Mushroom growth environment intelligent monitoring system based on Internet of Things

Abstract

The invention relates to the technical field of agricultural management, in particular to an intelligent mushroom growth environment monitoring system based on the Internet of Things, which comprises an environment sensing module, an edge computing node, a multi-strategy control execution module, a cloud platform server and a user interaction terminal, the environment sensing module collects growth environment parameters and state data in real time; the edge computing node completes data preprocessing and anomaly diagnosis; the multi-strategy control execution module adjusts the environment according to the instruction; the cloud platform server stores data, trains a model and generates a global instruction; the user interaction terminal realizes data visualization and remote control; the system adopts an improved double-flow convolutional neural network to identify a growth stage, utilizes a collaborative filtering algorithm to optimize cultivation parameters, and has strategies of timing and quantitative control, conflict resolution and the like. The mushroom growing environment can be accurately regulated and controlled, intelligent management is achieved, the labor cost is reduced, the mushroom yield and quality are improved, and intelligent development of the mushroom planting industry is promoted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural management, in particular to a mushroom growth environment intelligent monitoring system based on Internet of Things. BACKGROUND

[0002] In the process of modern agricultural development, as a high economic value agricultural product, the planting scale and market demand of mushrooms continue to grow. However, the traditional mushroom planting method has many limitations and cannot meet the current high-quality and large-scale production needs, which is embodied in the following aspects: Extensive environmental regulation: Traditional mushroom cultivation relies on manual experience to determine environmental parameters such as temperature and humidity, carbon dioxide concentration, and light intensity. The regulation method is extensive and lagging. For example, manual inspection interval is long, and it is difficult to timely perceive subtle changes in the environment. When the temperature and humidity are abnormal, it is difficult to adjust in time, which can lead to slow mushroom growth, quality decline, and even large-scale disease, resulting in yield loss; Growth state judgment depends on experience: The judgment of mushroom growth stage mainly relies on the experience observation of planting personnel, and lacks scientific and accurate evaluation means. Mushrooms at different growth stages have significantly different environmental requirements. Relying on experience to judge can easily lead to errors and cannot provide the most suitable environmental conditions for mushroom growth, affecting the yield and quality stability of mushrooms; Low management efficiency: With the expansion of mushroom planting scale, the disadvantages of manual management become more and more obvious. Manual monitoring and regulation require a lot of manpower and time cost, and it is difficult to realize real-time and comprehensive monitoring of large-scale planting areas. At the same time, during the manual management process, there are problems such as non-standard operation and inaccurate records, which are not conducive to the standardization and normalization of the planting process; Lack of intelligent decision support: In the traditional planting mode, planting decisions lack data support and intelligent analysis. It is difficult to predict the growth trend of mushrooms, optimize cultivation parameters, and respond to complex and variable environmental factors and market demand, which limits the sustainable development of the mushroom industry.

[0003] Therefore, a mushroom growth environment intelligent monitoring system based on Internet of Things is proposed to solve the above problems. SUMMARY

[0004] The purpose of the present application is to provide a mushroom growth environment intelligent monitoring system based on Internet of Things to solve the problems raised in the background art.

[0005] To achieve the above purpose, the present application provides the following technical scheme: A mushroom growth environment intelligent monitoring system based on Internet of Things, comprising: An environmental perception module includes a temperature and humidity sensor, a carbon dioxide concentration sensor, a light intensity sensor, and an image acquisition device distributed in the mushroom cultivation unit, and is configured to collect mushroom growth environment parameters and growth state data in real time. An edge computing node is connected to the environmental perception module and is configured to locally pre-process the collected data and perform abnormality diagnosis to generate environment control instructions. A multi-strategy control execution module includes a humidifier, a ventilation fan, a light supplement lamp, a roller blind device, and an ultrasonic atomization device, and is configured to receive instructions from the edge computing node and perform environment adjustment operations. A cloud platform server is connected to the edge computing node through an NB-IoT or WiFi communication module, and is configured to store environment data, train a growth stage recognition model, and generate global optimization instructions based on a timing strategy and a quantitative strategy. A user interaction terminal includes a PC terminal and a mobile APP, and is configured to provide an environment data visualization interface, a device remote control interface, and a growth stage adaptive parameter configuration function.

[0006] As a preferred solution, the local pre-processing of the edge computing node includes: Threshold comparison of temperature and humidity, CO2 concentration, and light intensity data is performed through an embedded expert rule base, and real-time alarms are triggered if the data exceeds the preset range. An improved double-flow convolutional neural network is used for growth stage recognition, and the algorithm process is as follows: Spatial feature flow calculation: after concatenating the RGB three-channel image features and the enhanced gray morphological features through the edge gradient operator, the spatial convolution kernel is convolved, and the spatial feature tensor is obtained through the Sigmoid activation function after adding the bias term. Time domain feature flow calculation: the feature difference matrix of adjacent time frames is convolved with the time domain convolution kernel, and the time domain feature tensor is obtained through the hyperbolic tangent activation function after adding the bias term. Growth stage probability output: after layer normalization processing, the spatial feature tensor and the time domain feature tensor are concatenated, multiplied by the full connection weight matrix, and the probability distribution of each growth stage is obtained through the Softmax function after adding the bias term.

[0007] As a preferred solution, the time domain convolution kernel size is 5x1x1, and the sliding step is 2 frames, which is used to capture the mycelium spread rate feature. An output layer connection attention mechanism module is used, and the calculation process is as follows: the hidden state of the previous moment and the spatio-temporal fusion feature vector of the current moment are processed using the LSTM network to obtain the attention score of the current moment, and the attention scores of all moments are exponentially normalized to obtain the attention weight of the current moment.

[0008] As a preferred solution, the cloud platform server includes a dynamic optimization engine that generates cultivation parameter recommendations using an improved collaborative filtering algorithm, the calculation process of which is as follows: Basic yield calculation: global average yield baseline plus variety bias term and growth stage bias term; Characteristic interaction term: inner product sum of variety characteristic vector and stage characteristic vector; Environmental impact term: environmental parameter vector is transformed by weight matrix and then passes through ReLU activation function, multiplied by environmental factor weight coefficient; Temperature correction term: input the temperature variance within 24 hours into the Sigmoid function and multiply by the temperature sensitivity coefficient; Final predicted yield: the sum of the above four results.

[0009] As a preferred solution, the dynamic optimization engine contains an abnormality suppression mechanism: When the environmental parameter fluctuation variance exceeds the threshold value, calculate the Euclidean distance between the variety characteristic vector and the stage characteristic vector; Multiply this Euclidean distance by the activation coefficient calculated by the Sigmoid function of the environmental variance as a regularization penalty term added to the optimization objective function.

[0010] As a preferred solution, the operation logic of the multi-strategy control execution module includes: Timing strategy: start the ventilation equipment at a fixed time period every day, with higher priority than the quantitative strategy; Quantitative strategy: when the humidity is lower than 75%, turn on the humidifier and floor fan; when it is higher than 95%, turn off; when the light intensity is lower than 500lx, start the light supplement lamp; when it is higher than 1000lx, trigger the sunshade roller shutter; Conflict resolution mechanism: ventilation instructions take priority over humidification instructions, and humidification instructions take priority over light adjustment instructions.

[0011] As a preferred solution, the user interaction terminal provides: Three-dimensional space monitoring view, showing the environmental parameter distribution thermograph of multi-layer mushrooms in the three-dimensional cultivation rack; Intelligent decision support interface, which returns growth trend assessment and picking time prediction after receiving the user uploaded mushroom image.

[0012] As a preferred solution, the edge computing node and the cloud platform server adopt a collaborative decision-making mechanism: The local node executes emergency control with a response delay of no more than 10 seconds; The cloud end issues an optimization model trained based on long-term data every week to update the edge node control parameters.

[0013] As a preferred solution, it also includes a security monitoring module: Infrared sensor detects illegal intrusion in real time, triggers local sound and light alarm and remote APP alarm. The standby battery automatically switches when the power supply is interrupted, maintaining the system running for at least 72 hours.

