Refrigerator temperature control system and method and refrigerator

Abstract

The invention provides a refrigerator temperature control system and method and a refrigerator, and relates to the technical field of refrigerator temperature control. The refrigerator temperature control system comprises an acquisition terminal and a processing terminal, and the acquisition terminal is used for acquiring multiple temperature and humidity data of a refrigerator in a refrigeration path and an environment where the refrigerator is located in a historical period and label data of the environment where the refrigerator is located; the processing terminal is used for analyzing change characteristics of the temperature and humidity data to obtain running state data of the refrigerator; a preset regression model is trained and tested, so that a regression relation between the various data and the historical temperature in the cabinet in the corresponding insensitive zone is constructed; inputting the current environment data of the refrigerator into the trained regression model, and obtaining a target in-cabinet temperature of the refrigerator based on an output result of the regression model; and controlling the refrigeration temperature of the refrigerator in the future time period according to the target temperature in the refrigerator. According to the technical scheme, different temperature control strategies are made according to different environments, and self-adaptive control over the temperature of the refrigerator in different scenes is achieved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of refrigerator temperature control, and in particular to a refrigerator temperature control system, method and refrigerator. BACKGROUND

[0002] In today's market, refrigerators have been widely used in food, medicine and other fields, and the temperature control accuracy will directly affect the preservation effect and energy efficiency of the goods. At present, in order to achieve the effect of intelligent temperature control, there are two kinds of mainstream solutions: the first kind is to control the temperature based on a strict rule base, and the start and stop of the compressor and the evaporative condenser fan are controlled by preset temperature threshold and temperature hysteresis range; the other kind is to adjust the compressor power by using PID (Proportional-Integral-Derivative) algorithm on the basis of the first kind of scheme, so as to realize dynamic temperature control.

[0003] Looking at the mainstream solutions, the existing technology mostly uses a temperature control system based on a rule engine, but it is difficult to dynamically adapt to the complex temperature control scene of the refrigerator, and the problem that the existing refrigerator cannot realize temperature self-adaptation has not been fundamentally solved. SUMMARY

[0004] The present disclosure provides a refrigerator temperature control system, method and refrigerator, which can realize self-adaptive control of the temperature of the refrigerator in different scenes.

[0005] The technical solution of the present disclosure is implemented as follows: In a first aspect, the present disclosure provides a refrigerator temperature control system, which comprises: A collection terminal is configured to collect a plurality of temperature and humidity data of a refrigeration path and an environment in which the refrigerator is located in a historical period, and label data of the environment; A processing terminal is connected to the collection terminal, and the processing terminal is configured to analyze the change characteristics of the plurality of temperature and humidity data to obtain running state data of the refrigerator; According to the plurality of temperature and humidity data, the label data and the running state data, a preset regression model is trained and tested, so as to construct a regression relationship between the various data in the corresponding insensitive band and the historical temperature in the cabinet; The current environment data of the refrigerator is input into the trained regression model, and the target cabinet temperature of the refrigerator is obtained based on the output result of the regression model; The refrigeration temperature of the refrigerator in a future period is controlled according to the target cabinet temperature, and the regression model is retrained after the future period to implement temperature regulation.

[0006] In an optional embodiment, the collection terminal comprises: A humidity sensor is connected to a first input end of the processing terminal, and the humidity sensor is installed on the refrigerator; the humidity sensor is configured to collect the environmental humidity of the refrigerator and output externally; A first temperature sensor is connected to a second input terminal of the processing terminal, and is installed on the refrigerator. The first temperature sensor is used to collect a first temperature of the environment where the refrigerator is located, and output the first temperature to the outside. A second temperature sensor is connected to a third input terminal of the processing terminal, and is installed inside the refrigerator and communicates with the storage space of the refrigerator. The second temperature sensor is used to collect a second temperature of the storage space inside the refrigerator, and output the second temperature to the outside. A third temperature sensor is connected to a fourth input terminal of the processing terminal, and is installed on the refrigerator at a position where a condenser is located. The third temperature sensor is used to collect a third temperature on the side of the condenser of the refrigerator, and output the third temperature to the outside. A fourth temperature sensor is connected to a fifth input terminal of the processing terminal, and is installed on the refrigerator at a position where an evaporator is located. The fourth temperature sensor is used to collect a fourth temperature on the side of the evaporator of the refrigerator, and output the fourth temperature to the outside.

[0007] In an optional embodiment, the processing terminal comprises: A server is configured to be communicatively connected to the collection terminal. The server is used to process various data output by the collection terminal to obtain a target cabinet temperature of the refrigerator. A controller is configured to be communicatively connected to the server. The controller is installed inside the refrigerator. The controller is used to control the refrigeration temperature of the refrigerator in a future period according to the target cabinet temperature.

[0008] In a second aspect, the embodiments of the present application also provide a refrigerator temperature control method, which is applied to the control system of any one of the first aspect, and the method comprises: Obtaining a plurality of temperature and humidity data of the refrigerator in a historical period in a refrigeration path and an environment where the refrigerator is located, and label data of the environment; Analyzing the change characteristics of the plurality of temperature and humidity data to obtain operation state data of the refrigerator; Training and testing a preset regression model according to the plurality of temperature and humidity data, the label data, and the operation state data, so as to construct a regression relationship between the various data in a corresponding insensitive band and a historical cabinet temperature; Inputting current environment data of the refrigerator into the regression model that has been trained, and obtaining a target cabinet temperature of the refrigerator based on an output result of the regression model; Controlling the refrigeration temperature of the refrigerator in a future period according to the target cabinet temperature, and retraining the regression model to implement temperature control after the future period.

