Environmental coupled multi-factor precise regulation and optimization for an artificial light plant factory based on a growth model

Abstract

The dissertation is dedicated to solving an urgent scientific and technological problem in the field of Mechanization and automation of agricultural in modern agricultural production: innovating multi factor coupling precise regulation and optimization technology for the environment inside artificial light plant factories, in order to improve comprehensive resource utilization and reduce crop’s industrial production costs. To meet the requirements of energy conservation and environmental protection and not be affected by external climate and land limitations, the construction of an artificial light plant factory in an enclosed and insulated chamber should be the best option. After research, we took the lead in proposing the concepts of modern building greenhouses and intelligent building greenhouses, and recommended building artificial light plant factories in urban areas and constructing larger scale intelligent building greenhouses plant factories to improve the building performance of plant factories, thereby ensuring permanent use and long-term production and operation. The urban intelligent plant factory is a highly intensive modern agricultural production system that can continuously provide the most suitable environment for plant growth and achieve high-quality and efficient production of plant products through precise environmental regulation techniques and mechanization, automation, digitization, intelligence, industrialization and factory technology. Moreover, this production method can adopt a "local production, local sales" operating model, continuously producing organic, green, clean, pollution-free, and fresh-eating plant products throughout the year, improving people's living standards, and ensuring the safety of the "vegetable basket" and food security. This is very important for modern Ukraine, for China, and even for all countries in the world. Object of research - theories and methods for constructing plant growth models based on deep learning algorithms; the overall composition, program architecture and development prospects of an artificial light plant factory; and the techniques and methods for mechanization, automation and intelligent regulation and optimization of the production environment. The subject of research - is the design and development of mechanized, intelligent, industrialized, factorized, periodical and modern plant production systems that can be built in urban areas, and the analysis and study of their system composition and architecture; the studies of the theories and methods for building plant growth models based on IoT, big data technologies and deep learning algorithms, which are different from traditional mathematical algorithms; the studies of the machines, means and methods for the coupled multi-factor precise regulation and optimization of the environments in the artificial lighting factory based on a plant growth model. Specifically, it includes three sub directions and subjects: firstly, investigate the construction form, system composition, development status, development trends and core technologies of the artificial light plant factory. The dissertation proposes recommendations for the construction of building greenhouses, intelligent building greenhouses, and intelligent building greenhouse plant factories. Second, research on methods and techniques for constructing plant growth models. The dissertation proposes a construction method and system architecture for plant growth models based on IoT and big data technologies. Furthermore, research on the relevant theories and core technologies of environmental regulation and optimization in artificial light plant factories. The dissertation proposes a multi-factor self-learning coupled precision regulation model, as well as research on the methods for constructing plant growth models based on deep learning models and the related studies on artificial lighting and nutrient solution regulation techniques. The purpose of the work is to create and improve modern, intensive plant production complexes and systems that can be constructed in urban areas, independent of geo-climatic and land resource constraints, and to study the theory law, methodology, and technology of mechanized, automated, intelligent, and precise control and optimization of plant growth and production environments of artificial light plant factories in buildings. The ultimate goal is to improve and optimize regulation strategies of the environment through intelligent and precise environmental regulation technologies, increase resource utilization efficiency, and reduce the cost of plant industrial production products. To achieve this goal, the following tasks need to be solved: 1. To analyze the current development status, trends, obstacles and opportunities of the artificial light plant factory, and to clarify the importance and direction of research. 2. To analyze and improve the existing forms of greenhouses, explore the optimal bearing form of artificial light plant factories, propose development recommendations for building greenhouses, intelligent building greenhouses, and intelligent building greenhouse plant factories, and study their strategic significance and development strategies. 3. To analyze the system composition and core technologies of artificial light plant factories and intelligent building plant factories, and to identify research topics and directions. 4. To investigate the construction methods for plant growth models and plant factory big data, to propose and design a systematic framework for building plant growth models based on IoT and big data technologies, and to develop a plant factory big data management systems and plant growth model analysis platforms. 