[0014] The technical solution provided by the above-mentioned application can be seen that the mushroom growing environment intelligent monitoring system based on the Internet of Things has the beneficial effects of: Precise environmental control to improve mushroom quality and yield: The environmental perception module collects key environmental parameters such as temperature, humidity, carbon dioxide concentration, and light intensity in real time and comprehensively. Combined with the local preprocessing of edge computing nodes and the global optimization strategy of cloud platform servers, the multi-strategy control execution module can accurately adjust the mushroom growing environment. For example, during the mycelium growth stage, the system can accurately control the temperature within the appropriate range and maintain the humidity at the optimal level, creating ideal conditions for mushroom growth, thereby effectively improving the yield and quality of mushrooms. Compared with traditional planting methods, the yield is expected to increase by 20%-30%, and the quality of mushrooms is better, with a significant reduction in the proportion of deformed and diseased mushrooms. Intelligent decision support to reduce labor management costs: Based on the improved double-flow convolutional neural network growth stage recognition algorithm and the cultivation parameter optimization model of the cloud platform server, the system can automatically identify the growth stage of mushrooms and provide accurate cultivation parameter recommendations. The intelligent decision support interface of the user interaction terminal can also receive user-uploaded mushroom images and quickly return growth trend evaluation and picking time prediction. These functions greatly reduce the reliance on human experience, reduce the difficulty and cost of manual management, and even people without professional planting experience can easily manage large-scale mushroom cultivation. Collaborative decision-making framework to ensure system stability and reliability: The collaborative decision-making mechanism of edge computing nodes and cloud platform servers enables the system to have both rapid response and long-term optimization capabilities. Local nodes implement emergency control with ≤10-second response delay for emergency situations, and the cloud end issues optimization model updates to control parameters every week to ensure that the system works stably and reliably in the face of sudden environmental changes and long-term operation optimization. The personnel intrusion detection and backup power supply functions of the security monitoring module further improve the safety and reliability of the system, reducing production interruptions and losses caused by unexpected situations. Multi-strategy control and conflict resolution to ensure effectiveness: The multi-strategy control execution module's timing strategy, quantitative strategy, and conflict resolution mechanism ensure the rationality and effectiveness of environmental regulation operations. By clearly defining the priority of instructions, such as ventilation instructions taking precedence over humidification instructions, conflicts between device operations are avoided, ensuring that all devices can work together to accurately adjust the environment and maintain the best environment for mushroom growth under complex environmental conditions. Data visualization and remote management improve management efficiency: The environmental data visualization interface and three-dimensional spatial monitoring view provided by the user interactive terminal enable users to intuitively and comprehensively understand the mushroom growth environment and equipment operating status; the remote control function supported by PC and mobile apps allows users to manage equipment and configure parameters anytime and anywhere, breaking time and space limitations, improving management efficiency, facilitating users to deal with various problems in a timely manner, and realizing intelligent and convenient cultivation management. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of the overall structure of an intelligent monitoring system for mushroom growth environment based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0018] like Figure 1 As shown, an embodiment of the present invention provides an intelligent monitoring system for mushroom growth environment based on the Internet of Things, comprising: Environmental perception module: This includes temperature and humidity sensors, carbon dioxide concentration sensors, light intensity sensors, and image acquisition devices distributed throughout the mushroom cultivation units, used to collect real-time data on mushroom growth environment parameters and growth status; Edge computing node: uses an STM32 microcontroller and is connected to an environmental perception module to perform local preprocessing and anomaly diagnosis on collected data and generate environmental control instructions; Multi-strategy control execution module: including humidifiers, ventilation fans, fill lights, roller blinds and ultrasonic atomizers, receives instructions from edge computing nodes and performs environmental adjustment operations; Cloud platform server: Connected to edge computing nodes via NB-IoT or WiFi communication modules, it is used to store environmental data, train growth stage recognition models, and generate global optimization instructions based on timing and quantitative strategies; User interaction terminal: including PC and mobile APP, providing environmental data visualization interface, equipment remote control interface and growth stage adaptive parameter configuration function.

[0019] In this embodiment, the environmental perception module serves as the "sensing nerve" of the IoT-based intelligent mushroom growth environment monitoring system, undertaking the core task of collecting real-time data on mushroom growth environment parameters and growth status. The following will describe this module in detail, from the overall perspective: 1. Overview of overall functions: The environmental perception module builds a comprehensive data collection network by distributing various types of sensors and image acquisition devices throughout the mushroom cultivation units. It continuously and in real time collects key parameters of the mushroom growth environment, such as temperature and humidity, carbon dioxide concentration, and light intensity, as well as image data of the mushroom growth status, providing basic data support for the system's subsequent data processing, analysis, and decision-making. The accuracy, timeliness, and comprehensiveness of the collected data directly affect the effectiveness of the entire monitoring system in regulating the mushroom growth environment and determining the growth stage. 2. Submodule composition and functions: (1) Environmental parameter collection unit: Temperature and humidity sensor deployment and data collection: Within the mushroom cultivation unit, temperature and humidity sensors are strategically placed based on the different cultivation areas and mushroom growth requirements. These high-precision digital sensors can collect real-time ambient temperature and relative humidity data at a frequency of minutes. For example, during the mycelial growth stage, sensors are placed at different heights on the culture racks to capture vertical temperature and humidity distribution, providing accurate data for subsequent environmental control. The collected data undergoes preliminary filtering to remove abnormal fluctuations and ensure data reliability. Working mechanism of the CO2 concentration sensor: CO2 concentration has a significant impact on the respiration and growth of mushrooms. The CO2 concentration sensor uses a non-dispersive infrared (NDIR) sensor, which can stably and accurately detect the CO2 concentration in the environment. The sensor converts the detected concentration signal into a digital signal and transmits it to the edge computing node via a communication protocol. During the peak growth period of mushrooms, the sensor increases the data collection frequency to timely capture changes in CO2 concentration, providing a basis for environmental adjustment operations such as ventilation. Light intensity sensor data acquisition: The light intensity sensor uses a silicon photocell sensor that can sense changes in light intensity in the environment in real time. Based on the different light requirements of mushrooms, such as some mushrooms requiring specific light intensities during the fruiting stage, the sensor transmits the collected light intensity data (unit: lx) to the system. The sensor adaptively adjusts its collection strategy during the day and night to ensure accurate data acquisition under different lighting conditions. (2) Growth status collection unit: Image acquisition device selection and arrangement: Select a high-resolution industrial camera as the image acquisition device and install it in a suitable position of the mushroom cultivation unit, such as above or beside the cultivation rack, to obtain clear images of the mushroom growth status; the camera supports multi-spectral imaging, which can simultaneously capture RGB images and near-infrared images to meet the needs of image feature analysis at different growth stages; in a three-dimensional cultivation rack, multiple cameras are deployed to achieve comprehensive coverage of the growth status of multiple layers of mushrooms; Image acquisition strategy and data processing: The image acquisition device acquires images at a set time interval (e.g., once an hour), and the acquisition interval is shortened during critical stages of mushroom growth (e.g., the fruiting stage); the raw image data collected are subjected to preliminary compression to reduce data transmission and storage pressure; at the same time, in order to facilitate subsequent growth stage recognition, the images are subjected to grayscale, noise reduction, and other preprocessing operations to highlight the morphological features of the mushrooms; (Three) Data transmission and preprocessing unit: Data transmission network construction: The data collected by the environmental perception module is transmitted to the edge computing node through wired or wireless communication; for sensors and image acquisition devices within close range, RS-485 bus or Ethernet is used for wired connection to ensure the stability and reliability of data transmission; for distributed sensors, low-power wireless communication technology (such as ZigBee) is used for data transmission to reduce system power consumption and wiring complexity; encryption protocols are used during data transmission to ensure data security and integrity; Data preprocessing mechanism: Before transmitting data to the edge computing node, preliminary verification and preprocessing of data are performed; for environmental parameter data collected by sensors, validity verification is performed to eliminate abnormal data outside the reasonable range; for image data, format conversion and size normalization are performed to meet the requirements of subsequent algorithm processing; the preprocessed data can be more efficiently received and processed by the edge computing node; Three, the working process of the module: (I) Initialization phase: After the environmental perception module starts, it performs self-checking on all sensors and image acquisition devices; checks whether the power supply of the sensors is normal, the communication interface is connected stably, and the lens of the image acquisition device is clean, the parameter setting is correct, etc.; Load the calibration parameters of the sensors and the configuration information of the image acquisition device, establish communication connection with the edge computing node, and ensure that data can be normally transmitted; (II) Data acquisition phase: The environmental parameter acquisition unit acquires temperature, humidity, carbon dioxide concentration, and light intensity data at a set acquisition frequency; the sensors convert the analog signals collected into digital signals and perform preliminary filtering; The growth state acquisition unit controls the image acquisition device to perform image acquisition according to a preset acquisition strategy; and the acquired original image data is compressed and preprocessed and then stored in a local cache; (Three) data transmission and preprocessing phase: The data transmission and preprocessing unit transmits the acquired environmental parameter data and image data to the edge computing node through the constructed communication network; in the transmission process, the data is encrypted and checked to ensure the safety and accuracy of the data; The transmitted data is preprocessed, including validity verification of the environmental parameter data and format conversion and size normalization of the image data, to prepare for data processing of the edge computing node; (Four) end phase: When the system stops running or receives a stop instruction, the environmental perception module stops data acquisition and transmission operations, turns off the power of the sensors and image acquisition device, saves relevant configuration parameters and operation records, and releases system resources.