[0009] In an optional embodiment, analyzing the change characteristics of the plurality of temperature and humidity data to obtain the operation state data of the refrigerator comprises: Based on the change of each temperature and humidity data over time, data curves for each temperature and humidity data are obtained. Among them, multiple data curves include the ambient humidity curve and the first temperature curve of the environment where the freezer is located, the second temperature curve of the storage space inside the freezer, the third temperature curve of the condenser side of the freezer, and the fourth temperature curve of the evaporator side of the freezer. The relationship between the ambient humidity curve, the first temperature curve, the second temperature curve, the third temperature curve, and the fourth temperature curve is analyzed to obtain the opening and closing frequency and operating load rate of the freezer. The frequency of door opening and closing and the operating load rate are determined as the operating status data of the freezer.

[0010] In an optional embodiment, before training and testing the preset regression model, the method further includes: When the freezer reaches a target state that meets multiple preset conditions, obtain the cooling power consumption of the freezer during the preset time of operation in the target state and the temperature change of the internal storage space of the freezer. Based on the refrigeration power consumption, temperature change, and the mass of newly added items in the storage space inside the freezer, the specific heat capacity data of the stored items inside the freezer are obtained. Configure specific heat capacity data as the target data for training and testing the regression model.

[0011] In one optional embodiment, determining that the freezer has reached a target state that satisfies multiple preset conditions includes: It is determined that the freezer is not running in energy-saving mode, the current door status is closed, and the door has completed the preset number of pull-down temperature cycles after the last closing. If so, then confirm that the freezer is operating in the target state.

[0012] In an optional embodiment, before training and testing the preset regression model, the method further includes: During the nighttime period, the current customer flow of the freezer is determined based on the second temperature change in the internal storage space of the freezer. When the current passenger flow is lower than the preset value, the regression model is trained and tested.

[0013] In an optional embodiment, before training and testing the preset regression model, the method further includes: Determine the working scenario characteristics of the freezer based on at least one of the following data: temperature and humidity data, label data, and operating status data. Based on the characteristics of the work scenario, corresponding training weights are configured for various types of data to improve the prediction accuracy of the regression model for the target cabinet temperature in the corresponding scenario.

[0014] Thirdly, embodiments of the present invention also provide a freezer, which includes any of the control systems described in the first aspect.

[0015] The present application has the following beneficial effects: The technical scheme of the present application comprises a collection terminal and a processing terminal. The collection terminal is used to collect a plurality of temperature and humidity data of a refrigeration path and an environment in which a refrigerator is located in a historical period, and label data of the environment, to provide original time sequence data reflecting changes in the refrigeration path and the environment. The processing terminal is used to analyze change characteristics of the plurality of temperature and humidity data to obtain running state data of the refrigerator. Since the plurality of temperature and humidity data, the label data and the running state data represent running characteristics of a use state and a disturbance of the refrigerator from different dimensions, a regression model is trained and tested, so that regression relationships are constructed between various data in corresponding insensitive bands and historical temperatures in the refrigerator, and the regression relationships are constructed and verified by taking multi-source characteristics as training basis. Current environment data of the refrigerator are input into the trained regression model, and a target temperature in the refrigerator is obtained based on an output result of the regression model. The refrigeration temperature of the refrigerator in a future period is controlled according to the target temperature in the refrigerator, and the regression model is retrained after the future period to implement temperature control. The technical scheme collects a large amount of data to determine the environment in which the refrigerator is located, and formulates different temperature control strategies according to different environments, so that the regression model obtains contributions of different data characteristics to the temperature in the insensitive bands, thereby correcting prediction errors and adjusting model parameters in a closed loop, to form a continuous, self-adaptive prediction, execution and update cycle. The technical scheme fundamentally avoids the limitation of single control of the compressor frequency by the PID algorithm, and is more efficient and has more reasonable control logic, so that the refrigerator realizes self-adaptive control of the temperature in different scenes. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A structural schematic diagram of a refrigerator temperature control system provided by an embodiment of the present application; Figure 2 A flowchart of a refrigerator temperature control method provided by an embodiment of the present application; Figure 3 A first principle diagram of training of a regression model provided by an embodiment of the present application; Figure 4 A second principle diagram of training of a regression model provided by an embodiment of the present application; Figure 5 An insensitive band diagram of a temperature in the refrigerator and an environment temperature provided by an embodiment of the present application; Figure 6 An insensitive band diagram of a temperature in the refrigerator and a fourth temperature of an evaporation side provided by an embodiment of the present application; Figure 7 An implementation architecture diagram of a refrigerator temperature control provided by an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the specific embodiments, structures, features and effects of a refrigerator temperature control system, method and refrigerator according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0019] For the temperature control strategy of the refrigerator, the first one uses strict rule base to control the temperature, for example, based on the sum of the set temperature and the temperature hysteresis to obtain the temperature threshold, when the temperature in the cabinet is greater than or equal to the temperature threshold, start refrigeration; when the temperature in the cabinet is less than or equal to the set temperature minus 1℃, stop refrigeration, and turn off the compressor. The second scheme is based on the first scheme superimposed PID control, which can only be applied to refrigerators equipped with variable frequency compressors, and has limitations in application. At present, although the refrigerator has been widely used in intelligent temperature controller, but it still depends on artificial experience to set fixed temperature threshold, and there are the following problems: 1. Poor environmental adaptability: when the external environment temperature fluctuates and the door opening and closing frequency changes, the fixed threshold value is easy to cause the internal temperature to exceed the limit, and cannot dynamically adapt to the change of the environment.

[0020] 2. Energy waste: excessive refrigeration or unreasonable temperature hysteresis setting leads to power loss.

[0021] 3. Lack of intelligence: lack of comprehensive analysis capability for multi-source data (such as load, item type).

[0022] In summary, the prior art cannot realize temperature adaptive control of the refrigerator in different scenarios when implementing temperature control of the refrigerator. The specific scheme of a refrigerator temperature control system, method and refrigerator provided by the present application is described in detail below in combination with the drawings.

[0023] Please refer to Figure 1 , Figure 1 The structure diagram of a refrigerator temperature control system provided by one embodiment of the present application is shown. The control system includes a collection terminal 1 and a processing terminal 2. The collection terminal 1 is connected to the processing terminal 2 to transmit the collected multi-type data to the processing terminal 2.

