Carbon dioxide modified atmosphere method and system for grain storage silos

Abstract

The present application mainly relates to the technical field of controlled atmosphere storage, in order to realize the integrated regulation and control of the pressure and carbon dioxide concentration in the warehouse during the process of carbon dioxide controlled atmosphere, improve the accuracy and stability of carbon dioxide controlled atmosphere, the present application provides a carbon dioxide controlled atmosphere method and system for grain storage warehouse, the core is: in the inflation stage, the fan speed is adjusted based on pressure feedback closed loop to maintain the set negative pressure and inhibit gas leakage; in the closed stage, the passive air supplement is limited to the negative pressure in the warehouse, and the structure safety is guaranteed; in the circulation stage, the carbon dioxide concentration of the warehouse top and the grain surface is taken as the main control variable, the model predictive control method based on attention mechanism time series prediction network is adopted, the candidate start-stop control sequence is constructed and the future concentration trajectory is predicted, the start-stop strategy of the circulation fan is optimized to maintain the concentration uniform and reduce the energy consumption, at the same time, the concentration inside the grain pile is taken as the safety constraint condition, the concentration is ensured to be not lower than the prevention threshold, finally, the inflation, closed and circulation stages are realized to be segmented and accurately regulated and controlled.

Description

Technical Field

[0001] This invention relates to the field of controlled atmosphere storage technology, and in particular to a carbon dioxide controlled atmosphere method and system for grain storage warehouses. Background Technology

[0002] Controlled atmosphere storage (CAPS) technology for grain has advantages such as good pest control effects, no chemical residues, and environmental friendliness, and has been widely used in the grain storage field. By filling the grain silo with a high concentration of carbon dioxide gas, a low-oxygen, high-carbon dioxide environment is created inside, thereby inhibiting or killing stored grain pests. This is one of the important technical approaches for green grain storage.

[0003] Shallow round silos and other grain storage silos have advantages such as large single-silo capacity and compact structure, but their large spatial dimensions and high vertical height still present many technical challenges when implementing controlled atmosphere storage (CAS) with carbon dioxide. CAS grain storage mainly includes three processes: the aeration stage, the sealing stage, and the circulation stage. In the aeration stage, traditional CAS methods typically involve directly injecting carbon dioxide into the silo, which can easily lead to gas leaks in gaps or weak points in the silo structure, wasting carbon dioxide and potentially affecting operational safety. In the sealing stage after aeration, due to the adsorption of carbon dioxide by the grain, the volume of carbon dioxide gas inside the silo decreases rapidly in a short time, easily leading to a significant increase in negative pressure. Without effective pressure control measures, this may adversely affect the structural safety of the silo. Furthermore, during CAS, because carbon dioxide is denser than air, it easily forms a vertical concentration gradient within the silo, with higher concentrations at the bottom and lower concentrations at the top, making it difficult to maintain the carbon dioxide concentration in some areas at the effective threshold required for pest control in the long term. In existing technologies, the timing of circulation is typically determined by manual experience, and fans are activated accordingly. However, concentration changes at different locations during circulation exhibit significant time lags and fluctuations. Improper control can lead to localized concentration drops below the threshold in a short period, affecting the control effect and increasing energy consumption. Therefore, there is an urgent need in this field for a shallow circular silo carbon dioxide controlled atmosphere control method and system that can integrate pressure and carbon dioxide concentration control during inflation, sealing, and circulation processes to improve the stability and automation level of the controlled atmosphere effect. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a carbon dioxide controlled atmosphere method and system for grain storage warehouses, with the aim of achieving integrated control of pressure and carbon dioxide concentration inside the warehouse during the controlled atmosphere process, thereby improving the accuracy and stability of the carbon dioxide controlled atmosphere effect.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0006] On one hand, the present invention provides a method for controlled atmosphere storage of carbon dioxide in grain storage warehouses, the method comprising:

[0007] S1: During the inflation phase, the chamber maintains a set negative pressure. When the carbon dioxide concentration at the top of the chamber reaches the set value, inflation ends and the chamber enters the sealing phase.

[0008] S2: During the closed-loop stage, air is supplied to the chamber, and once the negative pressure inside the chamber does not exceed the set threshold, the circulation stage begins.

[0009] S3: During the circulation phase, the carbon dioxide concentration values ​​at multiple locations within the chamber are detected. Using the carbon dioxide concentration at a set location as a safety constraint, the start-stop control sequence of the circulation fan within the chamber is predicted based on the established prediction model.