[0020] In this embodiment, the local preprocessing of the edge computing node includes: The threshold value of the temperature and humidity, CO2 concentration and light intensity data is compared through the built-in expert rule library, and if it exceeds the preset range, a real-time alarm is triggered; An improved double-flow convolutional neural network is used for growth stage recognition, and the algorithm process is as follows: Spatial feature flow calculation: after the RGB three-channel image features and the gray morphological features enhanced by the edge gradient operator are spliced, convolution operation is performed with a spatial convolution kernel, and then a bias term is added and a spatial feature tensor is obtained through a Sigmoid activation function; Time domain feature flow calculation: the feature difference matrix of adjacent time frames is convolved with a time domain convolution kernel, and then a bias term is added and a time domain feature tensor is obtained through a hyperbolic tangent activation function; Growth stage probability output: after the spatial feature tensor and the time domain feature tensor are spliced and layer normalized, they are multiplied by a fully connected weight matrix, and then a bias term is added and the probability distribution of each growth stage is obtained through a Softmax function; The size of the time domain convolution kernel is 5x1x1, and the sliding step is 2 frames, which is used to capture the mycelium spread rate feature; The output layer connects an attention mechanism module, and the calculation process is as follows: the hidden state of the previous moment and the spatio-temporal fusion feature vector of the current moment are processed by an LSTM network to obtain the attention score of the current moment, and then the attention scores of all moments are exponentially normalized to obtain the attention weight of the current moment; Furthermore, the edge computing node, serving as the "local brain" of the IoT-based intelligent monitoring system for mushroom growth environments, undertakes the key tasks of data preprocessing, anomaly diagnosis, and generating environmental control instructions. Its efficient operation ensures the system's real-time response and precise control of the mushroom growth environment. The following describes this node in detail from multiple perspectives: 1. Overview of overall functions: The edge computing node, with an STM32 microcontroller at its core, builds a localized data processing and decision-making unit. It is closely connected to the environmental perception module, receiving real-time environmental parameters such as temperature and humidity, carbon dioxide concentration, and light intensity, as well as image data on mushroom growth status. Using built-in algorithms and rule libraries, it rapidly pre-processes and deeply analyzes the data. When abnormal environmental parameters or mushroom growth problems are detected, control instructions are promptly generated and sent to the multi-strategy control execution module. Simultaneously, the processed data is uploaded to the cloud platform server, enabling collaborative decision-making with the cloud, effectively reducing the system's reliance on the network and improving response speed and stability. 2. Submodule composition and functions: (1) Data reception and preprocessing module: Multi-source data access: This module connects to the environmental sensing module through multiple communication interfaces (such as UART, SPI, and I²C). It can simultaneously receive digital signals from temperature and humidity sensors, carbon dioxide concentration sensors, and light intensity sensors, as well as image data transmitted by image acquisition devices. It also performs protocol parsing and conversion for data transmission protocols (such as Modbus and TCP / IP) of different sensor types to ensure accurate data access. Data preprocessing: Received environmental parameter data is filtered and a sliding average filter algorithm is used to remove random noise. Image data is formatted and adjusted to reduce the pressure of subsequent processing. All data is integrity checked to eliminate abnormal data and ensure the reliability of input data. (2) Local data analysis and diagnosis module: Threshold comparison driven by an expert rule library: A built-in expert rule library for mushroom growth environments covers standard thresholds for temperature, humidity, CO2 concentration, and light intensity for different mushroom varieties at various growth stages. Pre-processed environmental parameter data is compared with the thresholds in the rule library in real time. If a parameter is found to be outside the preset range, a real-time alarm mechanism is immediately triggered, with a local alarm signaled by flashing indicator lights, a buzzer alarm, and other means. The alarm information is also uploaded to the user interaction terminal. Growth stage recognition based on deep learning: An improved two-stream convolutional neural network is used to identify the growth stages of mushrooms. The algorithm model is as follows: (in, RGB three-channel image feature matrix (dimension H x W x 3) is used to obtain the color and appearance information of the mushroom; Gray morphological feature matrix (dimension H x W x 1) is used to highlight the outline of the mushroom; Mushroom cap edge morphological gradient operator is used to strengthen the edge features; Adjacent time frame feature difference matrix (dimension H x W x 1) is used to capture the dynamic changes of growth; Hadamard product is used to realize corresponding multiplication between feature elements; Convolution operation is used to extract image features; Feature splicing is used to fuse different dimensional features; Layer normalization operation is used to accelerate model convergence; 、 、 Weight matrix is used to control the direction and strength of feature extraction; 、 、 Bias vector is used to adjust the output result; Sigmoid activation function is used to map the output to the 0-1 interval; Hyperbolic tangent activation function is used to introduce non-linear relationship; Softmax activation function is used to output the probability distribution of each growth stage; Spatial feature representation, Temporal feature representation, Growth stage prediction probability distribution); Through the model, the spatial and temporal features of the mushroom image are combined to accurately determine the growth stage of the mushroom; Temporal feature capture and attention mechanism: temporal convolution kernel The size is set to 5x1x1, the sliding step is 2 frames, and it is specially used to capture the spreading rate feature of mycelium in the time dimension; the output layer is connected to the attention mechanism module, and the weight calculation method is: (Where, is the attention weight of the frame, which determines the importance of the frame feature in the final decision; LSTM hidden state records historical information; Spacetime fusion feature vector integrates spatial and temporal dimension features; Long short-term memory network processes time series data; Time series length; Exponential function amplifies the weight difference); Through the attention mechanism, the key growth features are focused, and the recognition accuracy is improved; (Three) instruction generation and output module: Environment regulation instruction generation: According to data analysis and diagnosis results, combined with the operation logic of the multi-strategy control execution module, specific environment regulation instructions are generated; for example, when the humidity is detected to be lower than 75% and in a certain stage of mushroom growth, instructions to turn on the humidifier and floor fan at the same time are generated; if the light intensity is lower than 500lx, instructions to start the light supplement lamp are generated; the instructions contain detailed information such as device type, operation mode (on / off / adjust parameter), duration, etc. Instruction priority processing: According to the conflict resolution mechanism of the multi-strategy control execution module, the generated instructions are prioritized and processed; for example, ventilation instructions are prioritized over humidification instructions, and when the two conflict, the ventilation instructions are executed first to ensure the rationality of environment regulation; Instruction output and communication: Through GPIO interface, PWM output, etc., the regulation instructions are sent to the corresponding devices of the multi-strategy control execution module; at the same time, using NB-IoT or WiFi communication module, key data and processing results are uploaded to the cloud platform server to realize information interaction with the cloud; (4) Collaborative decision-making interaction module: Local emergency regulation execution: For emergency situations that require high response delay (≤10 seconds), such as temperature sudden rise that may cause damage to mushrooms, the edge computing node does not need to wait for cloud instructions, but generates regulation instructions and executes them directly according to local rules and algorithms, quickly stabilizes the environment, and ensures the safety of mushroom growth; Cloud model reception and update: Receive the optimized model based on long-term data training from the cloud platform server every week, including growth stage identification model parameter update, environment regulation strategy optimization, etc.; update the control parameters of the edge node, so that the system can continuously adapt to new growth environment and demand changes; III. Key technical principles: (1) Microcontroller architecture and data processing principles: STM32 microcontroller is based on ARM Cortex-M core, with high performance and low power consumption; its rich peripheral resources (such as timers, ADC, DMA, etc.) support fast collection and processing of multi-source data; through reasonable memory management and task scheduling algorithm, parallel processing of data reception, analysis, instruction generation, etc. is realized to ensure efficient operation of the system; (2) Deployment principles of deep learning algorithms: Improved dual-flow convolutional neural network, LSTM, etc. Deep learning models are optimized through model compression, quantization, etc. to adapt to the limited computing resources and memory space of edge computing nodes; lightweight inference framework (such as TensorFlow Lite for Microcontrollers) is used to realize fast inference of the model locally, completing complex tasks such as mushroom growth stage identification; (Three) Principle of collaborative decision-making mechanism: Based on the concept of edge computing and cloud computing collaboration, the responsibilities of local nodes and the cloud are clearly divided; local nodes are responsible for real-time emergency control, using localized data and rules for rapid response; the cloud is based on a large amount of historical data for in-depth analysis and model training, providing optimization strategies for edge nodes regularly, realizing complementary advantages and improving the overall performance of the system; Four, the workflow of the module: (One) initialization phase: After the edge computing node is powered on, the STM32 microcontroller and each peripheral (communication interface, storage unit, etc.) are initialized and configured, and the hardware connection is detected to see if it is normal; Load built-in expert rule base, deep learning model initial parameters, establish communication connection with environment perception module, multi-strategy control execution module, cloud platform server, prepare to receive and process data; (Two) data receiving and preprocessing phase: Continuously monitor the data transmission of the environment perception module, receive various environmental parameters and image data through the corresponding communication interface; The received data is subjected to protocol analysis, filtering, format conversion and other preprocessing operations, and is stored in the local cache to provide clean and standardized data for subsequent analysis; (Three) data analysis and diagnosis phase: Compare the preprocessed environmental parameter data with the threshold value in the expert rule base to determine if there is an anomaly, and if there is an anomaly, trigger an alarm; The image data is subjected to growth phase recognition using an improved double-flow convolutional neural network, combined with time domain feature capture and attention mechanism, to output accurate growth phase judgment results; (Four) instruction generation and execution phase: According to the data analysis and diagnosis results, generate environmental control instructions according to the logic of the multi-strategy control execution module, and perform priority processing; Send the instructions to the multi-strategy control execution module to control the relevant devices to perform environmental adjustment operations; at the same time, upload the key data and processing results to the cloud platform server; (Five) collaborative decision-making and updating phase: Real-time monitoring of emergency situations, if the local emergency control conditions are met, immediately perform the corresponding operation; Regularly check if the cloud platform server has an optimization model to issue, if so, receive and update the local control parameters, and optimize the system operation strategy; (Six) End phase: When the system stops running or receives a stop instruction, the edge computing node stops data receiving, analysis and instruction generation operations, closes the communication connection, saves the running record and configuration parameters, and releases system resources.

