Freezer self-adaptive energy-saving control system based on multi-mode perception

Abstract

The invention relates to the technical field of refrigeration, in particular to a refrigerator self-adaptive energy-saving control system based on multi-mode perception, which comprises a data acquisition module, a refrigeration regulation module, a wind power auxiliary module, an air pressure auxiliary module and a data processing module, the data acquisition module is used for capturing temperature distribution, humidity conditions and air pressure conditions in the refrigerator and sending the data information to the data processing module. By arranging the thermal camera and the wind power auxiliary module, the thermal camera captures the position of an ice layer, the draught fan is aligned with the ice layer, sublimation of the surface of the ice layer is accelerated through air flowing, and when air passes through the surface of the ice layer, heat can be transferred to the ice layer with the lower temperature due to work of the evaporator; and the temperature of flowing air is reduced while ice layer disappearance is accelerated, so that the electric energy consumption of the refrigeration adjusting module is reduced, and the refrigeration quality is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of refrigeration technology, and in particular to a refrigerator adaptive energy-saving control system based on multi-modal perception. BACKGROUND

[0002] The refrigerator is a kind of refrigeration equipment widely used in family, supermarket and cold chain logistics field, mainly used for long-term preservation of food, with the improvement of environmental protection and energy saving demand, the intelligentization and energy saving of refrigerator become the key direction of technology development. The traditional refrigerator mainly realizes refrigeration through periodic start-stop of compressor, while the modern refrigerator improves the control system and refrigeration technology to improve the energy efficiency.

[0003] Ice layer is one of the common problems of refrigerator, and thick ice layer will hinder heat exchange, reduce refrigeration efficiency, increase compressor burden, and further increase energy consumption. The existing refrigerator defrosting method is mainly time-based defrosting, that is, the heating element is started to melt the ice layer on the evaporator of the refrigerator every fixed time. However, the thickness and distribution of ice layer vary due to humidity, door opening frequency and stored goods, and time-based defrosting may result in unnecessary defrosting at unnecessary time, and the heating element often has a large heating range, which cannot accurately defrost and wastes energy. In addition, the ice layer can store cold energy during accumulation, but the traditional defrosting method often regards the ice layer as an obstacle that must be removed, and does not fully utilize its low temperature characteristics, resulting in energy consumption during defrosting and inefficient use of resources.

[0004] In summary, the existing refrigerator has obvious deficiencies in ice removal, energy saving and ice layer utilization, and a more intelligent and energy-saving control scheme is needed to solve these problems. SUMMARY

[0005] In order to overcome the shortcomings of high power consumption and inability to utilize ice layer as a resource of the existing ice removal technology, the present application provides a refrigerator adaptive energy-saving control system based on multi-modal perception.

[0006] The technical scheme is: a refrigerator adaptive energy-saving control system based on multi-modal perception, comprising a data acquisition module, a refrigeration adjustment module, a wind power auxiliary module, a gas pressure auxiliary module and a data processing module, wherein: The data acquisition module comprises a temperature sensor, a thermal camera, a light sensor, a humidity sensor and a gas pressure sensor, which is used to capture the temperature distribution, humidity condition and gas pressure condition in the refrigerator, and send these data information to the data processing module; The refrigeration adjustment module comprises a compressor, a condenser, an expansion valve, a filter, an evaporator, an energy accumulator and a refrigerant, which is used to convert the refrigerant in the refrigeration adjustment module between liquid and gas state, so as to refrigerate the inside of the refrigerator, and the refrigeration speed is controlled by the data processing module; The wind auxiliary module includes a fan with an adjustable angle, which is used to circulate the air inside the refrigerator so that the refrigerated air in the refrigerator can flow fully throughout the refrigerator, thereby improving the refrigerator's cooling efficiency; The air pressure auxiliary module includes an electrically controlled air pressure valve, an air pressure pump, and an air storage chamber, which is used to adjust the air pressure inside the refrigerator to change the relative humidity and absolute humidity inside the refrigerator. When the relative humidity in the air storage chamber does not reach 100%, the relative humidity and temperature of the air in the air storage chamber remain consistent with those inside the refrigerator; The data processing module is used to receive data information from the data acquisition module, perform feature extraction and feature splicing on the data information, and obtain the characteristic values ​​of the state inside the refrigerator through the neural network, so as to timely adjust the temperature through the refrigeration adjustment module and the wind auxiliary module, and adjust the air pressure and humidity through the air pressure auxiliary module.

[0007] Preferably, the temperature sensor includes an external sensor and a built-in sensor. The external sensor is used to detect the external ambient temperature outside the refrigerator, and the built-in sensor is used to detect the refrigeration temperature inside the refrigerator. By comparing the temperature difference between the two built-in and external temperature sensors, a temperature adjustment basis is provided for the data processing module, so that the data processing module can adjust the refrigeration effect in time to prevent the temperature difference between the inside and outside of the refrigerator from being too small, resulting in frequent starting and stopping of the compressor or low-frequency operation, thereby increasing the power consumption of the compressor.

