Intelligent regional refrigeration adjusting system and method based on multi-mode perception

Abstract

The invention discloses an intelligent regional refrigeration adjusting system and method based on multi-mode sensing, relates to the technical field of refrigerated cabinets, and aims to solve the problem that a traditional refrigeration device only conducts passive adjustment through pressure and temperature and cannot sense material decay in the early stage. The system comprises a pressure / temperature acquisition module, a state sensing module (a color recognition unit and a smell recognition unit), a grading decision module and a cold supply adjusting module, and material states are divided into L0-L3 four levels; according to the method, multi-modal data are synchronously obtained through a depolarization method, color / smell change indexes are obtained after processing, state grades are judged, customized cold supply instructions (such as L1 cooling and wind speed increasing and L3 limit refrigeration and isolation) and early warning instructions are generated in combination with pressure and temperature, and the instructions are independent in different areas.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of refrigerated cabinets, and in particular to an intelligent sub-region refrigeration adjustment system and method based on multi-modal perception. BACKGROUND

[0003] Freezing preservation is one of the most widely used preservation methods with the longest application history. Traditional refrigeration devices usually operate based on a fixed set temperature and cannot make dynamic adjustments according to the actual state of the stored goods, resulting in high energy consumption and unsatisfactory preservation effect. Therefore, in the prior art (such as patent CN107314596A), the weight (pressure value) and the environmental temperature value of the food in the target region are obtained by pressure sensors and temperature sensors respectively, and the refrigeration temperature and air speed of the region are dynamically adjusted based on the preset pressure and temperature threshold combination. This method achieves energy saving and on-demand refrigeration to some extent, and has made significant progress compared with traditional cold storage.

[0004] However, the present inventors have found that the above prior art still has limitations. The spoilage process of food is a complex biochemical change process, and relying on only two parameters of weight (indirectly reflecting the load) and environmental temperature for judgment is relatively extensive. Weight and temperature cannot directly and early reflect the freshness state of the food itself, for example:

[0005] Meat may begin to discolor and spoil due to microbial contamination before reaching the weight or temperature threshold; the decay of fruits and vegetables often starts from a local part and is accompanied by the production of specific volatile gases (such as ethylene and hydrogen sulfide), but the overall weight and the average temperature in the cabin may not change significantly.

[0006] In other words, the existing control mechanism is a "passive response" mode, which only starts to adjust when the food load or the environmental temperature changes, and cannot "actively intervene" in the quality deterioration of the food. This results in the device being unable to respond in time in the early stage of food spoilage, which may miss the best preservation control opportunity and still has the risk of food spoilage and potential economic loss.

[0007] Therefore, there is an urgent need in the art for a new refrigeration device that can more directly perceive the freshness change of the food itself and actively adjust based on the same, in order to further reduce waste, ensure food safety and improve the level of intelligence. SUMMARY

[0008] An intelligent sub-region refrigeration adjustment system based on multi-modal perception, comprising,

[0009] a pressure acquisition module for acquiring a pressure value of a target region;

[0010] a temperature acquisition module configured to acquire a temperature value of the target area;

[0011] a cooling adjustment module configured to adjust a cooling parameter of the target area according to the pressure value of the target area, the temperature value of the target area, and a preset threshold value;

[0012] a state sensing module configured to acquire color change information and odor change information of the material in the target area;

[0013] a hierarchical decision module connected to the state sensing module and configured to calculate a state level of the material according to the color change information and the odor change information, the state level being used to represent a degree of spoilage change of the material;

[0014] The cooling adjustment module is also connected to the hierarchical decision module and configured to adjust the cooling parameter of the target area and / or start an early warning mechanism according to the state level.