[0024] The acquisition terminal 1 can be configured as a plurality of temperature and humidity sensors for collecting a plurality of temperature and humidity data of the refrigeration path and the environment of the refrigerator in the historical period and the label data of the environment. The acquisition terminal and the processing terminal can be built based on the structure of an embedded system, and a signal conditioning circuit and a data transmission circuit can also be configured at the output end of each temperature and humidity sensor. The signal conditioning circuit is used for filtering the original signal of each temperature and humidity data, and the data transmission circuit is used for transmitting the filtered original signal through a preset protocol. For example, a low-pass filter is used to reduce high-frequency noise interference in the acquisition process, and an amplification circuit is used for signal amplitude conditioning to meet the input requirements of an analog-to-digital converter. An analog-to-digital conversion unit can be provided on the acquisition terminal to convert the filtered and amplified analog signal into a digital signal, which is then transmitted to the processing terminal 2 through the data bus or wireless communication module (such as RS485, Bluetooth or Wi-Fi) of the data transmission circuit.

[0025] The processing terminal 2 is used to analyze the change characteristics of the plurality of temperature and humidity data to obtain the running state data of the refrigerator, which represents the door opening and closing frequency and load rate of the refrigerator. According to the plurality of temperature and humidity data, the label data and the running state data, the preset regression model is trained and tested, so that the various types of data construct a regression relationship with the historical temperature in the cabinet in the corresponding insensitive band. The regression model can be configured as an SVR (Support Vector Regression) model. Of course, it can also be configured as a decision tree model or a random forest model, which is not specifically limited here. The current environment data of the refrigerator is input into the trained regression model, and the target cabinet temperature of the refrigerator is obtained based on the output result of the regression model. The target cabinet temperature is used to control the refrigeration temperature of the refrigerator in the future period, and the regression model is retrained after the future period to implement temperature control.

[0026] The processing terminal can be integrated into the main control board of the refrigerator temperature controller, and its hardware structure includes a microprocessor, a memory and an interface circuit. The microprocessor is used for data analysis and regression model training operation, and the memory is used for saving various types of data and model parameters in the historical period. The processing terminal can be further configured with a separate power stabilizing circuit and a radiator to prevent calculation errors caused by power fluctuations or heat accumulation. The processing terminal can also be connected to the control circuit of the compressor, evaporator fan and condenser fan of the refrigerator at its output end, and execute the control instruction of the target cabinet temperature through PWM speed regulation or relay switching. Through the above hardware configuration, the acquisition terminal realizes high-precision acquisition and anti-interference processing of environmental data, and the processing terminal realizes integrated execution of data operation and temperature control, thereby ensuring that the entire temperature control system can stably and accurately implement temperature self-adaptive control in different operating environments.

[0027] In practical applications, if the local temperature of the local position of the refrigeration path of the refrigerator is collected, the prediction accuracy of the subsequent target cabinet temperature may be affected. Based on this, in a specific embodiment, the collection terminal includes a humidity sensor, a first temperature sensor, a second temperature sensor, a third temperature sensor, and a fourth temperature sensor. The model of each temperature and humidity sensor can be selected based on actual needs, for example, each temperature sensor uses 18B20 to implement temperature data collection. By setting each temperature and humidity sensor, data corresponding to each position can be obtained and stored and updated at a predetermined period. Since the heat exchange process of the refrigerator involves four typical heat nodes, namely the external environment, the storage space, the condensing side (i.e., the condenser side), and the evaporating side (i.e., the evaporator side), the dynamic changes of the temperatures of each node directly reflect the efficiency of the refrigeration cycle and the change law of the thermal load.

[0028] The humidity sensor is connected to the first input end of the processing terminal, and the humidity sensor is installed on the refrigerator and located outside the storage space. The humidity sensor is used to collect the environmental humidity of the refrigerator and output to the processing terminal. High humidity environment will cause the heat transfer efficiency of the refrigerator evaporator to decrease, and the influence of air water content on the refrigeration and frosting trend of the refrigerator can be obtained through the environmental humidity, so the humidity data is an important parameter for judging the environmental load.

[0029] The first temperature sensor is connected to the second input end of the processing terminal, and the first temperature sensor is installed on the refrigerator. The first temperature sensor is used to collect the first temperature of the environment in which the refrigerator is located and output to the processing terminal. The first temperature is the boundary condition for temperature control of the refrigerator, and is the basic data for calculating the heat dissipation capacity of the refrigeration system and the external heat penetration.

[0030] The second temperature sensor is connected to the third input end of the processing terminal, and the second temperature sensor is installed inside the refrigerator and communicates with the storage space of the refrigerator. The second temperature sensor is used to collect the second temperature of the storage space inside the refrigerator and output to the processing terminal. The change of the second temperature can determine whether the refrigeration of the refrigerator meets the set requirements, and can also determine the temperature fluctuation amplitude in the cabinet, thereby providing reference data for the output of the regression model and temperature control feedback.

[0031] The third temperature sensor is connected to the fourth input end of the processing terminal, and the third temperature sensor is installed at the position of the condenser of the refrigerator. The third temperature sensor is used to collect the third temperature at the condenser side of the refrigerator and output to the processing terminal. The third temperature represents the thermal state of the high-temperature end of the refrigeration cycle of the refrigerator, and the difference between the third temperature and the ambient temperature can represent the condensing efficiency and the heat dissipation condition. When the ambient temperature is high or the ventilation is poor, the temperature rise of the condenser will increase the energy consumption, so this data can reflect the running performance of the refrigerator under high load.

[0032] The fourth temperature sensor is connected to a fifth input end of the processing terminal, and is installed at a position of the evaporator of the refrigerator. The fourth temperature sensor is used to collect a fourth temperature at the side of the evaporator of the refrigerator and output the fourth temperature to the processing terminal. The fourth temperature represents the temperature at the low-temperature end of the refrigeration cycle, and the change of the value of the fourth temperature corresponds to the temperature characteristics of the suction end of the compressor, can reflect the heat absorption effect and the evaporation temperature fluctuation of the evaporator, and can indirectly reflect the refrigerant flow and the evaporation strength of the refrigerator.

