Refrigeration device and intelligent temperature control method thereof
By constructing multiple decision trees and combining them with a power consumption calculation model, the uncertainty problem of temperature control in refrigeration equipment was solved, enabling energy-saving temperature control that can flexibly adapt to different scenarios and user habits.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GREE ELECTRIC APPLIANCE INC OF ZHUHAI
- Filing Date
- 2023-12-22
- Publication Date
- 2026-07-21
Smart Images
Figure CN117847943B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of refrigeration equipment, and in particular to a refrigeration equipment and its intelligent temperature control method. Background Technology
[0002] Most refrigeration equipment currently relies on manual control. For example, household air conditioners are basically controlled manually. Since the comfort level of household air conditioners varies from person to person, manual control is still feasible. However, some refrigeration equipment has many problems when relying on manual control.
[0003] Taking cold storage as an example, current temperature control in cold storage largely relies on two methods: manual operation and remote control. Both methods depend on human intervention to some extent, which often leads to two situations. First, users may adjust the temperature too high to save electricity, resulting in poor freezing performance. Second, users may adjust the temperature too low to ensure freezing performance, which, while maintaining freezing efficiency, consumes a lot of electricity.
[0004] like Figure 1 As shown, the prior art with publication number CN111256428A proposes an adaptive fuzzy control method for cold storage temperature to solve this problem of relying on manual control. The method mainly adopts the following steps.
[0005] Step 1, Parameter Data Acquisition: Based on the factors affecting the temperature inside the cold storage, the total load of the cold storage, the temperature difference between inside and outside the cold storage, and the time point are determined as parameters affecting the temperature of the cold storage, and the above parameters are divided into several levels. Step 2, Calculate the membership degree of the parameter data: Substitute the total cold storage load, the temperature difference between inside and outside the cold storage, and the time point values into the membership function to calculate the membership data of the above parameters; Step 3, Establishment of the fuzzy control model: Establish fuzzy control models for the total load of the cold storage, the temperature difference between the inside and outside of the cold storage, and the relationship between time points and the required temperature of the cold storage. Step 4: Substitute the current parameter membership data into the fuzzy control model established in Step 3 to obtain the working level information of the cold storage refrigeration equipment that meets the current temperature requirements. The temperature inside the cold storage is divided into three levels: Level 1, Level 2, and Level 3. Level 1 means that the cold storage refrigeration equipment is outputting 50% of its power, Level 2 means that the cold storage refrigeration equipment is outputting 75% of its power, and Level 3 means that the cold storage refrigeration equipment is outputting 100% of its power.
[0006] As can be seen from the above steps, existing technologies for controlling the temperature of refrigeration equipment are based on the external environment, such as heat exchange rate and ambient temperature. Although these external environments are subject to change, the patterns of change are predictable, making it relatively easy to obtain corresponding control models. However, in actual use, refrigeration equipment is also affected by human usage habits. For example, in cold storage, the number of times the door is opened and closed will affect the temperature of the refrigeration equipment. In addition, for refrigeration equipment used to store goods, different goods require different temperatures. For example, the temperature required for vaccines is completely different from that required for chicken. However, existing cold storage technologies do not consider these factors because these factors have a lot of uncertainty, making it difficult to obtain corresponding models to control the temperature of the refrigeration equipment.
[0007] Therefore, how to provide a new intelligent temperature control method for refrigeration equipment that can eliminate manual control, take into account various uncertain factors, and achieve energy saving is a technical problem to be solved. Summary of the Invention
[0008] To address the technical problems of existing refrigeration equipment temperature control typically having only one calculation model, which limits its application to a single scenario and results in insufficient energy saving accuracy, this invention proposes a refrigeration equipment and its intelligent temperature control method.
[0009] The intelligent temperature control method for refrigeration equipment proposed in this invention includes: Pre-extract the parameters that affect the power consumption of the refrigeration equipment; Based on the different influencing factor parameters of refrigeration equipment and the corresponding historical power consumption data, a power consumption calculation model is obtained; Different decision trees were constructed using different impact factor parameters as key impact factors. When controlling refrigeration equipment, the areas of the ROC curves and AUC curves of the output results of different decision trees are combined to obtain the common area as the control range of the target temperature of the refrigeration equipment. Then, based on the power consumption calculation model, the temperature with the lowest power consumption that falls within the control range is taken as the target temperature.