[0021] In this embodiment, the operation logic of the multi-strategy control execution module includes: Timing strategy: start the ventilation equipment in a fixed time period every day, with higher priority than the quantitative strategy; Quantitative strategy: when the humidity is lower than 75%, turn on the humidifier and the floor fan; when the humidity is higher than 95%, turn off; when the light intensity is lower than 500lx, turn on the light supplement lamp; when the light intensity is higher than 1000lx, trigger the sunshade roller shutter; Conflict resolution mechanism: ventilation instruction is prior to humidification instruction, and humidification instruction is prior to light adjustment instruction; Further, the multi-strategy control execution module, as the "action center" of the mushroom growing environment intelligent monitoring system based on the Internet of Things, undertakes the key task of converting the control instructions of the edge computing node and the cloud platform server into actual environment adjustment operations; through the cooperation of multiple devices and flexible control strategies, it realizes the precise control of the mushroom growing environment, and the following will be a comprehensive and detailed description of the module: I. Overall function overview: The multi-strategy control execution module is composed of a humidifier, a ventilation fan, a light supplement lamp, a roller shutter device, and an ultrasonic atomization device, etc. Through receiving the environment control instructions generated by the edge computing node and the global optimization instructions of the cloud platform server, according to the preset timing strategy, quantitative strategy and conflict resolution mechanism, it adjusts the parameters such as temperature and humidity, carbon dioxide concentration, light intensity, etc. in the mushroom cultivation environment, creates suitable environmental conditions for mushroom growth, and guarantees the yield and quality of mushrooms; at the same time, the module can feedback the device running state in real time, ensuring the reliability and effectiveness of the whole control process; II. Submodule composition and function: (1) Environment adjustment device unit: Humidifier and ultrasonic atomization device: the humidifier uses ultrasonic atomization technology to atomize water into small particles through high-frequency oscillation, increasing air humidity; when receiving the instruction of the edge computing node, and the quantitative strategy determines that the environmental humidity is lower than 75%, the humidifier starts, and simultaneously links the floor fan to accelerate the diffusion of water mist, making the humidity evenly distributed in the mushroom cultivation area; the ultrasonic atomization device can control the water mist particle size more finely, further improving the humidification efficiency and uniformity, meeting the special needs of mushrooms for high humidity environment; when the humidity reaches 95%, the humidifier is turned off according to the strategy to avoid excessive humidity causing mushroom diseases; Ventilation fan: The ventilation fan is used to regulate the air flow and carbon dioxide concentration in the mushroom cultivation environment. According to the timing strategy, it is automatically started at fixed time periods every day (such as 10:00-11:00 am and 3:00-4:00 pm) to ensure regular air renewal. When the edge computing node detects that the carbon dioxide concentration is too high or the oxygen content is insufficient, it will also trigger the ventilation fan to operate. The ventilation fan has different speed gears, which can be adjusted according to actual needs to ensure air flow while avoiding excessive wind force affecting mushroom growth. In addition, the ventilation instruction has the highest priority, and when it conflicts with other instructions, the ventilation operation is executed first. Supplementary light and roller shutter equipment: The supplementary light uses LED light source, which can adjust the light intensity and spectrum according to the needs of different mushroom varieties at different growth stages. When the light intensity is lower than 500 lx, the supplementary light automatically turns on to supplement the light. When the light intensity is higher than 1000 lx, the roller shutter equipment starts to adjust the opening degree of the sunshade roller shutter to block the strong light and prevent the mushrooms from being damaged by strong light. The roller shutter equipment can also automatically adjust the position of the sunshade roller shutter according to weather conditions and time changes to achieve dynamic control of light intensity. Other auxiliary equipment: In addition to the above main equipment, the unit may also include heating or cooling equipment (configured according to actual needs) for adjusting temperature, gas injection device for adjusting air composition, etc. These devices work together to achieve comprehensive adjustment of the mushroom growing environment. (II) Instruction receiving and analysis module: Multi-source instruction receiving: This module connects with the edge computing node and cloud platform server through communication interfaces (such as RS-485, WiFi, etc.) to receive real-time control instructions from both ends. Whether it is an emergency control instruction generated by the edge computing node based on local data processing or an instruction generated by the cloud platform server according to long-term data and global optimization strategy, it can be accurately received. Instruction analysis and conversion: The received instructions are analyzed, and the device operation information (such as device type, operation mode, operating parameters, etc.) in the instructions is converted into specific control signals. For example, the instruction "turn on the humidifier and run for 30 minutes" is converted into the corresponding electrical signal and transmitted to the control circuit of the humidifier to ensure that the device can accurately operate according to the instruction requirements. At the same time, the priority information in the instruction is identified to provide a basis for subsequent conflict resolution. (III) Control strategy execution module: Timing strategy execution: The built-in timer automatically triggers the operation of devices such as ventilation fans according to the preset fixed time period every day. The timing strategy has high priority, and once the set time is reached, the corresponding device operation will be executed regardless of the current environmental parameter state, ensuring regular air renewal and maintaining a good gas environment for mushroom growth. Quantitative strategy execution: According to the environmental parameter data provided by the edge computing node, it is judged in real time whether the environmental parameters meet the quantitative strategy conditions; when the humidity is lower than 75%, the humidifier and the floor fan are started at the same time; when the humidity is higher than 95%, the humidifier is turned off; when the light intensity is lower than 500lx, the light supplement lamp is started; when the light intensity is higher than 1000lx, the roller blind device is triggered to act; through this precise quantitative control, the temperature, humidity, light intensity and other parameters of the mushroom growing environment are always kept within the appropriate range; Conflict resolution mechanism execution: when multiple instructions arrive at the same time or conflict occurs during instruction execution, the preset conflict resolution mechanism is used for processing; the ventilation instruction is prior to the humidification instruction, and the humidification instruction is prior to the light regulation instruction; for example, when the ventilation instruction and the humidification instruction exist at the same time, the ventilation operation is executed first, and then the humidification operation is executed after the ventilation is completed, so as to ensure the rationality and effectiveness of the device operation and avoid environmental regulation confusion caused by instruction conflict; (4) Device state feedback module: Real-time state monitoring: sensors or state monitoring circuits are installed on each environmental regulation device to monitor the running state of the device in real time, including whether the device is started, running parameters (such as the humidifier's mist output, the ventilation fan's speed, the light supplement lamp's brightness, etc.), whether the device has faults (such as motor overload, circuit short circuit, etc.); State feedback and processing: the running state information of the device is fed back to the edge computing node and the user interactive terminal through the communication interface; the edge computing node can adjust the subsequent regulation and control strategy according to the device state, such as issuing an alarm in time and enabling a standby device when the device fails; the user interactive terminal displays the device state to the user in an intuitive way, so that the user can master the device running condition in real time and perform remote monitoring and management; III. Workflow of the module: (1) Initialization stage: After the multi-strategy control execution module is powered on, the initialization configuration of the hardware such as device control circuit, communication interface and state monitoring circuit is completed, the device connection is detected to ensure that the device can normally respond to the control instruction; Load the preset configuration information of the timing strategy, the quantitative strategy and the conflict resolution mechanism, establish the communication connection with the edge computing node and the cloud platform server, and prepare to receive the regulation and control instruction; (2) Instruction receiving and analysis stage: Continuously monitor the communication interface to receive the regulation and control instruction from the edge computing node and the cloud platform server in real time; Parse the received instruction, extract the device operation information and priority information in the instruction, convert it into a specific control signal, and store it in the instruction buffer for execution; (3) Control strategy execution stage: Timing policy check: The timer detects the current time in real time, and when the preset timing policy execution time is reached, the control operation of the corresponding device is triggered, and the timing policy instruction is executed preferentially; Quantitative policy judgment: Read the environmental parameter data transmitted by the edge computing node, and judge according to the quantitative policy condition; if the condition is met, generate the corresponding device control instruction, and process according to the instruction priority; Conflict resolution processing: When there are multiple instructions to be executed, the priority of the instructions is sorted according to the conflict resolution mechanism, the high-priority instructions are executed preferentially, and the low-priority instructions are queued or adjusted to ensure that the device executes the operation in order and reasonably; Instruction execution: Send the processed control signal to the corresponding environmental regulation device to control the device to start, stop or adjust the operating parameters, and realize the regulation of the mushroom growing environment; (Four) Device state feedback stage: The device state monitoring circuit collects the running state information of each environmental regulation device in real time, including device operating parameters, fault state, etc.; The device state information is fed back to the edge computing node and the user interaction terminal through the communication interface, so as to carry out subsequent policy adjustment and device management; (Five) End stage: When the system stops running or receives a stop instruction, the multi-policy control execution module stops receiving instructions, closes all running environmental regulation devices, saves device running records and configuration parameters, releases system resources, and completes the work flow.

