method and a technical base are provided for the problem of energy flow distribution state prediction for the regional integrated energy system.

Description

technical field [0001] The invention relates to a method for predicting energy flow distribution of a regional comprehensive energy system. Background technique [0002] Energy flow distribution prediction is an important part of energy system layout planning and operation optimization, and is the basic link to effectively improve energy utilization efficiency and alleviate energy and environmental problems. Moreover, compared with the traditional single energy system, the regional integrated energy system is a complex multi-network flow system formed by the cooperative coupling of multiple energy networks. Spatial correlation, spatiotemporal modeling and dynamic prediction are very difficult. Therefore, under the integrated demand response of the multi-energy flow system, an integrated prediction method based on multi-energy flow spatio-temporal modeling and multi-source heterogeneous data fusion is needed to predict the future energy flow distribution state. [0003] At ...

AI Technical Summary

Problems solved by technology

With the development and breakthrough of sensor measurement technology, network communication technology, computer technology and data processing methods, massive spatio-temporal data have been accumulated. For the task of mining spatial correlation and temporal correlation, the deep learning model developed in recent years can Exhibiting good information expression ability, robustness and generalization, existing research shows that recurrent neural network is very effective for deep mining of time series (J.Cao et al., Science China Information Sciences, 2017, 60(3 ):032201.), in which the long-term short-term memory network is a special recurrent neural network, which has great advantages in processing and predicting time series with long intervals and delays, but it cannot understand and reflect the various energy sources in each energy node. Influence of spatial interaction coupling on the time series variation of each energy flow state

Deep convolutional neural networks are very effective for spatial data mining (A.Krizhevsky et al., Advances in Neural Information Processing Systems, 2012, 1097-1105.), and have been extended to many scenarios by combining with cyclic neural networks , but in the field of regional integrated energy systems, the interactive coupling relationship in the spatial distribution of various energy flow networks and the periodic characteristics on different time scales have not been considered

At the same time, in the face of multi-source heterogeneous time, space, and external factor characteristics and attributes, it has not yet been considered based on semantic construction that can fully mine the timing, proximity, periodicity, trend, and spatial characteristics of spatio-temporal data in terms of time attributes. The deep learning integrated model of distance, hierarchy, coupling characteristics and the global influence relationship of external factors, but cannot map and process multi-source heterogeneous data uniformly based on its network topology and time series, and mine more comprehensive features to improve prediction accuracy and generalization performance

Images