[0033] Through the cooperative arrangement of the four temperature sensors and the humidity sensor, the temperature control system can establish a temperature and humidity distribution database at four key thermal nodes of the external environment, the condenser side, the evaporator side and the storage space, and obtain temperature gradient information throughout the refrigeration path. The collected multi-dimensional data represent the heat cycle state of the refrigerator, and also provide sufficient data basis for the processing terminal to analyze the running state data and establish a regression relationship, so that the refrigerator can realize accurate temperature control and self-adaptive adjustment under different environmental and load conditions.

[0034] After collecting the above-mentioned characteristic data, the data is stored in the memory in the form of a two-dimensional data group, with the time sequence of collection as the row label and the data value of each type of data as the column label, waiting to be called by the processing terminal. It should be noted that due to the space problem of the circuit board on the refrigerator, most of the temperature controllers on the market do not even have a flash memory chip, and the capacity of the flash memory equipped in some high-end temperature controllers is also about 4Mb. Therefore, the memory capacity limitation of the temperature controller should be considered when designing the data storage logic. Taking 4Mb as an example, according to the actual measurement, if the data is stored once every half hour, the flash memory can support three months of data storage. After comprehensive consideration, the total amount of collected data is set to 72 hours, and the data is collected every ten minutes. This has the advantages of having enough data for machine learning of the regression model, and not affecting the normal historical data storage function of the temperature controller. After the 72-hour data storage capacity is full, the newly collected data will gradually replace the data with earlier time labels, ensuring that the data stored in the memory is the latest 72-hour data, thereby ensuring the adaptability of the data to the seasonal environment.

[0035] Since the training of the regression model requires a device with a large amount of calculation power, but the hardware configured by the refrigerator is to balance energy consumption and cost, and mostly uses hardware with low calculation power. Based on this, in a specific embodiment, the processing terminal includes a server and a controller, and the server can be configured as a cloud server or a local server through network communication.

[0036] The server is configured to be in communication connection with the collection terminal, and the server is used to process each item of data output by the collection terminal to obtain the target cabinet temperature of the refrigerator. The server has high computing power and can run a support vector machine regression algorithm to fit and predict multi-source feature data, and then accurately output the target cabinet temperature of the refrigerator.

[0037] The controller is configured to be in communication connection with the server, and the controller is installed in the interior of the refrigerator. The controller is used to control the refrigeration temperature of the refrigerator in the future period according to the target cabinet temperature. The controller realizes accurate control of the internal temperature of the refrigerator by adjusting the operation of the compressor, the evaporator fan and the condenser fan. The distributed design of the server and the controller makes the data calculation and the execution control independent of each other, which not only ensures high-performance data analysis at the calculation level, but also ensures the real-time performance and stability of the device end control.

[0038] Based on the same technical concept as the control system, the embodiment of the present application also provides a refrigerator temperature control method. Please refer to Figure 2 , Figure 2 The flowchart of the refrigerator temperature control method is shown in the figure. The method is applied to any of the above control systems. The control method can be run based on the processing terminal of the control system. The control method comprises the following steps: S11, obtaining a plurality of temperature and humidity data of the refrigerator in the historical period in the refrigeration path and the environment, and label data of the environment.

[0039] Specifically, the historical period can be configured based on the temperature control accuracy requirement of the refrigerator, for example, 72 hours before training the regression model. Each temperature and humidity data represents the temperature or humidity of the refrigerator at the corresponding node in the refrigeration path or in the environment, which can be obtained by various sensors; the label data can represent the season and / or geographical location of the environment. When the label data represents the season, the season label is calculated by the built-in clock of the temperature controller according to the time and date; when the label data represents the geographical location, the geographical location coordinates are obtained by the built-in communication unit of the temperature controller from the network base station. A 72-hour sliding window can be used to save the obtained recent data, so as to balance the storage limit and the training sample size requirement of the regression model. After obtaining various data through various sensors, clock and network base station coordinates, the method proceeds to step S12.

[0040] S12, analyzing the change characteristics of the plurality of temperature and humidity data to obtain the running state data of the refrigerator.

[0041] Specifically, multiple temperature and humidity data can represent the change of the running state of the refrigerator, for example, whether the door is opened can be determined based on the change characteristics of the temperature in the cabinet, and the frequency of opening and closing the door can be determined by counting the number of times the door is opened. Of course, the opening and closing of the door can also be identified based on the temperature mutation and recovery curve to count the opening and closing frequency. Since the temperature and humidity data change over time, time domain features can be extracted and event detection can be performed based on the temperature and humidity data, for example, the first order difference, short time maximum value, minimum value, recovery time and energy consumption of the running period of the temperature / humidity in the cabinet, etc. The running state data is characterized by these data, and the running load rate is estimated based on the recovery time and amplitude of the evaporation side temperature in the compressor cycle. The extracted running state data is added to the feature set in the form of standardized numerical values as a reflection of the refrigerator running disturbance and load index.

[0042] For example, step S12 includes sub-steps S12-1 to S12-3, which are described as follows: S12-1, according to the change amount of each temperature and humidity data in the time dimension, obtain the data curve of each temperature and humidity data, wherein the multiple data curves include the environmental humidity curve and the first temperature curve of the environment where the refrigerator is located, the second temperature curve of the internal storage space of the refrigerator, the third temperature curve of the condenser side of the refrigerator, and the fourth temperature curve of the evaporator side of the refrigerator. Each temperature curve is a curve of corresponding temperature data changing with time. By analyzing each temperature curve, the running event occurring during the running of the refrigerator can be analyzed.

