Edible mushroom environment collaborative regulation method based on internet of things and industrial control

Abstract

The application relates to the cross field of computer technology and industrial control systems, and discloses an edible mushroom environment collaborative regulation method based on an Internet of Things and industrial control. The method comprises the following steps: collecting environment and mushroom state data in real time through a multi-source heterogeneous sensor array; using an edge side event trigger discrimination engine to identify abnormalities and generate priority regulation events; using a multi-target collaborative decision unit of an industrial control core to combine a physiological response model and execution constraints to generate a multi-actuator collaborative action instruction set; and using a time-sensitive network to realize hard real-time instruction transmission and synchronous execution. The application realizes precise closed-loop regulation with high timeliness, high energy efficiency and high robustness, and improves the stability and intelligent level of edible mushroom cultivation.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of computer technology and industrial control systems, specifically involving a method for the coordinated regulation of the edible fungi environment based on the Internet of Things and industrial control. Background Technology

[0002] With the deep integration of the Internet of Things (IoT) and industrial control technology in modern agriculture, the industrialized cultivation of edible fungi places higher demands on the intelligence level of environmental control systems. The growth process of edible fungi is highly dependent on the precise coordination of multi-dimensional environmental parameters such as temperature, humidity, carbon dioxide concentration, and light intensity, and their physiological state exhibits significant stages and sensitivities. Traditional environmental control systems generally employ periodic sampling and fixed-frequency control strategies, that is, collecting sensor data and executing control commands according to preset time intervals. While this method is structurally simple, it is essentially an "egalitarian" resource allocation model, unable to distinguish the severity and urgency of environmental changes, resulting in a large amount of computing, communication, and execution resources being consumed in redundant operations during stable operation phases.

[0003] Event-driven intelligent control has become a key direction for improving system responsiveness. An "event" specifically refers to a sudden change in state that has a decisive impact on the growth process of edible fungi or system safety, such as the completion of mycelial cover, the large-scale formation of primordia, a sharp rise in CO2 concentration exceeding the threshold, or abnormal data from key sensors—all high-value biological or equipment events. These events often have short windows of opportunity and far-reaching impacts; failure to identify and respond promptly can lead to yield losses, quality degradation, or even batch contamination. However, existing industrial control systems lack proactive sensing and priority processing mechanisms for such events. Their control logic is fixed in cyclical loops, making it difficult to interrupt routine processes, switch control strategies, and reallocate system resources within millisecond to second timescales.

[0004] Existing technologies generally suffer from the following problems: data acquisition and control execution are strictly bound to a fixed clock cycle, making it impossible to dynamically adjust the sampling frequency and actuator response priority according to the urgency of the event; control algorithms and parameter sets are statically fixed, lacking a dedicated control mode library for different biological events, resulting in coarse response strategies and poor adaptability; system resources are allocated according to the principle of equal distribution, making it impossible to achieve instantaneous focusing and reconstruction of local resources when an event is triggered. These problems are particularly prominent when dealing with sudden biological events or equipment failures, easily causing control lag or failure, severely restricting the stability and intelligence level of industrialized edible fungi production. Therefore, there is an urgent need for a collaborative control method that can integrate event perception, mode switching, and resource reallocation capabilities, enabling the system to have rapid and accurate response characteristics similar to neural reflexes. Summary of the Invention

[0005] This invention provides a collaborative control method for the edible fungi environment based on the Internet of Things (IoT) and industrial control. It aims to solve the problems of wasted computing resources, network bandwidth, and energy caused by the fixed-frequency data acquisition and command issuance in traditional periodic sampling control during edible fungi cultivation. It also overcomes the technical shortcomings of traditional periodic sampling control, such as slow response, delayed control, and even control failure when faced with sudden biological events or equipment malfunction alarms. This invention constructs a dynamic collaborative control architecture that deeply integrates event-driven and state-aware technologies, achieving precise closed-loop control of the edible fungi growth microenvironment with high timeliness, high energy efficiency, and high robustness.

[0006] This invention provides a method for the coordinated regulation of the edible fungi environment based on the Internet of Things and industrial control, comprising:

[0007] A multi-source heterogeneous sensor array deployed in the edible fungus cultivation space is used to collect raw sensing data in real time that characterizes the growth status of edible fungi and environmental parameters.

