Refrigerator door non-tight-closing detection system based on similarity
By obtaining the refrigerator's historical fault feature data and real-time feature data and generating a similarity curve, the detection delay and failure problems caused by relying on sensors and fixed parameter values in the existing technology are solved, and continuous and accurate detection of refrigerator door openings is achieved.
Patent Information
- Application Number
- CN202510860524.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-25
- Publication Date
- 2025-09-05
AI Technical Summary
Existing refrigerator door open detection technology relies on the sensitivity of the sensor and fixed parameter values, resulting in detection delays or failures, and is unable to continuously and accurately detect whether the refrigerator door is closed.
By acquiring historical fault feature data and real-time feature data, an effective historical fault data curve is generated, and the similarity between the real-time feature data and the historical fault data is calculated to determine the tightness of the refrigerator door.
It achieves the goal of not relying on sensors and fixed parameter values, and can continuously and accurately detect whether the refrigerator door is closed tightly, thereby improving the timeliness and adaptability of detection.
Smart Images
Figure CN120593469A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of refrigerator control, and in particular to a similarity-based refrigerator door ajar detection system. Background Art
[0002] Refrigerators are a common refrigeration appliance that requires constant operation in daily life. However, the widespread problem of doors not being securely closed can directly impact resource conservation and user experience, potentially leading to energy waste and a sharp decline in food preservation. With the increasing adoption of smart homes and improvements in user quality of life, refrigerator energy efficiency and food preservation technologies have become key concerns for consumers. As refrigerators are used more frequently and their years of service increase, the problem of cold air leaking through the door gap increases. This typically occurs when a user closes the refrigerator door, perhaps due to improper force or damaged door seals, leaving a small gap. This can cause the door sensor to detect the door as closed when it is actually not.
[0003] The current existing technical solutions mainly focus on adding corresponding sensors at appropriate positions on the refrigerator compartment door body, and judging whether the refrigerator door seal is tightly fitted with the cabinet body according to the different signals received by the sensors, and then judging whether the refrigerator door is not closed tightly; or using corresponding sensors to analyze the door closing signal, obtain the required compartment temperature change characteristics and compare them with the preset characteristic values of the door not closed tightly, and judge whether the compartment temperature change meets the temperature curve model or the preset characteristic values, and then judge whether the door is not closed tightly.
[0004] These methods for detecting doors that are not tightly closed require the installation of corresponding sensors at specific locations on the refrigerator door. Based on the different information received by the sensors and the differences in this information, different methods for determining whether the door is not tightly closed are adopted. This method relies on the sensitivity of the sensor and is combined with the control system to determine whether the door is not tightly closed. If the sensitivity of the sensor is not enough, it will not be able to receive key signals in time during detection, resulting in detection delays. In addition, when receiving the temperature changes in the compartment of the sensor, the preset characteristics or curve models are compared, which all rely on the values of certain fixed parameters. For different refrigerator models and different refrigerator operating life cycles, these fixed parameter values need to be adjusted in time under certain conditions. During the adjustment process, it is likely that different scenario experiments will need to be conducted based on the new refrigerator model to obtain parameter values that can be uniformly represented. Summary of the Invention
[0005] The present application provides a similarity-based refrigerator door not-closed detection system to solve the problem that the existing refrigerator door closing technology needs to rely on inherent settings, such as setting sensors, to achieve detection, making it impossible to continue detection once such means fail.
[0006] The system comprises:
[0007] a data acquisition module configured to acquire historical fault characteristic data and real-time characteristic data; the historical fault characteristic data is used to characterize the multi-dimensional situation when the refrigerator door is not closed due to a fault, and the real-time characteristic data is used to characterize the real-time multi-dimensional situation after the refrigerator door is closed;
[0008] a fault standard generating module, the fault standard generating module being configured to generate a valid historical fault data curve according to the historical fault characteristic data;
[0009] A similarity calculation module is configured to calculate the similarity between the real-time feature data and the effective historical fault data curve; the similarity is used to characterize the tightness of the current refrigerator door after closing.