[0008] Preferably, the wind-assisted module uses a fan to control the temperature of the refrigerator, blows the cold air generated by heat exchange at the evaporator of the refrigeration regulation module into the refrigerator, accelerates the circulation of cold air in the refrigerator, and the fan pushes all the cold air around the evaporator into the interior of the refrigerator, so that the cold air is evenly distributed in the refrigerator.

[0009] Preferably, the electrically controlled air pressure valve in the air pressure auxiliary module is used to control the air inside the refrigerator and the air from the external environment to enter and exit the air storage cavity. The air storage cavity separates the interior of the refrigerator from the external environment, thereby avoiding direct exchange of gas with the external environment when adjusting the air pressure inside the refrigerator, which will cause a large loss of temperature inside the refrigerator and an increase in humidity.

[0010] Preferably, during cooling, the data processing module will splice the light-sensitive image and the thermal image into a four-channel image containing three-color channels and a thermal channel after a pre-processing process of resolution adjustment, image cropping, and normalization. The neural network of the data processing module obtains the image features of the four-channel image through a convolution-then-linear method, and expands it to obtain a one-dimensional image feature matrix. The one-dimensional image feature matrix is ​​merged with the data information of the temperature sensor, the humidity sensor, and the air pressure sensor to form a one-dimensional multimodal matrix. The one-dimensional multimodal matrix obtains the refrigerator adjustment data features through the linear layer operation and nonlinear transformation of the neural network. The refrigerator adjustment data features include the target power of the compressor, the target power of the fan, and the target air pressure in the refrigerator. Adjusting the power of the compressor can adjust the cooling module in the refrigerator. The refrigeration effect inside the cabinet, the greater the power of the compressor, the stronger the cooling effect of the refrigeration module. The flow of cold air in the freezer is adjusted by adjusting the fan power, thereby adjusting the distribution of cold air in the freezer. The air pressure, relative humidity and absolute humidity in the freezer are adjusted by controlling the flow of air in the air storage cavity. When the air pressure in the freezer increases, the moisture density in the freezer increases, thereby increasing the absolute humidity and relative humidity of the air in the freezer before the saturated water vapor pressure is reached. At this time, the high-humidity air in the freezer will inhibit the evaporation of moisture in the food. When the freezer door is opened, under the influence of the air pressure difference, the water vapor is quickly released outward, and then the air pressure in the freezer is restored, thereby reducing the relative humidity and absolute humidity. The cold air will stay around the freezer door, thereby inhibiting the high-humidity air from the outside from entering the freezer and keeping the freezer dry.

[0011] Preferably, when local air circulation in the refrigerator is difficult due to the placement of food in the refrigerator, the position of the food in the refrigerator can be collected by a light-sensing camera, and the temperature of the surface of the food in the refrigerator can be collected by a thermal camera. Based on the temperature conditions in the densely populated areas, the visual neural network of the data processing module can be used to quickly determine the obstructed position of air circulation, and the fan angle of the wind-assisted module can be adjusted in time, so that cold air can circulate more smoothly from the densely populated areas, thereby improving the refrigeration effect inside the refrigerator, reducing the power required for refrigeration, and ensuring longer storage of food.

[0012] Preferably, when the temperature of a local area in the refrigerator needs to be lower than that of other areas, the position of the food in the refrigerator can be captured by a light-sensing camera, so that the cold air guided by the fan can actively blow to the area where local low temperature is required, causing the temperature of the required area in the refrigerator to be lower than that of other areas, and the air pressure in the refrigerator is adjusted by storing or releasing air in the air storage cavity of the air pressure auxiliary module in the refrigerator, so that the air pressure in the refrigerator is increased, and the moisture density in the refrigerator is increased, thereby adjusting the relative humidity and absolute humidity of the air in the refrigerator, inhibiting the transpiration of water in the food, reducing the evaporation of water caused by blowing cold air in the area where local low temperature is required, and ensuring the quality of food in the area where local low temperature is required.

[0013] Preferably, thermal images are collected by a thermal camera and sent to a data processing module for analysis. Based on the low temperature distribution in the freezer and a convolutional neural network, the unformed ice layer on the inner wall of the freezer can be identified, and the position of the unformed ice layer in the freezer can be obtained. The wind auxiliary module is controlled by the data processing module to aim the fan at the unformed ice layer. Only a small amount of ice in the unformed ice layer is attached to the inside of the freezer. The air flows on the surface of the unformed ice layer, thereby accelerating the sublimation of the surface of the unformed ice layer without affecting the refrigeration effect. When the air passes over the surface of the unformed ice layer, it transfers heat to the unformed ice layer whose temperature is lower due to the operation of the evaporator, accelerating the disappearance of the unformed ice layer while reducing the temperature of the flowing air, thereby utilizing the unformed ice layer to share the refrigeration burden of the refrigeration regulation module, reducing the power consumption of the refrigeration regulation module, and improving the refrigeration quality.