[0008] An event-triggered discrimination engine, set up at an edge computing node, is used to perform online feature extraction and abnormal pattern recognition on the raw sensing data to generate a control event signal with priority identifier.

[0009] A multi-objective collaborative decision-making unit configured in the core of industrial control is used to receive the control event signal and generate a set of multi-actuator collaborative action instructions based on the preset physiological response model of edible fungi and the constraints of equipment execution capability.

[0010] A deterministic communication bus connected to various environmental control actuators is used to transmit the multi-actuator coordinated action instruction set in hard real-time mode, driving the actuators to complete the joint regulation of temperature, humidity, carbon dioxide concentration, light intensity and ventilation rate.

[0011] Preferably, the multi-source heterogeneous sensor array includes a distributed temperature and humidity sensor, an infrared thermal imaging array, a carbon dioxide concentration detection module, a spectrally tunable light source feedback unit, an airflow velocity sensor, and a visual monitoring device for mycelial growth status.

[0012] Distributed temperature and humidity sensors are installed in a grid layout on each layer of the cultivation rack; an infrared thermal imaging array covers the entire cultivation area.

[0013] The carbon dioxide concentration detection module uses the non-dispersive infrared principle;

[0014] The spectrally tunable light source feedback unit integrates multi-channel photodiodes, corresponding to the blue light, red light, and far-infrared light bands respectively;

[0015] The airflow velocity sensor is a thermal anemometer;

[0016] The mycelial growth status visual monitoring device consists of a high-resolution industrial camera and a ring LED light source.

[0017] Preferably, the event-triggered discrimination engine performs sliding window segmentation on the raw sensing data stream from the multi-source heterogeneous sensor array;

[0018] Wavelet packet decomposition is performed on the data in each window to extract energy entropy, sample entropy and singular values ​​as time-frequency domain feature vectors.

[0019] The time-frequency domain feature vector is input into a pre-trained lightweight convolutional neural network model, which includes three convolutional layers, a global average pooling layer, and a fully connected output layer. The number of nodes in the output layer is equal to the total number of predefined event categories.

[0020] The event categories include events such as sudden changes in mycelial growth rate, localized high temperature and humidity condensation events, sudden increases in carbon dioxide concentration events, lighting system failure events, fan shutdown events, and comprehensive environmental deviation events.

[0021] When the confidence score of any event category is greater than a preset threshold, a corresponding control event signal is generated and a priority identifier mapped from the confidence score is attached.

[0022] Preferably, the multi-objective collaborative decision-making unit maintains a dynamic priority queue to cache all pending regulatory event signals;

[0023] The multi-objective collaborative decision-making unit periodically retrieves the highest priority regulatory event signal from the queue and queries the built-in edible fungi physiological response model library;

[0024] The edible fungi physiological response model library contains multi-component piecewise linearized state-space equations for different fungal species and different growth stages. Each set of equations describes the quantitative relationship between environmental parameter disturbances and mycelial biomass accumulation rate and fruiting body differentiation probability.

[0025] Preferably, the multi-objective collaborative decision-making unit matches the most suitable state-space equation from the model library based on the disturbance type, disturbance amplitude, and disturbance location indicated by the current control event signal; then it constructs a multivariable model predictive control problem, with the objective function being to minimize the weighted quadratic integral of the environmental parameters deviating from the set trajectory, while simultaneously satisfying the hard constraints of actuator action amplitude, action rate, and action coordination.

[0026] By solving the optimization problem, the following sequences are generated for a specified time period in the future: temperature control valve opening sequence, humidifier power sequence, carbon dioxide solenoid valve on / off sequence, light source spectral ratio sequence, and fan speed sequence. This sequence is the multi-actuator coordinated action instruction set.

[0027] Preferably, the deterministic communication bus uses a time-sensitive networking protocol stack, and the physical layer is gigabit Ethernet;

[0028] All environmental control actuators are equipped with embedded controllers that support time-sensitive networking, and have internal hardware-level timestamp units and instruction prefetch buffers.

[0029] The multi-executor cooperative action instruction set is encapsulated into a time-stamped scheduling frame before being sent.

[0030] Time-sensitive network switches forward data within a pre-allocated time gating window based on the time stamp in the scheduling frame.