5. to systematically analyze the production environment factors of artificial light plant factories and their effect on plant growth, study the coupling effect of multiple factors on plant growth, to propose a multi factor self-learning coupling precise regulation model, and to develop a production management systems and environmental precise regulation platforms for artificial light plant factories. 6. To research on algorithms and implementation techniques for constructing plant growth models based on deep learning models. To research on object detection of tomato fruits for the artificial light plant factory using an improved YOLO deep learning model and instance segmentation of plant seedlings for the artificial light plant factory using a modified Mask R-CNN and Transformer deep learning models, and to lay the theoretical and technical foundation for constructing plant growth models and environmental control models. 7. To experimental verify on the effects of different lighting conditions of LED artificial light and nutrient solution formulations on plant growth and quality, to improve to environmental regulation technology and means in artificial light plant factories, and to lay the foundation for theoretical research, model construction, and implementation technology of precise regulation of environmental multi factor coupling. In the introduction, the choice of the topic of the dissertation and scientific tasks is substantiated, the purpose and tasks of research are formulated, the scientific novelty and practical value of the received results are defined, and also the information on approbation, structure and volume of work is resulted. In the first Section, systematic literature analysis, social demand studies, and related theoretical research have been conducted on topics such as plant factories, intelligent building greenhouses, artificial light plant factories, plant growth model construction, and environmental regulation and optimization of plant factories. The necessity, importance and development prospects of the research topic of artificial light plant factories are clarified. A comprehensive analysis of the problems, obstacles and intelligent development needs encountered in their research and industrialization at the current stage has been carried out. Further research has been carried out on the core technologies for the development of artificial photoproduction plants, and a research topic based on growth models for the precise regulation and optimization of multi-factor couplings has been proposed. All this allowed the applicant to formulate the purpose, objectives and tasks of the dissertation. In the second section, the laboratory situation, experimental conditions, experimental equipment, experimental materials, etc. of the scientific work are presented in a comprehensive and systematic manner. It also showcases image acquisition system platforms designed to successfully achieve scientific research goals and tasks, as well as the experimental design and research status of screening illumination formulations and nutrient solution formulations. In the third section, basic research related to the construction of plant growth models is carried out to provide theoretical support for the construction of growth models and the precise regulation of the environment. collecting environmental data through the intelligent monitoring platform for plant factory environment and growth, constructing big data for plant factory environment, analysing plant growth parameters using different deep learning models and algorithms, and constructing big data for plant factory growth are studied. Then data mining techniques and deep learning models to construct a sub-model for plant growth is used, and a comprehensive model for plant growth is summarized and constructed. In the fourth section, the effects of light environment, nutrient solution environment, and the comprehensive regulation of various environmental factors in plant factories on plant growth are studied, and the effectiveness of environmental regulation is experimentally verified. In the fifth section, the full text is summarized, conclusions are drawn, recommendations for future research are made, and technical recommendations for implementing the results of the study into production are developed. In accordance with the set goal and tasks, the following results were obtained in the work: 1. Ukraine is a vast, sparsely populated territory with a scattered rural population and densely populated urban areas. Moreover, most of the country has a temperate continental climate, with an average temperature of -7.4 ℃ in January and 19.6 ℃ in July. The mean annual temperature is relatively low. Planting vegetables in the open air, on the one hand, is difficult to meet the balanced supply of fresh vegetables throughout the year, and on the other hand, it is also difficult to meet the diverse supply of fresh vegetable varieties. The intelligent building greenhouse plant factory proposed by the applicant and the improved all artificial light plant factory can improve the situation in Ukraine, and develop micro or small artificial light plant factories in sparsely populated and dispersed rural areas to meet the perennial fresh vegetable needs of the rural population; The construction of large-scale intelligent building greenhouses and plant factories in densely populated cities will enhance the demand of urban residents for the quality of fresh vegetables. The results of this work have been highly recognized by peer experts, government managers, and entrepreneurs, and have received government project funding. At the same time, it could serve as a reference for Ukraine. 