[0015] The state level includes a normal period (LO), a slight change period (L1), an obvious change period (L2), and a severe change period (L3) from light to heavy;

[0016] The cooling adjustment module is specifically configured to execute a customized adjustment strategy corresponding to the state level, and when the state level is LO, the cooling parameter is adjusted according to the pressure value and the temperature value of the target area;

[0017] When the state level is L1, the cooling temperature setting value is reduced and the cooling air speed is increased on the basis of the adjustment according to the pressure value and the temperature value;

[0018] When the state level is L2, the cooling temperature setting value is controlled to drop to a first preset low temperature value, and the cooling air speed is controlled to continuously run at the highest gear;

[0019] When the state level is L3, the cooling temperature setting value is controlled to drop to a minimum value allowed by the equipment or a second preset low temperature value, and the cooling air speed is maintained at the highest gear, and meanwhile, an isolation mechanism is started.

[0020] The state sensing module includes,

[0021] a color recognition unit configured to acquire an image of the material in the target area through an image sensor, and extract color features of the image to generate the color change information;

[0022] an odor recognition unit configured to acquire component data of a gas in the target area through a gas sensor array, and analyze the component data through a pattern recognition algorithm to generate the odor change information.

[0023] The target area comprises a plurality of independent sub-areas, each of which is independently configured with the pressure acquisition module, the temperature acquisition module, the state sensing module, the hierarchical decision module and the cold supply adjustment module, so that independent sensing and adjustment of sub-areas are realized.

[0024] Preferably, the application further comprises an intelligent sub-area refrigeration adjustment method based on multi-modal sensing, which comprises the following steps,

[0025] acquiring the pressure value, the temperature value, the color change information and the odor change information of the target area;

[0026] calculating the state level of the material based on the color change information and the odor change information;

[0027] generating corresponding cold supply parameter adjustment instructions and early warning instructions based on the state level, the pressure value and the temperature value;

[0028] S1, acquiring a plurality of preset sensor data, transmitting and recording the data into the system;

[0029] S2, based on the data acquired in step S1, processing and analyzing the data in the system,

[0030] S3, generating final cold supply parameter adjustment instructions and early warning instructions according to the Q and T values acquired in S1 and the state level L determined in S203,

[0031] S4, sending the final cold supply temperature instructions and air speed instructions generated in S302 to the refrigeration unit and the fan frequency converter of the target area, executing temperature and air speed adjustment, continuously collecting and monitoring data, forming a closed loop control, until the target area is emptied or the system is shut down.

[0032] The beneficial effects of the application are as follows:

[0033] Beneficial effects

[0034] 1. Early active intervention, significantly reducing spoilage loss

[0035] Through color + odor dual-mode sensing, it can be detected and trigger intensive refrigeration in the early stage of spoilage (L1 stage), compared with the traditional T / P threshold mode, the average can delay the spoilage start time of fruits and vegetables / meat by 18-25h; in 30 days of continuous test, the damage rate is reduced from 8.9% to 2.1%, which is equivalent to a reduction of about 75% in economic loss.

[0036] 2. Sub-area independent regulation, energy saving

[0037] Each sub-area independently calculates the refrigeration capacity according to the "real-time load Q + real-time state L", avoiding overcooling of the whole warehouse; test 4x2m 3Sub-regions run simultaneously, saving energy by 12-15%.

[0038] 3. Four-stage progressive adjustment method, considering preservation and quality

[0039] LO normal preservation, L1 micro 1℃ drop + wind speed increase, L2 forced deep cooling, L3 extreme deep cooling + isolation, which can not only prolong the shelf life, but also avoid cross contamination in L3 stage.

[0040] 4. Threshold self-adaptation, easy to transplant

[0041] Color threshold C1-C3, odor threshold O1-O3 can be updated by OTA, and different batches and different categories of materials only need to be calibrated once to be online, with low maintenance cost.

[0042] 5. Closed-loop control and human-computer cooperation

[0043] MCU completes millisecond-level closed loop; L2 / L3 automatically alarms and pushes to mobile phone / background, realizing "unattended" or "few people on duty" cold chain scene, and labor inspection time is reduced by 60%.

[0044] 6. Strong system versatility

[0045] Hardware only adds low-cost CMOS and MOS gas sensor array, which can be quickly modified in traditional cold storage, convenience store refrigerator, cold chain transport compartment, and is compatible with existing refrigeration systems. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 It is an intelligent sub-regional refrigeration adjustment method based on multi-modal perception. DETAILED DESCRIPTION

[0047] Embodiment 1

[0048] The intelligent sub-regional refrigeration adjustment system based on multi-modal perception comprises,

[0049] The pressure acquisition module is used for acquiring the pressure value of the target region.

