An intelligent control system and method for nitrogen filling in an agricultural product warehouse

Abstract

The application discloses a kind of agricultural product warehouse nitrogen filling intelligent control system and method, its system includes perception layer, execution layer, edge computing control terminal and cloud management platform;Perception layer is used to real-time acquisition in pile environment parameter, curtain distension distance and nitrogen making machine working condition;Execution layer is used to pass through electric regulating valve, two-way ball valve and frequency double nitrogen making machine set dynamic control nitrogen flow and filling and discharging time sequence;Edge computing control terminal fuses multi-source data, solves the internal state of stack, generates optimized nitrogen filling strategy based on the stack state prediction model issued by cloud, realizes accurate collaborative control;Cloud platform uses historical data to continuously optimize the stack state prediction model and issues updated parameters.The application realizes the fine collaborative control to nitrogen making machine operating state and each stack nitrogen filling valve, can improve the real-time performance, reliability and energy efficiency level of nitrogen filling control, reduce manual intervention and operating cost, with good engineering applicability and popularization value.

Description

Technical Field

[0001] This invention relates to the field of precision agricultural product maintenance technology, and in particular to an intelligent control system and method for nitrogen filling agricultural product warehouses. Background Technology

[0002] Agricultural products are highly susceptible to the effects of oxygen, ambient temperature and humidity, and harmful organisms during long-term aging and storage, leading to mold, insect infestation, and quality loss. Nitrogen, as a green and pollution-free preservation method, creates a low-oxygen environment by filling sealed stacks with high-purity nitrogen, effectively inhibiting the respiration of agricultural products and the breeding of pests. This is of great significance for ensuring the safe storage of agricultural products and improving the quality of aging. Achieving automation, precision, and energy efficiency optimization in the nitrogen filling process has become an urgent need for upgrading warehousing technology in the industry.

[0003] Existing nitrogen filling systems for agricultural product warehouses mostly employ on / off control based on preset thresholds, meaning the nitrogen generator is simply started and stopped based on the oxygen concentration measured by sensors. However, this approach reveals significant drawbacks in practical applications: 1. Due to the extremely tight stacking of agricultural products, gas diffusion is severely delayed, resulting in surface oxygen concentrations meeting standards while residual oxygen remains in the core area; 2. Existing control schemes neglect the viscoelastic deformation of the flexible tent under nitrogen filling pressure, making it highly susceptible to tent bursting or sealing failure due to pressure overshoot; 3. When multiple stacks are filled with nitrogen in parallel, there is a lack of effective energy efficiency scheduling algorithms, causing the nitrogen generator units to operate at fixed power, resulting in significant energy waste.

[0004] Therefore, it is necessary to propose an intelligent control system and method for nitrogen filling in agricultural product warehouses, which can solve the problem of delayed oxygen diffusion inside the stack during the existing nitrogen filling control process, making it difficult to eliminate residual oxygen in the core area, avoid excessive deformation or even rupture of the flexible tent caused by improper nitrogen filling pressure control, and improve the safety, stability and economy of nitrogen filling maintenance process in agricultural product warehouses. Summary of the Invention

[0005] In view of this, the present invention provides an intelligent control system and method for nitrogen filling in agricultural product warehouses, which solves the technical problems of excessive nitrogen filling and high energy consumption caused by existing nitrogen filling systems for agricultural product warehouses due to their reliance on fixed thresholds and lack of stacking status perception and dynamic optimization capabilities.

[0006] In a first aspect, the present invention provides an intelligent control system for nitrogen filling in agricultural product warehouses, including an edge computing control terminal, wherein the edge computing control terminal is connected to a perception layer, an execution layer and a cloud management platform; The perception layer is used to collect real-time environmental parameters of agricultural product stacking, tent deformation characteristics, and nitrogen generator operating conditions. It includes an environmental composite sensor deployed at each stack location, an ultrasonic ranging sensor deployed on the tent surface, and an edge acquisition module that monitors the operating status of the nitrogen generator unit. The execution layer is used to dynamically adjust the nitrogen production, the timing and flow of intake and exhaust, including electric regulating valves arranged in the main pipeline, bidirectional electric ball valves arranged at the intake and exhaust ends of each stack, and duplex nitrogen generators with variable frequency speed control function. The edge computing control terminal is used to calculate stacking characteristic parameters in real time based on the data from the perception layer, and generate a real-time nitrogen filling control strategy based on the stacking state prediction model, so as to realize the coordinated scheduling of nitrogen generator power and the opening degree of each stacking valve. The cloud management platform is used to acquire historical nitrogen filling operation data, train and update the stacking status prediction model, and send the updated model parameters to the edge computing control terminal.