[0022] In this embodiment, the cloud platform server includes a dynamic optimization engine that generates cultivation parameter suggestions using an improved collaborative filtering algorithm, and the calculation process is as follows: Basic yield calculation: The global average yield base plus the variety bias term and the growth stage bias term; Characteristic interaction term: Inner product summation of variety characteristic vector and stage characteristic vector; Environmental influence term: The environmental parameter vector is transformed by the weight matrix and then passed through the ReLU activation function, multiplied by the environmental factor weight coefficient; Temperature correction term: The temperature variance within 24 hours is input into the Sigmoid function and multiplied by the temperature sensitivity coefficient; Final predicted yield: The sum of the above four results is obtained; The dynamic optimization engine includes an anomaly suppression mechanism: When the environmental parameter fluctuation variance exceeds the threshold value, calculate the Euclidean distance between the variety characteristic vector and the stage characteristic vector; Multiply this Euclidean distance by the activation coefficient calculated by the Sigmoid function from the environmental variance, and add it as a regularization penalty term to the optimization objective function; Further, the cloud platform server as the "wisdom hub" of the mushroom growing environment intelligent monitoring system based on the Internet of Things carries the core functions of data storage, model training, global decision-making, etc. Through the integration and analysis of massive data, it provides optimization strategies and intelligent decision support for the system. The following will be elaborated from multiple aspects: I. Overall function overview: The cloud platform server establishes a connection with the edge computing nodes through the NB-IoT or WiFi communication module, receives and stores real-time data from the environment perception module and the processing results of the edge computing nodes. Relying on powerful computing resources and algorithm models, it conducts in-depth mining and analysis of data, trains growth stage recognition models and cultivation parameter optimization models, generates global optimization instructions based on timing strategies and quantitative strategies, and realizes macro-control of the entire mushroom planting system. At the same time, it provides data interfaces for user interaction terminals, supporting remote viewing and management by users; II. Submodule composition and function: (1) Data storage and management module: Multi-source data access and storage: This module supports receiving environmental parameter data such as temperature, humidity, carbon dioxide concentration, and light intensity uploaded by edge computing nodes, as well as mushroom growth state image data and device operation status information. It uses a distributed storage architecture (such as Hadoop Distributed File System HDFS) to classify and store massive data, ensuring data security and scalability. For example, real-time environmental data is stored by time sequence, and image data is classified and archived by mushroom variety and growth stage; Data indexing and retrieval: An efficient data indexing mechanism is established to achieve fast retrieval and query of data through multi-dimensional indexing such as timestamp, mushroom variety, and cultivation area. When a user requests to view mushroom growth data in a certain time period or area on the interaction terminal, the system can respond within seconds and return the results, providing convenience for data analysis and decision-making; Data backup and recovery: Regularly backup stored data, adopt a strategy combining offsite backup and local backup to prevent data loss. When the system fails or data is damaged, the data recovery process can be quickly started to ensure data integrity and availability; (2) Model training and optimization module: Growth stage recognition model training: Based on a large number of mushroom growth image data uploaded by edge computing nodes, an improved dual-stream convolutional neural network is trained using a deep learning framework (such as TensorFlow, PyTorch). By adjusting network parameters and optimizing loss functions, the accuracy of the model in recognizing mushroom growth stages is continuously improved. For example, using mushroom images in different growth stages as training samples, the model learns the features of each stage such as shape and color, achieving accurate recognition; Cultivation parameter optimization model construction: an improved collaborative filtering algorithm is used to construct the cultivation parameter optimization model, and the optimization model is: (Wherein, is the predicted yield of the variety in the growth stage; is the global average yield base; is the variety bias term, reflecting the inherent yield difference of different varieties; is the growth stage bias term, reflecting the influence of each stage on yield; , are the variety feature vector and stage feature vector (dimension , respectively, used to characterize the characteristics of the variety and the characteristics of the growth stage; is the environmental factor weight coefficient (range 0.8-1.2), adjusting the influence degree of environmental factors on yield prediction; is the rectified linear unit activation function, introducing a nonlinear relationship; is the environmental factor weight matrix; is the environmental factor vector of the variety in the growth stage; is the temperature fluctuation correction coefficient; is the temperature fluctuation correction term, is the temperature fluctuation sensitivity coefficient, is the temperature variance within 24 hours, used to correct the influence of temperature fluctuation on yield prediction); By analyzing historical yield data, environmental parameters and mushroom variety information, the model is trained to predict the optimal cultivation parameters under different conditions; Model updating and optimization: periodically update the model according to the newly collected data, use online learning algorithm, so that the model can adapt to the dynamic changes of mushroom growing environment and the introduction of new varieties; At the same time, through cross-validation, model evaluation indicators (such as accuracy, mean square error, etc.) to evaluate the performance of the model, constantly optimize the model structure and parameters, improve the prediction accuracy and decision reliability; (Three) Global decision and instruction generation module: Data analysis and trend prediction: deep analysis of the stored historical data, using time series analysis, data mining and other algorithms to predict the change trend of mushroom growing environment parameters, the development process of growth stage and the yield change; For example, by analyzing the seasonal variation of temperature and humidity data, predict the environmental parameter fluctuation in the future period, and develop control strategies in advance; Global optimization instruction generation: based on timing strategy and quantitative strategy, combined with model training results and data analysis conclusions, generate global optimization instructions; in terms of timing strategy, according to the growth cycle of fungi and environmental change law, formulate periodic equipment operation plan (such as adjusting ventilation time once a week); in terms of quantitative strategy, when it is predicted that the temperature and humidity in a certain area will deviate from the appropriate range, generate instructions to adjust the operation parameters of humidifier and ventilation fan; the instructions contain specific device operation, parameter setting and execution time, etc. Instruction issuing and collaborative management: through the communication module, the global optimization instructions are issued to the edge computing nodes, and at the same time, the local regulation and control strategy of the edge computing nodes is coordinated to ensure the effective execution of the instructions; after the edge computing nodes feedback the execution results, the execution effect is evaluated, if the expected target is not reached, the instructions or optimization strategy are adjusted again; (4) Safety and operation management module: Data security protection: adopt data encryption (such as SSL / TLS encryption transmission, AES data encryption storage), access control (role-based permission management), firewall and other technical means to ensure the security of data in the process of transmission and storage, prevent data leakage, tampering and illegal access; System monitoring and fault early warning: real-time monitoring of the running state of cloud platform server, including server resource usage (CPU, memory, disk space), network connection state, service response time, etc.; when abnormal conditions (such as server load too high, service interruption) are detected, the alarm mechanism is triggered immediately, and the operation and maintenance personnel are notified through email, SMS and other ways, and detailed fault diagnosis information is provided to facilitate quick positioning and problem solving; System operation and maintenance and upgrading: regular maintenance of servers, including software update, patch installation, performance optimization and other operations to ensure the stability and efficiency of the system; according to business needs and technology development, upgrade and expand the architecture and functions of cloud platform server, such as adding new data analysis algorithms, optimizing model training process, etc. III. Workflow of modules: (1) Initialization phase: After the cloud platform server starts, it completes the initialization configuration of server hardware, operating system, database, application service, etc., and detects whether the network connection and service running state are normal; Load initial parameters such as data storage rules, model training configuration, global decision strategy, establish communication connection with edge computing nodes and user interaction terminals, and prepare to receive and process data; (2) Data receiving and storage phase: Continuously monitor the communication interface and receive various data uploaded by edge computing nodes in real time, including environmental parameters, image data, device status, etc. The received data is verified for format, integrity checked, and after removing noise and outliers through data cleaning, stored in the database according to the preset storage rules, and a data index is established; (Three) Model training and optimization phase: Periodically extract historical data from the database as a sample set for model training; according to different model requirements (growth stage identification model, cultivation parameter optimization model, etc.), preprocess and feature engineering are performed on the data; Use deep learning and machine learning frameworks to train the model, adjust model parameters and optimize algorithms to achieve the best performance; after training, evaluate and verify the model, and if it meets the requirements, save the model; otherwise, continue to optimize and train; Based on newly collected data, periodically update and iterate the model to ensure that the model can adapt to environmental changes and business needs; (Four) Global decision-making and instruction generation phase: Analyze the stored historical data and use data analysis algorithms to predict the growth environment and yield trend of mushrooms; According to the timing strategy and quantitative strategy, combined with the model prediction results, generate global optimization instructions; after reasonable checking and priority sorting of the instructions, send them to the edge computing node through the communication module; Receive the feedback of the edge computing node on the instruction execution results, and evaluate the execution effect. If the expected goal is not achieved, adjust the strategy and generate new instructions; (Five) Security and operation phase: Real-time monitoring of server running status, detection of network attacks, data abnormal access and other security threats, and timely protection and processing; Periodically maintain and upgrade the server, including software updates, data backups, performance optimizations and other operations to ensure stable operation of the system; (Six) End phase: When the system stops running or receives a stop instruction, the cloud platform server stops data reception, model training and instruction generation, closes the communication connection, saves the running records and configuration parameters, and releases system resources.