[0050] The temperature acquisition module is used for acquiring the temperature value of the target region.

[0051] The cooling adjustment module is used for adjusting the cooling parameter of the target region according to the pressure value of the target region, the temperature value of the target region and the preset threshold value.

[0052] Adjust the cooling parameter of the target region.

[0053] The state perception module is configured to acquire color change information and smell change information of the materials in the target area; and the hierarchical decision module is connected with the state perception module and configured to calculate a state grade of the materials according to the color change information and the smell change information, the state grade being used to represent a degree of spoilage change of the materials.

[0054] The cooling supply adjustment module is further connected with the hierarchical decision module and configured to adjust a cooling supply parameter of the target area and / or start a warning mechanism according to the state grade.

[0055] By adopting the decision framework of "environmental physical parameters (pressure, temperature) + material spoilage state (color, smell)", a breakthrough is realized in the limitation of traditional refrigeration systems which only rely on temperature and pressure as single environmental parameters for adjustment. The pressure parameter can reflect the material stacking density and heat transfer load, the temperature parameter represents the current refrigeration environment benchmark, the color and smell parameters directly capture the early spoilage signals of the materials, and the four parameters together provide a comprehensive basis for "environmental adaptation + spoilage warning" for cooling supply adjustment, effectively avoiding the problems of "fresh materials being over-cooled and energy being wasted" or "insufficient refrigeration accelerating spoilage in the early stage of spoilage" caused by only looking at environmental parameters, and ensuring that the cooling strategy not only fits the environmental load but also accurately matches the freshness demand of the materials, thereby improving the refrigeration efficiency and the preservation period of the materials.

[0056] The state grade includes a normal period LO, a slight change period L1, an obvious change period L2 and a serious change period L3 from light to heavy;

[0057] The cooling supply adjustment module is specifically configured to execute a customized adjustment strategy corresponding to the state grade, when the state grade is LO, the cooling supply parameter is adjusted according to the pressure value and the temperature value of the target area;

[0058] When the state grade is L1, the cooling temperature set value is reduced and the cooling air speed is increased on the basis of the adjustment according to the pressure value and the temperature value;

[0059] When the state grade is L2, the cooling temperature set value is controlled to drop to a first preset low temperature value and the cooling air speed is controlled to continuously run at the highest gear;

[0060] When the state grade is L3, the cooling temperature set value is controlled to drop to the lowest value allowed by the equipment or a second preset low temperature value, and the cooling air speed is maintained at the highest gear, and a isolation mechanism is started.

[0061] The state perception module includes,

[0062] The color recognition unit is configured to acquire an image of the materials in the target area through an image sensor, and extract color features of the image to generate the color change information;

[0063] An odor recognition unit is configured to collect component data of gas in the target area by a gas sensor array, and analyze the component data by a pattern recognition algorithm to generate the odor change information.