[0010] Preferably, the data acquisition module includes:
[0011] a fault data acquisition unit configured to acquire historical fault data; the historical fault data is used to characterize multi-dimensional situations in which the temperature inside the refrigerator undergoes abnormal changes;
[0012] a fault data screening unit, the fault data screening unit being configured to perform data screening based on the historical fault data to obtain target historical fault data; the target historical fault data is used to characterize a multi-dimensional situation when the refrigerator door is not tightly closed;
[0013] a fault data analysis unit, the fault data analysis unit being configured to perform feature analysis based on the target historical fault data to obtain the historical fault feature data;
[0014] A real-time data acquisition unit is configured to acquire the real-time feature data.
[0015] Preferably, the fault data analysis unit is further configured to:
[0016] Determining whether a first temperature feature exists in a temperature variation cycle of a corresponding refrigerator internal compartment in the target historical fault data;
[0017] If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a second temperature feature; the temperature cycle sequence corresponding to the second temperature feature is after the temperature cycle sequence corresponding to the first temperature feature;
[0018] If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a third temperature feature; the temperature cycle sequence corresponding to the third temperature feature is after the temperature cycle sequence corresponding to the second temperature feature;
[0019] If so, the historical fault feature data is generated according to the first temperature feature, the second temperature feature, the third temperature feature, and the target historical fault data.
[0020] Preferably, the first temperature characteristic, the second temperature characteristic and the third temperature characteristic are temperature characteristics within the same temperature period;
[0021] The first temperature characteristic includes that after the refrigerator door is opened and closed once, the temperature of the internal compartment of the refrigerator rises to a first temperature within a first time; the first temperature characteristic includes that the temperature of the internal compartment of the refrigerator is higher than a second temperature and runs for a second time; the third temperature characteristic includes that after the refrigerator door is opened and closed next time, the temperature of the internal compartment of the refrigerator drops by a third temperature within a third time.
[0022] Preferably, the fault standard generation module includes:
[0023] a category screening unit, the category screening unit being configured to screen refrigerator categories according to the historical fault feature data to obtain a plurality of groups of screening data;
[0024] A curve generating unit is configured to generate a plurality of valid historical fault data curves according to each set of the screening data; the valid historical fault data curves correspond one to one with the screening data.
[0025] Preferably, the refrigerator category includes refrigerator refrigeration system category, refrigerator model category and refrigerator door gap size category.
[0026] Preferably, the multi-dimensional conditions include compartment temperature conditions, compartment evaporator temperature conditions, compartment set temperature conditions, compartment temperature and evaporator temperature difference conditions, and compartment temperature and set temperature difference conditions.
[0027] Preferably, the similarity calculation module is further configured to:
[0028] The similarities between the real-time feature data and the first temperature feature, the second temperature feature, and the third temperature feature are calculated in sequence, and similarity intervals corresponding to the first temperature feature, the second temperature feature, and the third temperature feature are output.
[0029] Preferably, the similarity calculation module is further configured to:
[0030] generating a first similarity matrix based on the real-time feature data and the first temperature feature, calculating a first similarity between the real-time feature data and the first temperature feature based on the first similarity matrix, and outputting a first similarity interval corresponding to the first similarity and belonging to the real-time feature data;
[0031] removing the first similar interval from the real-time feature data to obtain first real-time feature data;
[0032] generating a second similarity matrix based on the first real-time feature data and the second temperature feature, calculating a second similarity between the first real-time feature data and the second temperature feature based on the second similarity matrix, and outputting a second similarity interval corresponding to the second similarity and belonging to the first real-time feature data;
[0033] removing the second similar interval from the first real-time feature data to obtain second real-time feature data;
[0034] A third similarity matrix is generated based on the second real-time feature data and the third temperature feature, a third similarity between the second real-time feature data and the third temperature feature is calculated based on the third similarity matrix, and a third similarity interval corresponding to the third similarity and belonging to the second real-time feature data is output.
[0035] Preferably, the system further comprises:
[0036] The early warning module is configured to generate early warning information when the similar interval exists, and feed back the early warning information to the user.