[0031] Preferably, after generating the multi-actuator cooperative action instruction set, the method also includes execution effect verification and online model calibration steps;

[0032] Specifically, the multi-source heterogeneous sensor array collects the actual response values ​​of environmental parameters within a specified time period after the command is executed;

[0033] The multi-objective collaborative decision-making unit calculates the residual between the actual response value and the model prediction value. If the Euclidean norm of the residual is greater than the preset tolerance threshold three times in a row, the online update process of the model parameters is triggered.

[0034] The online parameter update process of this model adopts the recursive least squares method, which uses the latest input and output data to iteratively correct the system matrix and input matrix of the currently active state space equation.

[0035] Preferably, the multi-objective collaborative decision-making unit is further configured with an equipment health status monitoring module; the equipment health status monitoring module continuously receives diagnostic data from the current sensor, voltage sensor and vibration sensor built into the actuator; when the diagnostic data shows that the actuator has signs of jamming, overload or slow response, the equipment health status monitoring module immediately generates a degraded operation plan for the equipment and injects it into the dynamic priority queue as the highest priority control event signal.

[0036] Preferably, the current sensor is used to detect abnormal motor starting current;

[0037] Voltage sensors monitor fluctuations in the power supply voltage;

[0038] The vibration sensor is a piezoelectric accelerometer, which is mounted on the actuator bearing housing;

[0039] The criteria for determining jamming are that the vibration amplitude is greater than the threshold, the criteria for overload are that the current is continuously greater than 120% of the rated value, and the criteria for slow response are that the actual action delay is greater than 5 milliseconds.

[0040] Preferably, the equipment degradation operation plan specifies the activation strategy of the backup actuator or the safe operation boundary of the main actuator; the backup actuator activation strategy includes switching to redundant fans.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] 1. This invention abandons the traditional fixed-cycle polling control architecture and instead adopts a dynamic collaborative control mechanism that deeply integrates event-driven and state-aware technologies. This eliminates the overhead of ineffective data acquisition, transmission, and computation during stable environmental periods, thereby reducing the overall energy consumption and network load of the system.

[0043] 2. By deploying a lightweight online anomaly detection model, this invention ensures millisecond-level perception and response capabilities to sudden biological events and equipment failure alarms, avoiding production losses or quality degradation caused by control lags.

[0044] 3. The multivariable model predictive control framework introduced by the multi-objective collaborative decision-making unit can take into account the strong coupling characteristics between multiple environmental parameters and generate truly collaborative actuator action commands, overcoming the parameter conflict and oscillation problems that are easily caused by traditional single-loop independent control.

[0045] 4. The introduction of a deterministic communication bus provides hard real-time assurance for the precise synchronous execution of instructions, ensuring the reliable implementation of complex control strategies.

[0046] 5. The execution effect verification and online model correction mechanism endows the system with the ability to continuously learn and adaptively evolve, enabling it to maintain high-precision regulation performance over a long period of time and cope with the inherent challenges such as nonlinearity and time-varying nature in the growth process of edible fungi. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention;

[0048] Figure 2 This is a schematic diagram of the core principle framework of the event triggering discrimination engine in this invention;

[0049] Figure 3 This is a logical flowchart of the multi-objective collaborative decision-making unit in this invention;

[0050] Figure 4 This is a schematic diagram of the data flow and time synchronization mechanism for the coordinated control of the deterministic communication bus and the actuator in this invention;

[0051] Figure 5 This is a closed-loop feedback logic framework diagram for performance verification and online model correction in this invention;

[0052] Figure 6This is a schematic diagram of the fault tolerance interaction between equipment health status monitoring and degraded operation plan generation in this invention. Detailed Implementation

[0053] refer to Figures 1 to 6 This invention provides a method for the coordinated regulation of the edible fungi environment based on the Internet of Things and industrial control. The core of this method lies in constructing a dynamic coordinated regulation architecture that deeply integrates event-driven and state-aware technologies to achieve precise closed-loop control of the edible fungi growth microenvironment with high timeliness, high energy efficiency, and high robustness. The method deploys a multi-source heterogeneous sensor array, sets up an event-triggered discrimination engine for edge computing nodes, configures a multi-objective collaborative decision-making unit for the industrial control core, and relies on a deterministic communication bus to complete the hard real-time transmission and execution of instructions, thereby completely eliminating the resource waste and response lag inherent in traditional periodic sampling control.