[0023] In this embodiment, the user interaction terminal provides: A three-dimensional space monitoring view that displays a three-dimensional cultivation rack environment parameter distribution heat map of multiple layers of mushrooms; An intelligent decision support interface that returns growth trend assessment and picking time prediction after receiving user-uploaded mushroom images; Further, the user interaction terminal serves as a "bridge" between the mushroom growing environment intelligent monitoring system based on the Internet of Things and the user, and undertakes core functions such as data visualization presentation, device remote control, and parameter individualization configuration. Through convenient and intuitive interaction, the user can easily manage the whole process of mushroom planting. The following will elaborate on it from multiple dimensions: I. Overall function overview: The user interaction terminal covers PC and mobile APP, and connects with the cloud platform server and edge computing nodes through the network to obtain mushroom growing environment data, device running status, and system analysis results in real time. It provides functions such as environment data visualization display, device remote control, growth stage parameter configuration, and intelligent decision support, enabling users to intuitively understand mushroom planting conditions and flexibly adjust system settings to achieve intelligent and precise mushroom planting management. II. Submodule composition and functions: (1) Data visualization module: Real-time data display: Real-time presentation of mushroom cultivation environment parameters such as temperature and humidity, carbon dioxide concentration, and light intensity through intuitive charts and interfaces. PC uses large-screen visualization design, displaying data in various chart forms such as dashboards, line charts, and column charts. Mobile APP uses a simple card layout, allowing users to swipe left and right to switch between different parameter pages and quickly view current environment data. Three-dimensional space monitoring view: Based on the layout information of the three-dimensional cultivation rack, a three-dimensional virtual scene is constructed to display the environmental parameter distribution heat map of multi-layer mushrooms. Users can view parameter differences such as temperature and humidity, light intensity, and other parameters in different areas from different angles through mouse dragging and zooming (PC) or gesture operation (mobile APP), and intuitively locate abnormal areas for precise control. Historical data query and analysis: Supports users to query historical environment data and device running records according to time, area, mushroom variety, etc. Through the generation of data comparison charts such as temperature and humidity change curve comparison in different time periods and yield data comparison in different areas, it helps users analyze the influence of environmental factors on mushroom growth and summarize planting experience. (2) Device remote control module: Device status real-time monitoring: Real-time display of the running status of humidifiers, ventilation fans, light supplement lamps, and roller screen devices on the interaction interface, and intuitive display of device opening, running parameters, etc. through icon color, text prompts, etc. For example, a green icon indicates normal device operation, and a red icon indicates device failure or shutdown status. Remote control operation: Users can remotely send instructions to control the start and stop of the device, adjust the operating parameters through PC or mobile APP, such as remotely turning on the light supplement lamp and adjusting the brightness, setting the running time and speed of the ventilation fan, etc. The operation interface is simple and easy to understand, providing clear operation buttons and parameter adjustment sliders for easy user operation. Batch control and scene mode setting: Support batch control of multiple devices, users can select multiple similar devices in different cultivation areas at the same time and send control instructions uniformly. In addition, scene modes such as "mushroom production mode" and "mycelium growth mode" can be set, and the corresponding device combination and parameter settings in this mode can be started with one key, realizing fast and convenient environmental control. (Three) Parameter configuration and management module: Growth stage adaptive parameter configuration: According to the growth characteristics of different mushroom varieties, provide growth stage division and corresponding environmental parameter configuration function; users can customize the temperature and humidity, carbon dioxide concentration, light intensity and other parameter thresholds of each growth stage, and the system will adjust the environment according to these settings and alarm; The configuration interface provides detailed parameter explanations and recommended value references to help users set parameters scientifically; System basic parameter setting: Allow users to set the basic parameters of the system, such as data refresh frequency, alarm notification method (SMS, APP push, email, etc.), device communication parameters, etc.; Users can flexibly adjust these parameters according to their own needs and network environment to optimize system use experience; User permission management: Support multi-user management, set different user roles and permission levels; For example, administrator users have full operational permissions and can control devices, configure parameters, and manage users; Ordinary users can only view data and receive alarm information to ensure the safety and standardization of system operation; (Four) Intelligent decision support module: Growth trend evaluation: After users upload mushroom images, the system uses the growth stage recognition model trained by the cloud platform server to analyze and evaluate the growth state of the mushrooms, providing information such as the current growth stage and health status of the mushrooms, and providing corresponding care suggestions such as whether to adjust environmental parameters or prevent pests and diseases; Picking time prediction: Based on mushroom growth stage data, environmental parameter history records, and yield prediction models, the system predicts the best picking time for users, helping them plan picking schedules and improve mushroom quality and economic benefits; Cultivation parameter optimization suggestions: Based on data analysis results and cultivation parameter optimization models on the cloud platform server, the system provides personalized cultivation parameter optimization suggestions, such as how to adjust temperature and humidity, light intensity, etc. to achieve higher yield and better quality under current environmental conditions; (five) alarm and notification module: Real-time alarm display: When the system detects abnormal environmental parameters (such as high temperature, low humidity), equipment failure or illegal intrusion of personnel, etc., an alarm window (PC end) or an alarm notification (mobile APP) is popped up in real time on the user interaction terminal, and the user is prompted with a prominent color and sound; the alarm information includes alarm type, occurrence time, specific location and other detailed content, making it easy for users to quickly understand the situation; Alarm history query: Save all alarm records, users can query historical alarm information at any time, quickly locate and analyze alarm events through filtering conditions (such as time range, alarm type), summarize the law, and take preventive measures to avoid similar problems from happening again; Notification management: In addition to alarm notifications, system message notifications (such as new feature online, software update prompt), operation result notifications (device control success or failure prompt) are also supported; users can choose the type and method of receiving notifications in the settings to avoid information interference; Three, the working process of the module: (1) initialization phase: After the user opens the PC program or mobile APP, the user identity is verified, and the communication connection with the cloud platform server is established after the verification; Load user personalized configuration information (such as interface layout, parameter settings, permission information, etc.), initialize the data visualization interface and each function module, and prepare to obtain and display data; (2) data acquisition and display phase: Send data requests to the cloud platform server regularly to obtain the latest mushroom growing environment data, device running status and other information; The received data is analyzed and processed, and the user is displayed in the form of charts, views, etc. through the data visualization module, while the device state display is updated in real time; (3) user operation processing phase: Listen to user operations on the interactive interface, such as device control operations, parameter configuration modifications, data query requests, image uploads, etc. Verify and process user operations, send device control instructions, parameter configuration data, etc. to the cloud platform server, and execute the corresponding operations by the server coordinating the edge computing nodes; for data query and image analysis requests, transmit the data to the cloud for processing, and then return the results to the user for display; (4) alarm and notification processing phase: Real-time receive alarm information and system notifications pushed by the cloud platform server, and display or push according to the user's notification settings; Save alarm information and notification records to the local database for user to query historical records; (V) End stage: When the user exits the program or closes the APP, the communication connection with the cloud platform server is disconnected, the user's operation record and temporary configuration information during use are saved, and the system resources are released.