[0064] By adopting the four-grade gradientized material state level division (LO-L3), the precise quantification of the material from "completely fresh" to "severe corruption" is realized. Compared with the traditional "normal / abnormal" two-grade determination, the division can capture the progressive characteristics of corruption: LO corresponds to fresh material, no additional intervention is needed; L1 corresponds to slight color change or trace of corruption gas, slight refrigeration is needed; L2 corresponds to obvious color change and increased corruption gas, heavy refrigeration is needed; L3 corresponds to severe deterioration, extreme refrigeration + isolation is needed. This fine classification provides a non-ambiguous and implementable determination standard for the subsequent refrigeration adjustment and early warning mechanism, avoiding the problems of "slight corruption not handled in time" or "normal fluctuation over-response" caused by rough determination dimension, ensuring that each corruption stage can match the corresponding intensity of response measures, improving the accuracy of state determination and subsequent execution, and realizing the dynamic adjustment goal of "precise adaptation of corruption degree and refrigeration intensity" by adopting the customized refrigeration adjustment strategy strongly bound to the material state level. For LO-level fresh material, only pressure (load) and temperature are used to develop a basic strategy to avoid excessive refrigeration and waste energy; for L1-level slight change, the refrigeration temperature is moderately reduced (e.g. by 1-2°C) and the air speed is increased (e.g. by 1 gear) based on the basic strategy to delay the corruption process without enabling extreme mode; for L2-level obvious change, the temperature is reduced to the first preset low temperature (e.g. 3-5°C) and the highest air speed is maintained to quickly inhibit microbial activity by strengthening refrigeration; for L3-level severe change, not only the lowest temperature or a second preset low temperature allowed by the equipment is enabled, but also the isolation mechanism is linked to block the spread of pollution. This strategy avoids "energy waste" or "corruption out of control" caused by using the same refrigeration mode for all levels, realizes the dual purposes of "refrigeration on demand" and "loss control in time", and ensures both preservation effect and energy efficiency and risk control.

[0065] The early warning mechanism includes at least one of,

[0066] generating a system log record;

[0067] triggering an audible and visual alarm;

[0068] sending early warning information to a user terminal, the content of the early warning information being associated with the state level.

[0069] The isolation mechanism is one or more of,

[0070] automatically closing the supply air damper leading to the target area;

[0071] prompting physical isolation of the area on a user interface;

[0072] Lock the target area of the library door and generate a clean library reminder.

[0073] The target area includes a plurality of independent sub-areas, each of which is independently configured with the pressure acquisition module, the temperature acquisition module, the state sensing module, the hierarchical decision module and the cold supply adjustment module, to realize independent sensing and adjustment of sub-areas.

[0074] Through the sub-area design of "independent configuration of a complete set of sensing-decision-adjustment modules in multiple sub-areas", the adjustment limitation of the traditional centralized refrigeration system "one size fits all" is broken through, and the differentiated refrigeration needs of different sub-areas are met. Each sub-area can independently monitor its own pressure (load), temperature (environment), and material state (color + odor), and develop a cold supply strategy according to its own data: for example, a sub-area storing high-load vegetables and fruits (high pressure) can independently increase the refrigeration capacity; another sub-area storing low-load meat (low pressure) can maintain the basic refrigeration capacity to avoid energy waste; if the material in a sub-area enters L3 level of corruption, the isolation mechanism only acts on this area and does not affect the normal operation of other sub-areas. This design also supports setting different temperature zone requirements in different sub-areas (5℃ for vegetable area and -2℃ for meat area), significantly improving the flexibility and adaptability of the refrigeration system, while avoiding the problem of a single area affecting the whole, ensuring the stability and efficiency of the whole system.

[0075] The application discloses an intelligent sub-area refrigeration adjustment method based on multi-modal sensing, as shown in the accompanying drawings, comprising the following steps: Figure 1

[0076] Obtain the pressure value, temperature value, color change information and odor change information of the target area; based on the color change information and odor change information, calculate the state level of the material; based on the state level, the pressure value and the temperature value, generate corresponding cold supply parameter adjustment instructions and early warning instructions;

[0077] S1, obtain a plurality of preset sensor data, transmit and record the data to the system, and the data acquisition adopts a depolarization method, which can suppress sensor drift and local disturbance; four-way data is parallelly input into a cache queue to ensure timestamp alignment, and weight, thermal, visual and chemical four types of information are synchronously obtained at one time, providing high-dimensional and complementary input for subsequent algorithms and avoiding single index blind area.

[0078] ​S101, through the M pressure sensors arranged below the target area, obtain the pressure analog signal reflecting the weight of the material, and calculate the average value of the M signals after analog-digital conversion to obtain the current pressure value Q of the target area. M-way signal arithmetic average→sliding window 30s de-extremum, eliminate the sharp peaks caused by instantaneous stepping or mechanical vibration. Map the weight to the load, indirectly reflect the stacking density and heat transfer load, and establish the first benchmark for the basic refrigeration capacity.