[0043] S12-2, analyze the change relationship between the environmental humidity curve, the first temperature curve, the second temperature curve, the third temperature curve and the fourth temperature curve to obtain the opening and closing frequency and the running load rate of the refrigerator. When the refrigerator door is opened, the temperature in the cabinet will instantaneously rise due to the entry of external hot air; after the door is closed, the evaporation side temperature will rapidly decrease to restore the set temperature. Therefore, the first derivative of the fourth temperature curve representing the change of the evaporation side temperature data in the continuous period can be calculated. When a positive jump in temperature is detected in a short period of time, for example, the temperature change amplitude is greater than 2°C, and the difference with the ambient temperature decreases, it is determined that the door is opened. After the door is closed, the first derivative changes from positive to negative and returns to stable, which is defined as the closing event. The opening and closing frequency in a unit of time is obtained by counting the occurrence period of the opening and closing event. Since the temperature change caused by opening and closing the door is simultaneously represented in the environmental humidity curve, the first temperature curve, the second temperature curve and the third temperature curve, the change characteristics represented by the fourth temperature curve can be verified by other curves to accurately analyze the opening and closing frequency.

[0044] Further, the operation load of the refrigerator is related to the heat balance of the refrigeration system. The greater the operation load, the higher the compressor start frequency and the operation time. The load level can be analyzed by characterizing the fourth temperature curve of the evaporation side temperature change and the third temperature curve of the condensation side temperature change. For example, in each compressor operation cycle, the operation time from the compressor start time to the stop time is measured , the compressor stop time is the operation time when the temperature in the cabinet returns to the set target value; at the same time, the temperature drop amplitude of the evaporator in the cycle is calculated ; the effective refrigeration power per unit time is defined as the ratio of the cycle operation time to the temperature drop amplitude, that is , is the effective refrigeration power; the operation load rate corresponding to the operation cycle is calculated by the ratio of the historical no-load reference power to the effective refrigeration power , that is , . The calculation method is based on the energy balance equation of the refrigeration system, and it is considered that the refrigeration capacity of the compressor is constant under the same conditions, so the longer the temperature recovery time and the lower the refrigeration efficiency, that is, the greater the operation load.

[0045] It should be noted that the calculation of the load rate can also be implemented by using a fitting model such as linear regression or recurrent neural network, based on the corresponding curves to obtain the evaporator temperature drop rate, the condenser temperature rise amplitude and the ambient temperature gradient, taking the evaporator temperature drop rate, the condenser temperature rise amplitude and the ambient temperature gradient as input features, fitting the functional relationship between the compressor operation cycle and the refrigeration rate, and thus estimating the operation load rate of the refrigerator per unit time.

[0046] S12-3, the door opening and closing frequency and the operation load rate are determined as the operation state data of the refrigerator. The door opening and closing frequency represents the user's use behavior of the refrigerator, and its change directly affects the heat exchange between the refrigerator and the external environment. The operation load rate reflects the matching between the internal storage heat capacity of the refrigerator and the output capacity of the refrigeration system, and is a parameter for describing the internal thermal load level of the equipment. These two types of parameters can dynamically reflect the thermal disturbance degree and refrigeration intensity of the refrigerator under different working conditions, and are key state quantities directly affecting the temperature change in the cabinet.

[0047] At this point, the calculation of the operation state data is completed, and step S13 is entered.

[0048] S13, according to the plurality of temperature and humidity data, the label data and the operation state data, the pre-set regression model is trained and tested, so that the regression relationship between the various data and the historical temperature in the cabinet is constructed in the corresponding insensitive band.

[0049] Specifically, the step includes distinguishing the training set from the test set, and training the regression model using the training set data. To adapt the temperature control of the refrigerator to seasonal changes, such as increased ambient humidity in summer, the regression model can be retrained every 24 hours using the latest 72 hours of data to adapt the constructed regression relationship to the new working scenario of the refrigerator. The input data of the regression model includes the ambient humidity of the environment where the refrigerator is located, the first temperature representing the ambient temperature, the third temperature on the condenser side, the fourth temperature on the evaporator side, the door opening frequency and the operating load rate of the refrigerator, as well as the season label and the geographic location coordinates. The output of the regression model is the second temperature representing the temperature inside the cabinet.

[0050] When dividing the training set and the test set, all the collected data is divided into the training set and the test set according to a ratio of 70% to 30%. The training set accounts for 70% to ensure that the model has enough data for training, and because the training set contains not only input data features but also output data features, the model can find the corresponding relationship according to the input and output data features, thereby completing the training. The remaining 30% of the data is used as the test set. Because the test set has the same data structure as the training set and also contains output features, the input features in the test set can be used to test the model to obtain the prediction results, which are then compared with the output features to calculate the prediction accuracy of the model. When the prediction accuracy is greater than a predetermined accuracy threshold, it indicates that the regression model can accurately predict the temperature inside the cabinet, and the training of the regression model is completed.

[0051] Because the working scenario of the refrigerator is diversified, the influence of different working scenarios on the refrigeration performance is different. For example, when the geographic location coordinates show that the refrigerator is working in a tropical region, the season is not clearly distinguished in the tropical region, and the winter is also relatively hot, similar to the summer. If the regression model treats all features equally during the training phase, it will result in prediction deviation in a specific scenario.

[0052] Based on this, in a specific embodiment, before training and testing the predetermined regression model, the method further includes: According to at least one of the plurality of temperature and humidity data, label data and operating state data, the working scenario feature of the refrigerator is determined. The working scenario feature is used to identify the environmental category and operating mode in which the refrigerator is currently located, such as high humidity, high load, frequent door opening or seasonal transition, etc. According to the working scenario feature, the corresponding training weight is configured for each type of data to improve the prediction accuracy of the target cabinet temperature in the corresponding scenario by the regression model. For features with significant influence, the weight needs to be increased; otherwise, for features with relatively small influence, the weight needs to be reduced. For example, the humidity data in a high humidity environment and the load rate data in a high load environment need to increase the weight; the season label in a tropical region needs to reduce the weight.

[0053] It can be understood that normally, the weight of all data features should remain the same, i.e. 1; but in some special scenarios, the weight adjustment of part of the data should follow the change. For example, in tropical regions, the seasonal division is not obvious, and in fact, the influence of season on the temperature of the refrigerator is very small, so in this case, the weight of the season label needs to be reduced to 0.5. By introducing the scene feature recognition and weight dynamic allocation mechanism, the regression model can focus on the main influencing factors in different working scenes, thereby effectively suppressing the error amplification caused by the difference between scenes.