[0037] As can be seen from the above content, the present application provides a refrigerator door not closed tightly detection system based on similarity, the system includes a data acquisition module, the data acquisition module is configured to acquire historical fault characteristic data and real-time characteristic data; the historical fault characteristic data is used to characterize the multi-dimensional situation when the refrigerator door is not closed tightly due to a fault, and the real-time characteristic data is used to characterize the real-time multi-dimensional situation after the refrigerator door is closed; a fault standard generation module, the fault standard generation module is configured to generate a valid historical fault data curve based on the historical fault characteristic data; a similarity calculation module, the similarity calculation module is configured to calculate the similarity between the real-time characteristic data and the valid historical fault data curve; the similarity is used to characterize the tightness of the current refrigerator door after closing. The present application solves the problem that the existing refrigerator door closing technology needs to rely on inherent settings, such as setting sensors, to achieve detection, so that once such means fail, continuous detection cannot be achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0039] Figure 1 This is a schematic diagram of a refrigerator door not-closed detection system based on similarity in this application;
[0040] Figure 2 This is a schematic diagram of a data acquisition module in a refrigerator door not tightly closed detection system based on similarity in this application;
[0041] Figure 3 This is a schematic diagram of a fault standard generation module in a refrigerator door not closed tightly detection system based on similarity in this application. DETAILED DESCRIPTION
[0042] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.
[0043] It should be noted that the brief descriptions of terms in this application are only for the purpose of facilitating the understanding of the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise specified, these terms should be understood according to their ordinary and usual meanings.
[0044] It should be noted that, in this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described in this application as "exemplary" or "for example" should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.
[0045] Figure 1 This is a schematic diagram of a refrigerator door not-closed detection system based on similarity in this application.
[0046] See also Figure 1 It can be seen that this embodiment provides a refrigerator door not tightly closed detection system based on similarity, the system comprising:
[0047] The data acquisition module 100 is configured to acquire historical fault characteristic data and real-time characteristic data; the historical fault characteristic data is used to characterize the multi-dimensional situation when the refrigerator door is not closed tightly due to a fault, and the real-time characteristic data is used to characterize the real-time multi-dimensional situation after the refrigerator door is closed.
[0048] Specifically, in this embodiment, the data acquisition module 100 is used to acquire refrigerator fault data, wherein the fault data is the historical fault characteristic data of the refrigerator, refrigerators of the same model, or refrigerators of other models.
[0049] Since this embodiment detects all types of refrigerators for doors not being tightly closed, the acquired historical fault feature data should be fault data of all types of refrigerators and should be classified so as to be applicable to all types of refrigerators.
[0050] Among them, considering that a single data evaluation standard cannot accurately detect whether the refrigerator door is not closed tightly, the historical fault feature data obtained in this embodiment is multi-dimensional data. If accurate detection is desired, the corresponding real-time feature data obtained is also multi-dimensional data, and the real-time feature data has the same data dimension as the historical fault feature data.
[0051] The system further comprises:
[0052] The fault standard generating module 200 is configured to generate a valid historical fault data curve according to the historical fault characteristic data.
[0053] Specifically, in this embodiment, the effective historical fault data curve generated by the fault standard generation module 200 is relied upon to judge whether the refrigerator door is closed tightly. The effective historical fault data curve can be understood as a curve composed of the historical fault characteristic data. By searching for a state similar to the current refrigerator state on the curve, it is possible to detect whether the refrigerator door is closed tightly.
[0054] The system further comprises:
[0055] The similarity calculation module 300 is configured to calculate the similarity between the real-time feature data and the effective historical fault data curve; the similarity is used to characterize the tightness of the current refrigerator door after closing.
[0056] Specifically, in this embodiment, since there are certain differences between the current fault state and the previous fault, direct comparison is not adopted. Instead, the similarity between the real-time feature data and the valid historical fault data curve is calculated, and the tightness of the refrigerator door after closing is measured by the similarity, so as to determine whether the refrigerator door is closed tightly.