[0010] Furthermore, a multi-channel pump sampling method is used to detect carbon dioxide concentration, which specifically includes: a flushing stage, in which residual air in the sampling pipeline is flushed with a large flow rate; and a detection stage, in which carbon dioxide is sampled with a small flow rate. When the range of multiple consecutive sampling data is less than a set threshold, the average value of the multiple detections is taken as the carbon dioxide concentration detection result of the corresponding detection point.

[0011] Furthermore, S3 includes:

[0012] Construct multiple candidate sequences for starting and stopping circulating fans;

[0013] The historical data of carbon dioxide concentration in the warehouse, the historical status of the start and stop of the circulating fan, and the candidate sequence of the start and stop of the circulating fan are used as inputs to the prediction model to predict the trajectory of carbon dioxide concentration change in the warehouse corresponding to each candidate sequence within a set time domain in the future.

[0014] Based on the carbon dioxide concentration change trajectory, an optimization objective function is constructed, and each candidate sequence is optimized and solved. The candidate sequence with the smallest objective function value is selected as the optimal circulating fan start-up and shutdown sequence.

[0015] Furthermore, the prediction model is as follows:

[0016] ;

[0017] in, The predicted trajectory of carbon dioxide concentration change in the warehouse corresponding to each candidate sequence; Historical carbon dioxide concentration , This represents the carbon dioxide concentration at the top of the storage tank. The concentration of carbon dioxide at the grain surface; This represents the historical start-up and shutdown status of the circulating fan. Candidate sequence for starting and stopping the circulating fan. For the prediction model function; To predict the length of the time domain, For the current moment, The length of the historical time domain;

[0018] The prediction model function This includes modal prediction structures, modal matching structures, and modal comparison structures;

[0019] The modal prediction structure is used to encode and decode historical carbon dioxide concentrations and historical circulating fan start-stop states to obtain predicted carbon dioxide concentration values. The modal prediction structure is as follows:

[0020] Encoder: , ;in, For encoder output, Historical carbon dioxide concentration For the input weight matrix, For input bias, Code for the operating status of the circulating fan. For position encoding, For encoding functions, This indicates that the encoder output dimension is , The feature dimension output by the encoder;

[0021] Decoder: , ;in, For predicting carbon dioxide concentration, For regularization functions, For decoding function, For vector encoding functions, To output the weight matrix, For output bias, This indicates that the output dimension of the decoder is D is the feature dimension of the decoder output;

[0022] The modality comparison structure is used to calculate the ratio of similarity between positive samples to similarity between negative samples, thereby narrowing the feature distance between different modalities of the same sample. The modality comparison structure is as follows:

[0023] ;in To compare the losses, This represents the carbon dioxide concentration at the top of the storage tank. The concentration of carbon dioxide on the grain surface. The projected weight of the carbon dioxide concentration at the top of the warehouse is used. The projected weight of the carbon dioxide concentration on the grain surface is used. For positive sample pairs, For sample index, The number of samples;

[0024] The modality matching structure is used to determine whether two modality data originate from the same sample. The modality matching structure is as follows:

[0025] ; To match scores, To match classification weights.

[0026] Furthermore, the optimization objective function includes carbon dioxide concentration threshold constraints at different depths of the grain pile. and the energy consumption constraints of the circulating fan .

[0027] Furthermore, the optimization objective function is:

[0028] ;

[0029] ;

[0030] ;

[0031] in, To preset the carbon dioxide concentration threshold, This is the start / stop sequence of the circulating fan. This indicates the start / stop status of the circulating fan; 0 means the circulating fan is off, and 1 means the circulating fan is on. In the candidate start-stop control sequence The predicted concentration value at the i-th future time step obtained under the influence of the action. For the time change, and The weighting coefficients are set.

[0032] Furthermore, the method also includes: according to the set weights, in the early stage of the controlled atmosphere circulation phase, using the carbon dioxide concentration at the top of the silo as the main evaluation basis after the start-stop control of the circulating fans, maintaining the carbon dioxide concentration in the silo at no lower than a preset threshold; in the later stage of the controlled atmosphere circulation phase, using the carbon dioxide concentration at the grain surface as the main evaluation basis after the start-stop control of the circulating fans, maintaining the carbon dioxide concentration at the grain surface and below the grain surface at no lower than a preset threshold.