[0010] Furthermore, it also includes: Classify refrigeration equipment; The influencing factor parameters, power consumption, and target temperature of different types of refrigeration equipment are collected to form corresponding algorithm models; When it is necessary to control a new refrigeration device, the target temperature that is closest to the influencing factor parameters and has the lowest power consumption is found based on the corresponding sampling model according to the category of the new refrigeration device, and is used as the target temperature of the new refrigeration device.
[0011] Furthermore, the sampling model is a negative sampling model.
[0012] Furthermore, the influence factor parameters are obtained using the following steps: Obtain historical data from refrigeration equipment; The correlation between various parameters in historical data and power consumption was obtained using the Pearson correlation coefficient. The N parameters with the strongest correlation are used as the influencing factors of power consumption, where N > 2.
[0013] Furthermore, the power consumption calculation model is obtained using the following steps: Select basic influencing factors from the parameters affecting power consumption, and construct a power consumption table based on the basic influencing factors; The corresponding power consumption model is obtained based on the basic influencing factors; Based on the power consumption table constructed from the basic influencing factors, each key influencing factor is superimposed, and other different power consumption models are obtained by changing the values of the corresponding key influencing factors. By fitting all the power consumption models together, a power consumption calculation model is obtained.
[0014] Furthermore, before fitting the power consumption model, the power consumption model is also calibrated. During calibration, the standard deviation and variance of the prediction results of different power consumption models and the actual power consumption are calculated, and the corresponding power consumption model is calibrated based on the obtained calculation results.
[0015] Furthermore, the basic influencing factors include at least one of the following: evaporation temperature and ambient temperature of the space requiring cooling or heating.
[0016] Furthermore, the key influencing factors include one or more of the following: the number of times the doors of the space requiring cooling or heating are opened and closed, the number of people or the types of goods stored in the space requiring cooling or heating, and the ambient temperature.
[0017] Furthermore, the random forest algorithm is used to construct different decision trees.
[0018] The refrigeration equipment proposed in this invention includes a control module, which uses the intelligent temperature control method for refrigeration equipment described in the above technical solution to control the target temperature of the refrigeration equipment.
[0019] Furthermore, the refrigeration equipment includes a cold storage room.
[0020] This invention employs multiple decision trees to control the temperature of refrigeration equipment, fully considering various factors, thus enabling temperature control to be adapted to a wide range of scenarios. Furthermore, it uses the common area of the ROC and AUC curves of different decision trees to determine the target temperature range, and then finds the target temperature with the lowest power consumption, thereby satisfying specific temperature requirements while minimizing power consumption. In addition, this invention considers power consumption from multiple perspectives and integrates the results of multiple power consumption models, then fits the results of these models to obtain the final predicted power consumption, which is more accurate than existing power consumption prediction methods. Attached Figure Description
[0021] The present invention will now be described in detail with reference to the embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart of the temperature control process in existing technology.
[0022] Figure 2 This is the main flowchart of the present invention; Figure 3 This is a flowchart of a cold storage temperature control system according to an embodiment of the present invention.
[0023] Figure 4 This is a power consumption meter according to an embodiment of the present invention.
[0024] Figure 5 This is a flowchart of the decision-making process of multiple decision trees according to an embodiment of the present invention.
[0025] Figure 6 This is a schematic diagram of the area of ROC and AUC according to an embodiment of the present invention.
[0026] Figure 7 This is a flowchart of a power consumption calculation model according to an embodiment of the present invention. Detailed Implementation
[0027] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0028] Therefore, a feature pointed out in this specification is used to describe one feature of one embodiment of the invention, and does not imply that every embodiment of the invention must have the described feature. Furthermore, it should be noted that this specification describes many features. Although certain features may be combined to illustrate possible system designs, these features may also be used in other combinations not explicitly stated. Therefore, unless otherwise stated, the described combinations are not intended to be limiting.
[0029] like Figure 2 As shown, the intelligent temperature control method for refrigeration equipment proposed in this invention mainly includes the following steps.
