A critical systematic review on spectral-based soil nutrient prediction using machine learning

Environmental Monitoring and Assessment, Journal Year: 2024, Volume and Issue: 196(8)

Published: July 4, 2024

Language: Английский

---

Explainable AI for Earth observation: current methods, open challenges, and opportunities

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown, P. 115 - 152

Published: Jan. 1, 2024

Language: Английский

---

Half-hourly electricity price prediction model with explainable-decomposition hybrid deep learning approach

Energy and AI, Journal Year: 2025, Volume and Issue: unknown, P. 100492 - 100492

Published: March 1, 2025

Language: Английский

---

Prediction of Crops Based on a Machine Learning Algorithm

2022 International Conference on Computer Communication and Informatics (ICCCI), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 23, 2023

Agriculture is crucial to any country's economy. Farmers around the world face a constant challenge in trying keep up with rising demand for food crops of fluctuating climates and an alarming rise population. One most widely grown cereals, wheat supplies significant portion world's main supply. This heat-sensitive crop being severely harmed by unusual environmental temperature decrease amount rainfall. Scientists from all have been looking at what are called Climate Sustainable Practices effort boost yields while reducing impact on environment. Predicting before harvest can assist scientists farmers evaluate risks implement preventative actions maintain consistent agricultural harvest. There two types models used predict yields: growth data-driven models. The time, money, accuracy costs associated using stem fact that these methods sensitive variables. So, farmer can't do anything nick time his crop's production. With advent machine learning algorithms, become even more effective fraction cost traditional empirical Machine come long way, but they haven't completely precise output forecasting. this study, authors want provide reliable method estimating future harvests one India's Punjab provinces. For timely yield prediction, KNN DT hybrid model proposed. To further improve model's performance, researchers genetic algorithm tune KNN-two DT's hyper parameters: size its window number neurons hidden layer. study has also examined factors order isolate important parameters regulation monitoring accurately yield. proposed predicting was tested battery trials. effectiveness suggested validated through comparative comparison state-of-the-art approaches prediction. Farmers, policymakers, planners benefit greatly improving their ability make informed decisions take corrective action yields.

Language: Английский

---

A hyperspectral metal concentration inversion method using attention mechanism and graph neural network

Ecological Informatics, Journal Year: 2024, Volume and Issue: 83, P. 102792 - 102792

Published: Aug. 25, 2024

Soil heavy metal contamination has emerged as a global environmental concern, posing significant risks to human health and ecosystem integrity. Hyperspectral technology, with its non-invasive, non-destructive, large-scale, high spectral resolution capabilities, shows promising applications in monitoring soil pollution. Traditional methods are often time-consuming, labor-intensive, expensive, limiting their effectiveness for rapid, large-scale assessments. This study introduces novel deep learning method, SpecMet, estimating concentrations naturally contaminated agricultural soils using hyperspectral data. The SpecMet model extracts features from data convolutional neural networks (CNNs) achieves end-to-end prediction of by integrating attention mechanisms graph networks. Results demonstrate that the OR-SpecMet model, which utilizes raw data, optimal performance, significantly surpassing traditional machine such multiple linear regression, partial least squares support vector regression lead (Pb), copper (Cu), cadmium (Cd), mercury (Hg). Moreover, training specialized models individual metals better accommodates unique spectral-concentration relationships, enhancing overall estimation accuracy while achieving 20.3 % improvement predicting low-concentration mercury. method showcases superior performance extensive application potential techniques precise pollution monitoring, offering new insights reliable technical prevention protection. code datasets used this publicly available at: https://github.com/zhang2lei/metal.git.

Language: Английский

---

Machine Learning for Smart Agriculture: A Comprehensive Survey

IEEE Transactions on Artificial Intelligence, Journal Year: 2023, Volume and Issue: 5(6), P. 2568 - 2588

Published: Dec. 20, 2023

As communication technologies and equipment evolve, smart assets become smarter. The agricultural industry is also evolving in line with the implementation of modern protocols, intelligent sensors, equipment. This evolution enabling large-scale production processes to operate independently, thus, securing food supply chain for an ever-growing population. Data processing such a system multiple heterogeneous sources requires proper management effective operations. Recognizing advantages Machine Learning(ML) performing data processing, researchers are investigating ML design architecture. aim this paper provide thorough analysis state-of-the-art agriculture, open challenges, guidelines development further enhanced agriculture systems. Specifically, we describe how used create systems supported by technology.