[0054] In practical applications, the existing temperature control method often ignores the difference in heat capacity of the stored goods in the refrigerator, while the specific heat capacity of the stored goods has a related influence on the refrigeration rate and energy consumption. Under the same refrigeration conditions, the difference in heat capacity of different stored goods (such as beverages, fruits, and meat) will cause different temperature response speeds in the cabinet. If this factor is not considered, the prediction of the regression model will be biased, resulting in excessive refrigeration or insufficient refrigeration.

[0055] Based on this, in a specific embodiment, before training and testing the preset regression model, the control method further comprises: determining that the refrigerator is running to meet a plurality of preset conditions, obtaining the refrigeration power consumption of the refrigerator running for a preset time length in the target state and the temperature change of the storage space in the cabinet. The plurality of preset conditions include that the refrigerator is not running in energy-saving mode, and the current cabinet door state is closed, and the cabinet door has been closed for the last time and completed a preset number of pull-down temperature cycles. The preset number can be set to 3 times, i.e. from the current temperature to the set target temperature in the cabinet, the temperature rises to the set target temperature in the cabinet after the difference, and then decreases to the set target temperature in the cabinet, and so on for 3 times. Because the door is opened by default, the temperature of the measured goods should be consistent with the local environment temperature before being put in, and should be lowered to the set temperature in the refrigerator at the beginning of the calculation, and there is generally a temperature difference of about 20°C between them, which needs to be lowered to the measured temperature through at least 3 refrigeration cycles.

[0056] After meeting the plurality of preset conditions, record the refrigeration power consumption of the refrigerator refrigeration system in the preset time ; and monitor the temperature change in the cabinet through the temperature sensor to obtain the temperature change of the storage space . The formula for calculating the specific heat capacity is: , wherein C is the specific heat capacity, and m is the mass of the goods. It should be noted that the specific heat capacity of the food in the refrigerator is measured using the most direct calorimetry method, which calculates the specific heat capacity by measuring the temperature change and heat exchange of the goods in the refrigerator. This method requires that the initial temperature of the measured goods be consistent with the set temperature in the refrigerator, so the temperature of the measured goods needs to be lowered to the temperature in the cabinet or close to the value before the method is applied. Since the specific heat capacity of the goods stored in the refrigerator is related to the temperature setting in the cabinet, the specific heat capacity data needs to be configured as the target item data for the regression model to implement training and testing.

[0057] In the daily operation of the refrigerator, training the regression model requires computing resources, and if training is performed during periods of high passenger flow or frequent door opening, it can cause system response delays or temperature control instability. In addition, the higher the passenger flow, the more frequent the refrigerator temperature fluctuations, and the greater the data noise, which, if directly used for model training, can easily cause over-model fitting or prediction bias.

[0058] Based on this, before training and testing the preset regression model, the control method further comprises: During the night period, the current passenger flow of the refrigerator is determined according to the second temperature change of the storage space inside the refrigerator. The night period can be set as the early morning period, and the current passenger flow of the refrigerator can be calculated by the change rate and fluctuation frequency of the second temperature. When the current passenger flow is lower than the preset threshold, it is determined that the refrigerator is in a stable operation state, and then the training and testing process of the regression model is started. At this time, the temperature change is stable, the representativeness and continuity of the collected data are higher, and the training process will not affect the real-time temperature control execution of the refrigerator, improving the stability of the refrigerator operation.

[0059] After the basic situation of the training set and the test set is clear, the training set is imported into the regression model for training. The model selected here is a support vector machine regression model. The core of this regression model is to fit the data by minimizing the prediction error. Taking two-dimensional data as an example, as shown in Figure 3 , the regression model will define a boundary composed of a center line and two parallel boundary lines (dotted lines). The distance between these boundary lines is called the margin , which can be freely set; the area outside the margin is recorded as K. The goal of the regression model is to find a function that makes most of the data points fall within the margin and minimizes the prediction error of the data points that fall in the area K outside the margin. These data points that fall in the area K outside the margin are called support vectors. The data within the margin will be ignored, and only the support vectors outside the margin will affect the trend of the model.

[0060] The conditions that the points within the boundary in the regression model need to satisfy are: , x is the input feature vector, such as environmental humidity, labeled data, specific heat capacity, and load rate, etc.; y is the predicted temperature in the cabinet; is the regression coefficient; b is the bias term; is the width parameter of the insensitive interval. The goal to be optimized by the regression model is: , that is, to construct an insensitive band with a width of , so that the distance of the data points farthest from the center line is minimized, thereby capturing the distribution trend of the data.

[0061] Please refer to Figure 4The final result of the regression model processing high-dimensional data is to fit a curve L0, and the insensitive band is denoted as S0, so as to realize the prediction function of the model. In the actual model fitting process, the behavior of the regression model can be understood as follows: a certain error tolerance range is set, most of the data are within the tolerance range, and the influence of the support vector on the prediction center line is reduced as much as possible. In the embodiment of the present application, the learning process of the regression model is automatically completed by the computer, and part of the logical process is selected for explanation.

[0062] Please refer to Figure 5 The second temperature of the environment in which the refrigerator is located and the historical temperature in the cabinet are selected as examples, and the data is captured at a fixed time, so that the environmental temperature at a certain time and the historical temperature in the cabinet have a one-to-one correspondence, which can be placed in a two-dimensional coordinate system.