[0057] Figure 2 This is a schematic diagram of a data acquisition module in a refrigerator door not tightly closed detection system based on similarity in this application.
[0058] See also Figure 2 It can be seen that, further, in some embodiments, the data acquisition module 100 includes:
[0059] The fault data acquisition unit 110 is configured to acquire historical fault data; the historical fault data is used to characterize the multi-dimensional situation of abnormal changes in the temperature inside the refrigerator.
[0060] Specifically, in this embodiment, the fault data acquisition unit 110 is used to directly obtain the historical fault data. It should be noted that the historical fault data is different from the historical fault characteristic data. The historical fault data is all situations where the internal temperature of the refrigerator is abnormal, while the historical fault characteristic data is only for situations where the internal temperature of the refrigerator is abnormal due to the refrigerator door not being closed tightly.
[0061] The data acquisition module 100 further includes:
[0062] The fault data screening unit 120 is configured to perform data screening based on the historical fault data to obtain target historical fault data; the target historical fault data is used to characterize the multi-dimensional situation when the refrigerator door is not closed tightly.
[0063] Specifically, in this embodiment, since the historical fault data characterizes all situations in which the internal temperature of the refrigerator is abnormal, it needs to be screened before it can be used as data indicating that the refrigerator door is not closed tightly. Therefore, the historical fault data is screened by the fault data screening unit 120 to obtain the target historical fault data.
[0064] It should be noted that the target historical fault data and the historical fault characteristic data are also different to a certain extent. The target historical fault data does not include the specific temperature characteristic data for examining whether the refrigerator door is tightly closed in this embodiment.
[0065] The data acquisition module 100 further includes:
[0066] The fault data analysis unit 130 is configured to perform feature analysis based on the target historical fault data to obtain the historical fault feature data.
[0067] Specifically, in this embodiment, the historical fault characteristic data is obtained only after the target historical fault data passes the characteristic analysis of the fault data analysis unit 130 and the characteristic data obtained by the analysis is added to the target historical fault data.
[0068] The data acquisition module 100 further includes:
[0069] The real-time data acquisition unit 140 is configured to acquire the real-time feature data.
[0070] Specifically, in this embodiment, the real-time feature data and the historical fault feature data are acquired through two channels, so it is necessary to set up another real-time data acquisition unit 140 to acquire the real-time feature data to avoid data confusion.
[0071] Furthermore, in some embodiments, the fault data analysis unit 130 is further configured to:
[0072] Determining whether a first temperature feature exists in a temperature variation cycle of a corresponding refrigerator internal compartment in the target historical fault data;
[0073] If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a second temperature feature; the temperature cycle sequence corresponding to the second temperature feature is after the temperature cycle sequence corresponding to the first temperature feature;
[0074] If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a third temperature feature; the temperature cycle sequence corresponding to the third temperature feature is after the temperature cycle sequence corresponding to the second temperature feature;
[0075] If so, the historical fault feature data is generated by combining the first temperature feature, the second temperature feature, the third temperature feature and the target historical fault data.
[0076] Specifically, in this embodiment, whether the refrigerator door is closed tightly is set as the first temperature characteristic, the second temperature characteristic, and the third temperature characteristic. These three temperature characteristics can be understood as temperature changes in three sub-periods within the same temperature period.
[0077] The fault data analysis unit 130 judges the target historical fault data with the entire temperature change cycle, thereby judging in turn whether it has the first temperature characteristic, the second temperature characteristic and the third temperature characteristic, and uses the target historical fault data having these three temperature characteristics as the valid data in the valid historical fault data curve, and the data in the valid historical fault data curve is supplemented with the first temperature characteristic, the second temperature characteristic and the third temperature characteristic.
[0078] The first temperature characteristic, the second temperature characteristic, and the third temperature characteristic are temperature characteristics within the same temperature period.
[0079] The first temperature characteristic includes that after the refrigerator door is opened and closed once, the temperature of the internal compartment of the refrigerator rises to a first temperature within a first time.
[0080] The first temperature characteristic includes that the temperature of the inner compartment of the refrigerator is higher than a second temperature and is operated for a second time.