[0014] Preferably, when the wind-assisted module sublimates the unformed ice layer, moisture from the unformed ice layer enters the air in the refrigerator, thereby increasing the humidity in the refrigerator. To prevent excessive humidity in the refrigerator from causing mold growth and food spoilage, the data processing module can control the air pressure assist module based on humidity information data collected by the humidity sensor. When the humidity in the refrigerator is normal, the data processing module will lower the air pressure in the refrigerator to reduce the moisture density in the refrigerator, thereby accelerating the sublimation process on the surface of the ice layer. As the ice layer sublimates, the humidity in the refrigerator gradually approaches saturation. At this time, the data processing module controls the electrically controlled air pressure valve of the air pressure assist module to discharge the high-humidity air in the refrigerator into the air storage chamber. The air storage chamber is pressurized by an air pressure pump, causing the moisture in the air storage chamber to exceed the saturated water vapor pressure, thereby causing the moisture to condense into water droplets and be discharged from the device before condensing into ice. At this time, the air in the air storage chamber still has a relatively low temperature and can be used to continue to maintain the humidity in the refrigerator, thereby reducing energy waste during the de-icing process.

[0015] Preferably, the data processing module divides a day into multiple time segments and records the number of times the refrigerator door is opened and closed in different time segments. An exponential decay curve that gradually tends to 0 is fitted based on the number of times the refrigerator door is opened and closed in different time segments. The exponential decay curve is used to indicate the possibility that the refrigerator user will form a habit of opening and closing the refrigerator door at fixed time segments every day. The exponential decay curve is used to assign weights to the user's behavior of opening and closing the refrigerator, thereby predicting the time segments when the refrigerator user opens and closes the refrigerator, and adjusting the refrigerator temperature, humidity and air pressure based on the prediction to reduce the entry of moisture into the refrigerator and the loss of cold air in the refrigerator caused by the opening and closing of the refrigerator door.

[0016] The present invention is provided with a thermal camera and a wind-assisted module. The thermal camera captures the position of the ice layer, and the fan is directed at the ice layer. Air flows over the surface of the ice layer, thereby accelerating the sublimation of the ice layer without affecting the cooling effect. When the air passes over the surface of the ice layer, it transfers heat to the ice layer, which is at a lower temperature due to the operation of the evaporator, thereby accelerating the disappearance of the ice layer and reducing the temperature of the flowing air. In this way, the ice layer is used to share the cooling burden of the cooling adjustment module, reducing the power consumption of the cooling adjustment module and improving the cooling quality. The present invention is equipped with a light-sensing camera, a thermal camera, and a wind-assisted module. The light-sensing camera collects the position of food placed in the refrigerator, and the thermal camera collects the surface temperature of food in the refrigerator. Based on the temperature of the food-intensive area, the visual neural network of the data processing module can quickly determine the location of air circulation obstruction and timely adjust the fan angle of the wind-assisted module, so that cold air can flow more smoothly from the food-intensive area, improve the cooling effect inside the refrigerator, reduce the power required for cooling, and ensure that food can be stored for a longer period of time. The present invention divides a day into multiple time segments through a data processing module, and records the number of times the refrigerator door is opened and closed in different time segments, assigns weights to the user's behavior of opening and closing the refrigerator, thereby predicting the time segments when the refrigerator user opens and closes the refrigerator, and adjusts the refrigerator temperature, humidity and air pressure based on the prediction to reduce the entry of moisture into the refrigerator and the loss of cold air in the refrigerator caused by the opening and closing of the refrigerator door. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the overall structure of the present invention; Figure 2 This is a refrigeration workflow diagram of the data processing module of the present invention; Figure 3 This is a de-icing workflow diagram of the data processing module of the present invention. DETAILED DESCRIPTION

[0018] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments, but this does not limit the present invention.