[0024] In this embodiment, the system further comprises a security monitoring module: The infrared sensor detects illegal intrusion of personnel in real time, triggers the local sound and light alarm and remote APP alarm; The standby battery automatically switches when the power supply is interrupted, maintaining the continuous operation of the system for at least 72 hours; Further, the security monitoring module serves as the "security line" of the mushroom growing environment intelligent monitoring system based on the Internet of Things, and through real-time monitoring of personnel activities and system power supply status, a comprehensive security protection system is built. The following describes the function architecture and technical principles in detail: I. Overall function overview: The security monitoring module is composed of two core sub-modules: personnel intrusion detection and power supply guarantee. The infrared sensor monitors the personnel activities in the mushroom cultivation area in real time. Once illegal intrusion is detected, the local sound and light alarm is triggered and a remote alarm is sent to the user interaction terminal. At the same time, a standby battery system is provided to automatically switch when the main power supply is interrupted, ensuring that the key equipment of the system operates continuously for ≥72 hours, ensuring uninterrupted data acquisition and environmental control, and providing a safe and stable operating environment for mushroom cultivation; II. Sub-module composition and function: (I) Personnel intrusion detection unit: Infrared sensing monitoring: Infrared sensors are deployed at key positions such as entrances and exits, surrounding boundaries, etc. in the mushroom cultivation area, using active or passive infrared detection technology. The active infrared sensor emits an infrared light beam. When a person or object blocks the light beam, the receiver detects the change in light intensity, triggering an alarm. The passive infrared sensor identifies personnel activities by detecting changes in human infrared radiation energy. The detection angle and distance can be adjusted according to the actual scene to achieve full coverage of the cultivation area. Alarm triggering mechanism: When the infrared sensor detects illegal intrusion of personnel, it immediately sends a trigger signal to the edge computing node. After receiving the signal, the edge computing node controls the local sound and light alarm to start, emitting a high-decibel alarm sound and flashing light, intimidating the intruder. On the other hand, through the NB-IoT or WiFi communication module, the alarm information (including alarm time, location coordinates, etc.) is pushed to the user interaction terminal in the form of pop-up windows, vibration, voice broadcast, etc. to remind the user, so that the user can take timely measures; Video recording evidence function (optional extension): can integrate high-definition camera and infrared sensor linkage, automatically start recording function when alarm is triggered, record the whole process of the intrusion event; recorded video data is stored in local storage device or uploaded to cloud platform server, providing strong evidence for subsequent event investigation and responsibility tracing; (II) Power supply guarantee unit: Power state monitoring: real-time monitoring of input voltage, current, frequency and other parameters of the main power supply through voltage sensor and current sensor to determine whether the power supply is normal; once the main power supply voltage is lower than the normal threshold (such as 80% of the rated voltage) or the current fluctuates abnormally, it is immediately identified as a main power supply failure, triggering the standby power supply switching process; Backup battery system: equipped with large-capacity lead-acid battery or lithium battery pack as backup power supply, its capacity is strictly calculated to ensure that it can continuously power the edge computing nodes, environmental perception modules, communication modules and other key devices for ≥72 hours when the main power supply is interrupted; the battery system has built-in charge and discharge management module, with overcharge protection, overdischarge protection, short circuit protection and other functions, prolonging the service life of the battery and ensuring the stability of power supply; Automatic switching and recovery: use intelligent switching switch (such as static transfer switch STS) to realize seamless switching between main power supply and backup power supply; when the main power supply fails, the switching switch switches the load to the backup power supply within milliseconds to ensure uninterrupted operation of the system equipment; when the main power supply returns to normal, the switching switch automatically switches the load back to the main power supply and starts the charging program for the backup battery to restore it to full power, preparing for the next power failure; (III) Security data management module: Alarm record storage: detailed records of personnel intrusion alarms, power failure alarms and all security events, including event occurrence time, type, processing status and other information, are stored in the local database or synchronized to the cloud platform server; supports quick search and query by time, type and other conditions, facilitating user review and analysis of security events; Security state display: set up a security monitoring special interface on the user interaction terminal to display the running state of the current security monitoring module with intuitive charts and status indicator lights; for example, a green indicator light means the system is safe and normal, and a red indicator light means an alarm event has occurred; a column chart is used to count the number of alarms in different time periods to help users understand the trend of security conditions; Security policy configuration: allows users to customize security monitoring policies on the interaction terminal, such as setting the arming / disarming time of the infrared sensor, adjusting the power alarm threshold, selecting the alarm notification method, etc.; users can flexibly configure according to actual needs to improve the pertinence and effectiveness of security monitoring; III. Workflow of the module: (I) Initialization phase: After the security monitoring module is started, self-checking is performed on hardware devices such as infrared sensors, power monitoring sensors, and backup battery systems, to check whether device connections are normal and parameter configurations are correct; Load security monitoring policy configuration information (such as alarm thresholds, notification methods, etc.), establish communication connections with edge computing nodes, cloud platform servers, and user interaction terminals, and prepare to enter a monitoring state; (2) Real-time monitoring phase: The infrared sensor continuously monitors the personnel activity in the cultivation area, and the power status monitoring module detects the operating parameters of the main power supply in real time, and transmits the collected data to the edge computing node for analysis and processing; The edge computing node performs real-time judgment on the data, and if illegal intrusion of personnel or main power supply failure is found, the corresponding alarm and processing procedures are triggered immediately; (3) Alarm and processing phase: Personnel intrusion alarm: When illegal intrusion of personnel is detected, the edge computing node controls the local sound and light alarm to start, and sends alarm information to the user interaction terminal; after receiving the notification, the user can view the alarm details through the interaction terminal, and take processing measures (such as contacting security personnel) according to the actual situation; Power failure processing: When the main power supply fails, the intelligent switch automatically switches the load to the backup battery power supply to ensure continuous system operation; the edge computing node sends power failure alarm information to the user interaction terminal, prompting the user of the main power supply interruption and the current backup power supply state; (4) Recovery and recording phase: Personnel intrusion event processing: After the user processes the personnel intrusion event, the user can confirm and mark the alarm information on the interaction terminal, and the system records the event processing result; Power recovery: After the main power supply is restored to normal, the intelligent switch automatically switches the load back to the main power supply, and starts the backup battery charging program; the edge computing node sends power recovery notification to the user interaction terminal, and records the whole process information of power failure and recovery; (5) End phase: When the system stops running or receives a stop instruction, the security monitoring module stops data collection and monitoring, turns off the power of related devices, saves running records and configuration parameters, and releases system resources.