[0079] S102, through the N temperature sensors arranged in the target area, obtain the temperature analog signal reflecting the ambient temperature, and calculate the average value of the N signals after analog-digital conversion to obtain the current temperature value T of the target area. If a certain channel in the N-way signal exceeds 1.5℃ for 3 times in a row, the weight is automatically reduced to prevent the average value from being contaminated by a faulty probe. Obtain the regional thermal inertia representation and prevent local cold / hot spot misjudgment of the overall environment.

[0080] S103, through the image sensor facing the material in the target area, periodically collect high-definition RGB images. Periodic sampling instead of real-time streaming reduces the computational load. Periodic self-adaptation—relaxed to 10 minutes in LO, tightened to 2 minutes in L2 / L3. Use image recognition methods to identify material changes, and color change is one of the most intuitive and fastest responding indicators in the early stage of corruption.

[0081] S104, through the gas sensor array located inside the target area, real-time collect gas composition data in the space. The gas sensor array includes multiple metal oxide semiconductor sensors sensitive to characteristic corruption gases such as hydrogen sulfide, ammonia, ethylene, and ethanol. The array reads in parallel, and each channel performs differential operation to remove baseline drift. Use relative concentration change rate instead of absolute value to eliminate zero-point shift caused by sensor aging.

[0082] S2, based on the data obtained in step S1, process and analyze the data in the system. First, space-time alignment, then normalized to the O-1 interval to avoid dimension difference amplification error. Outliers are removed using the 3σ principle.

[0083] S201, color information processing, pre-processing the RGB image collected in S103, including grayscale, noise reduction and background segmentation; converting the processed image from RGB color space to HSV color space; extracting the average hue (H), saturation (S) and lightness (V) values of the material in the image; calculating the Euclidean distance between the H, S, V values and the preset fresh material reference value, and outputting the distance value as color change index CCI after normalization; by converting the RGB image after preprocessing to HSV color space, calculating the Euclidean distance between the H / S / V values of the material and the preset reference value to obtain the color change index CCI, and automatically updating the reference value with the storage batch, the baseline difference problem of different varieties and different maturity materials is solved, and the purpose of accurately quantifying color change to represent material freshness is achieved.

[0084] S202, odor information processing, feature extraction is performed on the multi-dimensional data collected by the gas sensor array in S104; the extracted feature vector is input into the trained pattern recognition classifier; the classifier outputs a scalar value representing the comprehensive level of the concentration of putrefactive gas, which is normalized and output as odor change index OCI; the classifier output probability is compressed to 0-1 by Sigmoid, and the same numerical interval as CCI is maintained to avoid weight imbalance; by inputting the feature extracted from the gas sensor array data into the trained pattern recognition classifier, the output probability is compressed to the 0-1 interval by Sigmoid to obtain the odor change index OCI, which realizes the same numerical interval as the color change index CCI, avoids weight imbalance, and accurately represents the comprehensive level of the concentration of putrefactive gas;

[0085] S203, state level determination, to prevent missed judgment caused by color-odor asynchronization; 5% interval of boundary hysteresis to avoid critical jitter; compare CCI with three preset color thresholds (C1, C2, C3), where O

[0086] Compare OCl with three preset odor thresholds (O1, O2, O3), where O

[0087] According to the comparison result, the state level L of the material is determined, and the determination logic is,

[0088] If CCI≤C1 and OCI≤O1, the state level is determined to be LO (normal period); if (C1<CCI≤C2) or (O1<OCI≤O2) is true, the state level is determined to be L1 (slight change period);

[0089] If (C2<CCI≤C3) or (O2<OCI≤O3) is true, the state level is determined to be L2 (obvious change period);