[0019] An adaptive energy-saving control system for refrigerators based on multimodal perception, such as Figure 1 As shown, it includes a data acquisition module, a refrigeration adjustment module, a wind auxiliary module, an air pressure auxiliary module and a data processing module, wherein: The data acquisition module includes a temperature sensor, a thermal camera, a light camera, a humidity sensor, and an air pressure sensor, which are used to capture the temperature distribution, humidity, and air pressure inside the refrigerator and send this data information to the data processing module; The refrigeration regulation module includes a compressor, a condenser, an expansion valve, a filter, an evaporator, an accumulator, and a refrigerant. It is used to continuously convert the refrigerant in the refrigeration regulation module between liquid and gaseous states, thereby cooling the interior of the refrigerator. The refrigeration speed is controlled by the data processing module. The wind auxiliary module includes a fan with an adjustable angle, which is used to circulate the air inside the refrigerator so that the refrigerated air in the refrigerator can flow fully throughout the refrigerator, thereby improving the refrigerator's cooling efficiency; The air pressure auxiliary module includes an electrically controlled air pressure valve, an air pressure pump, and an air storage chamber, which is used to adjust the air pressure inside the refrigerator to change the relative humidity and absolute humidity inside the refrigerator. When the relative humidity in the air storage chamber does not reach 100%, the relative humidity and temperature of the air in the air storage chamber remain consistent with those inside the refrigerator; The data processing module is used to receive data information from the data acquisition module, perform feature extraction and feature splicing on the data information, and obtain the characteristic values ​​of the state inside the refrigerator through the neural network, so as to timely adjust the temperature through the refrigeration adjustment module and the wind auxiliary module, and adjust the air pressure and humidity through the air pressure auxiliary module.

[0020] Specifically, the temperature sensor is used to capture the temperature of the air in a certain area in the refrigerator, the thermal camera is used to capture the temperature distribution of all items in a certain area in the refrigerator, and the light camera is used to capture the structure of all food in a certain area. The data processing module can identify the type of food and select the appropriate refrigerator temperature and humidity for the food. The humidity sensor is used to detect the humidity changes in the refrigerator, providing a basis for the data processing module to control the humidity in the refrigerator, preventing the humidity in the refrigerator from being too high, causing mold growth and food deterioration, or too low humidity, causing food to lose water and reduce quality.

[0021] Specifically, the temperature sensor includes an external sensor and a built-in sensor. The external sensor is used to detect the external ambient temperature outside the refrigerator, and the built-in sensor is used to detect the refrigeration temperature inside the refrigerator. By comparing the temperature difference between the two built-in and external temperature sensors, the data processing module is provided with a temperature adjustment basis, so that the data processing module can adjust the refrigeration effect in time to prevent the temperature difference inside and outside the refrigerator from being too small, which will cause the compressor to start and stop frequently or work at a low frequency, thereby increasing the power consumption of the compressor. Among them, the traditional compressor will start and stop frequently when the temperature difference inside and outside the refrigerator is too small, and frequent start and stop will consume a lot of energy, resulting in energy waste of the compressor. The variable frequency compressor will run at a low speed when the temperature difference inside and outside the refrigerator is too small. Too low a speed will cause the variable frequency compressor to fail to form a good lubrication effect, resulting in wear of components.

[0022] Specifically, when the refrigeration regulation module is started, the compressor will compress the gaseous refrigerant into liquid. The compressed liquefaction of the refrigerant will generate a large amount of heat and enter the condenser to release heat to the external environment. The refrigerant then enters the expansion valve. When flowing out of the expansion valve, the volume of the refrigerant expands to form a gas-liquid mixture. After the impurities in the refrigerant are filtered out by the filter, the refrigerant absorbs heat in the freezer in the evaporator and continues to evaporate. Finally, it absorbs energy at the accumulator, causing the refrigerant to be completely vaporized. At this time, one refrigeration cycle of the refrigeration regulation module is completed, and the refrigerant in the accumulator can enter the compressor for subsequent multiple refrigeration cycles.

[0023] It is worth noting that in order to ensure the normal operation of the compressor, when the freezer is used as a refrigerator, the temperature difference between the freezer and the external environment should be at least 5 degrees Celsius to prevent the growth of mold and toxins in cooked food and dairy ingredients. When the freezer is used as a freezer, the temperature difference between the freezer and the external environment should be at least 15 degrees Celsius to prevent frozen food from thawing.

[0024] The wind-assisted module uses a fan to control the temperature of the freezer, blowing the cold air generated by heat exchange at the evaporator of the refrigeration regulation module into the freezer to speed up the cold air circulation in the freezer. The fan will push all the cold air around the evaporator into the inside of the freezer, so that the cold air is evenly distributed in the freezer, avoiding the cold air gathering at the evaporator, resulting in the low temperature difference between the cold air and the evaporator, reducing the cooling efficiency, and effectively reducing the energy waste of cooling.

[0025] Specifically, the electrically controlled air pressure valve in the air pressure auxiliary module is used to control the air inside the refrigerator and the air from the external environment to enter and exit the air storage cavity. The air storage cavity separates the interior of the refrigerator from the external environment, thereby avoiding direct exchange of gas with the external environment when adjusting the air pressure inside the refrigerator, resulting in a large loss of temperature inside the refrigerator and increased humidity.

