Analysing Near Earth Object & Visualizing Hazards

Swathi K Amirthavarshini P M
Vinora A
Department Of Information Department Of Information
Department Of Information
Technology Technology
Technology
Velammal College Of Engineering And Velammal College Of Engineering And
Velammal College Of Engineering And
Technology Technology
Technology
Madurai,TamilNadu Madurai,TamilNadu
Madurai,TamilNadu
[email protected] [email protected]
[email protected]
Nancy Deborah R SivaKarthi G
Soundarya M
Department Of Information Department Of Information
Department Of Information
Technology Technology
Technology
Velammal College Of Engineering And Velammal College Of Engineering And
Velammal College Of Engineering And
Technology Technology
Technology
Madurai,TamilNadu Madurai,TamilNadu
Madurai,TamilNadu
[email protected] [email protected]
[email protected]

Abstract—Analyzing Near-Earth Objects (NEOs) and Arizona ,NASA-Jet Proplusion Lab and industry partners. The
Predicting Hazards combines data exploration and machine project has advanced critical components and spacecraft
learning techniques to analyze data from NASA's NEO design, positioning it well for the upcoming detailed design
database. It aims to understand the characteristics of NEOs, phase, aligning with NASA's planetary defense goals[2].
predict potential hazards, and visualize the results.Key steps
include data preprocessing, exploratory analysis, machine Author L. Z. Ladeira et al introduced NEO-NDA (NEO
learning model training, and hazard predictions. The project Natural Language Data Augmentation) , a versatile tool for
offers insights into NEOs, provides a predictive model for generating and rebalancing textual datasets. Unlike many
identifying potentially hazardous objects, and also presents existing approaches, NEO-NDA operates with multiple
visualizations showcasing the distribution of NEOs and their languages, offering a wide range of modifications for
characteristics over time. generating novel data samples. Their system demonstrate that
it significantly enhances the performance of machine learning
Keywords— Near-Earth Objects (NEOs, data exploration, models, often doubling their effectiveness when compared to
data preprocessing, distribution analysis, characteristic analysis original datasets without augmentation[5].
R. N. Ranaweera et al employs deep neural networks to
I. INTRODUCTION
analyze the orbital data of Near-Earth Asteroids (NEAs),
Analysis and Classification of Near-Earth Objects (NEOs) allowing the model to learn intricate patterns and features
for Planetary Defense and Space Exploration: In this within the dataset. Through this approach, it classifies objects
ambitious and multidimensional undertaking, we leverage a as either harmful or harmless based on the acquired insights
rich dataset encompassing a wide array of Near-Earth Objects from the data, enhancing the efficiency of identifying and
(NEOs), including asteroids and comets. Employing a characterizing these celestial bodies to prevent potential
combination of data processing, deep learning, exploratory impacts on Earth[6].
analysis and ML methodologies, we delve into the realm of
NEOs to unearth profound insights into their characteristics, C. Rosu et al introduced a convolutional neural network model
orbital dynamics, potential threat assessment (Potentially for classifying astronomical images to detect asteroids. The
Hazardous Asteroids - PHAs), and in-depth classification. system is trained on original data from the Telescope of
This project's goal is to not only contribute significantly to the Newton in La Palma. The primary aim is to develop an
field of planetary defense but also to fuel the broader mechanized system for accurate asteroid detection while
exploration of NEOs in our celestial vicinity, opening doors to minimizing false negatives, achieving a 94.01% recall rate
new scientific frontiers and enhancing our understanding of when undergone training with solitary images[9].
the cosmos." This paper by V. Bahel et al underscores AI's growing
importance in space science for mitigating asteroid-related
II. RELATED WORKS risks. It conducts a meticulous understanding on Potentially
The Author G. Deva ram ganesh et al utilizes CNN Harmful Asteroids and employs supervised ML to assess
(Convolutional Neural Network) to catergorize hyperspectral asteroid hazard based on specific parameters. Among various
images based on both spectral and spatial features, enhancing classification algorithms, Random Forest stands out with an
classification accuracy. The CNN strategically processes average F1-score of 99.22% and an exceptional 99.9%
spectral and spatial properties within 13x13 patches. While accuracy offering a robust solution for asteroid risk
CNN demands more computation time than other classifiers, evaluation[10].
it outperforms them, achieving the best hyperspectral image H. Lin et al introduced an innovative ensemble method,
classification accuracy in simulation results[1]. (OBBStacking) designed for geo-spatial sensing object
T. Hoffman et al During the preliminary design phase, the identification with oriented bounding boxes (OBBs).
NEOS Project has made substantial progress towards its OBBStacking not only effectively combines OBB-based
objectives of identifying, characterizing, and cataloging Near- detection results but also leverages deep learning object
Earth Objects (NEOs). Collaborating with the University of detectors' confidence indicators. In the 2021, it outperformed
Gaofen Challenge on mechanised Earth Observation Image potential threat to Earth. This risk assessment is a crucial
Interpretation with High-Resolution, showcasing enhanced component of planetary defense efforts.