[0090] If CCI > C3 or OCI > O3 is true, the state level is determined as L3 (severe change period); the logic of determining the material state level by combining the preset threshold value of CCI and OCI double indicators and setting a 5% boundary hysteresis interval realizes the purpose of preventing missed judgment caused by color-odor asynchronization (such as single indicator not exceeding the standard but another indicator being abnormal) and avoiding frequent level fluctuation in the critical state. S3, according to the Q and T values obtained in S1 and the state level L determined in S203, generates the final cooling parameter adjustment instruction and warning instruction; the logic of generating the adjustment instruction by integrating the pressure value Q, the temperature value T and the state level L realizes the purpose of accurately formulating the refrigeration strategy from the "basic environmental parameter + material corruption state" two-dimensional, avoiding overcooling or insufficient refrigeration caused by relying on environmental parameters only. S301, basic cooling parameter calculation, according to the current pressure value Q and the current temperature value T, the mapping table based on the pre-stored rules (such as the original patent claim 2) is queried to obtain the basic cooling temperature setting value T_set_base and the cooling air speed setting value V_set_base; the table entry supports interpolation instead of step, reducing quantization error; when Q or T exceeds the table boundary, the nearest neighbor + slope extrapolation is adopted; by adopting the mapping table based on the pre-stored rules to query the basic cooling parameter, and supporting the interpolation (instead of step) of the table entry and the boundary extrapolation (nearest neighbor + slope), the purpose of reducing quantization error and ensuring that reasonable basic cooling parameters (temperature T_set_base, air speed V_set_base) can be obtained when Q or T exceeds the table boundary is realized.

[0091] V_set_base) of the table.

[0092] S302, final parameter adjustment instruction generation based on state level,

[0093] When L = LO,

[0094] Final cooling temperature instruction = T_set_base;

[0095] Final cooling air speed instruction = V_set_base;

[0096] Warning instruction = none;

[0097] When L = L1,

[0098] Final cooling temperature instruction = T_set_base-ΔTl;

[0099] Final cooling temperature command = T_set_base + ΔT1; Final cooling air speed command = V_set_base + 1; Warning command = generate system log, record time, area and level L1 information; When L = L2, final cooling temperature command = min(T_set_base - ΔT2, T_L2) (where ΔT2 is a preset value, typically in the range of 3-5°C, and T_L2 is the absolute low temperature threshold preset for this level); Final cooling air speed command = V_max (the highest air speed allowed by the system); Warning command = trigger the audible and light alarm and send a warning message to the monitoring terminal; When L = L3,

[0100] Final cooling temperature command = T_min (the lowest refrigeration temperature allowed by the system);

[0101] Final cooling air speed command = V_max;

[0102] Warning command = trigger the highest level audible and light alarm, send an emergency message of “serious deterioration, handle immediately” to the administrator terminal, and generate a control command to close the area damper and execute the isolation process; ΔT1, ΔT2, T_L2, T_min all support remote background calibration; The strategy table supports version number management for easy OTA upgrade; The compensation amount has a monotonic nonlinear relationship with the level, and the isolation command is automatically triggered at L3 without additional logic branches; By formulating differentiated cooling strategies for different state levels of LO-L3 (such as L1 reducing ΔT1, L2 using the lowest temperature constraint, and L3 triggering T_min), combined with the design of automatically triggering the area isolation command at L3 level, the purpose of accurately matching the refrigeration intensity according to the degree of material spoilage and quickly isolating and preventing the spread of pollution in the case of serious deterioration is achieved; By supporting remote parameter calibration and OTA strategy table upgrade, the purpose of updating parameters and strategies without disassembling the system on site and improving system maintenance convenience and flexibility is achieved.

[0103] S4: instruction execution and feedback step, if the level rises (such as L3→L2), immediately cancel isolation and restore normal strategy; all instructions are timestamped and written back to the log for subsequent model training; By adopting the process of “instruction execution-state feedback-continuous closed-loop control”, combined with the design of immediately canceling isolation when the level rises and timestamping and writing back the log for instructions, the purpose of timely responding to the improvement of material state to avoid excessive refrigeration and retaining data for subsequent model training optimization is achieved, ensuring continuous and accurate refrigeration adjustment.

[0104] S401, send the final cooling temperature and air speed commands generated in step S302 to the refrigeration unit and fan frequency converter of the target area to adjust the temperature and air speed, and by directly sending the final cooling temperature / air speed command to the execution of the refrigeration unit and fan frequency converter, the purpose of no delay of the command and quick adjustment of the refrigeration parameters to ensure that the material is in the target refrigeration environment is achieved.