[0026] Specifically, the air storage chamber in the air pressure auxiliary module is located inside the refrigerator and shares the same refrigeration adjustment module with the refrigerator for temperature control, so that the temperature inside the refrigerator is consistent with the temperature in the air storage chamber. Since the saturated moisture pressure is only related to the temperature, the lower the temperature, the lower the saturated moisture pressure. The outside air temperature is higher than the temperature in the air storage chamber, which makes the saturated moisture pressure of the outside air higher and has a higher absolute humidity. When the outside air enters the air storage chamber, the temperature of the outside air in the air storage chamber decreases, which reduces the saturated water vapor pressure and condenses the water vapor in the outside air, resulting in a lower absolute humidity and a higher relative humidity in the air storage chamber.

[0027] like Figure 2As shown, during cooling, the data processing module will splice the light-sensitive image and the thermal image into a four-channel image after a pre-processing process of resolution adjustment, image cropping, and normalization. The neural network of the data processing module obtains the image features of the four-channel image through a convolution-then-linear method, and expands it to obtain a one-dimensional image feature matrix. The one-dimensional image feature matrix is ​​merged with the data information of the temperature sensor, the humidity sensor, and the pressure sensor to form a one-dimensional multimodal matrix. The one-dimensional multimodal matrix obtains the refrigerator adjustment data features through the linear layer operation and nonlinear transformation of the neural network. The refrigerator adjustment data features include the target power of the compressor, the target power of the fan, and the target air pressure in the refrigerator. Adjusting the power of the compressor can adjust the cooling effect of the refrigeration module in the refrigerator. The greater the power of the compressor, the stronger the cooling effect of the refrigeration module. The flow of cold air in the refrigerator is adjusted by adjusting the fan power, thereby adjusting the cooling The distribution of cold air in the cabinet is achieved by controlling the flow of air in the air storage cavity to adjust the air pressure, relative humidity and absolute humidity in the freezer. When the air pressure in the freezer increases, the moisture density in the freezer increases, thereby increasing the absolute humidity and relative humidity of the air in the freezer before reaching the saturated water vapor pressure. At this time, the high-humidity air in the freezer will inhibit the evaporation of moisture in the food. When the freezer door is opened, under the influence of the air pressure difference, the water vapor is quickly released outward, and then the air pressure in the freezer is restored, thereby reducing the relative humidity and absolute humidity. The cold air will stay around the freezer door, thereby inhibiting the high-humidity air from the outside from entering the freezer, keeping the freezer dry, and avoiding temperature fluctuations in the freezer. Compared with ordinary freezers, the stay of cold air at the freezer door can reduce the humidity fluctuation in the freezer by about 60% and the temperature change in the freezer by about 69% when the freezer door is opened, reduce the cold air loss by about 40%, and reduce the compressor energy consumption by about 15%.

[0028] When local air circulation in the refrigerator is difficult due to the placement of food in the refrigerator, the position of the food in the refrigerator can be collected through a light-sensing camera, and the surface temperature of the food in the refrigerator can be collected through a thermal camera. Based on the temperature conditions in the densely populated areas, the visual neural network of the data processing module can quickly determine the obstructed position of air circulation, and timely adjust the fan angle of the wind auxiliary module, so that cold air can circulate more smoothly from the densely populated areas, thereby improving the refrigeration effect inside the refrigerator, reducing the power required for refrigeration, and ensuring longer storage of food.

[0029] When the temperature of a local area in the refrigerator needs to be lower than other areas, the position of the food in the refrigerator can be captured by a light-sensing camera, so that the cold air guided by the fan can actively blow to the area where local low temperature is required, causing the temperature of the required area in the refrigerator to be lower than other areas. The air pressure in the refrigerator is adjusted by storing or releasing air in the air storage cavity of the air pressure auxiliary module in the refrigerator, so that the air pressure in the refrigerator is increased, and the moisture density in the refrigerator is increased, thereby inhibiting the transpiration of water in the food, reducing the evaporation of water caused by cold air blowing in the area where local low temperature is required, and ensuring the quality of food in the area where local low temperature is required.

[0030] like Figure 3 As shown, thermal images are collected by a thermal camera and sent to a data processing module for analysis. Based on the low temperature distribution inside the freezer and a convolutional neural network, the unformed ice layer on the inner wall of the freezer can be identified and the position of the unformed ice layer inside the freezer can be obtained. The wind auxiliary module is controlled by the data processing module to align the fan with the unformed ice layer. Only a small amount of ice in the unformed ice layer is attached to the inside of the freezer. When air flows on the surface of the unformed ice layer, the sublimation of the surface of the unformed ice layer is accelerated without affecting the refrigeration effect. When the air passes over the surface of the unformed ice layer, it transfers heat to the unformed ice layer, which has a lower temperature due to the operation of the evaporator, accelerating the disappearance of the unformed ice layer while reducing the temperature of the flowing air. In this way, the unformed ice layer is used to share the refrigeration burden of the refrigeration regulation module, reducing the power consumption of the refrigeration regulation module and improving the refrigeration quality.