[0081] The third temperature characteristic includes that the temperature of the internal compartment of the refrigerator drops by a third temperature within a third time after the refrigerator door is opened and closed next time.
[0082] Figure 3 This is a schematic diagram of a fault standard generation module in a refrigerator door not closed tightly detection system based on similarity in this application.
[0083] See also Figure 3 It can be seen that, further, in some embodiments, the fault criterion generating module 200 includes:
[0084] The category screening unit 210 is configured to screen refrigerator categories according to the historical fault feature data to obtain several groups of screening data.
[0085] Specifically, in this embodiment, since the historical fault characteristic data includes the fault conditions of refrigerator doors not being tightly closed for all types of refrigerators, if one wants to detect only the condition of refrigerator doors not being tightly closed for a certain type of refrigerator, it is necessary to filter out the characteristic data for this type of refrigerator from all the historical fault characteristic data.
[0086] The fault criterion generating module 200 further includes:
[0087] The curve generating unit 220 is configured to generate a plurality of valid historical fault data curves according to each set of the screening data; the valid historical fault data curves correspond to the screening data one by one.
[0088] Specifically, in this embodiment, a valid historical fault data curve is generated for each type of the screening data, so as to be suitable for all types of refrigerators.
[0089] The refrigerator category may include a refrigerator refrigeration system category, a refrigerator model category, and a refrigerator door gap size category.
[0090] The multi-dimensional conditions include compartment temperature conditions, compartment evaporator temperature conditions, compartment set temperature conditions, compartment temperature and evaporator temperature difference conditions, and compartment temperature and set temperature difference conditions.
[0091] Furthermore, in some embodiments, the similarity calculation module 300 is further configured to:
[0092] The similarities between the real-time feature data and the first temperature feature, the second temperature feature, and the third temperature feature are calculated in sequence, and similarity intervals corresponding to the first temperature feature, the second temperature feature, and the third temperature feature are output.
[0093] Specifically, in this embodiment, the similarity is the similarity between the real-time feature data and the first temperature feature, the second temperature feature and the third temperature feature in the valid historical fault data curve, that is, the real-time feature data and the valid historical fault data curve both have the first temperature feature, the second temperature feature and the third temperature feature, and the similarity of these three temperature features between the two is calculated to obtain the similarity, and after the similarity is calculated, the similarity interval between the two needs to be output.
[0094] The similarity interval can be understood as an interval of temperature change cycles on the real-time feature data that is similar to the effective historical fault data curve, and the first temperature feature, the second temperature feature, and the third temperature feature each correspond to a similarity interval.
[0095] The steps for calculating the similarity between the real-time feature data and the valid historical fault data curve are as follows:
[0096] generating a first similarity matrix based on the real-time feature data and the first temperature feature, calculating a first similarity between the real-time feature data and the first temperature feature based on the first similarity matrix, and outputting a first similarity interval corresponding to the first similarity and belonging to the real-time feature data;
[0097] removing the first similar interval from the real-time feature data to obtain first real-time feature data;
[0098] generating a second similarity matrix based on the first real-time feature data and the second temperature feature, calculating a second similarity between the first real-time feature data and the second temperature feature based on the second similarity matrix, and outputting a second similarity interval corresponding to the second similarity and belonging to the first real-time feature data;
[0099] removing the second similar interval from the first real-time feature data to obtain second real-time feature data;
[0100] A third similarity matrix is generated based on the second real-time feature data and the third temperature feature, a third similarity between the second real-time feature data and the third temperature feature is calculated based on the third similarity matrix, and a third similarity interval corresponding to the third similarity and belonging to the second real-time feature data is output.
[0101] Furthermore, in some embodiments, the system further comprises:
[0102] The warning module 400 is configured to generate warning information when the similar interval exists, and feed back the warning information to the user.
[0103] Specifically, in this embodiment, when the existence of the similar interval is detected, it indicates that the refrigerator door is not tightly closed, and the user is warned through the warning module 400.