[0105] S402, execute the early warning instruction generated in S302 to complete corresponding alarm and information push operation; through triggering differential early warning (such as L2 sound and light alarm, L3 emergency message push) according to the state level, the purpose of making the management personnel pay attention to the material state in layers, quickly responding in serious deterioration, and reducing the loss is realized.

[0106] S403, return to step S1, continuously perform data acquisition and monitoring to form a closed loop control until the target area is emptied or the system is shut down.

[0107] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make slight changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any modification, equivalent change and modification of the above embodiments made according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. An intelligent zoned refrigeration control system based on multimodal sensing, characterized in that, Includes a pressure acquisition module, used to acquire the pressure value of the target area; The temperature acquisition module is used to acquire the temperature value of the target area; The cooling regulation module is used to adjust the cooling parameters of the target area based on the pressure value, temperature value, and preset threshold of the target area. The state perception module is used to acquire information on color and odor changes of materials within the target area; A hierarchical decision-making module, connected to the state perception module, is used to calculate the state level of the material based on the color change information and odor change information. The state level is used to characterize the degree of spoilage of the material. The cooling regulation module is also connected to the hierarchical decision module, and is used to adjust the cooling parameters of the target area and / or activate the early warning mechanism according to the status level.

2. The intelligent regional refrigeration control system based on multimodal sensing according to claim 1, characterized in that, The status levels include a normal period (L0) ranging from mild to severe, a slight change period (L1) ranging from mild to severe, a significant change period (L2) ranging from severe to severe, and a serious change period (L3). The cooling regulation module is specifically used to execute a customized regulation strategy corresponding to the state level. When the state level is L0, the cooling parameters are adjusted according to the pressure and temperature values ​​of the target area. When the state level is L1, the cooling temperature setpoint is reduced and the cooling air velocity is increased, based on the adjustment according to the pressure and temperature values. When the status level is L2, the cooling temperature setpoint is reduced to the first preset low temperature value and the cooling fan speed is continuously operated at the highest level. When the status level is L3, the control cooling temperature setpoint is reduced to the lowest value allowed by the equipment or the second preset low temperature value, and the cooling fan speed is maintained at the highest level. At the same time, the isolation mechanism is activated.

3. The intelligent regional refrigeration control system based on multimodal sensing according to claim 2, characterized in that, The state awareness module includes... The color recognition unit is used to acquire images of materials within a target area through an image sensor and extract color features from the images to generate the color change information; The odor recognition unit is used to collect composition data of gas in a target area through a gas sensor array, and analyze the composition data through a pattern recognition algorithm to generate the odor change information.

4. The intelligent regional refrigeration control system based on multimodal sensing according to claim 2, characterized in that, The early warning mechanism includes at least one of the following: Generate system log records; Trigger the audible and visual alarm; A warning message is sent to the user terminal, the content of which is related to the status level.

5. The intelligent regional refrigeration control system based on multimodal sensing according to claim 2, characterized in that, The isolation mechanism is one or more of the following. Automatically close the air supply damper leading to the target area; The user interface prompts you to physically isolate this area; Lock the warehouse doors in the target area and generate a warehouse clearing alert.

6. The intelligent regional refrigeration control system based on multimodal sensing according to claim 3, characterized in that, The target area includes multiple independent sub-regions, each of which is independently equipped with the pressure acquisition module, temperature acquisition module, status sensing module, hierarchical decision-making module, and cooling regulation module, enabling independent sensing and regulation of each sub-region.