[0007] Furthermore, the perception layer and execution layer adopt a hybrid heterogeneous communication architecture, including: The wireless sensing subnet includes an environmental composite monitoring module and a stacking nitrogen filling status monitoring module that are wirelessly connected via a LoRa gateway, which is connected to the edge computing control terminal via a wired connection. The dual-mode execution subnet allows the electric regulating valves of the nitrogen-filling pipeline to be configured with two control modes based on their location: the main pipeline electric valve is connected to the electric valve control cabinet on each floor and is controlled via PLC in a wired manner; the stack end electric valve is connected to a wireless valve controller and is controlled via the LoRaWAN protocol, and the wireless valve controller has a logic protection circuit with an automatic interlock function when the network is disconnected. The industrial bus device network allows the PLC controller of the duplex nitrogen generator to communicate directly with the serial / network port of the edge computing gateway via Modbus or OPC UA protocol, enabling millisecond-level power command issuance and device status feedback.

[0008] Secondly, this invention provides an intelligent control method for nitrogen filling in agricultural product warehouses, which is implemented using the above-mentioned technical solution and includes: The edge computing control terminal receives data from the perception layer, including the temperature and humidity of the stacking environment, oxygen concentration, and tent bulging distance, and constructs the current stacking physical state matrix. Based on the stacking status prediction model issued by the cloud management platform, the core oxygen concentration of the stack and the internal moisture migration rate of agricultural products are calculated according to the stacking physical state matrix. With the objective function of maintaining the oxygen concentration of the entire stack below the biological survival threshold and nitrogen production energy consumption, and combining the core oxygen concentration, the moisture migration rate of agricultural products and the working status characteristics of the nitrogen generator, a collaborative control sequence is generated that includes the target frequency of the nitrogen generator and the pulse duty cycle of each stack electric valve. At the same time, the control sequence is safely corrected in real time based on the tent deformation characteristics fed back by the ultrasonic ranging sensor. The execution layer adjusts the gas production of the duplex nitrogen generator and the opening degree and opening and closing sequence of the inlet and outlet valves of each stack according to the control sequence.

[0009] Furthermore, the stacking physical characteristic prediction model includes: a porous medium diffusion sub-model, a thermodynamically coupled moisture migration sub-model, and a tent viscoelasticity sub-model; The viscoelastic mechanical sub-model of the tent is used to calculate the internal air pressure field formed by the tent tension feedback based on the measured tent bulging distance, and to update the equivalent porosity of the stack in real time according to the tent expansion volume. The thermodynamically coupled moisture migration sub-model is used to calculate the fluid mass transfer characteristics on the surface and between layers of agricultural products based on real-time temperature and humidity, output the internal moisture migration rate of the agricultural products, and dynamically compensate the diffusion coefficient in the porous medium diffusion sub-model based on the changes in gas composition caused by moisture migration. The porous media diffusion sub-model is used to use the pressure field provided by the tent viscoelasticity sub-model as the convection driving force, and to solve the unsteady convection-diffusion equation by combining the equivalent porosity and diffusion coefficient, to simulate the infiltration process of nitrogen into the deep layers of the stack, and to obtain the predicted value of the core oxygen concentration of the stack.

[0010] Furthermore, the objective function, which is to maintain the oxygen concentration of the entire stack below the biological survival threshold and nitrogen generation energy consumption, combines the core oxygen concentration, the moisture migration rate of agricultural products, and the operating status characteristics of the nitrogen generator to generate a coordinated control sequence that includes the target frequency of the nitrogen generator and the pulse duty cycle of each stack's electric valves. This sequence includes: A multi-objective cost function is constructed based on core oxygen concentration, agricultural product dry loss due to moisture migration, and nitrogen generator energy consumption. Using the tent displacement acceleration as a constraint, a predictive control algorithm is used to calculate the control vector sequence that minimizes the multi-objective cost function in each sampling period; Based on the control vector sequence, the ratio of nitrogen generator output power to the opening and closing time of the terminal valve is determined, and a coordinated control scheme is generated.