performance in experiments on the DOTA and FAIR1M
datasets[3]. 4. Deep Learning and Machine Learning Models: The
project explores the use of deep learning and ML models to
R. A. Mohammed et al discusses the increasing asteroid make predictions and classifications based on the dataset.
hazards in the asteroid belt located b/w Mars & Jupiter, within Random Forest, Artificial Neural Networks (ANNs) &
a range of 2.2 to 3.2 astronomical hjunits (AU). It highlights Gradient Boosting are employed to enhance the accuracy of
the use of planetary radars for tracking celestial objects and risk assessment & classification.
safeguarding Earth, detailing a simulation-based approach to
analyze radar cross-sectional data and model signal 5. Planetary Defense and Space Exploration Implications:
processing with MATLAB and Simulink to examine the Understanding NEOs and identifying potential threats is not
behavior of moving asteroids[13]. only vital for planetary defense but also for advancing space
exploration. By gaining a better understanding of these
celestial bodies, this project contributes to humanity's
III. PROPOSED METHODOLOGY knowledge of our solar system and our ability to prepare for
The proposed system aims to enhance the accuracy and and mitigate potential hazards.
efficiency of Near-Earth Object (NEO) classification using 6. Safety and Education: The project has implications for
advanced machine learning techniques. By leveraging state- safety and education. By sharing the results and insights with
of-the-art algorithms, including Random Forest, Neural the public, educational institutions, and space agencies,
Networks & Gradient Boosting, the system aims to improve this initiative contributes to raising awareness about NEOs
the predictive capabilities for identifying Potentially Harmful and the importance of planetary defense.
Asteroids among NEOs.The system's methodology involves
feature extraction, preprocessing, and training the machine 7. International Collaboration: Given the global nature of
learning models on a comprehensive dataset containing key the NEO threat, this project may involve international
NEO attributes. This approach enables the development of a collaboration, as organizations like NASA, ESA, and other
robust classification system capable of categorizing NEOs space agencies actively monitor NEOs.
based on various physical and orbital parameters.To address
challenges such as missing data and categorical variables, the PSEUDOCODE:
system employs data preprocessing techniques, ensuring the # Input: NEO Dataset
quality and completeness of the input dataset. The utilization
of machine learning algorithms facilitates more accurate # Output: Prediction of Potentially Hazardous Asteroids
predictions, contributing to the early identification of PHAs
Step 1: Load Dataset
and enhancing planetary defense capabilities.The proposed
system not only advances the accuracy of NEO classification neo_data = load_neo_dataset()
but also provides a scalable framework for ongoing research
Step 2: Data Preprocessing
and monitoring of celestial objects. By combining
astronomical data with cutting-edge machine learning neo_processed = preprocess_data(neo_data)
methodologies, the system stands to significantly impact our
Step 3: Split Dataset
understanding of NEOs and improve our ability to assess
potential threats from these cosmic bodies. train_set, test_set = split_dataset(neo_processed)
The Advanced Analysis and Classification of Near-Earth Step 4: Random Forest Classifier
Objects (NEOs) project is a multidisciplinary endeavor aimed
rf_model = initialize_random_forest_classifier()
at enhancing our understanding of NEOs, which are comets &
asteroids that come in close proximity to the Earth. The rf_train_acc, rf_test_acc = train_and_evaluate(rf_model,
primary objectives of this project are to: train_set, test_set)
Step 5: Gradient Boosting Classifier
1. Data Collection and Preparation: The project begins gb_model = initialize_gradient_boosting_classifier()
with gathering and processing a comprehensive dataset of
gb_train_acc, gb_test_acc = train_and_evaluate(gb_model,
NEOs. This dataset contains crucial information about these
celestial objects, including their names, sizes, albedo, orbital train_set, test_set)
parameters, and potential threat levels (Potentially Hazardous Step 6: Neural Network Model
Asteroids - PHAs).
nn_model = initialize_neural_network_model()
2. Exploratory Data Analysis (EDA): To acquire deeper
nn_train_acc, nn_test_acc = train_and_evaluate(nn_model,
understanding of the dataset, exploratory data analysis is
conducted. EDA involves data visualization and statistical train_set, test_set)
techniques to identify patterns and correlations among NEO Step 7: Display Results
attributes. This step helps in better understanding the
characteristics of NEOs. display_results(rf_train_acc, rf_test_acc, gb_train_acc,
gb_test_acc, nn_train_acc, nn_test_acc)
3. Classification and Risk Assessment: The project
involves utilizing machine learning algorithms to classify
NEOs into different categories based on their characteristics. # Input: NEO Dataset with Hazard Information
Particular attention is given to identifying PHAs that pose a
# Output: Visualization of Potentially Hazardous Asteroids diameter- Diameter of the object (km), albedo- Albedo
(reflectivity)
Step 1: Load Dataset with Hazard Information
neo_hazard_data = load_neo_dataset_with_hazard_info()
Step 2: Filter Potentially Hazardous Asteroids
pha_data =
filter_potentially_hazardous_asteroids(neo_hazard_data)
Step 3: Visualize Potentially Hazardous Asteroids
plot_potentially_hazardous_asteroids(pha_data).