2. According to a questionnaire survey from China, the main limiting factors for the development of plant factories at present are high construction costs (73.3%) and high operating costs (66.4%). Consumers are particularly concerned and anxious about the unbearable price (70.6%). Consumers prioritize cleanliness and pollution free (39.3%), green and healthy (30.3%), high freshness (17.6%), product quality (8.8%), and nutritional index (3.7%) when making purchases. While the situation in Ukraine and China may not be exactly the same, the demand for vegetable quality should be consistent. The artificial light plant factory scheme proposed by the applicant and the precise control system architecture of environmental multi factor coupling can solve the following three problems: (1) the production environment of the artificial light plant factory is clean and pollution-free, and the vegetables produced are clean, pollution-free, and green and healthy. (2) Through precise environmental regulation and optimization, the production cost of vegetables will be greatly reduced, making it affordable for ordinary people to consume. (3) Although the construction cost of intelligent building greenhouse plant factories and artificial light plant factories is high, their materials are stable and their structure is firm, and they can be built for almost long-term use. The results of this work can address the balanced supply of fresh vegetables and hunger threats faced by Ukraine, China, and even the world, ensuring the "vegetable basket" and food security. 3. The dynamic and constantly changing growth environment of plants, with diverse growth characteristics and complex growth processes, makes the construction of plant growth models quite complex and difficult. It is almost impossible to construct a complete and perfect mathematical model. The applicant's proposal to build crop growth models based on IoT and big data technologies transforms the intractable problem of simulating complex systems with mathematical formulas into the study of a divide and conquer relational correlation problem. Relational models of complex systems have been constructed using data mining algorithms and deep learning models. The applicants obtained a large amount of big data on growing environments and plant growth during their work, and extensive preparations were made to model the relationship between environmental factors and growth indicators. 4. Intelligent regulation and optimization of plant growth environments is the most complex and central scientific and technological problem in artificial light plant factories. Environmental regulation is not precise and accurate enough, and the most immediate results can affect plant growth, yield and quality. In addition, it may result in significant waste of production materials and integrated resources. The proposed system architecture of a multi-factor environmental regulation platform for an artificial light plant factory based on a growth model has been applied in the development of a comprehensive control system for the plant factories and has also been tried and tested by enterprises. The results of this work can improve the utilization rate of water resources by 10%, save water soluble fertilizer by 8%, and comprehensively reduce electricity by 18%, with huge market prospects. 5. The biomass accumulation of plants is closely related to water replenishment, lighting, fertilizers, CO2, and even environmental temperature and humidity. Target detection of plant fruits and instance segmentation of seedlings can be used in intelligent monitoring systems for plant growth processes, obtaining real-time growth status information, perceiving growth trends, predicting biomass growth, intelligently and accurately controlling water and fertilizer replenishment, regulating environmental variables such as light, CO2 concentration, temperature and humidity, providing the best environment for plant growth. The proposed improvements YOLOv5_MT algorithm is used for tomato fruit detection in artificial light plant factories, improving the detection accuracy of dense and obstructed tomatoes. The proposed CMRDF instance segmentation algorithm that integrates RGB-D multi channel image data is used for the segmentation of plant seedling leaves in artificial light factories, with a PA of 93% and an IoU of 93.4%. These two research results were both used in the control and management system of plant factories. 6. Three experimental studies were conducted in the laboratory of an artificial light plant factory, including the experimental study on illumination screening and uniformity simulation of hydroponic lettuce, the experimental study on the effect of light quality on the quality of hydroponic Cichorium endivia L., and the screening study on the formulation of nutrient solution for hydroponic green leaf lettuce. The control techniques for light and nutrient solutions have been improved and the effectiveness of environmental multi-factor control techniques has been demonstrated. The results of the work achievements have been applied in multiple government scientific research projects undertaken by the applicant and in the products of related enterprises, with broad market prospects and enormous economic benefits. 