Exploring Deep Learning Models for Environmental Optimization in Morel Mushroom Cultivation

Abstract

ABSTRACT Morchella mushrooms ( Morchella spp.) are highly valued edible fungi due to their distinctive flavor and medicinal benefits. However, their artificial cultivation remains challenging because growth is extremely sensitive to environmental fluctuations and complex interactions among soil, moisture, and temperature. Conventional control approaches are largely manual, inefficient, and unable to respond dynamically to microclimatic variations, resulting in unstable yields. To address these limitations, this study proposes a Hybrid Deep Environmental Optimization Framework (HyDEOF) that integrates Deep Belief Networks (DBN) with an Adaptive Neuro‐Fuzzy Inference System (ANFIS) for intelligent environmental regulation during Morel cultivation. DBN models nonlinear patterns from sensor data, while ANFIS enables real‐time adaptive control of humidity, CO 2 concentration, and soil temperature. Experimental results from controlled cultivation chambers demonstrate that HyDEOF improves yield consistency by 21%, reduces energy consumption by 19%, and enhances substrate moisture retention by 17% compared with conventional systems, highlighting its potential for sustainable and efficient Morel production. Key stressors influencing Morel cultivation include humidity fluctuation, CO 2 imbalance, and soil temperature variability, critically affecting fruiting stability. The methodology utilizes sensor‐based environmental data collection using DBN‐based feature extraction and ANFIS‐based adaptive control within a closed loop system. This approach is tested using experimental studies that evaluate environmental stability, productivity, and energy consumption.