Table.3.1 Accuracy table

The table 3.1 represents the accuracy of gradient boosting

classifier, ada boost classifier, Random forest classifier,
decision tree classifier , logistic Regression, K Neighbors
classifier & SVC. Among this it is clearly concluded that
Random forest classifier has higher accuracy than other
algorithms

Fig 3.1 feature importance for imporantant parameters

The above barchart 3.1 shows the feature importance for

imporantant parameters such as a-Semi-major axis,q-
Perihelion distance, i- Inclination (deg), e-Orbital eccentricity,
diameter-Diameter of the object (km) , tp- Time of perihelion
passage, & albedo-Albedo (reflectivity).

Fig 3.3 Precison score for all parameters

The bar chart 3.3 shows the Precison score for all
parameters,from that we observe that random forest classifier
has the highest percison score,so we do the analysis by it.

Fig 3.2 feature importance for all parameters

The barcharts 3.2 shows the feature importance of all the

parameters such as a- Semi-major axis , q- Perihelion distance,
e- Orbital eccentricity, i- Inclination (degrees), om- Longitude
of the ascending node (degrees), w- Argument of perihelion
(degrees), ma- Mean anomaly (degrees), ad:- aphelion
distance, n-Mean motion (degrees per day), tp- Time of
perihelion passage, moid- Minimum Orbit Intersection,
 objects. Future research should focus on refining data
accuracy, improving classification, and developing advanced
prediction models to enhance our preparedness for NEO
encounters. Public awareness and international collaboration
Fig 3.4 Scatterplot are also essential components of this ongoing effort.
V CONCLUSION AND FUTURE ENHANCEMENT
The figure 3.4 shows the representation of the Near-Earth The NEO project has successfully classified NEOs and
object as non-hazardous or hazardous using a scatterplot. identified potentially hazardous asteroids. It provides vital
data on orbital and physical characteristics, improving our
IV. RESULT AND DISCUSSION preparedness for NEO encounters. The inclusion of epoch
data refines our ability to predict NEO movements,
The NEO (Near-Earth Object) project has yielded important enhancing early warning systems. It Gather more detailed and
results and insights into the characteristics and dynamics of accurate data on NEOs, including size, shape, and surface
these celestial bodies. Here's a summary of the project's key properties. Also, it Develops advanced prediction models to
findings and a discussion of their implications: improve tracking and forecasting. It Collaborate with
international space agencies and observatories for a broader
NEO Classification: dataset, Increases public awareness about NEOs and potential
- The project successfully classified NEOs into distinct impacts and fund space missions for up-close study of NEOs
categories, including Amor, Apollo, Aten, Atira, and Main to gather invaluable insights.
Belt Asteroids (MBA). This classification aids in
understanding the variety of NEOs and their orbital REFERENCES
characteristics.
[1] G. Deva ram ganesh, P. Vidyullatha, M. Ravi krishna, S. T.
Potentially Hazardous Asteroids (PHAs): Prapulla, A. Pavan Saran and P. Ramya, "Machine Vision
- The project identified and characterized PHAs among based Object Detection using Deep Learning Techniques,"
2023 3rd International Conference on Smart Data Intelligence
NEOs. These are objects with the potential to impact Earth. (ICSMDI), Trichy, India, 2023, pp. 466-470, doi:
The data collected is crucial for assessing and mitigating 10.1109/ICSMDI57622.2023.00088.
potential threats. [2] T. Hoffman et al., "Near-Earth Object Surveyor Project
Preliminary Design," 2023 IEEE Aerospace Conference, Big
Orbital and Physical Characteristics: Sky, MT, USA, 2023, pp. 1-19, doi:
10.1109/AERO55745.2023.10115663.
- Detailed information about NEOs' orbital elements, such [3] H. Lin, C. Sun and Y. Liu, "OBBStacking: An Ensemble
as perihelion distance, semi-major axis, eccentricity and Method for Remote Sensing Object Detection," in IEEE
physical properties like diameter & albedo, has been Journal of Selected Topics in Applied Earth Observations and
compiled. These parameters provide valuable insights into Remote Sensing, vol. 16, pp. 2112-2120, 2023, doi:
10.1109/JSTARS.2023.3243168.