[0104] For example, the entire process of the system provided in this embodiment is as follows:
[0105] Acquisition of historical fault feature data: Obtaining a dataset of fault label curves for doors not fully closed can be done in the following three ways:
[0106] The first method involves obtaining data from a refrigerator fault detection big data system. This refrigerator diagnostic system feeds offline data from regular time periods for various refrigerator fault types into corresponding diagnostic algorithms to diagnose whether a particular fault occurred within that time period. However, the fault diagnosis results for some fault types contain historical trend data whose fault characteristics do not correspond to those of the fault type. This dataset was selected based on refrigerators from different systems and models that share common characteristics.
[0107] The second method is to obtain the data through experimental testing. Based on refrigerators of different models and systems, different door gap sizes are selected to reproduce the data characteristics of doors not being closed tightly in various situations.
[0108] The third method: Combining the first and second methods, the data sets obtained by the two methods are compared and analyzed, and representative data sets with obvious characteristics of each category of refrigerator door not closed tightly are selected.
[0109] Fault type confirmation: The selected dataset of refrigerator door failures was discussed with industry experts to confirm whether they were genuine failures. This generated a historical dataset of refrigerator door failures across different systems, models, and door gap sizes. This historical dataset of refrigerator door failures in various situations served as a label for subsequent diagnostic datasets.
[0110] Data feature analysis: fully analyze the label data set, explore the characteristics of each dimension, such as compartment temperature, set temperature, ambient temperature, compartment evaporator temperature, temperature changes within a certain period of time, temperature anomaly duration, etc., analyze the correlation between the characteristics, and screen the key core dimension characteristics. This series of features includes: when there is door switch information, the compartment temperature fluctuates and rises by m degrees within a short time t1 hour (sharp heating stage R1), and the compartment temperature is kept higher than the set temperature n degrees for continuous operation t hours, and then before returning to normal, it becomes a periodic temperature change trend. The overall temperature difference between the compartment temperature and the evaporator in each cycle increases (periodic maintenance stage R2), until there is door switch information, the temperature fluctuates and drops by k degrees in a short time t2 (temperature recovery stage R3). The above process is called a complete door not closed fault.
[0111] Filtering historical label sets: In the historical label big data set, there are multiple samples of refrigerators of different systems and models. Among the samples of the same type of equipment, the temperature change stages of each equipment sample in the fault area are extracted. In the area with multiple faults, the big data selects multiple dimensions such as compartment temperature, compartment evaporator temperature, compartment set temperature, the difference between compartment temperature and evaporator temperature, and the difference between compartment temperature and set temperature based on the correlation method. The fault area with the greatest correlation between each stage of the multiple fault area and the corresponding stage of other equipment samples is found as the label data. In this way, the best label data set (R1, R2, R3) can be filtered out among the equipment of the same system, model and door gap size. R1 represents the trend historical labels of each dimension in the rapid temperature rise stage in the area where a single fault has just occurred, R2 represents the trend historical labels of each dimension in the periodic maintenance stage in the area after the single fault has occurred, and R3 represents the trend historical labels of each dimension in the temperature recovery stage in the area where the single fault is about to disappear.
[0112] Similarity algorithm selection: Combined with historical label data, the accuracy of multiple similarity algorithms was tested, evaluated and compared, and the cosine similarity algorithm was finally selected as the application detection algorithm in this embodiment.
[0113] Detailed testing process: Get the data set to be tested within a period of time (including at least one defrost cycle),
[0114] Calculate the similarity between this dataset and the labeled datasets in the order R1, R2, and R3. First, generate a similarity matrix M1 with R1. Set the similarity to a threshold value, such as 0.99, and filter similar intervals with a duration greater than a certain time, such as 2 hours. Generate multiple continuous similar intervals. When the time interval between two similar intervals is less than a threshold value, such as 10 minutes, merge the similar intervals. If they exist, obtain the final similar interval set S1, each of which contains the start and end time of the fault. Subsequently, calculate the similarity between the data set under test after deducting S1 and R2, using the same calculation process as R1, to obtain the final similar interval with R2. The similarity calculation method with R3 is the same as above.