[0011] Furthermore, the step of using the tent displacement acceleration as a constraint and employing a predictive control algorithm to calculate the control vector sequence that minimizes the multi-objective cost function within each sampling period includes: Using the current nitrogen generator frequency, regulating valve opening, and ball valve pulse duty cycle as initial values, and combining them with historical control trajectories, an initial control sequence for a preset time period is generated. Calculate the gradients of the multi-objective cost function and the tent displacement acceleration constraint with respect to each control variable; At the current working point, the cost function is approximated as a quadratic function through Taylor expansion, and the acceleration constraint is linearized to the first order, thus transforming the multi-objective problem into multiple local quadratic programming sub-functions. Using the interior point method, under the constraints of ensuring that the tent displacement acceleration does not exceed the limit and that each actuator is physically limited, the optimal control increment step size is iteratively searched along the negative gradient direction to update the control vector sequence. After obtaining the optimal control vector sequence that meets the convergence condition, the control vector at the first sampling moment in the sequence is extracted and mapped to the nitrogen generator target frequency signal, the opening control command of each branch regulating valve, and the pulse modulation timing signal of the electric ball valve at the inlet and outlet gas ends. This signal is then sent to the execution layer as a collaborative control scheme.

[0012] Furthermore, the cooperative control scheme employs a variable frequency pulse width modulation strategy, including: Based on the physical state matrix of the stack, the porosity characteristics of the stack are extracted, and the target frequency that causes the pressure wave inside the stack cavity to resonate is calculated. Based on the target frequency, and combined with the pulse duty cycle output by the model predictive control algorithm, a square wave signal sequence for controlling the opening and closing of the electric ball valve is generated. The execution layer drives the electric ball valve at the air inlet to perform high-frequency pulse action according to the square wave signal sequence, forming a fluctuating air pressure at the air inlet. The mechanical impact force of the pressure wave breaks the boundary layer gas film trapped between the gaps in agricultural products, forcing nitrogen to migrate to the compacted area in a convection diffusion mixing mode, thereby shortening the time for the spatial oxygen concentration to reach equilibrium.

[0013] Furthermore, the frequency conversion pulse width modulation strategy also includes: Based on the physical layout of the nitrogen filling pipeline network, calculate the fluid resistance imbalance caused by the difference in pipe length between the near-end and far-end stacked branches. Based on the branch impedance, the opening pulse width of the electric ball valve of the far stack is automatically increased, while the opening degree of the electric regulating valve of the main pipeline of the near stack is reduced, so as to achieve physical consistency of the charging pressure of all stacks in the warehouse.

[0014] Furthermore, the real-time security correction of the control sequence includes: The system monitors the humidity at the nitrogen inlet and the humidity gradient inside the stack in real time. When the slope of the humidity drop in the stack caused by nitrogen replacement exceeds the preset threshold, the system adjusts the micro-humidification device at the end of the nitrogen generator or reduces the instantaneous flow rate of the pulse nitrogen charging.

[0015] The present invention also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the intelligent nitrogen filling control method for agricultural product warehouses described in any of the above technical solutions.

[0016] Compared with existing technologies, the intelligent nitrogen-filling control system and method for agricultural product warehouses proposed in this invention have the following advantages: (1) By integrating and calculating the stacking environment parameters, tent deformation characteristics and nitrogen generator operating conditions on the edge computing control terminal side, characteristic parameters reflecting the real physical state of the stack are constructed. A real-time nitrogen filling control strategy is generated using a prediction model, thereby achieving coordinated scheduling of nitrogen generator power and the opening degree of each stack valve, significantly improving the accuracy and adaptability of nitrogen filling control, and avoiding local oxygen concentration runaway or over-nitrogen filling.

[0017] (2) Historical nitrogen filling operation data is centrally managed and analyzed through a cloud management platform, the stacking status prediction model is continuously trained and updated, and the updated model parameters are sent to the edge computing control terminal to achieve continuous self-optimization of the control strategy. The control objectives simultaneously introduce the full stack oxygen concentration constraint and nitrogen production energy consumption constraint, so that the system can dynamically optimize the nitrogen generator operating frequency and valve pulse duty cycle under the premise of meeting the conditions for inhibiting biological survival, thereby reducing the overall energy consumption. This not only improves the long-term economic efficiency of the system, but also enhances the system's adaptability to different storage scales and agricultural product characteristics.