[0063] According to the data in Figure 5 , the temperature of the refrigerator fluctuates between 3℃ and 12℃, and during this period, the environmental temperature fluctuates between about 22℃ and 28℃. The insensitive band width is set to 3℃, and the prediction result given by the regression model under this condition is a straight line L1 of y=25. According to Figure 5 , a 3℃ interval has almost included most of the data points, and the support vectors are evenly distributed around the insensitive band S1, so the prediction trend of the regression model is accurate and reliable. For the second temperature and the historical temperature in the cabinet, the prediction of the regression model is that the environmental temperature is maintained at 25℃ during this period of data time, and is not affected by the change of the cabinet temperature. After obtaining the prediction of the environmental temperature and the historical temperature in the cabinet, another set of two-dimensional data representing the fourth temperature of the evaporator side and the historical temperature in the cabinet is selected for regression prediction.

[0064] As shown in Figure 6 , the data points of the evaporator side temperature and the historical temperature in the cabinet are distributed in the range of -5℃~5℃ of the evaporator side and 3℃~12℃ of the cabinet temperature within this time range, which corresponds to the normal working cooling process of the evaporator. The regression model performs regression analysis on these data points, and the insensitive band S2 is set at an interval of 3℃, and finally the regression line is a straight line L2. According to the data points, the equation of the straight line can be written as , where x is the historical temperature in the cabinet. Based on the analysis, it can be obtained that the evaporator side temperature and the historical temperature in the cabinet have a significant positive correlation within the above time points, and the regression equation is , where x is the historical temperature in the cabinet, and the historical temperature in the cabinet can be calculated using the evaporator side temperature.

[0065] In the same way, all data features are combined two by two, and the regression model will derive the regression relationship between each set of data features and express it in an equation. Because this scheme involves 9 sets of data features, including environmental humidity, first temperature, third temperature, fourth temperature, door opening frequency, load rate, geographic location coordinates, seasonal label, and specific heat capacity, 9 corresponding regression relationships with the historical temperature in the cabinet are involved. Therefore, the prediction equation output after the fitting process is complete should be a high-order polynomial. As previously analyzed, the last regression equation of the two-dimensional array will contain two data features, so the polynomial obtained based on the 9 sets of data features will contain all the data features, and the prediction result is more comprehensive and accurate.

[0066] After completing the fitting process, the final model obtained is a support vector machine regression model, which is trained using the latest 72 hours of data, and is retrained every 24 hours based on updated data to ensure adaptability to the environment. Because the entire process is independent of the temperature control system, data collection, data updating, and model training can be completely unresponsive, and the training process, which occupies a large amount of system resources, will be set during the early morning hours when there is less traffic, and will not affect the operation of the entire machine.

[0067] At this point, the training and testing of the regression model have been completed, and step S14 is entered.

[0068] S14, input the current environmental data of the refrigerator into the trained regression model, and obtain the target cabinet temperature of the refrigerator based on the output result of the regression model.

[0069] Specifically, the current environmental data includes the data directly collected by each sensor, including the current environmental humidity, the current first temperature, etc. The current environmental data is input into the regression model, and the target cabinet temperature required for the refrigerator to control is obtained. This step can be implemented based on the interaction between the temperature controller and the regression model, including the reading of environmental data by each sensor, and the reading of real-time current environmental data by the temperature controller through the integrated sensor. The current environmental data is imported into the regression model, which is different from the ordinary rule-based temperature controller system. The temperature controller will interact with the trained regression model and transfer the data to the model for prediction processing. Based on the imported current environmental data, the regression model will output the prediction result, which is the current most suitable target cabinet temperature recognized by the model. This temperature value will be returned to the temperature control system and enter step S15.

[0070] S15, control the refrigeration temperature of the refrigerator in the future period according to the target cabinet temperature, and retrain the regression model after the future period to implement temperature regulation.

[0071] Specifically, the predicted temperature value output by the regression model is taken as the target cabinet temperature, which will be accepted by the temperature control system and used to replace the original set temperature. Since the set temperature is specified, the temperature controller will call all resources, such as the compressor, evaporator fan, and condenser fan, to adjust the temperature to the target cabinet temperature. The temperature adjustment process can set a corresponding implementation frequency, for example, adjusting the set temperature once an hour, and the more frequent the adjustment, the better the temperature controller's cooling strategy adapts to the ambient temperature. It should be noted that the future period can be set based on actual conditions, for example, set to a fixed value of 24 hours; of course, it can also be set to a dynamic value of 24-48 hours based on seasonal changes or changes in ambient temperature, which is not specifically limited here.

[0072] After the refrigerator runs for 24 hours based on the target cabinet temperature, the regression model is retrained using the latest 72 hours of data and a new target cabinet temperature is predicted. The temperature controller arranged in the refrigerator performs temperature adjustment based on the updated target cabinet temperature in the next 24 hours.

[0073] Please refer to Figure 7 , Figure 7 for the implementation architecture diagram of the refrigerator temperature control. The server performs collection of real-time feature data, i.e., various data associated with the temperature setting in the refrigerator; stores and organizes various historical data and updates based on a pre-set time period; imports various historical data into the regression model for training and verification; and obtains a trained regression model. The temperature controller imports the current environmental data collected by various sensors into the trained regression model, so that the regression model outputs the target cabinet temperature of the refrigerator in the future period, and the temperature controller sets the temperature data based on the target cabinet temperature to implement the adaptive temperature control strategy of the refrigerator.

[0074] Based on the same technical concept as the control system, the embodiments of the present application also provide a refrigerator, which comprises the control system according to any one of the above embodiments.

[0075] The technical solution of the present application is based on the technical problems of the prior art, combines the computing power advantage of the regression model, calculates the real-time target cabinet temperature required through the changes of various variables of the environment around the refrigerator, takes it as a temperature threshold and establishes a corresponding temperature control strategy based on it. The traditional passive temperature control is converted into active temperature control, which fundamentally solves a series of problems caused by the method of setting fixed temperature threshold by artificial experience, such as frequent start-stop or temperature overshoot. For the strategy of adjusting the frequency of the compressor by the PID algorithm, due to the solidification of parameters, the influence of seasonal changes on heat dissipation efficiency is not considered, so the problems of easy overcooling in winter and insufficient refrigeration in summer are prone to occur, and it only controls the compressor itself, and the frequency of the compressor does not have a direct logical relationship with refrigeration. The present scheme collects a large amount of environmental data to judge the external environment, and formulates different temperature control strategies for different environments, which fundamentally bypasses the limitations of single control of the frequency of the compressor by the PID algorithm, and directly acts on the temperature setting, so the efficiency is higher and the control logic is more reasonable.