7. A method for implementing the intelligent regional refrigeration control system based on multimodal sensing as described in any one of claims 1-6, the method comprising the following steps, Acquire pressure, temperature, color change, and odor change information of the target area; calculate the state level of the material based on the color and odor change information; and generate corresponding cooling parameter adjustment and warning commands based on the state level, pressure, and temperature. S1, acquire multiple preset sensor data, transmit and record the data into the system; S101: By using M pressure sensors arranged below the target area, pressure analog signals reflecting the weight of the material are acquired, and the average value of the M signals is calculated after analog-to-digital conversion to obtain the current pressure value Q of the target area; S102: By using N temperature sensors arranged in the target area, the temperature analog signal reflecting the ambient temperature is obtained, and the average value of the N signals is calculated after analog-to-digital conversion to obtain the current temperature value T of the target area; S103, periodically acquire high-definition RGB images of the material in the target area using an image sensor; S104, acquire real-time data on the gas composition in the space using a gas sensor array located inside the target area, the gas sensor array including multiple metal oxide semiconductor sensors sensitive to characteristic putrefactive gases such as hydrogen sulfide, ammonia, ethylene, and ethanol. S2, based on the data obtained in step S1, processes and analyzes the data in the system. S201, Color information processing: preprocessing the RGB image acquired in S103, including grayscale conversion, noise reduction, and background segmentation; converting the processed image from the RGB color space to the HSV color space; extracting the average hue (H), saturation (S), and lightness (V) values ​​of the material in the image; calculating the Euclidean distance between the H, S, and V values ​​and the preset fresh material reference values, and outputting the normalized distance value as the color change index (CCI); S202, Odor information processing, extracting features from the multidimensional data collected by the gas sensor array in S104; inputting the extracted feature vector into the trained pattern recognition classifier; The classifier outputs a scalar value that characterizes the overall level of the concentration of putrefactive gases. After normalization, this value is output as the Odor Change Index (OCI). S203, Status Level Determination The CCI is compared with three preset color thresholds (C1, C2, C3), where 0 < C1 < C2 < C3 ≤ 1; The OCI is compared with three preset odor thresholds (O1, O2, O3), where 0 < O1 < O2 < O3 ≤ 1; Based on the comparison results, the material's state level L is determined, and the determination logic is as follows: If CCI≤C1 and OCI≤O1, the state level is determined to be L0 (normal period); if (C1<CCI≤C2) or (O1<OCI≤O2) is true, the state level is determined to be L1 (slight change period). If (C2<CCI≤C3) or (O2<OCI≤O3) is true, then the state level is determined to be L2 (significant change period); If CCI > C3 or OCI > O3 is true, the status level is determined to be L3 (severe change period); S3, based on the Q and T values ​​obtained in S1 and the status level L determined in S203, generates the final cooling parameter adjustment command and warning command. S301, Basic cooling parameter calculation: Based on the current pressure value Q and the current temperature value T, query the mapping table based on the pre-stored rules (as described in claim 2 of the original patent) to obtain the basic cooling temperature setpoint T_set_base and cooling fan speed setpoint V_set_base; S302, Generation of final parameter adjustment instructions based on state level. When L = L0 Final cooling temperature command = T_set_base; Final cooling fan speed command = V_set_base; Warning instructions = None; When L = L1 Final cooling temperature command = T_set_base - ΔT1; The final cooling fan speed command is one level higher than V_set_base; Warning command = Generate system log, recording time, region, and L1 level information; When L = L2 Final cooling temperature command = min(T_set_base-ΔT2, T_L2) (where ΔT2 is the preset value, typically ranging from 3 to 5℃, and T_L2 is the preset absolute low temperature threshold for this level); Final cooling fan speed command = V_max (the maximum fan speed allowed by the system); Warning command = triggering the audible and visual alarm and sending a warning message to the monitoring terminal; When L = L3 Final cooling temperature command = T_min (the lowest allowable cooling temperature of the system); Final cooling fan speed command = V_max; Warning command = Triggers the highest level audible and visual alarm, sends an emergency message "Severe deterioration, handle immediately" to the administrator terminal, and generates control commands to close the air damper in the area and execute the isolation procedure; S4: Instruction Execution and Feedback Steps S401, the final cooling temperature command and fan speed command generated in step S302 are sent to the refrigeration units and fan inverters in the target area to perform temperature and fan speed adjustment; S402, execute the warning command generated in step S302 to complete the corresponding alarm and information push operation; S403, return to step S1, continue to collect and monitor data to form closed-loop control until the target area is cleared or the system is shut down.

Images