[0031] When the wind-assisted module sublimates the unformed ice layer, the moisture in the unformed ice layer will enter the air of the freezer, thereby increasing the humidity in the freezer. In order to prevent the excessive humidity in the freezer from causing mold growth and food spoilage, the data processing module can control the air pressure auxiliary module according to the humidity information data collected by the humidity sensor. When the humidity in the freezer is normal, the data processing module will lower the air pressure in the freezer, reduce the water density in the freezer, and lower the melting point of the ice layer, thereby accelerating the sublimation process of the ice layer surface. As the ice layer sublimates, the freezer The humidity inside the cabinet gradually approaches saturation. At this time, the data processing module controls the electrically controlled air pressure valve of the air pressure auxiliary module to discharge the high-humidity air inside the refrigerator into the air storage chamber. The air storage chamber is pressurized by the air pressure pump, causing the moisture in the air storage chamber to be higher than the saturated water vapor pressure. The relative humidity in the air storage chamber temporarily exceeds 100%, resulting in oversaturation, causing the moisture to condense into water droplets and be discharged out of the device before condensing into ice. At this time, the air in the air storage chamber still has a relatively low temperature and can be used to continue to maintain the humidity in the refrigerator, thereby reducing energy waste during the de-icing process.

[0032] The data processing module divides a day into multiple time segments and records the number of times the refrigerator door is opened and closed in different time segments. An exponential decay curve that gradually tends to 0 is fitted based on the number of times the refrigerator door is opened and closed in different time segments. This exponential decay curve is used to indicate the possibility that refrigerator users will form a habit of opening and closing the refrigerator door at fixed time segments every day. The exponential decay curve is used to assign weights to the user's behavior of opening and closing the refrigerator, thereby predicting the time segments when the refrigerator user opens and closes the refrigerator. Based on the prediction, the refrigerator temperature, humidity and air pressure are adjusted to reduce the ingress of moisture into the refrigerator and the loss of cold air in the refrigerator caused by the opening and closing of the refrigerator door.

[0033] Specifically, in this embodiment, the data processing module will divide a day into 144 segments on average, that is, the number of times the refrigerator door is opened and closed will be recorded every 10 minutes, and a weight will be assigned to each time segment based on the number of times the refrigerator door is opened and closed and the Ebbinghaus forgetting curve. The greater the weight, the higher the probability that the refrigerator door is opened in the time segment, so that the refrigeration adjustment module adjusts the refrigeration in advance, slightly reducing the refrigeration effect, and the temperature of the refrigerator is increased by 1 to 2 degrees Celsius in the time segment before the time segment of the current time segment. Based on the ideal gas state equation, when the temperature in the refrigerator rises, the air pressure will also rise, and the air pressure auxiliary module will adjust so that when the refrigerator door is opened, the cold air in the refrigerator will hinder the outside air from entering the refrigerator, so as to greatly reduce the outside moisture from entering the refrigerator, and the slight increase in the temperature in the refrigerator will reduce the energy consumption caused by the outflow of cold air when the refrigerator door is opened, further reducing the energy consumption of the refrigerator.

[0034] Specifically, the calculation formula for the time segment weight is: ;

[0035] in, A sequence representing a time segment of a day, Indicates that on each day The time segment weight of each time period, Indicates the total number of days the data processing module remembers the refrigerator door switch. In this embodiment, it is determined that the total number of days the refrigerator door switch is remembered is 30 days. Indicates in Tiandi The probability of maintaining the refrigerator door open behavior in a time segment is calculated based on the exponential decay curve fitted by the number of opening and closing times of the refrigerator door in different time segments. Indicates in Tiandi The number of times the refrigerator door is opened in a time segment. It should be understood that all variables mentioned in the formula are pure numbers and do not involve physical unit conversion.

[0036] By setting the time segment weights, the open state of the freezer is abstracted into a probability. The freezer door can be in both open and closed states at the same time. The power of the compressor in the refrigeration module is adjusted based on the size of the probability, thereby changing the temperature in the freezer. The degree of temperature increase in the freezer is proportional to the probability. Therefore, when the freezer door is open, the temperature loss caused by the overflow of cold air is reduced, thereby reducing the energy waste of the freezer.