[0018] In summary, this invention achieves coordinated and refined control of the nitrogen generator's operating status and the nitrogen filling valves of each stack by integrating multi-source sensing data, stack status prediction models, and a hybrid heterogeneous communication architecture. This improves the real-time performance, reliability, and energy efficiency of nitrogen filling control, reduces manual intervention and operating costs, and has good engineering applicability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of the intelligent nitrogen-filling control system for agricultural product warehouses provided by the present invention; Figure 2 This invention provides a structural diagram of the communication architecture of each module; Figure 3 A schematic diagram of the architecture for wireless valve control provided by the present invention; Figure 4 This is a schematic diagram of the process for intelligent nitrogen filling control in agricultural product warehouses provided by the present invention. Detailed Implementation

[0020] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0021] Example 1

[0022] Please see Figure 1 This embodiment provides an intelligent nitrogen-filling control system 100 for agricultural product warehouses, including: The perception layer 101 is used to collect stacking environmental parameters, tent deformation characteristics and nitrogen generator operating conditions in real time. It includes an environmental composite sensor deployed at each stack position, an ultrasonic ranging sensor deployed on the tent surface, and an edge acquisition module for monitoring the operating status of the nitrogen generator unit. The execution layer 102 is used to dynamically adjust the nitrogen production, the timing and flow of the intake and exhaust, including an electric regulating valve arranged in the main pipeline, a two-way electric ball valve arranged at the intake and exhaust ends of each stack, and a duplex nitrogen generator with variable frequency speed control function. The edge computing control terminal 103 is used to calculate stacking characteristic parameters in real time based on the data of the perception layer, and generate a real-time nitrogen filling control strategy based on the stacking state prediction model to realize the coordinated scheduling of nitrogen generator power and the opening degree of each stacking valve. The cloud management platform 104 is used to acquire historical nitrogen filling operation data, train and update the stacking status prediction model, and send the updated model parameters to the edge computing control terminal.

[0023] The system in this embodiment receives stacking environmental parameters, oxygen concentration, and tent deformation information collected by the perception layer through an edge computing control terminal. It constructs a stacking physical state matrix and calculates the core oxygen concentration and agricultural product moisture migration rate based on a stacking state prediction model distributed from the cloud. Under the premise of optimizing stacking safety and nitrogen production energy consumption, it generates a collaborative control sequence including the target frequency of the nitrogen generator and the pulse duty cycle of each stacking valve, achieving dynamic adjustment of the nitrogen generator's gas production and the air intake and exhaust processes of each stack. This system can adaptively adjust the nitrogen filling strategy according to the actual stacking state, improving the accuracy and stability of nitrogen filling control.

[0024] Since agricultural products must meet the standards of being free of insects, mold, and odors during storage, in some embodiments, the sensing layer also includes an insect monitoring module, which is used to detect signs of live insect activity on the surface of the stack or in the gaps of the tent using visual devices. The insect monitoring module includes a high-definition infrared camera and an edge inference unit deployed around the stack, which are used to capture and identify the morphological characteristics and movement trajectory of adult insects in real time in low-light environments, and generate an insect pest risk level signal based on the number of insects, frequency of occurrence, and location distribution, so that the edge computing control terminal can dynamically adjust the nitrogen filling intensity and maintenance time of the stack.

[0025] In a preferred embodiment, the perception layer and the execution layer adopt a hybrid heterogeneous communication architecture, including: The wireless sensing subnet includes an environmental composite monitoring module and a stacking nitrogen filling status monitoring module that are wirelessly connected via a LoRa gateway, which is connected to the edge computing control terminal via a wired connection. The dual-mode execution subnet allows the electric regulating valves of the nitrogen-filling pipeline to be configured with two control modes based on their location: the main pipeline electric valve is connected to the electric valve control cabinet on each floor and is controlled via PLC in a wired manner; the stack end electric valve is connected to a wireless valve controller and is controlled via the LoRaWAN protocol, and the wireless valve controller has a logic protection circuit with an automatic interlock function when the network is disconnected. The industrial bus device network allows the PLC controller of the duplex nitrogen generator to communicate directly with the serial / network port of the edge computing gateway via Modbus or OPC UA protocol, enabling millisecond-level power command issuance and device status feedback.

[0026] Specifically, such as Figure 2 As shown, Figure 2 The system communication architecture diagram in a practical application is shown.

[0027] As a specific example, the warehouse is divided into multiple layers, and the stacks may be located on different floors. In order to control the stacks on different floors, each stack position is equipped with two electric control valves, one inlet and one outlet. The electric valve control lines enter the valve control cabinets on each floor. The control cabinets on each floor are connected to the server via network cables through an edge computing gateway, which supports remote control of each electric valve at the on-site control / system terminal on each floor.