7. The technology of artificial light micro plant factories, such as LED supplementary light type planting cabinet (Chinese Patent NO. ZL 2023 2 0899208.6), a multispectral crop phenotype analysis platform for plant factories (Chinese Patent NO. ZL 2021 2 1596146.7), an assembled aerospace culture layer shelf with adjustable layer height for the plant factory with artificial light (Chinese Patent NO. ZL 2022 2 0821668.2), has been protected by Chinese utility model patents. The technology named "A general real-time detection and counting method for eggplant frost in plant factory (Chinese Patent NO. 202210152745.4)" is protected by Chinese invention patents. These patented technologies have been put into production, earning considerable economic benefits and generating enormous social value. Scientific novelty of the obtained results: 1. For the first time, the concepts of intelligent building greenhouses and intelligent building greenhouses plant factories were proposed, with clear definitions. Extensive social demand research and literature analysis were conducted to systematically and scientifically demonstrate their strategic significance. The development strategy of "3-Positions and 1-Entity" was studied, providing an innovative model and systematic solution for the sustainable and clean plant production system in urban development. 2. For the first time, it is proposed to use the physiological mechanisms and biological theories of plant light regulation as the theoretical basis for artificial light plant factory light environment regulation, improve the technical means of light environment regulation, and regulate the production process of plants through the role of light in photosynthesis, growth and development, morphological construction, material metabolism, gene expression, and nutritional quality, in order to adapt to market changes. 3. For the first time, a flat IoT solution using multiple sensors and controllable work units has been provided for artificial light plant factories. A system architecture for constructing scientific big data for plant factories has been proposed, and the process and methods of comprehensively utilizing IoT, big data, and deep learning technologies to construct plant growth models have been systematically studied. The plant factory big data platform and crop growth model service system constructed using this method can provide data and model services for plant factory industrial enterprises through cloud services. 4. For the first time, the architecture and framework of a multi-factor environmental regulation platform for artificial light plant factories based on growth models were proposed, and control system software was designed, developed, and tested. The system software can automatically obtain plant growth model files from the cloud, and intelligently and accurately regulate the environment of plant growth based on the plant growth model, to obtain high-quality and high-yield plant products with minimal cost. 5. An improved YOLOv3 deep learning model and algorithm have been proposed for target detection of hydroponic tomato fruits in artificial light plant factories, providing theoretical foundation and technical support for yield estimation, robotic picking, and precise regulation of growing environments. This method can classify and detect the growing tomato fruits, obtain the quantity of green fruits, color changing fruits, and red fruits, as a basis for precise regulation of light environment and nutrient solution concentration, thereby effectively reducing water, electricity, nutrient solution waste and sewage discharge, improving resource comprehensive utilization rate and yield. 6. For the first time, a CMRDF algorithm for plant seedling instance segmentation was proposed, which integrates RGB-D multi-channel image data to improve the accuracy of seedling instance segmentation. It is used to analyze plant phenotypic data in artificial light plant factories, to construct crop growth models, and to provide theoretical and technical support for plant intelligent growth monitoring, disease and pest detection, production management, yield estimation, robotic operations, and environmental regulation. 7. For the first time, experimental studies on illumination screening and uniformity simulation of hydroponic lettuce, experimental study on the effect of light quality on the quality of hydroponic Cichorium endivia L., and screening study on the formulation of nutrient solution for hydroponic green leaf lettuce are conducted in an artificial light plant factory, providing technical references for precise regulation of environmental multi-factor coupling. The practical significance of the results obtained lies in providing a series of systematic theoretical achievements for the intelligent control and optimization of the artificial light plant factory environment, and proposing comprehensive technical suggestions. At the same time, it also provides a complete set of solutions and suggestions for sustainable and clean plant production systems for urban development. The use of these suggestions will improve the intelligence and intensification of plant factories, improve the utilization rate of comprehensive resources such as land, water, electricity, and fertilizers, reduce plant production costs, and generate considerable economic benefits and immeasurable social value. The theoretical achievements and technical solutions obtained in the work are protected by 4 patents and 7 computer software copyrights, and have been implemented by the high-tech enterprise "ZSP" Electronic Technology Co., Ltd. in Henan Province, China. After preliminary testing, it can improve the water resource utilization rate by 10%, save water soluble fertilizer by 8%, comprehensively reduce electricity by 18%, significantly reduce the production cost of plant, and bring huge economic benefits and social value. The main results of the dissertation, generalizations, scientific prescriptions and conclusions that constitute the essence of the work were independently obtained and formulated by the applicant.