[0037] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A refrigerator adaptive energy-saving control system based on multimodal perception, characterized in that: It includes data acquisition module, refrigeration regulation module, wind auxiliary module, air pressure auxiliary module and data processing module, among which: The data acquisition module includes a temperature sensor, a thermal camera, a light camera, a humidity sensor, and an air pressure sensor, which are used to capture the temperature distribution, humidity, and air pressure inside the refrigerator and send this data information to the data processing module; The refrigeration regulation module includes a compressor, a condenser, an expansion valve, a filter, an evaporator, an accumulator, and a refrigerant. It is used to continuously convert the refrigerant in the refrigeration regulation module between liquid and gaseous states, thereby cooling the interior of the refrigerator. The refrigeration speed is controlled by the data processing module. The wind auxiliary module includes a fan with an adjustable angle, which is used to circulate the air inside the freezer so that the refrigerated air in the freezer can flow fully throughout the freezer, thereby improving the refrigeration efficiency of the freezer; The air pressure auxiliary module includes an electrically controlled air pressure valve, an air pressure pump, and an air storage chamber, which is used to adjust the air pressure inside the refrigerator to change the relative humidity and absolute humidity inside the refrigerator. When the relative humidity in the air storage chamber does not reach 100%, the relative humidity and temperature of the air in the air storage chamber remain consistent with those inside the refrigerator; The data processing module is used to receive data information from the data acquisition module, perform feature extraction and feature splicing on the data information, and obtain the characteristic values ​​of the state inside the refrigerator through the neural network, so as to timely adjust the temperature through the refrigeration adjustment module and the wind auxiliary module, and adjust the air pressure and humidity through the air pressure auxiliary module.

2. The refrigerator adaptive energy-saving control system based on multimodal perception according to claim 1 is characterized in that: The temperature sensor includes an external sensor and a built-in sensor. The external sensor is used to detect the external ambient temperature outside the refrigerator, and the built-in sensor is used to detect the refrigeration temperature inside the refrigerator. By comparing the temperature difference between the two built-in and external temperature sensors, the data processing module is provided with a temperature adjustment basis, so that the data processing module can adjust the refrigeration effect in time to prevent the temperature difference between the inside and outside of the refrigerator from being too small, resulting in frequent starting and stopping of the compressor or low-frequency operation, which increases the power consumption of the compressor.

3. The adaptive energy-saving control system for refrigerators based on multimodal perception according to claim 2, characterized in that: The wind-assisted module uses a fan to control the temperature of the freezer, blowing the cold air generated by heat exchange at the evaporator of the refrigeration regulation module into the freezer, accelerating the cold air circulation in the freezer. The fan will push all the cold air around the evaporator into the interior of the freezer, making the cold air evenly distributed in the freezer.

4. The adaptive energy-saving control system for refrigerators based on multimodal perception according to claim 3 is characterized in that: The electrically controlled air pressure valve in the air pressure auxiliary module is used to control the air inside the freezer and the air from the external environment to enter and exit the air storage chamber. The air storage chamber separates the interior of the freezer from the external environment, thereby avoiding direct exchange of gas with the external environment when adjusting the air pressure inside the freezer, which would cause a large loss of temperature inside the freezer and an increase in humidity.

5. The adaptive energy-saving control system for refrigerators based on multimodal perception according to claim 4 is characterized in that: During cooling, the data processing module will splice the light image and the thermal image into a four-channel image containing three color channels and a thermal channel after a pre-processing process of resolution adjustment, image cropping, and normalization. The neural network of the data processing module obtains the image features of the four-channel image through the method of convolution followed by linearization, and expands it to obtain a one-dimensional image feature matrix. The one-dimensional image feature matrix is ​​merged with the data information of the temperature sensor, humidity sensor, and pressure sensor to form a one-dimensional multimodal matrix. The one-dimensional multimodal matrix obtains the refrigerator adjustment data features through the linear layer operation and nonlinear transformation of the neural network. The refrigerator adjustment data features include the target power of the compressor, the target power of the fan, and the target air pressure in the refrigerator. Adjusting the power of the compressor can adjust the refrigeration module in the refrigerator. The cooling effect, the greater the power of the compressor, the stronger the cooling effect of the refrigeration module, by adjusting the fan power to adjust the flow of cold air in the freezer, thereby adjusting the distribution of cold air in the freezer, by controlling the flow of air in the air storage cavity to adjust the air pressure, relative humidity and absolute humidity in the freezer, when the air pressure in the freezer increases, the moisture density in the freezer increases, thereby increasing the air humidity in the freezer before the saturated water vapor pressure is reached, the absolute humidity and relative humidity in the freezer both increase, at this time the high humidity air in the freezer will inhibit the evaporation of moisture in the food, when the freezer door is opened, under the influence of the air pressure difference, the water vapor is quickly released outwards, and then the air pressure in the freezer is restored, thereby reducing the relative humidity and absolute humidity, the cold air will stay around the freezer door, thereby inhibiting the high humidity air from the outside from entering the freezer, keeping the freezer dry.