[0030] Pre-extract the parameters that affect the power consumption of the refrigeration equipment; Based on the different influencing factor parameters of refrigeration equipment and the corresponding historical power consumption data, a power consumption calculation model is obtained; Different decision trees were constructed using different impact factor parameters as key impact factors. When controlling refrigeration equipment, the common area of the ROC curve and AUC curve of the output results of different decision trees is combined to obtain the common area as the control range of the target temperature of the refrigeration equipment. Then, based on the power consumption calculation model, the temperature with the lowest power consumption that falls within the control range is taken as the target temperature.
[0031] While the power consumption influencing factors for different refrigeration equipment are roughly similar, their effects vary depending on the specific circumstances. For example, refrigerators and cold storage facilities storing food or goods share similar influencing factors, including the number of door openings, ambient temperature, and the temperature and humidity of the stored items. However, refrigerators, being indoors, typically experience relatively stable ambient temperatures over extended periods. Cold storage facilities, on the other hand, are affected by factors such as sunny days, cloudy days, sunrise, and sunset, resulting in significant daily variations. Furthermore, even with identical influencing factors, the target temperature may differ depending on the location of the cold storage facility. For instance, a cold storage facility in the shade will have a different ambient temperature than one in direct sunlight, thus requiring a different target temperature.
[0032] Therefore, the present invention uses historical data of refrigeration equipment to obtain a calculation model for the power consumption of the refrigeration equipment, which is usually quite accurate. That is, the calculation model for the power consumption of different refrigeration equipment is different, but the method of obtaining the power consumption calculation model is the same. That is, the power consumption calculation model of the refrigeration equipment is obtained by using the influencing factor parameters and the different power consumptions obtained when these influencing factor parameters change.
[0033] After obtaining a relatively accurate calculation model of the power consumption of the refrigeration equipment, since temperature control is the most important aspect of refrigeration equipment, this invention constructs different decision trees using different influencing factor parameters as key influencing factors. Existing decision trees typically consist of only one type, applicable only to relatively simple application scenarios. However, in reality, the target temperature required by refrigeration equipment is affected by its different uses, locations, and user habits. Therefore, this invention establishes multiple decision trees, each with different key influencing factors, meaning each decision tree has a different maximum information entropy. Entropy is a measure of the branch selection in a decision tree, used to describe the expected information. In probability theory, information entropy provides a way to measure uncertainty, used to measure the uncertainty of random variables; entropy is the expected value of information. Simply put, each decision tree focuses on the influence of one key influencing factor. Although other influencing factor parameters exist, their information entropy is relatively low.
[0034] This invention, through decision trees with different focuses, can encompass more application scenarios; however, determining the specific target temperature remains a challenge. To address this challenge, this invention, when actually controlling the refrigeration equipment, combines the common area of the ROC and AUC curves of all decision tree outputs. The output of each decision tree represents the target temperature of the refrigeration equipment. As time progresses, the influence factors of each decision tree change, thus changing the target temperature output by each tree. This invention first calculates the area of the ROC and AUC curves of each decision tree's output, then combines the areas of all decision trees' ROC and AUC curves, taking the common area (also called the common region). Based on this common region, the upper and lower limits of the target temperature obtained by different decision trees can be obtained.
[0035] Although the above methods yield the upper and lower limits of the target temperature, in actual control, a lower or higher target temperature is not necessarily better. To find the most suitable target temperature, this invention uses the previously obtained power consumption calculation model to calculate the corresponding power consumption. The temperature corresponding to the current influencing factor with the lowest power consumption (taking the cold storage temperature as an example) is taken as the final target temperature for controlling the refrigeration equipment.
[0036] Through the above description of the technical solution of the present invention, it can be seen that the present invention fully considers various influencing factors of refrigeration equipment, constructs different decision trees for the refrigeration equipment under different application environments and different usage habits, and then takes the intersection of the output results of the decision trees to determine the upper and lower limits of the target temperature. Then, based on the historical data of the refrigeration equipment, an accurate power consumption calculation model for the refrigeration equipment is obtained. Finally, a temperature control method with flexible applicability and relatively accurate power consumption calculation can be obtained.
[0037] The above embodiments of the present invention are particularly suitable for refrigeration equipment with historical data, thus enabling the most accurate and suitable target temperature for energy saving. In another embodiment, although the decision tree part of the present invention can be applied to new refrigeration equipment, an accurate power consumption calculation model cannot be obtained due to the lack of historical data. To solve this problem, the present invention performs the following processing for refrigeration equipment with historical data.