[0033] On the other hand, the present invention also provides a carbon dioxide controlled atmosphere system for grain storage silos, the system comprising:

[0034] An inflation module is used to fill the chamber with carbon dioxide;

[0035] The pressure control module includes a fan and a pressure sensor installed on the air extraction pipeline of the storage chamber. It is used to adjust the fan speed based on the pressure signal obtained by the pressure sensor during the inflation stage so as to maintain a set pressure state inside the grain storage chamber.

[0036] A closed pressure regulating module, including a one-way suction valve and a valve control component that works with it, is used to limit the negative pressure inside the grain storage silo during the closed-loop stage.

[0037] The circulation module includes a circulation pipe and a circulation fan installed on the circulation pipe, used to create gas circulation in the grain storage silo;

[0038] The concentration detection module includes multiple gas sampling channels and carbon dioxide concentration sensors located at different positions within the grain storage silo.

[0039] The control module is used to execute the carbon dioxide controlled atmosphere method for grain storage silos described in this invention.

[0040] Furthermore, the concentration detection module includes a solenoid valve group, a diaphragm pump, a second three-way solenoid valve, a first three-way solenoid valve, a pulse damper, and a carbon dioxide concentration sensor connected in sequence.

[0041] Furthermore, the system also includes a one-way exhaust valve installed on the top of the grain storage silo. The one-way exhaust valve is used to automatically open and release pressure when the pressure control module fails during the inflation stage and the pressure inside the silo exceeds a preset threshold.

[0042] The beneficial effects of this invention are:

[0043] 1) By constructing candidate sequences for the start and stop of circulating fans and predicting the future concentration change trajectory corresponding to each candidate sequence, the start and stop decision of circulating fans is dynamically updated, realizing a forward-looking evaluation of different start and stop strategies of circulating fans, thereby selecting the start and stop sequence of circulating fans that meets the concentration safety constraints and has the best energy consumption, avoiding the carbon dioxide concentration from falling below the prevention and control threshold, while reducing energy consumption and improving the intelligence level and robustness of atmospheric control.

[0044] 2) By using the carbon dioxide concentration at different depths inside the grain pile and the energy consumption of the circulating fan as safety constraints for the circulation process, rather than predictive inputs, we can avoid misjudgments in control due to inconsistent concentration trends in different areas during the circulation process. This fully utilizes the physical properties of carbon dioxide, such as its high density and easy settling, to improve the stability of the controlled atmosphere effect.

[0045] 3) By introducing a closed-loop control method based on pressure feedback during the inflation stage, the chamber is kept under a set negative pressure, which effectively suppresses carbon dioxide leakage during the inflation process and reduces gas consumption.

[0046] 4) By setting up a one-way suction valve in the sealed stage and cooperating with valve control, the negative pressure inside the warehouse can be actively limited, avoiding the problem of excessive negative pressure inside the warehouse caused by the grain adsorbing carbon dioxide, and improving operational safety.

[0047] 5) By detecting carbon dioxide concentrations at the top of the silo and the grain surface, combined with pump sampling and stability criteria, the reliability and representativeness of the concentration detection results can be improved. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of a carbon dioxide controlled atmosphere system for grain storage silos according to the present invention.

[0049] Figure 2 This is a schematic diagram of the concentration detection module and the gas sampling pipeline;

[0050] In the diagram: 1-First electric valve; 2-Second electric valve; 3-Third electric valve; 4-One-way suction valve; 5-Centrifugal fan; 6-Fourth electric valve; 7-Fifth electric valve; 8-Circulating fan; 9-One-way exhaust valve; 10-Concentration detection point on the top of the silo; 11-Concentration detection point on the grain surface; 12-Concentration detection point inside the grain pile; 13-Pressure sensor; 14-Pulse damper; 15-Carbon dioxide concentration sensor; 16-First three-way solenoid valve; 17-Second three-way solenoid valve; 18-Diaphragm pump; 19-Solenoid valve assembly. Detailed Implementation

[0051] In existing technologies, the carbon dioxide controlled atmosphere process in grain storage warehouses suffers from problems such as severe gas leakage during the filling stage, difficulty in controlling negative pressure inside the warehouse during the sealing stage, and reliance on manual experience for circulation control, which can easily lead to local concentration fluctuations.