6. The adaptive energy-saving control system for refrigerators based on multimodal perception according to claim 5, characterized in that: When local air circulation in the refrigerator is difficult due to the placement of food in the refrigerator, the position of the food in the refrigerator can be collected through a light-sensing camera, and the surface temperature of the food in the refrigerator can be collected through a thermal camera. Based on the temperature conditions in the densely populated areas, the visual neural network of the data processing module can quickly determine the obstructed position of air circulation, and timely adjust the fan angle of the wind auxiliary module, so that cold air can circulate more smoothly from the densely populated areas, thereby improving the refrigeration effect inside the refrigerator, reducing the power required for refrigeration, and ensuring longer storage of food.

7. The adaptive energy-saving control system for refrigerators based on multimodal perception according to claim 6, characterized in that: When the temperature of a local area in the refrigerator needs to be lower than other areas, the position of the food in the refrigerator can be captured by a light-sensing camera, so that the cold air guided by the fan can actively blow to the area where local low temperature is required, causing the temperature of the required area in the refrigerator to be lower than other areas. The air pressure in the refrigerator is adjusted by storing or releasing air in the air storage cavity of the air pressure auxiliary module in the refrigerator, so that the air pressure in the refrigerator is increased and the moisture density in the refrigerator is increased, thereby adjusting the relative humidity and absolute humidity of the air in the refrigerator, inhibiting the transpiration of food, reducing the evaporation of water caused by cold air blowing in the area where local low temperature is required, and ensuring the quality of food in the area where local low temperature is required.

8. The refrigerator adaptive energy-saving control system based on multimodal perception according to claim 7, characterized in that: Thermal images are collected by a thermal camera and sent to a data processing module for analysis. Based on the low temperature distribution inside the freezer and a convolutional neural network, the unformed ice layer on the inner wall of the freezer can be identified, and the position of the unformed ice layer inside the freezer can be obtained. The wind auxiliary module is controlled by the data processing module to aim the fan at the unformed ice layer. Only a small amount of ice in the unformed ice layer is attached to the inside of the freezer. The air flows on the surface of the unformed ice layer, thereby accelerating the sublimation of the surface of the unformed ice layer without affecting the refrigeration effect. When the air passes over the surface of the unformed ice layer, it transfers heat to the unformed ice layer, which has a lower temperature due to the operation of the evaporator, accelerating the disappearance of the unformed ice layer while reducing the temperature of the flowing air, thereby utilizing the unformed ice layer to share the refrigeration burden of the refrigeration regulation module, reducing the power consumption of the refrigeration regulation module, and improving the refrigeration quality.

9. The refrigerator adaptive energy-saving control system based on multimodal perception according to claim 8, characterized in that: When the wind-assisted module sublimates the unformed ice layer, moisture from the unformed ice layer enters the air inside the freezer, causing the humidity in the freezer to increase. To prevent excessive humidity in the freezer from causing mold growth and food spoilage, the data processing module can control the air pressure assist module based on the humidity information data collected by the humidity sensor. When the humidity in the freezer is normal, the data processing module will lower the air pressure in the freezer to reduce the moisture density inside the freezer, thereby accelerating the sublimation process on the surface of the ice layer. As the ice layer sublimates, the humidity in the freezer gradually approaches saturation. At this time, the data processing module controls the electronically controlled air pressure valve of the air pressure assist module to discharge the high-humidity freezer air into the air storage chamber. The air storage chamber is pressurized by an air pump, causing the moisture in the air storage chamber to exceed the saturation water vapor pressure. This causes the moisture to condense into water droplets and be discharged from the device before it condenses into ice. At this time, the air in the air storage chamber still has a relatively low temperature and can be used to continue to maintain the humidity in the freezer, thereby reducing energy waste during the de-icing process.

10. The refrigerator adaptive energy-saving control system based on multimodal perception according to claim 9, characterized in that: The data processing module divides a day into multiple time segments and records the number of times the refrigerator door is opened and closed in different time segments. An exponential decay curve that gradually tends to 0 is fitted based on the number of times the refrigerator door is opened and closed in different time segments. This exponential decay curve is used to indicate the possibility that refrigerator users will form a habit of opening and closing the refrigerator door at fixed time segments every day. The exponential decay curve is used to assign weights to the user's behavior of opening and closing the refrigerator, thereby predicting the time segments when the refrigerator user opens and closes the refrigerator. Based on the prediction, the refrigerator temperature, humidity and air pressure are adjusted to reduce the ingress of moisture into the refrigerator and the loss of cold air in the refrigerator caused by the opening and closing of the refrigerator door.

Images