[0115] Detection result analysis: Analyze the detection results based on the specific situation. If the detection purpose is to analyze the fault conditions during refrigerator operation and calculate the frequency of door failure, then during diagnosis, the similarity between the test dataset and the labeled datasets R1, R2, and R3 can be calculated according to the process, and the corresponding fault interval can be output. If the detection purpose is to detect possible door failures during refrigerator operation in real time, diagnose and detect faults in a timely manner, and issue a timely background warning, then during detection, it is only necessary to calculate the similarity interval between the test dataset and the labeled datasets R1 and R2. When similar intervals exist, a timely background system warning will be issued. When the background warnings accumulate to a certain amount, the user can be notified of the door failure through the connected app at an appropriate time, such as during non-working hours.
[0116] Generalization of this detection algorithm: Using a labeled dataset of a certain model from a certain system, we applied the above detection process to a dataset of historical refrigerator failures from other models of the same system to test whether the algorithm can correctly detect the fault interval. Currently, the generalization effect has been good. Whether this solution can be generalized to the same model in other systems, or to different models of refrigerators in other systems, can be tested in real-world scenarios.
[0117] This embodiment has the following advantages:
[0118] 1. This embodiment's algorithm logic is simple. It only requires obtaining a historically confirmed dataset of labeled cases of refrigerator doors not being properly closed across different systems and models, and then calculating the similarity interval between the test data and this labeled dataset, which is greater than a threshold. This eliminates the need for traditional rule-based refrigerator diagnosis algorithms to analyze various thresholds for varying dimensions within the fault area, which often require reanalysis and optimization of data as the refrigerator ages. This also improves the timeliness and accuracy of diagnosis.
[0119] 2. When the door not closing feature changes slightly with changes in refrigerator model and operating time, the similarity method verifies its similarity with the historical label set. If the similarity is high and greater than a certain threshold, the label set is not updated. If it is less than a certain threshold, the label dataset for the corresponding temperature change stage is added to the label set. Rule-based diagnostic methods, on the other hand, require re-analysis of big data to add the various dimensional features of the door not closing, and feature engineering analysis to determine the dimensions required for the new rules. Fault area curve analysis and statistical analysis of the thresholds of each dimensional feature are performed to optimize and update various thresholds. The setting of the threshold is limited by the size of the label dataset. Therefore, the similarity method improves the generalization of the algorithm and the flexibility of update iterations.
[0120] For ease of explanation, the above description has been made in conjunction with specific embodiments. However, the above discussion of some embodiments is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Based on the above teachings, various modifications and variations can be obtained. The above embodiments are selected and described to better explain the content of this disclosure, thereby enabling those skilled in the art to better use the embodiments.
Claims
1. A refrigerator door not closed tightly detection system based on similarity, characterized in that: The system comprises: A data acquisition module (100), the data acquisition module (100) being configured to acquire historical fault characteristic data and real-time characteristic data; the historical fault characteristic data being used to characterize the multi-dimensional situation when the refrigerator door is not tightly closed due to a fault, and the real-time characteristic data being used to characterize the real-time multi-dimensional situation after the refrigerator door is closed; A fault standard generation module (200), the fault standard generation module (200) being configured to generate an effective historical fault data curve based on the historical fault characteristic data; A similarity calculation module (300) is configured to calculate the similarity between the real-time characteristic data and the effective historical fault data curve; the similarity is used to characterize the tightness of the current refrigerator door after closing.
2. The refrigerator door not closed tightly detection system based on similarity according to claim 1, characterized in that: The data acquisition module (100) comprises: A fault data acquisition unit (110), the fault data acquisition unit (110) being configured to acquire historical fault data; the historical fault data being used to characterize multi-dimensional situations in which the temperature inside the refrigerator undergoes abnormal changes; A fault data screening unit (120), the fault data screening unit (120) being configured to perform data screening based on the historical fault data to obtain target historical fault data; the target historical fault data is used to characterize a multi-dimensional situation when the refrigerator door is not tightly closed; a fault data analysis unit (130), the fault data analysis unit (130) being configured to perform feature analysis based on the target historical fault data to obtain the historical fault feature data; A real-time data acquisition unit (140) is configured to acquire the real-time feature data.
