Synthetic resampling strategies and machine learning for digital soil mapping in Iran

Taghizadeh‐Mehrjardi, Ruhollah; Schmidt, Karsten; Eftekhari, Kamran; Behrens, Thorsten; Jamshidi, Mohammad; Davatgar, Naser; Toomanian, Norair; Scholten, Thomas

doi:10.1111/ejss.12893

Cited by 47 publications

(29 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the system cannot capture enough information on the transitions between health states to improve forecasting horizon while ensuring good prediction accuracy. Most common machine learning algorithms usually work well on balanced training sets, that is, datasets in which all classes are approximately represented equally [50]. Because these algorithms treat all misclassifications equally, they bias classes with many instances, resulting in false accuracy estimates.…”

Section: F Classification Resultsmentioning

confidence: 99%

“…Because these algorithms treat all misclassifications equally, they bias classes with many instances, resulting in false accuracy estimates. Therefore, the accuracy estimates may be unreliable and classes with only a few values are often misclassified or neglected [50]. This issue is known as a class imbalance problem in machine learning.…”

Section: F Classification Resultsmentioning

confidence: 99%

“…This issue is known as a class imbalance problem in machine learning. Datasets that do not meet this criterion are referred to as imbalanced data [50]. Most condition monitoring datasets are imbalanced data in some extent.…”

Section: F Classification Resultsmentioning

confidence: 99%

“…Time series of vibration excursions collected from heavy machinery in non-stationary operations under typical operating conditions in mine sites usually are not equal in data density for each class. It can additionally be influenced by the sampling strategy [50].…”

Section: F Classification Resultsmentioning

confidence: 99%

See 3 more Smart Citations

Data-Driven Condition Monitoring of Mining Mobile Machinery in Non-Stationary Operations Using Wireless Accelerometer Sensor Modules

et al. 2021

View full text Add to dashboard Cite

This paper presents the development of an easy-to-deploy and smart monitoring IoT system that utilizes vibration measurement devices to assess real-time condition of bulldozers, power shovels and backhoes, in non-stationary operations in the mining industry. According to operating experience data and the type of mining machine, total loss failure rates per machine fleet can reach up to 30%. Vibration analysis techniques are commonly used for condition monitoring and early detection of unforeseen failures to generate predictive maintenance plans for heavy machinery. However, this maintenance strategy is intensively used only for stationary machines and/or mobile machinery in stationary operations. Today, there is a lack of proper solutions to detect and prevent critical failures for non-stationary machinery. This paper shows a cost-effective solution proposal for implementing a vibration sensor network with wireless communication and machine learning data-driven capabilities for condition monitoring of non-stationary heavy machinery in mining operations. During the machine operation, 3-axis accelerations were measured using two sensors deployed across the machine. The machine accelerations (amplitudes and frequencies) are measured in two different frequency spectrums to improve each sensing location's time resolution. Multiple machine learning algorithms use this machine data to assess conditions according to manufacturer recommendations and operational benchmarks Proposed data-driven machine learning models classify the machine condition in states according to the ISO 2372 standards for vibration severity: Good, Acceptable, Unsatisfactory, or Unacceptable. After performing field tests with bulldozers and backhoes from different manufacturers, the machine learning algorithms are able to classify machine health status with an accuracy between 85% -95%. Moreover, the system allows early detection of ''Unacceptable'' states between 120 to 170 hours prior to critical failure. These results demonstrate that the proposed system will collect relevant data to generate predictive maintenance plans and avoid unplanned downtimes.

show abstract

Section: F Classification Resultsmentioning

confidence: 99%

Section: F Classification Resultsmentioning

confidence: 99%

Section: F Classification Resultsmentioning

confidence: 99%

Section: F Classification Resultsmentioning

confidence: 99%

See 2 more Smart Citations

Data-Driven Condition Monitoring of Mining Mobile Machinery in Non-Stationary Operations Using Wireless Accelerometer Sensor Modules

et al. 2021

View full text Add to dashboard Cite

show abstract

“…About 80% of the case studies used at least one tree-based algorithm such as regression tree (e.g. Taghizadeh-Mehrjardi et al, 2019b;Heung et al, 2016), random forest (e.g. Häring et al, 2012) 2017; Ramcharan et al (2018).…”

Section: Machine Learning Modelsmentioning

confidence: 99%

Machine learning for digital soil mapping: applications, challenges and suggested solutions