[0054] The method first includes a multi-source heterogeneous sensor array deployed in the edible fungus cultivation space. This multi-source heterogeneous sensor array consists of a distributed temperature and humidity sensor, an infrared thermal imaging array, a carbon dioxide concentration detection module, a spectrally tunable light source feedback unit, an airflow velocity sensor, and a visual monitoring device for mycelial growth status.

[0055] Distributed temperature and humidity sensors are installed in a grid layout on each layer of the cultivation rack, with no fewer than 9 measuring points on each layer. The horizontal and vertical spacing is no more than 1.5 meters. The sampling resolution is 0.1 degrees Celsius and 1% relative humidity, and the sampling frequency is once per second.

[0056] The infrared thermal imaging array covers the entire cultivation area, employing an uncooled microbolometer focal plane detector with a frame rate of 25 frames per second, thermal sensitivity better than 0.05 degrees Celsius, and a spatial resolution of 640×480 pixels, used to capture subtle changes in the temperature distribution on the surface of the mushroom bed.

[0057] The carbon dioxide concentration detection module uses the non-dispersive infrared principle, with a range of 0 to 5000 ppm, an accuracy of ±20 ppm, a response time of less than 2 seconds, and is installed at the center height of the bacterial bed to reflect the local gas concentration produced by bacterial respiration.

[0058] The spectrally tunable light source feedback unit integrates three-channel photodiodes, corresponding to the blue light band of 450 nm, the red light band of 660 nm, and the far-infrared light band of 730 nm, respectively, for real-time monitoring of the actual ratio and intensity of the light source output spectrum.

[0059] The airflow velocity sensor is a thermal anemometer with a range of 0 to 5 meters per second and a response time of less than 500 milliseconds. It is installed at the air supply and return vents to monitor the actual air velocity of the ventilation system.

[0060] The mycelial growth status visual monitoring device consists of a high-resolution industrial camera and a ring LED light source. The image resolution is 4912×3264 pixels, and the working wavelength is 850 nanometers in the near-infrared band to avoid interference from visible light on mycelial metabolism. It triggers a full-area image acquisition every 30 minutes to quantify mycelial coverage density, edge expansion rate, and surface texture characteristics.

[0061] After completing the acquisition of raw sensing data, the method enters the event-triggered discrimination stage. This discrimination stage is executed by the event-triggered discrimination engine located on the edge computing node. The event-triggered discrimination engine performs sliding window segmentation on the raw sensing data stream from the multi-source heterogeneous sensor array. The window length is fixed at 10 seconds, and the step size is 1 second, ensuring the continuity and timeliness of data processing.

[0062] For the data within each window, the event-triggered discrimination engine performs wavelet packet decomposition, using the db4 wavelet basis function and a four-level decomposition layer, thus decomposing the original signal into 16 sub-bands. Subsequently, for each sub-band, the energy entropy, sample entropy, and singular values ​​are calculated to form a 48-dimensional time-frequency domain feature vector.

[0063] Energy entropy is used to characterize the uniformity of energy distribution of a signal across frequency bands, sample entropy is used to measure the complexity and regularity of a signal, and singular values ​​reflect the degree of energy concentration of the principal components of the signal matrix. This 48-dimensional time-frequency domain feature vector is input into a pre-trained lightweight convolutional neural network model. This lightweight convolutional neural network model contains three convolutional layers: the first layer has 32 kernels and a size of 1×3; the second layer has 64 kernels and a size of 1×3; and the third layer has 128 kernels and a size of 1×3. Each convolutional layer is followed by batch normalization and a ReLU activation function; then a global average pooling layer is applied to compress the feature map into a 128-dimensional vector; finally, a fully connected output layer is passed, with the number of nodes equal to the total number of predefined event categories, i.e., six categories.

[0064] The event categories include events such as sudden changes in mycelial growth rate, localized high temperature and humidity condensation, sudden increases in carbon dioxide concentration, lighting system failure, fan shutdown, and comprehensive environmental deviations. The output of the lightweight convolutional neural network model is the confidence score for each event category. When any score exceeds a preset threshold of 0.85, the event triggering engine generates the corresponding control event signal, appending a priority identifier mapped from the confidence score: a confidence score greater than 0.95 indicates high priority, between 0.85 and 0.95 indicates medium priority, less than 0.85 but greater than 0.75 indicates low priority, and less than 0.75 is considered noise and not triggered. The priority identifier serves as metadata for the control event signal and is transmitted to the industrial control core along with the signal.