3. The refrigerator door not closed tightly detection system based on similarity according to claim 2, characterized in that: The fault data analysis unit (130) is further configured to: Determining whether a first temperature feature exists in a temperature variation cycle of a corresponding refrigerator internal compartment in the target historical fault data; If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a second temperature feature; the temperature cycle sequence corresponding to the second temperature feature is after the temperature cycle sequence corresponding to the first temperature feature; If so, determining whether the temperature change cycle of the refrigerator internal compartment corresponding to the target historical fault data has a third temperature feature; the temperature cycle sequence corresponding to the third temperature feature is after the temperature cycle sequence corresponding to the second temperature feature; If so, the historical fault feature data is generated according to the first temperature feature, the second temperature feature, the third temperature feature, and the target historical fault data.
4. The refrigerator door ajar detection system based on similarity according to claim 3, characterized in that: The first temperature characteristic, the second temperature characteristic, and the third temperature characteristic are temperature characteristics within the same temperature period; The first temperature characteristic includes that after the refrigerator door is opened and closed once, the temperature of the internal compartment of the refrigerator rises to a first temperature within a first time; the first temperature characteristic includes that the temperature of the internal compartment of the refrigerator is higher than a second temperature and runs for a second time; the third temperature characteristic includes that after the refrigerator door is opened and closed next time, the temperature of the internal compartment of the refrigerator drops by a third temperature within a third time.
5. The refrigerator door not closed tightly detection system based on similarity according to claim 1, characterized in that: The fault standard generation module (200) comprises: A category screening unit (210), the category screening unit (210) being configured to screen refrigerator categories according to the historical fault characteristic data to obtain a plurality of groups of screening data; A curve generating unit (220) is configured to generate a plurality of valid historical fault data curves according to each set of the screening data; the valid historical fault data curves correspond one-to-one to the screening data.
6. The refrigerator door not closed tightly detection system based on similarity according to claim 5, characterized in that: The refrigerator category includes a refrigerator refrigeration system category, a refrigerator model category, and a refrigerator door gap size category.
7. The refrigerator door ajar detection system based on similarity according to claim 1, characterized in that: The multi-dimensional conditions include compartment temperature conditions, compartment evaporator temperature conditions, compartment set temperature conditions, compartment temperature and evaporator temperature difference conditions, and compartment temperature and set temperature difference conditions.
8. The refrigerator door ajar detection system based on similarity according to claim 4, characterized in that: The similarity calculation module (300) is further configured to: The similarities between the real-time feature data and the first temperature feature, the second temperature feature, and the third temperature feature are calculated in sequence, and similarity intervals corresponding to the first temperature feature, the second temperature feature, and the third temperature feature are output.
9. The refrigerator door ajar detection system based on similarity according to claim 8, characterized in that: The similarity calculation module (300) is further configured to: generating a first similarity matrix based on the real-time feature data and the first temperature feature, calculating a first similarity between the real-time feature data and the first temperature feature based on the first similarity matrix, and outputting a first similarity interval corresponding to the first similarity and belonging to the real-time feature data; removing the first similar interval from the real-time feature data to obtain first real-time feature data; generating a second similarity matrix based on the first real-time feature data and the second temperature feature, calculating a second similarity between the first real-time feature data and the second temperature feature based on the second similarity matrix, and outputting a second similarity interval corresponding to the second similarity and belonging to the first real-time feature data; removing the second similar interval from the first real-time feature data to obtain second real-time feature data; A third similarity matrix is generated based on the second real-time feature data and the third temperature feature, a third similarity between the second real-time feature data and the third temperature feature is calculated based on the third similarity matrix, and a third similarity interval corresponding to the third similarity and belonging to the second real-time feature data is output.
10. The refrigerator door not closed tightly detection system based on similarity according to claim 9, characterized in that: The system further comprises: The early warning module (400) is configured to generate early warning information when the similar interval exists, and feed back the early warning information to the user.