Gen_Z++

https://github.com/DINAMOHMD/Gen_Z-

https://www.youtube.com/watch?v=hG63a3G0ySI

Problem definition:

In the first six months of this year, it was observed that about 40 percent of the European Union's forests were destroyed by fire, at a rate of four times the average of the past twelve years as shown in fig.1.

The causes of these fires are divided into two groups, the first is natural and includes thunder strikes that release the first spark. while the second group is the fires caused by humans directly or indirectly, and these reasons differ according to the region because they are related to the nature of the land, type of soil, and climatic situation. Forest fires destroy ecosystems where only ash remains and lead to the death of living creatures inside and their migration to other places. The loss of valuable timber leads to great material losses for countries like what happened in Australia 2020 as shown in fig.2 and is a factor in the desert encroachment of continents. Fires also destroy many Residential areas adjacent to forests.

our solution:

The first protocol (Software Solution): -

We used M.L. to solve the problem in particular his library scikit-learn which is located inside python and through a dataset that contained scripts we used NLP and in particular label encoder to convert the string into an integer so that the algorithm can deal with the dataset and then we divided the data into a part train and a partial test.

Once we enter the data it will be divided into five classification algorithms and we compared scores on their own through model check and in particular cross-validation-score to find out which algorithm is the best to deal with this data, then we entered the best algorithm on metrics model where we used ROC AUC Score, classification report, confusion matrix, zero-one loss. then we saved the model, and its weights as declared in fig.3.

We have built an application using pyqt5 as shown in fig.4 and fig.5 which works to receive the factors that may cause a fire from the datasets and sensors. Then when the button is pressed, the user will be told whether a fire is expected or not and also has the capability of visualizing this prediction.

The second protocol (Hardware solution): -

The control unit makes the user overconfident. It has been attached to a solar panel that auto-feeds the system with power. The control unit will be placed all over the forests that send alarms and data to the server throws the GSM for example temperature, humidity, smoke, fire detection, location of the fire, rain detection, and the wind direction to track the fire

As well as to make a secure path plan for the animals in case of an emergency through the buzzing or beeping to keep the animals away from danger. The block diagram consists of our hardware solution.

Background and literal review:

We develop an application to chase the fire to make classification and predict the areas where the fire will spread using artificial intelligence. Controlled burns have become more important as fire suppression efforts have grown over the last century. Historically, smaller fires occurred in forests at regular intervals. When these fires are suppressed, flammable materials accumulate, insect infestations increase, forests become more crowded with trees and underbrush, and invasive plant species move in.

• Using satellites, we measure the thermal radiation of living organisms in the specified area, from which we determine the type, density, and condition of living organisms.
• By making a classification using artificial intelligence and based on the classification we determine if the plant is alive.
• By using MPU sensors, we can determine the wind speed, and by the direction of the wind, we determine the path of the fire.
• We made a Desktop and Hardware Application to Make a heat map based on the Satellites, Classification, and Sensor results to predict the location of fires and the specifications, and thus the fires can be controlled and faced.
• Sending Waves at certain frequencies will act as a warning system and move the animals away to safe areas outside the expected fire path.

Methodology:

Using FWI index and datasets provided by NASA such as fire archive M6_157333. We build a predictive model using machine learning algorithms to apply our idea. To enhance the results, we implemented an embedded hardware circuit with multi-sensors and actuators to collect more data. We use the incoming data as a dataset after cleaning and doing pre-processing for it to enhance our system prediction (Classification and Regression), and this is going to give our application more flexibility to represent the data in a good manner as visualized data.

ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

We analyzed and used :

fire_archive_M6_157333 data (From Nasa)

in 5 algorithms

Decision Tree Classifier: It is a tree-structured classifier, where internal nodes represent the features of a dataset, branches represent the decision rules, and each leaf node represents the outcome

Support Vector Classification (SVC): Uses a subset of training points in the decision function (called support vectors), so it is memory efficient

SCD Classifier: the gradient of the loss is estimated each sample at a time and the model is updated along the way with a decreasing strength schedule

Random Forest Classifier: It builds decision trees on different samples and takes their majority vote for classification 

GaussianNB: It’s specifically used when the features have continuous values. It's also assumed that all the features are following a gaussian distribution

ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

Throughout the NASA space app hackathon's journey Despite the little amount of time we had, we enjoyed it and had a great time completing the challenge. We got the inspiration for this challenge from the regular forest fire that happens every year at the beginning of the summer season as well as the climate change that acts as the main factor in the forest fire. For example, in 2020 in Australia, 2022 in Europe, 2022 in Europe, and 2022 in the Amazon, animal life was lost. During the hackathon, we had lots of trouble, such as the different opinions of the team members through our meetings on Discord and Google Meet, and that ended up wasting time for us. A big thank you to ENG. Omar Atef As a team leader, Omar Atef participated in the management of the group. Finally, I'd want to express my appreciation to the team members for the wonderful job they have done in such a short amount of time.

·ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

 P. Cortez and A. Morais. A Data Mining Approach to Predict Forest Fires using Meteorological Data. In J. Neves, M. F. Santos and J. Machado Eds, New Trends in Artificial Intelligence, December 2007.

· D. Stojanova, P. Panov, A. Kobler, S. Dzeroski, and K. Taskova. Learning to

· Fires with Different DataMining Techniques. In D. Mladenic and M. Grobelnik, editors.Predict Forest.

International multiconference Information Society (IS 2006), Ljubljana, Slovenia, 2006

ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

Resources

ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

1. Fig1. Sources: the researcher Anna Fleck (July. 9,2022), Statista organization.
2. Fig2.Source: The New York Times (Dec. 31, 2019), Australia 2019-2020 fire
3. Tools: Python Machine Learning - python App Designer-Atmel Studio
4. Coding Languages: python - Embedded C.
5. Software: python Libraries - Classification and Regression Trees for Machine Learning - Atmega328 SDK
6. Hardware: we design two circuits:

           I.  Sensor Circuit: Atmega328P - Smoke Sensor - MPU 6050.

   II. High-Frequency Circuit:555 IC-Capacitors - Resistors -Potential resistors – Piezo.

#FIRE #forest #forest_fire #nature #hardware #software #AI #ML #machine_learnning #data_analysis