[0038] Classify refrigeration equipment; The influencing factor parameters, power consumption, and target temperature of different types of refrigeration equipment are collected to form corresponding algorithm models; When it is necessary to control a new refrigeration device, the target temperature that is closest to the influencing factor parameters and has the lowest power consumption is found based on the corresponding sampling model according to the category of the new refrigeration device, and is used as the target temperature of the new refrigeration device.
[0039] In other words, this invention categorizes existing refrigeration equipment, such as cold storage units with model number XXX. It collects data on the target temperature and power consumption of these cold storage units under different external temperatures, different door opening and closing times, different stored goods (such as different types of goods and / or the volume or weight of goods), forming a corresponding data set. When a new cold storage unit with model number XXX is applied, an appropriate algorithm or model is used to find the target temperature with the closest conditions and the lowest power consumption from the data set of all cold storage units with model number XXX, and uses this as the target temperature for the new refrigeration equipment, so that the new refrigeration equipment can also achieve similar results.
[0040] In one specific embodiment, the target temperature with the lowest power consumption can be found from the closest conditions using a negative sampling model. However, it is not limited to the negative sampling model; other computational models can also be used, such as the hierarchical model in the prior art.
[0041] like Figure 7 As shown, the power consumption calculation model can be obtained using the following several embodiments.
[0042] In one specific embodiment, the influence factor parameters of the refrigeration equipment are obtained using the following steps.
[0043] Obtain historical data from refrigeration equipment; The correlation between various parameters in historical data and power consumption was obtained using the Pearson correlation coefficient. The N parameters with the strongest correlation are used as the influencing factors of power consumption, where N > 2.
[0044] Different refrigeration units have different influencing factors. To obtain a more accurate power consumption calculation model, this invention uses the Pearson correlation coefficient to find the parameters with the strongest correlation to power consumption as influencing factors, thus facilitating accurate power consumption calculation. Taking cold storage as an example, the influencing factors of power consumption in cold storage include: evaporation temperature, ambient temperature, number of door openings and closings, volume of goods, temperature, humidity, etc.
[0045] The specific implementation steps of the power consumption calculation model of the present invention are described below.
[0046] Select basic influencing factors from the parameters affecting power consumption, and construct a power consumption table based on the basic influencing factors; The corresponding power consumption model is obtained based on the basic influencing factors; Based on the power consumption table constructed from the basic influencing factors, each key influencing factor is superimposed, and other different power consumption models are obtained by changing the values of the corresponding key influencing factors. By fitting all the power consumption models together, a power consumption calculation model is obtained.
[0047] Similar to decision trees, this invention uses each influencing factor parameter as a variable to form different power consumption models when calculating power consumption. These models are then corrected and calibrated, and finally fitted together to obtain a power consumption calculation model that considers various influencing factor parameters, thus achieving accurate power consumption calculation. It should be noted that the above implementation steps are merely a preferred embodiment of a specific power consumption calculation model of this invention, but do not represent that this is the only embodiment of the invention. Existing technologies can also obtain corresponding power consumption calculation models based on historical data of refrigeration equipment using various algorithms (such as fuzzy control algorithms), but these do not consider as comprehensively as the preferred embodiment of this invention.
[0048] As described above, in a further preferred embodiment, the present invention further includes calibrating the power consumption model before fitting it. During calibration, the standard deviation and variance of the predicted results of different power consumption models compared to the actual power consumption are calculated, and the corresponding power consumption model is calibrated based on the obtained calculation results. Since different power consumption models have different focuses, and if they are inconsistent with the actual power consumption, the standard deviation and variance of the predicted results of the power consumption models compared to the actual power consumption can be calculated to correct each power consumption model.
[0049] The basic influencing factors in the above embodiments include at least one of the following: evaporation temperature and ambient temperature of the space requiring cooling or heating.
[0050] The key influencing factors in the above embodiments include one or more of the following: the number of times the doors of the space requiring cooling or heating are opened and closed, the number of people or the types of goods stored in the space requiring cooling or heating, and the ambient temperature.
[0051] In one specific embodiment, the present invention uses the random forest algorithm to construct different decision trees.
[0052] like Figure 3 As shown below, a cold storage facility is used as a specific application example of refrigeration equipment to fully illustrate the above-mentioned technical solution of the present invention.
[0053] First, a calculation model for the power consumption of the cold storage is needed.
