A Novel Hybrid Intelligent SOPDEL Model with Comprehensive Data Preprocessing for Long-Time-Series Climate Prediction
نویسندگان
چکیده
Long-time-series climate prediction is of great significance for mitigating disasters; promoting ecological civilization; identifying change patterns and preventing floods, drought typhoons. However, the general public often struggles with complexity extensive temporal range meteorological data when attempting to accurately forecast extremes. Sequence disorder, weak robustness, low characteristics interpretability are four prevalent shortcomings in predicting long-time-series data. In order resolve these deficiencies, our study gives a novel hybrid spatiotemporal model which offers comprehensive preprocessing techniques, focusing on decomposition, feature extraction dimensionality upgrading. This provides feasible solution puzzling problem long-term prediction. Firstly, we put forward Period Division Region Segmentation Property Extraction (PD-RS-PE) approach, divides into stationary series (SS) an Extreme Learning Machine (ELM) oscillatory (OS) Long Short-term Memory (LSTM) accommodate changing trend sequences. Secondly, new type input-output mapping mode three-dimensional matrix was constructed enhance robustness Thirdly, implemented multi-layer technique extract features high-speed input based Deep Belief Network (DBN) Particle Swarm Optimization (PSO) parameter searching neural network, thereby enhancing overall system’s learning ability. Consequently, by integrating all above innovative technologies, SS-OS-PSO-DBN-ELM-LSTME (SOPDEL) established improve quality forecasting. Five models featuring partial enhancements discussed this paper three state-of-the-art classical were utilized comparative experiments. The results demonstrated that majority evaluation indices exhibit significant optimization proposed model. Additionally, relevant system showed “Excellent Prediction” “Good exceeds 90%, no “Bad appear, so accuracy process obviously insured.
منابع مشابه
Preprocessing Time Series Data for Classification with Application to CRM
We develop an innovative data preprocessing algorithm for classifying customers using unbalanced time series data. This problem is directly motivated by an application whose aim is to uncover the customers’ churning behavior in the telecommunication industry. We model this problem as a sequential classification problem, and present an effective solution for solving the challenging problem, wher...
متن کاملVisual-Interactive Preprocessing of Time Series Data
Time series data is an important data type in many different application scenarios. Consequently, there are a great variety of approaches for analyzing time series data. Within these approaches different strategies for cleaning, segmenting, representing, normalizing, comparing, and aggregating time series data can be found. When combining these operations, the time series analysis preprocessing...
متن کاملPrediction of Rainfall Time Series Using Modular Artificial Neural Networks Coupled with Data Preprocessing Techniques
1 Dept. of Civil and Structural Engineering, Hong Kong Polytechnic University, 4 Hung Hom, Kowloon, Hong Kong, People’s Republic of China 5 6 2 Changjiang Institute of Survey, Planning, Design and Research, 7 Changjiang Water Resources Commission, 8 430010, Wuhan, HuBei, People’s Republic of China 9 10 3 Department of Civil Engineering, Ryerson University, 11 350 Victoria Street, Toronto, Ontar...
متن کاملHybrid Criteria for Nearest Neighbor Selection with Avoidance of Biasing for Long Term Time Series Prediction
Nearest neighbor is pattern matching method for time series prediction in which most recent values of the time series are compared with previous available values and forecasting is achieved by finding the best match pattern (nearest neighbor). Usually Euclidean distance is used to check the similarity of pattern. In this paper two hybrid criteria of pattern matching are being proposed and evalu...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Remote Sensing
سال: 2023
ISSN: ['2072-4292']
DOI: https://doi.org/10.3390/rs15071951