Research on air Pollution Gases Recognition Method Based on LSTM Recurrent Neural Network and Gas Sensors Array
| Autor: | Taiyang Xie, Qingfeng Wang, Shuaiqi Wang |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Scale (ratio)
Computer science business.industry Deep learning 010401 analytical chemistry Air pollution Process (computing) Pattern recognition 02 engineering and technology 021001 nanoscience & nanotechnology medicine.disease_cause 01 natural sciences 0104 chemical sciences Nonlinear system Identification (information) Recurrent neural network Pattern recognition (psychology) medicine Artificial intelligence 0210 nano-technology business |
| Zdroj: | 2018 Chinese Automation Congress (CAC). |
| DOI: | 10.1109/cac.2018.8623060 |
| Popis: | The integration of sensors array and pattern recognition to replace large scale analytical instruments is an important and feasible method for accurate and rapid measurement of atmospheric pollution gases. This paper proposed a quantitative detection method of mixed gases based on long short-term memory (LSTM) recurrent neural network, including data pre-processing, network structure design, model training and prediction process. This method can extract the deep characteristics of sensors array's responses automatically and match the complex nonlinear characteristics more accurately. It is proved that the proposed LSTM prediction model has a strong applicability and higher accuracy in the concentration identification of gas mixtures by comparing with the experiments of existing models. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|