[0025] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring system for mushroom growth environment based on the Internet of Things, characterized by: include: Environmental perception module: This includes temperature and humidity sensors, carbon dioxide concentration sensors, light intensity sensors, and image acquisition devices distributed throughout the mushroom cultivation units, used to collect real-time data on mushroom growth environment parameters and growth status; Edge computing node: uses an STM32 microcontroller, connected to the environmental perception module, for local preprocessing and abnormal diagnosis of collected data, and generating environmental control instructions; Multi-strategy control execution module: including a humidifier, ventilation fan, fill light, roller blind equipment and ultrasonic atomization device, receiving instructions from the edge computing node and performing environmental adjustment operations; Cloud platform server: Connected to edge computing nodes via NB-IoT or WiFi communication modules, it is used to store environmental data, train growth stage recognition models, and generate global optimization instructions based on timing and quantitative strategies; User interaction terminal: including PC and mobile APP, providing environmental data visualization interface, equipment remote control interface and growth stage adaptive parameter configuration function.

2. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: The local preprocessing of the edge computing node includes: The built-in expert rule library compares the thresholds of temperature, humidity, CO2 concentration and light intensity data, and triggers real-time alarms if they exceed the preset range; An improved two-stream convolutional neural network is used for growth stage recognition, and its algorithm process is as follows: Spatial feature flow calculation: After concatenating the RGB three-channel image features with the grayscale morphological features enhanced by the edge gradient operator, a convolution operation is performed with the spatial convolution kernel. After adding the bias term, the spatial feature tensor is obtained through the Sigmoid activation function. Time domain feature flow calculation: The feature difference matrix of adjacent time frames is convolved with the time domain convolution kernel, and after adding the bias term, the time domain feature tensor is obtained through the hyperbolic tangent activation function; Growth stage probability output: The spatial feature tensor and the time domain feature tensor are concatenated and normalized, then multiplied by the fully connected weight matrix, and the bias term is added and the Softmax function is used to obtain the probability distribution of each growth stage.

3. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 2, characterized in that: The temporal convolution kernel size is 5×1×1, and the sliding step is 2 frames, which is used to capture the characteristics of hyphae spreading rate; The output layer is connected to the attention mechanism module, and its calculation process is as follows: use the LSTM network to process the hidden state of the previous moment and the spatiotemporal fusion feature vector of the current moment to obtain the attention score of the current moment, and then perform exponential normalization on the attention scores of all moments to obtain the attention weight of the current moment.

4. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: The cloud platform server includes a dynamic optimization engine that uses an improved collaborative filtering algorithm to generate cultivation parameter recommendations. The calculation process is as follows: Basic yield calculation: global average yield base plus variety bias and growth stage bias; Feature interaction term: sum the inner products of the variety feature vector and the stage feature vector; Environmental impact term: The environmental parameter vector is transformed by the weight matrix and then passed through the ReLU activation function and multiplied by the environmental factor weight coefficient; Temperature correction term: The temperature variance within 24 hours is input into the Sigmoid function and multiplied by the temperature sensitivity coefficient; Final predicted output: obtained by adding the above four results.

5. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 4, characterized in that: The dynamic optimization engine includes an exception suppression mechanism: When the variance of environmental parameter fluctuation exceeds the threshold, the Euclidean distance between the variety feature vector and the stage feature vector is calculated; The Euclidean distance is multiplied by the activation coefficient calculated by the environmental variance through the Sigmoid function and added to the optimization objective function as a regularization penalty term.

6. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: The operation logic of the multi-strategy control execution module includes: Timing strategy: Start ventilation equipment at a fixed time every day, which has a higher priority than quantitative strategy; Quantitative strategy: When the humidity is below 75%, the humidifier and floor fan are turned on in conjunction, and when it is above 95%, they are turned off; when the light intensity is below 500lx, the fill light is activated, and when it is above 1000lx, the sunshade roller blinds are triggered; Conflict resolution mechanism: ventilation instructions take precedence over humidification instructions, and humidification instructions take precedence over light adjustment instructions.

7. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: The user interaction terminal provides: A three-dimensional spatial monitoring view shows a heat map of the environmental parameter distribution of multi-layer mushrooms in a three-dimensional cultivation rack; The intelligent decision support interface receives mushroom images uploaded by users and returns growth status assessment and picking time prediction.

8. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: The edge computing nodes and cloud platform servers adopt a collaborative decision-making mechanism: The local node performs emergency control with a response delay of no more than 10 seconds; The cloud sends out optimization models based on long-term data training every week to update edge node control parameters.

9. The intelligent monitoring system for mushroom growth environment based on the Internet of Things according to claim 1, characterized in that: Also includes security monitoring module: Infrared sensors detect illegal intrusions in real time, triggering local sound and light alarms and remote APP alarms; The backup battery automatically switches when the power is interrupted, keeping the system running for at least 72 hours.

Images