[0028] like Figure 3 As shown, the electric valve controller is connected to the server via a LoRaWAN gateway, supporting remote control of each electric valve from the field control on each floor / system terminal. An oxygen + temperature and humidity composite detector (monitoring oxygen concentration within the stack) and an ultrasonic ranging sensor (monitoring the degree of tent bulging) are installed inside the stack. When nitrogen filling operations are performed on multiple stack locations on a specific floor, the system automatically opens the corresponding floor and stack access electric control valves, monitors the oxygen concentration and tent bulging degree within the stack, and automatically opens and closes the corresponding electric valves based on the corresponding nitrogen filling business logic and configured linkage rules, thereby achieving automatic nitrogen filling and ensuring that the oxygen concentration meets the standard and the stack bulging state does not exceed the limit. In addition, this system also supports direct remote control of the nitrogen generator's opening and closing, adjustment of the nitrogen generator's output power, and opening and closing of corresponding floor and stack valves via system endpoint manual methods, thereby achieving manual nitrogen filling.

[0029] Example 2

[0030] like Figure 4 As shown, this embodiment of the invention also provides an intelligent control method for nitrogen filling in agricultural product warehouses, implemented using the intelligent control system for nitrogen filling in agricultural product warehouses described in Embodiment 1, comprising: Step S101: Receive the stacking environment temperature and humidity, oxygen concentration and tent swelling distance uploaded by the perception layer through the edge computing control terminal, and construct the current stacking physical state matrix; Step S102: Based on the stacking status prediction model issued by the cloud management platform, calculate the core oxygen concentration of the stack and the internal moisture migration rate of agricultural products according to the stacking physical state matrix. Step S103: With the objective function of maintaining the oxygen concentration of the entire stack below the biological survival threshold and nitrogen production energy consumption, and combining the core oxygen concentration, the moisture migration rate of agricultural products and the working status characteristics of the nitrogen generator, a collaborative control sequence containing the target frequency of the nitrogen generator and the pulse duty cycle of each stack electric valve is generated. At the same time, the control sequence is safely corrected in real time based on the tent deformation characteristics fed back by the ultrasonic ranging sensor. Step S104: The execution layer adjusts the gas production of the duplex nitrogen generator and the opening degree and opening and closing sequence of the inlet and outlet valves of each stack according to the control sequence.

[0031] As a preferred embodiment, the stacking physical characteristic prediction model includes: a porous medium diffusion sub-model, a thermodynamically coupled moisture migration sub-model, and a tent viscoelasticity sub-model; The viscoelastic mechanical sub-model of the tent is used to calculate the internal air pressure field formed by the tent tension feedback based on the measured tent bulging distance, and to update the equivalent porosity of the stack in real time according to the tent expansion volume. The thermodynamically coupled moisture migration sub-model is used to calculate the fluid mass transfer characteristics on the surface and between layers of agricultural products based on real-time temperature and humidity, output the internal moisture migration rate of the agricultural products, and dynamically compensate the diffusion coefficient in the porous medium diffusion sub-model based on the changes in gas composition caused by moisture migration. The porous media diffusion sub-model is used to use the pressure field provided by the tent viscoelasticity sub-model as the convection driving force, and to solve the unsteady convection-diffusion equation by combining the equivalent porosity and diffusion coefficient, to simulate the infiltration process of nitrogen into the deep layers of the stack, and to obtain the predicted value of the core oxygen concentration of the stack.

[0032] Furthermore, the aforementioned unsteady porous media diffusion model includes a compensation algorithm. By introducing physical laws, it achieves predictive control of invisible areas to address the problem of pseudo-nitrogen filling caused by gas permeation lag due to compacted agricultural product stacking, resulting in surface compliance but still high oxygen levels in the core. The compensation algorithm is expressed by the following formula: in, For the actual surface concentration measured by the sensor, and The diffusion retardation coefficient is obtained by training the cloud management platform based on historical data. For the actual surface concentration measured by the sensor, This refers to the porosity parameter of agricultural product stacks.

[0033] In a preferred embodiment, the objective function is to maintain the oxygen concentration of the entire stack below the biological survival threshold and nitrogen generation energy consumption. This is combined with core oxygen concentration, agricultural product moisture migration rate, and nitrogen generator operating status characteristics to generate a coordinated control sequence containing the nitrogen generator target frequency and the pulse duty cycle of each stack's electric valves. The sequence includes: A multi-objective cost function is constructed based on core oxygen concentration, agricultural product dry loss due to moisture migration, and nitrogen generator energy consumption. Using the tent displacement acceleration as a constraint, a predictive control algorithm is used to calculate the control vector sequence that minimizes the multi-objective cost function in each sampling period; Based on the control vector sequence, the ratio of nitrogen generator output power to the opening and closing time of the terminal valve is determined, and a coordinated control scheme is generated.