[0052] This invention provides a carbon dioxide controlled atmosphere method and system for grain storage silos. The core of this system lies in achieving integrated pressure and concentration control throughout the entire process of aeration, sealing, and circulation through phased collaborative control and model prediction optimization. Specifically, during the aeration phase, the fan speed is adjusted based on pressure feedback closed-loop control to maintain a set negative pressure and suppress gas leakage. During the sealing phase, passive gas replenishment via a one-way suction valve limits the negative pressure inside the silo, ensuring structural safety. During the circulation phase, the carbon dioxide concentration at the silo top and grain surface is used as the main control variable. A model prediction control method based on an attention mechanism time-series prediction network is employed. By constructing candidate start-stop control sequences and predicting future concentration trajectories, the start-stop strategy of the circulation fan is optimized to maintain uniform concentration and reduce energy consumption. Simultaneously, the concentration inside the grain pile is used as a safety constraint to ensure that the concentration does not fall below the prevention threshold. Ultimately, this achieves segmented and precise pressure regulation during the aeration, sealing, and circulation phases, significantly improving the stability, safety, and automation level of controlled atmosphere grain storage.

[0053] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following embodiments.

[0054] like Figure 1 and Figure 2 As shown, the carbon dioxide controlled atmosphere system for grain storage silos of the present invention includes an inflation module, a circulation pipeline, a pressure regulating device, a sealed pressure regulating module, a concentration detection device, and a control module.

[0055] The inflation module includes a carbon dioxide source and an inflation pipeline connected to the storage chamber for introducing carbon dioxide into the chamber. The pressure control module includes a fan and a pressure sensor mounted on the exhaust pipeline, used to adjust the fan speed based on the pressure signal obtained from the pressure sensor during the inflation phase, maintaining a set pressure within the grain storage chamber. The airtight pressure regulating module includes a one-way suction valve and a corresponding valve control assembly, used to limit negative pressure within the grain storage chamber during the airtight phase. The circulation module includes a circulation pipe and a circulation fan mounted on it, used to create gas circulation within the grain storage chamber. The concentration detection module includes multiple gas sampling channels located at different heights within the grain storage chamber and a carbon dioxide concentration sensor. The control module executes the carbon dioxide controlled atmosphere method for grain storage chambers described in this invention.

[0056] Specifically, a first electric valve 1 is installed on the inflation pipeline, and a second electric valve 2 and a fifth electric valve 7 are installed on the circulation pipeline. The sealed pressure regulating module includes a second electric valve 2, a third electric valve 3, and a one-way suction valve 4. A centrifugal fan 5 is installed on the exhaust pipeline at the top of the silo. A pressure sensor 13 is connected to the internal space of the silo via a PU tube to collect the internal pressure signal. A one-way exhaust valve 9 is also installed at the top of the silo to automatically release pressure when abnormal positive pressure occurs inside the silo. The one-way exhaust valve 9 adopts a spring preload structure, and its opening pressure is a preset safety threshold.

[0057] Multiple carbon dioxide concentration detection points are set up inside the warehouse, including at least a concentration detection point 10 on the top of the warehouse, a concentration detection point 11 on the grain surface, and concentration detection points 12 at different depths inside the grain pile. Each detection point is connected to a solenoid valve assembly 19 via a PU gas sampling tube. The solenoid valve assembly 19 is sequentially connected to a diaphragm pump 18, a second three-way solenoid valve 17, a first three-way solenoid valve 16, a pulse damper 14, and a carbon dioxide concentration sensor 15, forming a multi-channel pump-suction sampling structure.

[0058] Based on the multi-channel pump sampling structure, this embodiment uses any one channel of the solenoid valve group 19 as an example to explain the carbon dioxide concentration detection method; the detection process for the other channels is the same.

[0059] First, the flushing stage begins. The control module activates the corresponding solenoid valve in the solenoid valve assembly 19, and the diaphragm pump 18 operates at the first speed to flush the residual air in the sampling pipeline at a high flow rate. During this process, the gas sequentially passes through the solenoid valve assembly 19, the diaphragm pump 18, the second three-way solenoid valve 17, and the first three-way solenoid valve 16, finally returning to the chamber from port A of the first three-way solenoid valve 16. Next, the detection stage begins. The control module energizes the first three-way solenoid valve 16 and the second three-way solenoid valve 17 to switch the gas path. The diaphragm pump 18 operates at the second speed to perform low-flow detection on the sampled gas. During this process, the gas sequentially passes through the solenoid valve assembly 19, the diaphragm pump 18, the second three-way solenoid valve 17, the pulse damper 14, the carbon dioxide concentration sensor 15, and the first three-way solenoid valve 16, finally returning to the chamber from port A of the first three-way solenoid valve 16.