their trajectories and behavior.
[4] D. Khajuria, A. Sharma, N. Sharma and M. Mangla,
"Classification and Comparative Analysis of Earth's Nearest
Epoch Data: Objects using Machine Learning Models," 2023 10th
- The inclusion of epoch data allows researchers to track the International Conference on Computing for Sustainable Global
positions of NEOs at specific times, which is essential for Development (INDIACom), New Delhi, India, 2023, pp. 16-
23.
predicting their future positions and potential close
[5] L. Z. Ladeira, F. Santos, L. Cléopas, P. Buteneers and L. Villas,
approaches to Earth. "NEO-NDA: Neo Natural Language Data Augmentation,"
2022 IEEE 16th International Conference on Semantic
- The classification of NEOs offers astronomers and Computing (ICSC), Laguna Hills, CA, USA, 2022, pp. 99-102,
doi: 10.1109/ICSC52841.2022.00021.J. Clerk Maxwell, A
scientists a structured approach to studying these objects, Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Oxford:
allowing for more targeted research and predictions. Clarendon, 1892, pp.68–73.
- The identification of PHAs is crucial for early warning [6] R. N. Ranaweera and T. Fernando, "Prediction of Potentially
systems and planetary defense strategies. Knowing which Hazardous Asteroids using Deep Learning," 2022 2nd
NEOs pose a threat helps in devising strategies to mitigate International Conference on Advanced Research in Computing
(ICARC), Belihuloya, Sri Lanka, 2022, pp. 31-36, doi:
potential impacts. 10.1109/ICARC54489.2022.9753945.
- The project's data on orbital and physical characteristics [7] R. J. Wainscoat, K. Chambers, R. Weryk, Y. Ramanjooloo, M.
is a valuable resource for tracking NEOs, monitoring their Huber and E. Magnier, "Recent Advances in the Pan-STARRS
movements, and refining predictions of future close Search for Near-Earth Objects," 2022 IEEE Aerospace
approaches. Conference (AERO), Big Sky, MT, USA, 2022, pp. 1-6, doi:
10.1109/AERO53065.2022.9843625.
- Epoch data enhances the precision of NEO tracking, [8] T. Hromakina et al., "NEOROCKS project: surface properties
enabling better predictions for astronomical observations and of small near-Earth asteroids," in Monthly Notices of the Royal
potential space missions. Astronomical Society, vol. 520, no. 2, pp. 3143-3150, Sept.
2022, doi: 10.1093/mnras/stad330.
The NEO project is instrumental in increasing our [9] C. Rosu and V. Bacu, "Asteroid Image Classification Using
Convolutional Neural Networks," 2021 IEEE 17th
understanding of NEOs and their potential implications for International Conference on Intelligent Computer
Earth. This knowledge can aid in planetary defense, space Communication and Processing (ICCP), Cluj-Napoca,
exploration, and our overall comprehension of these celestial
 Romania, 2021, pp. 3-10, doi: [13] R. A. Mohammed and M. Y. Alradi, "Model of planetary radar
10.1109/ICCP53602.2021.9733484. for tracking near-Earth asteroids," 2017 Fifth International
[10] V. Bahel, P. Bhongade, J. Sharma, S. Shukla and M. Gaikwad, Conference on Aerospace Science & Engineering (ICASE),
"Supervised Classification for Analysis and Detection of Islamabad, Pakistan, 2017, pp. 1-4, doi:
Potentially Hazardous Asteroid," 2021 International 10.1109/ICASE.2017.8374262.
Conference on Computational Intelligence and Computing [14] C. P. Yang et al., "An architecture for mitigating near earth
Applications (ICCICA), Nagpur, India, 2021, pp. 1-4, doi: object's impact to the earth," 2017 IEEE Aerospace
10.1109/ICCICA52458.2021.9697222. Conference, Big Sky, MT, USA, 2017, pp. 1-13, doi:
[11] M. Veto et al., "Ball Aerospace Spacecraft and Systems for the 10.1109/AERO.2017.7943594.
Near-Earth Object Surveyor Mission," 2021 IEEE Aerospace [15] N. Rowell, S. Parkes and M. Dunstan, "Image Processing for
Conference (50100), Big Sky, MT, USA, 2021, pp. 1-14, doi: Near Earth Object Optical Guidance Systems," in IEEE
10.1109/AERO50100.2021.9438216. Transactions on Aerospace and Electronic Systems, vol. 49, no.
[12] L. Gentile, A. Capria, M. Conti, C. Moscardini and M. 2, pp. 1057-1072, APRIL 2013, doi:
Martorella, "Resident Space Object Passive Bistatic Radar 10.1109/TAES.2013.6494399.
Detection Using DVB-S2 Signals," 2018 19th International
Radar Symposium (IRS), Bonn, Germany, 2018, pp. 1-7, doi:
10.23919/IRS.2018.8448182.