[0034] Specifically, the multi-objective cost function is constructed as follows: in, For the predicted core oxygen concentration, For nitrogen generator energy consumption, This refers to the amount of dry loss due to moisture migration. These are the weighting coefficients.

[0035] Within each sampling period, using the tent displacement acceleration as a nonlinear constraint, the cost function is solved. The smallest control vector sequence includes: Calculate the target frequency of the nitrogen generator based on the instantaneous total gas demand of all nitrogen-filled stacks; Based on the fluid impedance differences of each stack branch, calculate the opening control amount of the electric regulating valve of each floor main pipeline to ensure uniform gas distribution between each stack position. By combining the predicted core oxygen concentration with the safety margin of the tent stress, the pulse duty cycle of the bidirectional electric ball valve at the gas inlet / outlet end of each stack is calculated. By controlling the timing ratio of opening and closing, nitrogen infiltration is accelerated by pressure pulsation while protecting the tent safety.

[0036] In a preferred embodiment, the step of using the tent displacement acceleration as a constraint and employing a predictive control algorithm to calculate the control vector sequence that minimizes the multi-objective cost function within each sampling period includes: Using the current nitrogen generator frequency, regulating valve opening, and ball valve pulse duty cycle as initial values, and combining them with historical control trajectories, an initial control sequence for a preset time period is generated. The gradients of the multi-objective cost function and the tent displacement acceleration constraint with respect to each control variable are calculated respectively; wherein, the tent displacement acceleration constraint is nonlinear, and this scheme linearizes it, and the acceleration is approximated as follows. ,in, The Jacobian matrix is ​​used to reflect the sensitivity of small changes in valve opening to tent acceleration, i.e., the effect of increasing the nitrogen generator frequency by 1 Hz on the core oxygen concentration. At the current working point, the nonlinear cost function is approximated as a quadratic function through Taylor expansion, and the nonlinear acceleration constraint is linearized to the first order, thereby transforming the original optimization problem into a series of local quadratic programming subproblems. Using the interior point method, under the constraints of ensuring that the tent displacement acceleration does not exceed the limit and that each actuator is physically limited, the optimal control increment step size is iteratively searched along the negative gradient direction, and the control vector sequence is updated. After obtaining the optimal control vector sequence that meets the convergence condition, only the control vector at the first sampling moment in the sequence is extracted and mapped to the nitrogen generator target frequency signal, the opening control command of each branch regulating valve and the pulse modulation timing signal of the electric ball valve at the inlet and outlet gas ends, and then sent to the execution layer as a cooperative control scheme.

[0037] In a preferred embodiment, the cooperative scheduling employs a variable frequency pulse width modulation strategy, including: The edge computing control terminal calculates the airflow permeation resonant frequency of the stack based on the stack's physical state matrix; The electric ball valve at the air inlet is controlled to open and close periodically at high speed using this resonant frequency. By generating micro-pressure waves, the boundary layer gas film between agricultural products is disrupted, forcing nitrogen to penetrate into the compact inner structure of the agricultural products and shortening the oxygen spatial gradient equilibrium time.

[0038] Specifically, for stacks where the core oxygen concentration decreases slowly, the electric ball valve at the air inlet is controlled to perform PWM opening and closing actions at a frequency of 0.1Hz~0.5Hz. The standing wave pressure effect generated in the pipeline by the high-frequency opening and closing of the valve is used to break the laminar boundary layer on the stack surface, forcing nitrogen to enter the deep gaps in a pulse form, so as to solve the problem of low gas replacement efficiency in high-density, high-compact agricultural product stacks.

[0039] In a preferred embodiment, the frequency conversion pulse width modulation strategy further includes: Based on the physical layout of the nitrogen filling pipeline network, calculate the fluid resistance imbalance caused by the difference in pipe length between the near-end and far-end stacked branches. Based on the branch impedance, the opening pulse width of the electric ball valve of the far stack is automatically increased, while the opening degree of the electric regulating valve of the main pipeline of the near stack is reduced, so as to achieve physical consistency of the charging pressure of all stacks in the warehouse.