[0060] During the detection phase, the carbon dioxide concentration sensor 15 collects concentration data at a sampling frequency of once per second. When the range of six consecutive sampling data is less than a set threshold, the average value is taken as the concentration detection result of the corresponding detection point.

[0061] After the test is completed, the solenoid valve group 19, diaphragm pump 18, first three-way solenoid valve 16 and second three-way solenoid valve 17 are reset, and the control module enters the sampling cycle of the next test channel.

[0062] Preferably, the control module performs time synchronization processing on the concentration data from different detection points to form a unified time-series concentration data sequence for subsequent circulation control decisions.

[0063] The following explanation, using a carbon dioxide controlled atmosphere system for grain storage silos as described in this invention, illustrates the carbon dioxide controlled atmosphere method for grain storage silos, specifically including the following processes:

[0064] S1: During the inflation stage, the gas input or output is adjusted according to the gas pressure inside the grain storage silo to maintain the pressure inside the grain storage silo. When the carbon dioxide concentration at the top of the grain storage silo reaches the set value, the inflation ends and the silo enters the sealing stage.

[0065] At the start of inflation, the control module opens the first electric valve 1 and the fourth electric valve 6, and the centrifugal fan 5 starts. Based on the pressure signal collected by the pressure sensor 13, the control module uses a closed-loop control method to adjust the speed of the centrifugal fan 5 in real time, maintaining the pressure inside the chamber within the preset target negative pressure range, thereby suppressing carbon dioxide leakage during inflation. If, during inflation, the pressure inside the chamber exceeds the preset positive pressure threshold due to a malfunction in the pressure control module or other reasons, the one-way exhaust valve 9 automatically opens against the spring preload, venting the gas inside the chamber and passively releasing the positive pressure. Simultaneously, the concentration detection device periodically acquires carbon dioxide concentration data at each detection point.

[0066] Preferably, when the carbon dioxide concentration at the top concentration detection point 10 reaches a preset threshold, the control module determines that the inflation stage has ended and enters the sealing stage.

[0067] S2: In the sealed stage, by replenishing air into the grain storage silo, the negative pressure inside the silo is adjusted to not exceed the set threshold, and the circulation stage begins.

[0068] When the carbon dioxide concentration at the top concentration detection point 10 reaches the preset threshold, aeration stops, the control module closes the first electric valve 1 and the fourth electric valve 6, and opens the second electric valve 2 and the third electric valve 3, putting the system into a sealed phase. Because grain adsorbs carbon dioxide, the negative pressure inside the silo gradually increases. When the negative pressure exceeds the working threshold of the one-way suction valve 4, the one-way suction valve 4 automatically opens, allowing outside air to enter the silo through the one-way suction valve 4, the third electric valve 3, and the second electric valve 2, thus limiting further increases in the absolute value of the negative pressure inside the silo. When the pressure sensor 13 detects that the negative pressure inside the silo is lower than the working threshold of the one-way suction valve 4, the control module closes the third electric valve 3 and opens the fifth electric valve 7, ensuring the circulation pipe is unobstructed and providing conditions for subsequent circulation control.

[0069] S3: During the circulation phase, based on the carbon dioxide concentration values ​​at multiple locations within the grain storage silo, and using the carbon dioxide concentration at a set location as a safety constraint, the start-stop control sequence of the circulation fan within the grain storage silo is obtained based on model prediction.

[0070] Before entering the circulation control phase, time-series data of carbon dioxide concentration in the silo for a preset time period are first collected, and historical state information required for model predictive control is constructed based on the time-series data. In this phase, a data-driven model predictive control method is used to control the start and stop of the circulation fan 8 to achieve stable maintenance of carbon dioxide concentration and energy consumption optimization in the silo. The control module operates in discrete time mode, making start and stop decisions for the circulation fan 8 according to a preset control cycle. In each control cycle, the control module first acquires carbon dioxide concentration data at the top of the silo and the grain surface, and constructs the current control state by combining historical concentration data and historical circulation fan operating states. The control module constructs multiple candidate sequences for the start and stop of the circulation fan based on the current control state. Each candidate start and stop control sequence includes a combination of the start and stop states of the circulation fan in several future time steps, where each time step corresponds to a circulation fan start and stop control command, multiple time step control commands constitute a control sequence, and multiple different control sequences constitute multiple candidate start and stop control sequences.