[0065] After receiving the control event signals, the industrial control core processes them through a multi-objective collaborative decision-making unit. This unit maintains a dynamic priority queue, where all pending control event signals are sorted from highest to lowest priority, and signals of the same priority are arranged in chronological order by timestamp. Every 10 milliseconds, the multi-objective collaborative decision-making unit retrieves the highest-priority control event signal from the queue. Subsequently, it queries a built-in edible fungi physiological response model library. This library contains multi-component piecewise linearized state-space equations for different fungal species (such as oyster mushrooms, shiitake mushrooms, and enoki mushrooms) and different growth stages (mycelial growth stage, primordia formation stage, and fruiting body expansion stage). Each set of equations is in the form of:

[0066] ;

[0067] ;

[0068] The rate of change of the state vector. The state vector contains latent variables such as mycelial biomass accumulation rate and fruiting body differentiation probability. The control input vector includes setpoints for temperature, humidity, carbon dioxide concentration, light intensity, and ventilation rate. This is the observable output vector, corresponding to the actual measured values ​​of environmental parameters; , , The system matrix, input matrix, and output matrix are shown below. This represents the external disturbance term.

[0069] The multi-objective collaborative decision-making unit matches the most suitable state-space equation from the model library based on the disturbance type, disturbance amplitude, and disturbance location indicated by the current regulatory event signal. For example, if the event is a "local high temperature and high humidity condensation event" and the location is on the east side of the third-layer cultivation rack, the state-space equation corresponding to the mycelial growth stage of shiitake mushrooms is matched, and the local environmental coupling parameters of the region are loaded.

[0070] After matching is completed, the multi-objective collaborative decision-making unit constructs a multivariate model predictive control problem. Its objective function is... Defined as:

[0071] ;

[0072] For a moment The observable output vector; For a moment The control input vector; The current moment; For the time domain of the prediction, set it to 15 seconds; The reference trajectory is determined by the standard environmental parameter curve of the current growth stage; Assign higher weights to temperature, humidity, and carbon dioxide concentration to the output error weighting matrix; The input weighting matrix is ​​used to control and suppress excessive actuator movement. The optimization problem must meet the following hard constraints: the opening rate of the temperature regulating valve is no more than 5% per second; the humidifier power output is no less than 10% or greater than 90% of the rated power; the on / off switching interval of the carbon dioxide solenoid valve is no less than 2 seconds; the proportion of blue light in the light source spectrum is no less than 15%; and the fan speed is no less than 300 revolutions per minute.

[0073] By solving the quadratic programming problem using the interior point method, the following sequences are generated for a specified future time period of 15 seconds: temperature control valve opening sequence, humidifier power sequence, carbon dioxide solenoid valve on / off sequence, light source spectral ratio sequence, and fan speed sequence. This is the multi-actuator coordinated action instruction set.

[0074] The generated multi-actuator cooperative action instruction set is transmitted to various environmental control actuators via a deterministic communication bus. This deterministic communication bus adopts a time-sensitive networking protocol stack, with a gigabit Ethernet physical layer and IEEE 802.1Qbv time-aware shaper and IEEE 802.1Qbu frame preemption mechanism enabled at the data link layer. All environmental control actuators are equipped with embedded controllers that support time-sensitive networking, and internally include hardware-level timestamp units and instruction prefetch buffers.

[0075] The multi-actuator coordinated action instruction set is encapsulated into a scheduling frame before transmission. The scheduling frame contains the instruction content, actuator address, and precise time stamp. The time stamp indicates the absolute moment when the instruction takes effect at the actuator with nanosecond precision. This absolute moment is generated synchronously by the master clock of the industrial control core.

[0076] Time-sensitive network switches forward data within a pre-allocated time-gated window based on the timestamps in the scheduling frames, ensuring that the end-to-end delay from the industrial control core issuing a command to the actuator starting to act is constant at two milliseconds, with jitter less than 10 microseconds. After receiving the scheduling frame, the actuator stores it in the command prefetch buffer and triggers the execution action at the time specified by the timestamp, achieving strict synchronization of multiple actuators in the time dimension.