[0040] As a preferred embodiment, the real-time safety correction of the control sequence also includes an anti-interference strategy: The system monitors the humidity at the nitrogen inlet and the humidity gradient inside the stack in real time. When the slope of the humidity drop in the stack caused by nitrogen replacement exceeds a preset threshold, the instantaneous flow rate of the pulse nitrogen charging is reduced.

[0041] In some embodiments, the method further includes an anomaly diagnosis step, comprising: When the valve signal feedback indicates that the valve is open and the nitrogen generator is outputting, but the displacement monitored by the ultrasonic sensor does not increase, it is determined that the branch pipeline is detached or the electric valve has a physical fault, triggering the corresponding branch's lockout protection and fault alarm.

[0042] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the intelligent nitrogen filling control method for agricultural product warehouses described in any of the above technical solutions.

[0043] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent nitrogen-filling control system for agricultural product warehouses, characterized in that, It includes an edge computing control terminal, which is connected to the perception layer, the execution layer and the cloud management platform; The perception layer is used to collect real-time environmental parameters of agricultural product stacking, tent deformation characteristics, and nitrogen generator operating conditions. It includes an environmental composite sensor deployed at each stack location, an ultrasonic ranging sensor deployed on the tent surface, and an edge acquisition module that monitors the operating status of the nitrogen generator unit. The execution layer is used to dynamically adjust the nitrogen production, the timing and flow of intake and exhaust, including electric regulating valves arranged in the main pipeline, bidirectional electric ball valves arranged at the intake and exhaust ends of each stack, and duplex nitrogen generators with variable frequency speed control function. The edge computing control terminal is used to calculate stacking characteristic parameters in real time based on the data from the perception layer, and generate a real-time nitrogen filling control strategy based on the stacking state prediction model, so as to realize the coordinated scheduling of nitrogen generator power and the opening degree of each stacking valve. The cloud management platform is used to acquire historical nitrogen filling operation data, train and update the stacking status prediction model, and send the updated model parameters to the edge computing control terminal.

2. The system according to claim 1, characterized in that, The perception layer and execution layer adopt a hybrid heterogeneous communication architecture, including: The wireless sensing subnet includes an environmental composite monitoring module and a stacking nitrogen filling status monitoring module that are wirelessly connected via a LoRa gateway, which is connected to the edge computing control terminal via a wired connection. The dual-mode execution subnet has two control modes for the electric regulating valves of the nitrogen filling pipeline, depending on their location: the electric valves of the main pipeline are connected to the electric valve control cabinets of each floor and are controlled by PLC in a wired manner; the electric valves at the stack end are connected to the wireless valve controller and are controlled by the LoRaWAN protocol, and the wireless valve controller has a logic protection circuit with an automatic lockout function when the network is disconnected. The industrial bus device network allows the PLC controller of the duplex nitrogen generator to communicate directly with the serial / network port of the edge computing gateway via Modbus or OPC UA protocol, enabling millisecond-level power command issuance and device status feedback.

3. A method for intelligent control of nitrogen filling in agricultural product warehouses, implemented using the system described in claim 1 or 2, characterized in that, include: The edge computing control terminal receives data from the perception layer, including the temperature and humidity of the stacking environment, oxygen concentration, and tent bulging distance, and constructs the current stacking physical state matrix. Based on the stacking status prediction model issued by the cloud management platform, the core oxygen concentration of the stack and the internal moisture migration rate of agricultural products are calculated according to the stacking physical state matrix. With the objective function of maintaining the oxygen concentration of the entire stack below the biological survival threshold and nitrogen production energy consumption, and combining the core oxygen concentration, the moisture migration rate of agricultural products and the working status characteristics of the nitrogen generator, a collaborative control sequence is generated that includes the target frequency of the nitrogen generator and the pulse duty cycle of each stack electric valve. At the same time, the control sequence is safely corrected in real time based on the tent deformation characteristics fed back by the ultrasonic ranging sensor. The execution layer adjusts the gas production of the duplex nitrogen generator and the opening degree and opening and closing sequence of the inlet and outlet valves of each stack according to the control sequence.