[0071] The control module inputs the operating status sequence corresponding to the candidate start-stop control sequence, along with the current and historical concentration data and the historical wind turbine operating status sequence, into the concentration prediction model. Based on the multi-location concentration sequence and operating status sequence within the historical time window, it performs multi-step prediction of the carbon dioxide concentration in the future prediction time domain, thereby obtaining the future concentration change trajectory corresponding to each candidate start-stop control sequence.

[0072] The prediction model is preferably implemented using a time-series prediction network based on an attention mechanism.

[0073] Let the current time be The historical time domain length is T, the prediction time domain length is N, and the multi-location concentration vector is... ,in, This represents the carbon dioxide concentration at the top of the storage tank. This represents the carbon dioxide concentration at the grain surface location.

[0074] The prediction model follows the following mapping relationship:

[0075] ;

[0076] in, This shows the predicted carbon dioxide concentration change trajectory for each candidate sequence. This is a sequence of historical start-up and shutdown states for the circulating fan. Candidate circulating fan start / stop sequence. This is the prediction model function.

[0077] It includes three model structures: modal prediction structure, modal matching structure, and modal comparison structure.

[0078] The modality prediction structure introduces a position vector at the input to represent the time step sequence information. The modality prediction structure includes the encoding and decoding process of the input modality, as shown in the following equation.

[0079] Encoder: , ;

[0080] in, For encoder output, Historical carbon dioxide concentration For the input weight matrix, For input bias, Code for the operating status of the circulating fan. For position encoding, For encoding functions, This indicates that the encoder output dimension is , The feature dimension output by the encoder;

[0081] Decoder: , ;

[0082] in, For predicting carbon dioxide concentration, For regularization functions, For decoding function, For vector encoding functions, To output the weight matrix, For output bias, This indicates that the output dimension of the decoder is D is the feature dimension of the decoder output.

[0083] The modality comparison structure is used to narrow the feature distance between positive samples and widen the feature distance between negative samples through contrastive learning, so as to enhance the feature consistency between different modalities, as shown in the following formula.

[0084] ;

[0085] in To contrast the loss, it is used to interpret the ratio between the similarity between positive samples and the similarity between the remaining negative samples. This represents the carbon dioxide concentration at the top of the storage tank. The concentration of carbon dioxide on the grain surface. The projected weight of the carbon dioxide concentration at the top of the warehouse is used. The projected weight of the carbon dioxide concentration on the grain surface is used. For positive sample pairs, For sample pair indexing, This represents the number of samples.

[0086] The modality matching structure is used to determine whether different modal input data come from the same time sample, as shown in the following formula.

[0087] ;

[0088] in, The matching score is used to characterize whether different modalities belong to the same sample. To match classification weights.

[0089] The prediction model, by jointly modeling the correlation between concentrations at different locations and the influence of operating status on the gas diffusion process, is used to characterize the dynamic coupling relationship between the operating status of the circulating fan and the changes in carbon dioxide concentration in the chamber.

[0090] The control module constructs an optimization objective function based on the future concentration prediction trajectory corresponding to each candidate start-stop control sequence. The optimization objective function includes at least a concentration threshold constraint term and a circulating fan energy consumption constraint term.

[0091] In the implementation method, assuming the prediction time domain length is N, the candidate start-stop control sequence is represented as follows: ,in This indicates the start / stop status of the circulating fan; 0 indicates the fan is off, and 1 indicates the fan is on.

[0092] To suppress frequent start-stop cycles of the circulating fan and reduce computational complexity, the control module uses an enumeration method to generate candidate start-stop control sequences, and limits each candidate sequence to at most one start-stop state transition within the prediction time domain, satisfying the following constraints: .

[0093] For any candidate sequence Based on the concentration prediction model, the concentration trajectory within the future prediction time domain is obtained, that is, the concentration sequence over N time steps within the prediction time domain. .

[0094] The control module constructs and optimizes the objective function: ; Carbon dioxide concentration threshold constraint, To constrain the energy consumption of the circulating fan, and The weighting coefficients are set.