[0054] To accurately determine the historical data of the cold storage facility, Pearson correlation coefficients were used to extract correlations from over thirty dimensions of data through cross-validation. This revealed that the fundamental influencing factors affecting the power consumption of the cold storage unit are evaporation temperature and storage temperature. Using these two parameters, a power consumption table for the cold storage was generated. This table has three columns: evaporation temperature, storage temperature, and power consumption. Each row contains the historical data for the cold storage facility. Specifically... Figure 4 As shown.
[0055] Based on the evaporation temperature, storage temperature, and power consumption of the cold storage, a corresponding power consumption model is obtained. This power consumption model is a linear regression model, where the evaporation temperature and storage temperature are used as input parameters, and the power consumption is used as the output parameter. The prediction result of this power consumption model is denoted as n1.
[0056] Based on the same evaporation temperature and storage temperature, the predicted power consumption model is denoted as n2 by changing the external ambient temperature.
[0057] Based on the same evaporation temperature and storage temperature, the frequency of door opening and closing is changed, and the predicted result of the power consumption model is denoted as n3.
[0058] Based on the same evaporation temperature and storage temperature, the type of goods is changed, and the predicted result of the power consumption model is denoted as n4.
[0059] The aforementioned power consumption model can specifically employ a linear regression model. After obtaining the power consumption model, it can be further modified. Specifically, the prediction results of the four power consumption models can be compared with the actual power consumption to calculate the standard deviation and variance, thereby modifying the prediction results of the four power consumption models. The modified power consumption model is then used to further refine the prediction results. Existing technologies typically only use variance for calibration. The advantage of incorporating standard deviation in this invention is that it effectively avoids overfitting caused by large variations in cargo volume or external temperature.
[0060] Finally, the four power consumption models are fitted to obtain the corresponding power consumption calculation model, such as the expression Predict_y = b0 + b1x1 + b2x2 + b3x3 + b4x4 + e, where Predict_y is the predicted power consumption, b0 to b4 are the corresponding coefficients, x1 to x4 represent the output results of different power consumption models (i.e., the predicted power consumption), and e represents the correction coefficient for balancing the four power consumption models.
[0061] The calculation model for the power consumption of each refrigeration unit is also updated regularly, for example, once a month, to ensure the availability of the fitting parameters and to achieve accurate prediction of power consumption.
[0062] Next, a random forest model was applied to the historical data of all cold storage facilities. Different decision trees were constructed based on different key influencing factors, and these factors determined the maximum information entropy in each decision tree. By combining the influence of multiple decision trees, a more accurate range of target temperatures could be obtained. This result is more objective and better takes into account the different operating environments of different equipment.
[0063] Four different decision trees need to be constructed based on different types of goods, different external temperatures, different numbers of door openings and closings, and different usage habits. The information entropy ratio of the key influencing factors in each tree is different.
[0064] Figure 5 The two logical trees shown are obtained based on the random forest algorithm.
[0065] One logic tree first determines if the goods are food. If not, it checks if they are medicine, vegetables, etc. If they are medicine, it further checks if they are vaccines. If they are vaccines, the target temperature should be set to M1; if they are other medicines, the target temperature should be set to M2. If they are vegetables, the target temperature is set to T1; if they are meat, it is set to T2; if they are fruit, it is set to T3; and if they are other foods, it is set to T4.
[0066] Another logic tree determines the user's door-opening habits. If the user prefers to open the door during the day, then the tree further determines the user's door-opening frequency, gives some door-opening thresholds X1, and then sets the target temperature to K1 or K2, or prunes the tree. Similarly, if the user prefers to open the door at night, the tree further determines the user's door-opening frequency, gives some door-opening thresholds X2, and then sets the target temperature to K2 or K3.
[0067] Press at this time Figure 5 Similar decision tree execution logic is used. After executing the decision logic of four different decision trees, the intersection of the different decision trees (that is, the control direction that different influencing factors point to together, i.e., the target temperature) is merged and output. The ROC curve and AUC curve of the output results of different decision trees are combined to find the common area and obtain the upper and lower limits of the target temperature.