[0077] After the command is executed, the method further includes execution effect verification and online model calibration steps. The multi-source heterogeneous sensor array collects the actual response values ​​of environmental parameters at specified time intervals of 5 seconds, 10 seconds, and 15 seconds after command execution. The multi-objective collaborative decision-making unit then processes the actual response values... Compared with model predictions Perform residual calculation; the residual vector is: If the Euclidean norm of the residuals If the residual value exceeds a preset tolerance threshold three times consecutively (e.g., temperature residual greater than 0.5 degrees Celsius, humidity residual greater than 3%), the online model parameter update process is triggered. This online model parameter update process uses the recursive least squares method, utilizing the latest input and output data to update the system matrix of the currently active state-space equation. With input matrix Perform iterative corrections. The correction formula is:

[0078] ;

[0079] ;

[0080] ;

[0081] The parameter vector contains and Element; The updated parameter vector; For regression vectors; This is the transpose of the regression vector; It is the gain vector; It is the covariance matrix; This is the updated covariance matrix; for The actual output at any given moment; It is the identity matrix; The forgetting factor is set to 0.99. The step size factor is fixed at 0.01 to ensure the model's tracking ability while maintaining numerical stability.

[0082] In addition, the multi-objective collaborative decision-making unit is also equipped with an equipment health status monitoring module. This module continuously receives diagnostic data from the current, voltage, and vibration sensors built into the actuator. The current sensor has a sampling frequency of 1000 Hz and is used to detect abnormal motor starting current; the voltage sensor monitors fluctuations in the power supply voltage; the vibration sensor is a piezoelectric accelerometer, installed in the actuator bearing housing, with a sampling frequency of 5000 Hz, used to identify mechanical jamming or imbalance. When diagnostic data shows that the actuator exhibits signs of jamming (vibration amplitude greater than the threshold), overload (current continuously greater than 120% of the rated value), or slow response (actual action delay greater than 5 milliseconds), the equipment health status monitoring module immediately generates a degraded operation plan for the equipment.

[0083] The equipment degradation operation plan specifies the activation strategy of the backup actuator (such as switching to the redundant fan) or the safe operating boundary of the main actuator (such as limiting the opening of the temperature control valve to below 50%), and injects it into the dynamic priority queue as the highest priority control event signal to ensure that the environmental control function can still maintain basic stability in degradation mode in the event of a single point of failure.

[0084] In summary, this embodiment constructs an efficient, robust, and adaptive edible fungi environment collaborative regulation system through event-driven data acquisition, lightweight online anomaly identification, multi-objective collaborative decision-making based on physiological models, hard real-time deterministic communication, and a closed-loop model self-correction mechanism. This effectively solves the fundamental defects of traditional periodic control in terms of resource utilization efficiency and emergency response capability.

[0085] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0086] Although embodiments of the invention 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 invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control, characterized in that, include: A multi-source heterogeneous sensor array deployed in the edible fungus cultivation space is used to collect raw sensing data in real time that characterizes the growth status of edible fungi and environmental parameters. An event-triggered discrimination engine, set up at an edge computing node, is used to perform online feature extraction and abnormal pattern recognition on the raw sensing data to generate control event signals with priority identifiers. A multi-objective collaborative decision-making unit configured in the core of industrial control is used to receive the control event signal and generate a set of multi-actuator collaborative action instructions based on the preset physiological response model of edible fungi and the constraints of equipment execution capability. A deterministic communication bus connected to various environmental control actuators is used to transmit the multi-actuator coordinated action instruction set in hard real-time mode, driving the actuators to complete the joint regulation of temperature, humidity, carbon dioxide concentration, light intensity and ventilation rate.

2. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 1, characterized in that, The multi-source heterogeneous sensor array includes a distributed temperature and humidity sensor, an infrared thermal imaging array, a carbon dioxide concentration detection module, a spectrally tunable light source feedback unit, an airflow velocity sensor, and a visual monitoring device for mycelial growth status. Distributed temperature and humidity sensors are installed in a grid layout on each layer of the cultivation rack; an infrared thermal imaging array covers the entire cultivation area. The carbon dioxide concentration detection module uses the non-dispersive infrared principle; The spectrally tunable light source feedback unit integrates multi-channel photodiodes, corresponding to the blue light, red light, and far-red light bands respectively; The airflow velocity sensor is a thermal anemometer; The mycelial growth status visual monitoring device consists of a high-resolution industrial camera and a ring LED light source.

3. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 2, characterized in that, The event-triggered discrimination engine performs sliding window segmentation on the raw sensing data stream from the multi-source heterogeneous sensing array; Wavelet packet decomposition is performed on the data in each window to extract energy entropy, sample entropy and singular values ​​as time-frequency domain feature vectors. The time-frequency domain feature vector is input into a pre-trained lightweight convolutional neural network model, which includes three convolutional layers, a global average pooling layer, and a fully connected output layer. The number of nodes in the output layer is equal to the total number of predefined event categories. The event categories include events such as sudden changes in mycelial growth rate, localized high temperature and humidity condensation events, sudden increases in carbon dioxide concentration events, lighting system failure events, fan shutdown events, and comprehensive environmental deviation events. When the confidence score of any event category is greater than a preset threshold, a corresponding control event signal is generated and a priority identifier mapped from the confidence score is attached.

4. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 3, characterized in that, The multi-objective collaborative decision-making unit maintains a dynamic priority queue to cache all pending control event signals; The multi-objective collaborative decision-making unit periodically retrieves the highest priority regulatory event signal from the queue and queries the built-in edible fungi physiological response model library; The edible fungi physiological response model library contains multi-component piecewise linearized state-space equations for different fungal species and different growth stages. Each set of equations describes the quantitative relationship between environmental parameter disturbances and mycelial biomass accumulation rate and fruiting body differentiation probability.

5. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 4, characterized in that, The multi-objective collaborative decision-making unit matches the most suitable state-space equation from the model library based on the disturbance type, disturbance amplitude, and disturbance location indicated by the current control event signal; then it constructs a multivariable model predictive control problem with the objective function being to minimize the weighted quadratic integral of the environmental parameters deviating from the set trajectory, while satisfying the hard constraints of actuator action amplitude, action rate, and action coordination. By solving the optimization problem, the following sequences are generated for a specified time period in the future: temperature control valve opening sequence, humidifier power sequence, carbon dioxide solenoid valve on / off sequence, light source spectral ratio sequence, and fan speed sequence. This sequence is the multi-actuator coordinated action instruction set.

6. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 5, characterized in that, The deterministic communication bus uses a time-sensitive networking protocol stack, and the physical layer is gigabit Ethernet; All environmental control actuators are equipped with embedded controllers that support time-sensitive networking, and have internal hardware-level timestamp units and instruction prefetch buffers. The multi-executor cooperative action instruction set is encapsulated into a time-stamped scheduling frame before being sent. Time-sensitive network switches forward data within a pre-allocated time gating window based on the time stamp in the scheduling frame.

7. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 6, characterized in that, After generating the multi-actuator cooperative action instruction set, the process also includes execution effect verification and online model calibration steps; Specifically, the multi-source heterogeneous sensor array collects the actual response values ​​of environmental parameters within a specified time period after the command is executed; The multi-objective collaborative decision-making unit calculates the residual between the actual response value and the model prediction value. If the Euclidean norm of the residual is greater than the preset tolerance threshold three times in a row, the online update process of the model parameters is triggered. The online parameter update process of this model adopts the recursive least squares method, which uses the latest input and output data to iteratively correct the system matrix and input matrix of the currently active state space equation.

8. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 7, characterized in that, The multi-objective collaborative decision-making unit is also equipped with an equipment health status monitoring module; the equipment health status monitoring module continuously receives diagnostic data from the current sensor, voltage sensor and vibration sensor built into the actuator; when the diagnostic data shows that the actuator has signs of jamming, overload or slow response, the equipment health status monitoring module immediately generates a degraded operation plan for the equipment and injects it into the dynamic priority queue as the highest priority control event signal.

9. The method for coordinated environmental control of edible fungi based on the Internet of Things and industrial control according to claim 11, characterized in that, The current sensor is used to detect abnormal motor starting current. Voltage sensors monitor fluctuations in the power supply voltage; The vibration sensor is a piezoelectric accelerometer, which is mounted on the actuator bearing housing; The criteria for determining jamming are that the vibration amplitude is greater than the threshold, the criteria for overload are that the current is continuously greater than 120% of the rated value, and the criteria for slow response are that the actual action delay is greater than 5 milliseconds.

10. The method for coordinated regulation of edible fungi environment based on Internet of Things and industrial control according to claim 9, characterized in that, The equipment degradation operation plan specifies the activation strategy for the standby actuator or the safe operating boundary of the main actuator; the standby actuator activation strategy includes switching to redundant fans.

Images