[0095] The concentration threshold constraint term is processed by a stepwise accumulation method on the concentration sequence, that is, the threshold default amount at each time step in the prediction time domain is calculated and accumulated to obtain the concentration threshold constraint term: , of which Threshold concentration, In the candidate start-stop control sequence The predicted concentration value at the i-th future time step obtained under the influence of the energy consumption constraint is: ,in It is the time-varying quantity, that is, a single control time step.

[0096] During execution, the control module executes only the current time step control command in the optimal start-stop control sequence in each control cycle, and reconstructs the candidate start-stop control sequence based on the latest collected concentration data in the next control cycle, predicts the future concentration change trajectory, and updates the optimal control sequence, thereby realizing the rolling optimization of the start-stop control of the circulating fan.

[0097] To adapt to the prevention and control objectives at different stages within the controlled atmosphere cycle, the control module adopts a phased control strategy: Based on experience-based weighting, in the early stage of the circulation phase (within one week from the start of the closed phase), the predicted carbon dioxide concentration sequence at the top of the silo is used as the main evaluation basis for the predictive model to predict controlled atmosphere control. The model predicts and controls the concentration at the top of the silo to maintain it above the preset threshold. In the later stage of the circulation phase (from one week after the start of the closed phase to the end of the controlled atmosphere), the predicted carbon dioxide concentration sequence at the grain surface is used as the main evaluation basis for the predictive model to predict controlled atmosphere control. The model predicts and controls the concentration at the grain surface and below to maintain it above the preset threshold, thereby reducing the energy consumption of the circulation fan while meeting the prevention and control requirements.

[0098] Furthermore, the carbon dioxide concentration at different depths within the grain pile is not used as an input variable for model predictive control, but rather as a safety constraint for the circulation process. During the operation of the circulation fan, when the carbon dioxide concentration at any detection point inside the grain pile drops to a preset threshold, the control module immediately stops the circulation fan. After circulation stops, the control module maintains the silo in a static, sealed state, allowing the carbon dioxide gas to settle naturally under gravity. When the concentration inside the grain pile rises back above the threshold, the control module re-enables the circulation fan's start / stop control.

Claims

1. A carbon dioxide modified atmosphere method for a grain storage bin, characterized by, The method includes: S1: During the inflation phase, the chamber maintains a set negative pressure. When the carbon dioxide concentration at the top of the chamber reaches the set value, inflation ends and the chamber enters the sealing phase. S2: During the closed-loop stage, air is supplied to the chamber, and once the negative pressure inside the chamber does not exceed the set threshold, the circulation stage begins. S3: During the circulation phase, the carbon dioxide concentration at multiple locations within the chamber is detected. Using the carbon dioxide concentration at a set location as a safety constraint, the start-stop control sequence of the circulation fans within the chamber is predicted based on the established prediction model; specifically including: Construct multiple candidate sequences for starting and stopping circulating fans; The historical data of carbon dioxide concentration in the warehouse, the historical status of the start and stop of the circulating fan, and the candidate sequence of the start and stop of the circulating fan are used as inputs to the prediction model to predict the trajectory of carbon dioxide concentration change in the warehouse corresponding to each candidate sequence within a set time domain in the future. Based on the carbon dioxide concentration change trajectory, an optimization objective function is constructed, and each candidate sequence is optimized and solved. The candidate sequence with the smallest objective function value is selected as the optimal circulating fan start-stop control sequence. The prediction model is as follows: ； wherein, is a predicted carbon dioxide concentration change trajectory in the silo corresponding to each candidate sequence; is a historical carbon dioxide concentration, , is a carbon dioxide concentration at the silo top, is a carbon dioxide concentration at the grain surface position; is a historical state of the circulation fan start-stop, is a circulation fan start-stop candidate sequence, is a prediction model function; is a prediction time domain length, is a current time, is a historical time domain length; The prediction model function comprising a modal prediction structure, a modal matching structure, and a modal contrast structure; The modal prediction structure is used to encode and decode historical carbon dioxide concentrations and historical circulating fan start-stop states to obtain predicted carbon dioxide concentration values. The modal prediction structure is as follows: Encoder: , ; wherein, is the encoder output, is the historical carbon dioxide concentration, is the input weight matrix, is the input bias, is the loop flow fan operating state encoding, is the location encoding, is the encoding function, denotes the encoder output dimension , is the feature dimension of the encoder output; Decoder: , ; wherein, a carbon dioxide predicted concentration, is a regularization function, is a decoding function, is a vector encoding function, is an output weight matrix, is an output bias, denotes the output dimension of the decoder is D is the feature dimension of the decoder output. The modality comparison structure is used to calculate the ratio of similarity between positive samples to similarity between negative samples, thereby narrowing the feature distance between different modalities of the same sample. The modality comparison structure is as follows: ； wherein, is a contrast loss, is a silo top carbon dioxide concentration, is a grain surface carbon dioxide concentration, is a silo top carbon dioxide concentration projection weight, is a grain surface carbon dioxide concentration projection weight, is a positive sample pair, is a sample index, is a sample number; The modality matching structure is used to determine whether two modality data originate from the same sample. The modality matching structure is as follows: ； in, To match scores, To match classification weights.