[0076] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0077] Each of the embodiments in the specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

Claims

1. A temperature control system for a freezer, characterized in that, The system includes: The data acquisition terminal is used to collect multiple temperature and humidity data of the freezer during historical periods, including the refrigeration path and the surrounding environment, as well as the label data of the surrounding environment. A processing terminal is connected to the acquisition terminal. The processing terminal is used to analyze the variation characteristics of the multiple temperature and humidity data to obtain the operating status data of the freezer. Based on the multiple temperature and humidity data, the label data, and the operating status data, the preset regression model is trained and tested to establish a regression relationship between various types of data and the historical temperature inside the cabinet in the corresponding insensitive zone. The current environmental data of the freezer is input into the trained regression model, and the target internal temperature of the freezer is obtained based on the output of the regression model. The refrigeration temperature of the freezer is controlled according to the target internal temperature in the future period, and the regression model is retrained after the future period to implement temperature regulation.

2. The freezer temperature control system according to claim 1, characterized in that, The data acquisition terminal includes: A humidity sensor is connected to the first input terminal of the processing terminal and is installed on the freezer; the humidity sensor is used to collect the ambient humidity of the freezer and output it to the outside. A first temperature sensor is connected to the second input terminal of the processing terminal and is installed on the freezer. The first temperature sensor is used to collect the first temperature of the environment in which the freezer is located and output it to the outside. The second temperature sensor is connected to the third input terminal of the processing terminal. The second temperature sensor is installed inside the freezer and connected to the storage space of the freezer. The second temperature sensor is used to collect the second temperature of the storage space inside the freezer and output it to the outside. The third temperature sensor is connected to the fourth input terminal of the processing terminal. The third temperature sensor is installed at the location of the condenser on the freezer. The third temperature sensor is used to collect the third temperature located on the side of the freezer condenser and output it to the outside. A fourth temperature sensor is connected to the fifth input terminal of the processing terminal. The fourth temperature sensor is installed at the location of the evaporator on the freezer. The fourth temperature sensor is used to collect the fourth temperature located on the evaporator side of the freezer and output it to the outside.

3. The freezer temperature control system according to claim 1, characterized in that, The processing terminal includes: A server is configured to communicate with the data acquisition terminal. The server is used to process the data output by the data acquisition terminal to obtain the target internal temperature of the freezer. A controller, configured to communicate with the server, is installed inside the freezer and is used to control the cooling temperature of the freezer in a future time period based on the target internal temperature.

4. A method for controlling the temperature of a freezer, characterized in that, Applied to the control system according to any one of claims 1-3, the method comprises: Acquire multiple temperature and humidity data of the freezer during historical periods, including the refrigeration path and the surrounding environment, as well as the label data of the surrounding environment; The changing characteristics of the multiple temperature and humidity data are analyzed to obtain the operating status data of the freezer; Based on the multiple temperature and humidity data, the label data, and the operating status data, the preset regression model is trained and tested to establish a regression relationship between various types of data and the historical temperature inside the cabinet in the corresponding insensitive zone. The current environmental data of the freezer is input into the trained regression model, and the target internal temperature of the freezer is obtained based on the output of the regression model. The refrigeration temperature of the freezer is controlled according to the target internal temperature in the future period, and the regression model is retrained after the future period to implement temperature regulation.

5. The freezer temperature control method according to claim 4, characterized in that, The step of analyzing the variation characteristics of the multiple temperature and humidity data to obtain the operating status data of the freezer includes: Based on the change of each temperature and humidity data over time, data curves for each temperature and humidity data are obtained. Among them, multiple data curves include the ambient humidity curve and the first temperature curve of the environment where the freezer is located, the second temperature curve of the storage space inside the freezer, the third temperature curve of the freezer condenser side, and the fourth temperature curve of the freezer evaporator side. The relationship between the ambient humidity curve, the first temperature curve, the second temperature curve, the third temperature curve, and the fourth temperature curve is analyzed to obtain the opening and closing frequency and operating load rate of the freezer. The door opening and closing frequency and the operating load rate are determined as the operating status data of the freezer.

6. The freezer temperature control method according to claim 4, characterized in that, Before training and testing the preset regression model, the method further includes: When the freezer is determined to operate to a target state that meets multiple preset conditions, the cooling power consumption of the freezer and the temperature change of the internal storage space of the freezer are obtained for a preset duration of operation in the target state. Based on the cooling power consumption, the temperature change, and the mass of newly added items in the storage space inside the freezer, the specific heat capacity data of the stored items in the freezer are obtained. The specific heat capacity data is configured as the target data for training and testing the regression model.

7. The freezer temperature control method according to claim 6, characterized in that, Determining the freezer to operate in a target state that meets multiple preset conditions includes: It is determined that the freezer is not running in energy-saving mode, and the current door status is closed, and the door has been closed for the last time and the temperature cycle has been pulled down a preset number of times. If so, then the freezer is determined to be operating in the target state.

8. The freezer temperature control method according to claim 4, characterized in that, Before training and testing the preset regression model, the method further includes: During the nighttime period, the current customer flow of the freezer is determined based on the second temperature change of the storage space inside the freezer. When the current passenger flow is lower than a preset value, the regression model is trained and tested.

9. The freezer temperature control method according to claim 4, characterized in that, Before training and testing the preset regression model, the method further includes: The working scenario characteristics of the freezer are determined based on at least one of the multiple temperature and humidity data, the label data, and the operating status data; Based on the characteristics of the work scenario, corresponding training weights are configured for various types of data so that the regression model can improve the prediction accuracy of the target cabinet temperature in the corresponding scenario.

10. A freezer, characterized in that, The freezer includes the control system described in any one of claims 1-3.

Images