Language: Английский

---

Convolutional Neural Networks Applied to Antimony Quantification via Soil Laboratory Reflectance Spectroscopy in Northern Portugal: Opportunities and Challenges

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(11), P. 1964 - 1964

Published: May 30, 2024

Antimony (Sb) has gained significance as a critical raw material (CRM) within the European Union (EU) due to its strategic importance in various industrial sectors, particularly textile industry for flame retardants and component of Sb-based semiconductor materials. Moreover, Sb is emerging potential alternative anodes used lithium-ion batteries, key element energy transition. This study explored feasibility identifying quantifying mineralisations through spectral signature soils using laboratory reflectance spectroscopy, non-invasive remote sensing technique, by employing convolutional neural networks (CNNs). Standard signal pre-processing techniques were applied data, analysed inductively coupled plasma mass spectrometry (ICP-MS). Despite achieving high R-squared (0.7) values an RMSE 173 ppm Sb, faces significant challenge generalisation model new data. limitations, this provides valuable insights into strategies future research field.

Language: Английский

---

Performance enhancement in hydroponic and soil compound prediction by deep learning techniques

PeerJ Computer Science, Journal Year: 2024, Volume and Issue: 10, P. e2101 - e2101

Published: June 10, 2024

The soil quality plays a crucial role in providing essential nutrients for crop growth and ensuring bountiful yield. Identifying the composition, which includes sand, silt particles, mixture of clay specific proportions, is vital making informed decisions about selection managing weed growth. Furthermore, pollution from emerging contaminants presents substantial risk to water resource management food production. Developing numerical models comprehensively describe transport reactions chemicals within both plants utmost importance crafting effective mitigation strategies. To address limitations traditional models, this paper devises an innovative approach that leverages deep learning predict hydroponic compound dynamics during plant This method not only enhances understanding how interact with their environment but also aids more agriculture, ultimately contributing sustainable efficient data needed perform developed prediction model acquired online resources. After that, forwarded feature extraction phase. weighted features, belief network (DBN) original features are achieved stage. get weights optimally obtained using Iteration-assisted Enhanced Mother Optimization Algorithm (IEMOA). Subsequently, these extracted fed into Multi-Scale fusion-based Convolution Autoencoder Gated Recurrent Unit (MS-CAGRU) prediction. Thus, attained end. Finally, performance evaluation suggested work conducted contrasted numerous conventional showcase system's efficacy.

Language: Английский

---

Estimation of lithium content in rock debris based on spectral feature coefficients

Ore Geology Reviews, Journal Year: 2024, Volume and Issue: 171, P. 106167 - 106167

Published: July 14, 2024

Hyperspectral remote sensing is a fast and non-destructive technology for identifying geological information, many successful cases have been achieved in mineral identification estimation of soil heavy metal content. However, there fewer studies on the application this to rare metals, especially detection lithium (Li) resources. Whether hyperspectral process can effectively identify Li anomalies significant expanding exploration To end, study explores potential techniques elemental content by collecting rock debris samples field extracting spectral feature coefficients using Gaussian Mixture Model (GMM). The results show that (1) parameter extraction technique based GMM quickly accurately extract absorption parameter. (2) Compared with reflectance, improve correlation content, constructed model more effective. (3) full-width at half maximum (FWHM) 1.93 μm most effective, determination (R2), relative root mean squared error (RRMSE), ratio performance deviation (RPD) 0.61, 0.516 1.601, respectively, which are significantly better than reflectance model. above use estimate debris, provides technical reference regional airborne support improving efficiency resources narrowing focus investigation.

Language: Английский

---

A New Porosity Prediction Method Based on Deep Learning of TabNet Algorithm

Published: Feb. 24, 2023

Porosity is one of the core parameters in process understanding subsurface fluid flow, reservoir characterization, and evaluation. Due to limitation number cores taken, conventional experimental analysis can only obtain a small amount porosity data. How improve accuracy prediction has always been an issue. It hotspots study parameter prediction. This paper proposes method based on Tab Net, compares it with traditional machine learning LSTM methods. The results show that RMSE three models are 3.48,3.67, 3.95 respectively. Therefore, this believes model TabNet network more effectively predict provide new for evaluation ideas

Language: Английский

---