2. The method for controlled atmosphere storage of grain using carbon dioxide according to claim 1, characterized in that, A multi-channel pump sampling method is used to detect carbon dioxide concentration. Specifically, it includes: a flushing stage in which residual air in the sampling pipeline is flushed with a large flow rate; and a detection stage in which carbon dioxide is sampled with a small flow rate. When the range of multiple consecutive sampling data is less than a set threshold, the average value of the multiple detections is taken as the carbon dioxide concentration detection result of the corresponding detection point.

3. The method for controlled atmosphere storage of grain using carbon dioxide according to claim 1, characterized in that, The optimization objective function includes carbon dioxide concentration threshold constraints at different depths of the grain pile. and the energy consumption constraints of the circulating fan .

4. A method for controlling carbon dioxide atmosphere in a grain storage silo according to claim 3, characterized in that, The optimization objective function is: ； ； ； in, The preset carbon dioxide concentration threshold, This is the start / stop sequence of the circulating fan. This indicates the start / stop status of the circulating fan; 0 means the circulating fan is off, and 1 means the circulating fan is on. In the candidate start-stop control sequence The predicted concentration value at the i-th future time step obtained under the influence of the action. For the time change, and The weighting coefficients are set.

5. A method for controlling carbon dioxide atmosphere in a grain storage silo according to claim 1, characterized in that, The method further includes: according to the set weights, in the early stage of the controlled atmosphere circulation phase, using the carbon dioxide concentration at the top of the silo as the main evaluation basis after the start-stop control of the circulating fan, maintaining the carbon dioxide concentration in the silo at no lower than a preset threshold; in the later stage of the controlled atmosphere circulation phase, using the carbon dioxide concentration at the grain surface as the main evaluation basis after the start-stop control of the circulating fan, maintaining the carbon dioxide concentration at the grain surface and below the grain surface at no lower than a preset threshold.

6. A carbon dioxide controlled atmosphere system for grain storage silos, characterized in that, The system includes: An inflation module is used to fill the chamber with carbon dioxide; The pressure control module includes a fan and a pressure sensor installed on the air extraction pipeline of the storage chamber. It is used to adjust the fan speed based on the pressure signal obtained by the pressure sensor during the inflation stage so as to maintain a set pressure state inside the grain storage chamber. A closed pressure regulating module, including a one-way suction valve and a valve control component that works with it, is used to limit the negative pressure inside the grain storage silo during the closed-loop stage. The circulation module includes a circulation pipe and a circulation fan installed on the circulation pipe, used to create gas circulation in the grain storage silo; The concentration detection module includes multiple gas sampling channels and carbon dioxide concentration sensors located at different positions within the grain storage silo. A control module for performing the carbon dioxide controlled atmosphere method for a grain storage silo as described in any one of claims 1 to 5.

7. A carbon dioxide controlled atmosphere system for grain storage silos according to claim 6, characterized in that, The concentration detection module includes a solenoid valve group, a diaphragm pump, a second three-way solenoid valve, a first three-way solenoid valve, a pulse damper, and a carbon dioxide concentration sensor connected in sequence.

8. A carbon dioxide controlled atmosphere system for grain storage silos according to claim 7, characterized in that, The system also includes a one-way exhaust valve installed on the top of the grain storage silo. The one-way exhaust valve is used to automatically open and release pressure when the pressure control module fails during the inflation stage and the pressure inside the silo exceeds a preset threshold.

Images