[0068] Figure 6 The figure shows the area of the ROC curve and the area of the AUC curve of a decision tree output. The horizontal axis represents the event and the vertical axis represents the temperature. The upper curve is the ROC curve and the lower curve is the AUC curve. The area between the ROC curve and the AUC curve is the area of the ROC curve and the AUC curve. By superimposing the areas of the ROC curve and the AUC curve of multiple decision trees at the same time, the corresponding target temperature range can be obtained.
[0069] Then, by combining the power consumption calculation model, the target temperature that has the lowest power consumption and best matches the actual situation of the current influencing factor parameters is found. For example, to meet the storage temperature condition of xx℃, when the outside temperature is (a1,a2,a3...), the power consumption is lowest when the storage temperature is set to (b1,b2,b3...).
[0070] After the above steps, a control command result can be obtained. This command will generate a variable, which is sent by the script program to the remote monitoring interface. After the interface recognizes the variable, it can automatically change the cooling temperature setpoint. To enhance intelligence, manual operation permissions should be retained, and the cooling temperature should be continuously monitored to ensure it does not exceed the expected cooling capacity or fall short of the target capacity.
[0071] This invention also protects refrigeration equipment, which includes a control module that uses the intelligent temperature control method of the above-described technical solution to control the target temperature of the refrigeration equipment.
[0072] The refrigeration equipment of the present invention includes, but is not limited to, cold storage.
[0073] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for intelligent temperature control of a refrigeration device, characterized in that, include: Pre-extract the parameters that affect the power consumption of the refrigeration equipment; Based on the different influencing factor parameters of refrigeration equipment and the corresponding historical power consumption data, a power consumption calculation model is obtained; Different decision trees were constructed using different impact factor parameters as key impact factors. When controlling refrigeration equipment, the areas of the ROC curves and AUC curves of the output results of different decision trees are combined to obtain the common area as the control range of the target temperature of the refrigeration equipment. Then, based on the power consumption calculation model, the temperature with the lowest power consumption that falls within the control range is taken as the target temperature; The power consumption calculation model is obtained through the following steps: Select basic influencing factors from the parameters affecting power consumption, and construct a power consumption table based on the basic influencing factors; The corresponding power consumption model is obtained based on the basic influencing factors; Based on the power consumption table constructed from the basic influencing factors, each key influencing factor is superimposed, and other different power consumption models are obtained by changing the values of the corresponding key influencing factors. By fitting all power consumption models, a power consumption calculation model is obtained; The basic influencing factors include at least one of the following: evaporation temperature and ambient temperature of the space requiring cooling or heating. The key influencing factors include one or more of the following: the number of times the doors of the space requiring cooling or heating are opened and closed, the number of people or the types of goods stored in the space requiring cooling or heating, and the ambient temperature.
2. The intelligent temperature control method for refrigeration equipment as described in claim 1, characterized in that, Also includes: Classify refrigeration equipment; The influencing factor parameters, power consumption, and target temperature of different types of refrigeration equipment are collected to form corresponding algorithm models; When it is necessary to control a new refrigeration device, the target temperature that is closest to the influencing factor parameters and has the lowest power consumption is found based on the corresponding sampling model according to the category of the new refrigeration device, and is used as the target temperature of the new refrigeration device.
3. The intelligent temperature control method for refrigeration equipment as described in claim 2, characterized in that, The sampling model is a negative sampling model.
4. The intelligent temperature control method for refrigeration equipment as described in claim 1, characterized in that, The impact factor parameters are obtained using the following steps: Obtain historical data from refrigeration equipment; The correlation between various parameters in historical data and power consumption was obtained using the Pearson correlation coefficient. The N parameters with the strongest correlation are used as the influencing factors of power consumption, where N > 2.
5. The intelligent temperature control method for refrigeration equipment as described in claim 1, characterized in that, Before fitting the power consumption model, the power consumption model is also calibrated. During calibration, the standard deviation and variance of the prediction results of different power consumption models and the actual power consumption are calculated, and the corresponding power consumption model is calibrated based on the obtained calculation results.
6. The intelligent temperature control method for refrigeration equipment as described in claim 1, characterized in that, The random forest algorithm is used to construct different decision trees.
7. A refrigeration device, comprising a control module, characterized in that, The control module uses the intelligent temperature control method for refrigeration equipment as described in any one of claims 1 to 6 to control the target temperature of the refrigeration equipment.
8. The refrigeration equipment as described in claim 7, characterized in that, The refrigeration equipment includes a cold storage room.