Wadoux¹,

Minasny²,

McBratney³

2020

Preprint

View full text Add to dashboard Cite

The uptake of machine learning (ML) algorithms in digital soil mapping (DSM) is transforming the way soil scientists produce their maps. Machine learning is currently applied to mapping soil properties or classes much in the same way as other unrelated fields of science. Mapping of soil, however, has unique aspects which require adaptations of the ML algorithms. These features are for example, but not limited to, the inclusion of pedological knowledge into the ML algorithm, the accounting of spatial structure present in the soil data, or the desire to increase our scientific understanding of the distribution and genesis of soil from a calibrated ML model. Tackling these challenges is critical for machine learning to gain credibility and scientific consistency in soil science. In this article, we review the current applications of machine learning in digital soil mapping and suggest improvements. We found a growing interest of the use of ML in DSM. Most studies focus on obtaining accurate maps and disregard the characteristics of soil data, such as spatial autocorrelation. Only a few studies account for existing soil knowledge or quantify the uncertainty of the predicted maps. We then discuss the challenges related to the application of ML for soil mapping and offer solutions from existing studies in the natural sciences. The challenges are organized as follows: sampling, resampling, accounting for the spatial information, multivariate mapping, uncertainty analysis, validation, integration of pedological knowledge and, interpretation of the models. We conclude that for future developments, machine learning should incorporate three core elements: plausibility, interpretability, and explainability, which will trigger soil scientists to move beyond model prediction and towards explanation of soil processes.

show abstract

3–4D soil model as challenge for future soil research: Quantitative soil modeling based on the solid phase

Gerke

Vogel

Weber

et al. 2022

J. Plant Nutr. Soil Sci.

Self Cite

View full text Add to dashboard Cite

A 3-4D soil model represents a logical step forward from one-dimensional soil columns (1D), two-dimensional soil maps (2D), and three-dimensional soil volumes (3D) toward dynamic soil models (4D), with time as the fourth dimension. The challenge is to develop modeling tools that account for the states of soil properties, including the spatial structure of solids and pores, as well as their dynamics, including soil mass and solute transfers in landscapes. Our envisioned 3-4D soil model approach aims at improving the capability to predict fundamental soil functions (e.g., plant growth, storage, matter fluxes) that provide ecosystem services in the socioeconomic context.This study provides a structured overview on current soil models, challenges, open questions, and urgent research needs for developing a 3-4D soil model. A 3-4D soil model should provide an inventory of spatially distributed and temporally variable soil properties. As basis for this, we propose a mass balance model for the solid phase, which needs to be supplemented by a model describing its structure. This should eventually provide adequate 3D parameter sets for the numerical modeling of soil functions (e.g., flow and transport). The target resolution is decameters in the horizontal plane and centimeters to decimeters in the vertical direction to represent characteristic soil properties and soil horizons. The actual state of soils and their properties can be estimated from spatial data that represent the soil forming factors, with the use of machine learning tools. Improved modeling of the dynamics of soil bulk density, biological processes, and the pore structure are required to relate the solid mass balance to matter fluxes.A 3-4D soil model can be built from several types of modeling approaches. We distinguish between (1) process models that simulate mass balances, fluxes and soil structure dynamics, (2) statistical pedometric models using machine learning and geostatistics to estimate the soil inventory within landscapes, and (3) pedotransfer functions to link observable attributes to specific model parameters required to simulate soil functions including water and matter fluxes. This should provide the prerequisites to predict the spatial distribution of soil functions and their changes in response to external forcing. This endeavor can draw upon many already established models and techniques, yet combining them into a newly created 3-4D soil model is a truly an ambitious, but promisingThis is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

show abstract

Synthetic resampling strategies and machine learning for digital soil mapping in Iran

Cited by 47 publications

References 48 publications

Data-Driven Condition Monitoring of Mining Mobile Machinery in Non-Stationary Operations Using Wireless Accelerometer Sensor Modules

Data-Driven Condition Monitoring of Mining Mobile Machinery in Non-Stationary Operations Using Wireless Accelerometer Sensor Modules

Machine learning for digital soil mapping: applications, challenges and suggested solutions

3–4D soil model as challenge for future soil research: Quantitative soil modeling based on the solid phase

Contact Info

Product

Resources

About