4. The method according to claim 3, characterized in that, The stacking physical characteristic prediction model includes: a porous medium diffusion sub-model, a thermodynamically coupled moisture migration sub-model, and a tent viscoelasticity sub-model. The viscoelastic mechanical sub-model of the tent is used to calculate the internal air pressure field formed by the tent tension feedback based on the measured tent bulging distance, and to update the equivalent porosity of the stack in real time according to the tent expansion volume. The thermodynamically coupled moisture migration sub-model is used to calculate the fluid mass transfer characteristics on the surface and between layers of agricultural products based on real-time temperature and humidity, output the internal moisture migration rate of the agricultural products, and dynamically compensate the diffusion coefficient in the porous medium diffusion sub-model based on the changes in gas composition caused by moisture migration. The porous media diffusion sub-model is used to use the pressure field provided by the tent viscoelasticity sub-model as the convection driving force, and to solve the unsteady convection-diffusion equation by combining the equivalent porosity and diffusion coefficient, to simulate the infiltration process of nitrogen into the deep layers of the stack, and to obtain the predicted value of the core oxygen concentration of the stack.

5. The method according to claim 3, characterized in that, The objective function is to maintain the oxygen concentration of the entire stack below the biological survival threshold and nitrogen production energy consumption. This is combined with core oxygen concentration, agricultural product moisture migration rate, and nitrogen generator operating status characteristics to generate a coordinated control sequence that includes the nitrogen generator target frequency and the pulse duty cycle of each stack's electric valves. A multi-objective cost function is constructed based on core oxygen concentration, agricultural product dry loss due to moisture migration, and nitrogen generator energy consumption. Using the tent displacement acceleration as a constraint, a predictive control algorithm is used to calculate the control vector sequence that minimizes the multi-objective cost function in each sampling period; Based on the control vector sequence, the ratio of nitrogen generator output power to the opening and closing time of the terminal valve is determined, and a coordinated control scheme is generated.

6. The method according to claim 5, characterized in that, The control vector sequence that minimizes the multi-objective cost function in each sampling period, using the tent displacement acceleration as a constraint and employing a predictive control algorithm, includes: Using the current nitrogen generator frequency, regulating valve opening, and ball valve pulse duty cycle as initial values, and combining them with historical control trajectories, an initial control sequence for a preset time period is generated. Calculate the gradients of the multi-objective cost function and the tent displacement acceleration constraint with respect to each control variable; At the current working point, the cost function is approximated as a quadratic function through Taylor expansion, and the acceleration constraint is linearized to the first order, thus transforming the multi-objective problem into multiple local quadratic programming sub-functions. Using the interior point method, under the constraints of ensuring that the tent displacement acceleration does not exceed the limit and that each actuator is physically limited, the optimal control increment step size is iteratively searched along the negative gradient direction to update the control vector sequence. After obtaining the optimal control vector sequence that meets the convergence condition, the control vector at the first sampling moment in the sequence is extracted and mapped to the nitrogen generator target frequency signal, the opening control command of each branch regulating valve, and the pulse modulation timing signal of the electric ball valve at the inlet and outlet gas ends. This signal is then sent to the execution layer as a collaborative control scheme.

7. The method according to claim 5, characterized in that, The collaborative control scheme employs a variable frequency pulse width modulation strategy, including: Based on the physical state matrix of the stack, the porosity characteristics of the stack are extracted, and the target frequency that causes the pressure wave inside the stack cavity to resonate is calculated. Based on the target frequency, and combined with the pulse duty cycle output by the model predictive control algorithm, a square wave signal sequence for controlling the opening and closing of the electric ball valve is generated. The execution layer drives the electric ball valve at the air inlet to perform high-frequency pulse action according to the square wave signal sequence, forming a fluctuating air pressure at the air inlet. The mechanical impact force of the pressure wave breaks the boundary layer gas film trapped between the gaps in agricultural products, forcing nitrogen to migrate to the compacted area in a convection diffusion mixing mode, thereby shortening the time for the spatial oxygen concentration to reach equilibrium.

8. The method according to claim 7, characterized in that, The frequency conversion pulse width modulation strategy also includes: Based on the physical layout of the nitrogen filling pipeline network, calculate the fluid resistance imbalance caused by the difference in pipe length between the near-end and far-end stacked branches. Based on the branch impedance, the opening pulse width of the electric ball valve of the far stack is automatically increased, while the opening degree of the electric regulating valve of the main pipeline of the near stack is reduced, so as to achieve physical consistency of the charging pressure of all stacks in the warehouse.

9. The method according to claim 3, characterized in that, The real-time security correction of the control sequence includes: The system monitors the humidity at the nitrogen inlet and the humidity gradient inside the stack in real time. When the slope of the humidity drop in the stack caused by nitrogen replacement exceeds a preset threshold, the instantaneous flow rate of the pulse nitrogen charging is reduced.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the intelligent nitrogen filling control method for agricultural product warehouses as described in any one of claims 3-9.

Images