Perancangan Sistem Pengkabutan Otomatis Berbasis Sensor SHT85 dan VEML7700 Menggunakan Logika Fuzzy pada Tanaman Tomat

Abstract

Fluctuating weather in open fields frequently triggers stress in tomato plants due to unstable temperature and humidity. An ESP32-based automated misting system was designed as a solution to execute the Mamdani fuzzy logic algorithm in real-time. The data acquisition process relies on SHT85 and VEML7700 sensors with a constant reading interval of 10 seconds. This specific interval was selected to ensure the system remains responsive in mitigating heat without compromising hardware memory stability. The control validation phase was conducted by comparing the ESP32 actuation duration against a Python simulation and mathematical calculations as the ground truth. The test results revealed a contrasting phenomenon regarding machine computational resolution. Under minimum and transitional weather conditions, the Python simulation aligned perfectly at 80 seconds, whereas the ESP32 experienced deviations to 119 and 90 seconds. The limitations of the microcontroller's 32-bit architecture triggered these deviations, as the system was forced to perform value truncation when executing defuzzification on narrow-angled fuzzy sets. The situation reversed during extreme weather conditions, where the ESP32 was highly precise, reaching exactly 510 seconds, while Python experienced a minor 3-second discretization error due to wide trapezoidal calculations. These actuation discrepancies stem purely from differences in machine computational resolution, rather than a control system failure. The actuation time differences remain well within safe agronomic tolerances. Ultimately, the implementation of this system successfully executed fuzzy logic within a temperature range of 16–29°C, 60–90% humidity, and 10,000–30,000 lux light intensity to mitigate the risk of crop failure.

Perubahan cuaca lahan terbuka sering memicu stres tanaman tomat akibat ketidakstabilan suhu dan kelembapan. Sistem pengkabutan otomatis ESP32 dirancang sebagai solusi pengeksekusi algoritma logika Fuzzy Mamdani secara real-time. Proses akuisisi data mengandalkan sensor SHT85 dan VEML7700 dengan interval pembacaan konstan 10 detik. Interval ini dipilih agar sistem responsif meredam panas tanpa mengorbankan stabilitas memori perangkat keras. Tahap validasi kendali dilakukan dengan mengkomparasi durasi aktuasi ESP32 melawan simulasi Python dan perhitungan matematis sebagai ground truth. Hasil pengujian mengungkap fenomena resolusi komputasi mesin yang kontras. Kondisi cuaca minimum dan transisi menunjukkan hasil Python sejalan sempurna di angka 80 detik, sedangkan ESP32 mengalami deviasi menjadi 119 dan 90 detik. Limitasi arsitektur 32-bit mikrokontroler memicu deviasi ini karena sistem terpaksa melakukan pembulatan nilai (truncation) saat mengeksekusi defuzzifikasi pada area himpunan bersudut sempit. Situasi berbalik saat cuaca ekstrem, dimana ESP32 sangat presisi menyentuh 510 detik, sementara Python mengalami discretization error minor 3 detik akibat kalkulasi trapesium lebar. Fluktuasi selisih ini murni perbedaan resolusi komputasi mesin, bukan kegagalan kendali. Selisih aktuasi tersebut masih berada dalam toleransi aman agronomis. Implementasi sistem ini sukses mengeksekusi logika fuzzy pada suhu 16–29°C, kelembapan 60–90%, dan intensitas cahaya 10.000–30.000 lux guna menekan risiko gagal panen.