Refrigerator fault positioning method and device based on feature similarity, medium and equipment

Abstract

The application relates to a refrigerator fault positioning method and device based on feature similarity, a medium and equipment. The method comprises the following steps: acquiring collection data of each sensor uploaded by a refrigerator in a current unit time; extracting feature data corresponding to the current unit time from the collection data of each sensor; determining whether the refrigerator has a fault according to feature data corresponding to each unit time in a time window in which the current unit time is located and standard feature data obtained by big data statistics; and if it is determined that the refrigerator has a fault, determining a fault type of the refrigerator according to feature similarity between feature data corresponding to the current unit time and feature data corresponding to each type of fault obtained by big data statistics. The embodiment of the application can improve fault positioning efficiency.

Description

Technical Field

[0001] This invention relates to the field of fault handling technology, and in particular to a method, apparatus, medium, and equipment for locating refrigerator faults based on feature similarity. Background Technology

[0002] With the development of society and the economy and the improvement of people's living standards, refrigerators have become a necessity in every household. Current refrigerators are controlled by an MCU (Microcontroller Unit). During operation, the MCU uses sensor feedback and related logic to determine if some components are malfunctioning. However, refrigerator MCUs typically only perform simple logical checks, so the accuracy of fault diagnosis is not high. When the device malfunctions and cannot function properly, users must contact after-sales personnel for a detailed inspection before diagnosis and repair can be performed. Summary of the Invention

[0003] To solve the above-mentioned technical problems, or at least partially solve them, the present invention provides a refrigerator fault location method, device, medium, and equipment based on feature similarity.

[0004] In a first aspect, embodiments of the present invention provide a refrigerator fault location method based on feature similarity, comprising:

[0005] Acquire the data collected by each sensor uploaded by the refrigerator within the current unit of time;

[0006] The feature data corresponding to the current unit of time is extracted from the data collected by each sensor;

[0007] Based on the characteristic data corresponding to each unit time within the time window where the current unit time is located, and the standard characteristic data obtained in advance through big data statistics, it is determined whether the refrigerator has a malfunction.

[0008] If the refrigerator is determined to be faulty, the fault type of the refrigerator is determined based on the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault obtained through big data statistics.

[0009] Secondly, embodiments of the present invention provide a refrigerator fault location device based on feature similarity, comprising:

[0010] The data acquisition module is used to acquire the data collected by the various sensors uploaded by the refrigerator within the current unit of time.

[0011] The feature extraction module is used to extract the feature data corresponding to the current unit time from the data collected by each sensor;

[0012] The fault determination module is used to determine whether the refrigerator has a fault based on the characteristic data corresponding to each unit time within the time window where the current unit time is located and the standard characteristic data obtained in advance through big data statistics.

[0013] The type determination module is used to determine the fault type of the refrigerator if it is determined that the refrigerator has a fault, based on the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault obtained through big data statistics.

[0014] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method provided in the first aspect.

[0015] Fourthly, embodiments of the present invention provide a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method provided in the first aspect.

[0016] The refrigerator fault location method, apparatus, medium, and equipment based on feature similarity provided in this invention have the following beneficial effects when combined:

[0017] (1) Acquire the data collected by each sensor uploaded by the refrigerator within the current unit time. Extract the feature data corresponding to the current unit time from the data collected by each sensor. Based on the feature data corresponding to each unit time within the time window of the current unit time and the standard feature data obtained in advance through big data statistics, determine whether the refrigerator has a fault. If the refrigerator has a fault, determine the fault type of the refrigerator based on the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault obtained through big data statistics. It can be seen that the method provided by this embodiment of the invention performs a series of analyses based on the data collected by each sensor to determine whether the refrigerator has a fault, and if so, what type of fault it is. This method can locate faults without requiring on-site inspection by maintenance personnel, which can greatly improve the efficiency of fault location.

[0018] (2) In one embodiment, the feature data of each unit time within the time window is compared with the standard feature data. The refrigerator is judged to have a fault by comparing the results of multiple consecutive unit times, which can improve the accuracy of fault judgment.

[0019] (3) In one embodiment, before feature extraction, missing value processing and outlier processing are performed on the collected data, which can improve the reliability of the collected data and thus improve the accuracy of fault judgment. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a refrigerator fault location method based on feature similarity in one embodiment of the present invention.

[0023] Figure 2 This is a structural block diagram of a refrigerator fault location device based on feature similarity in one embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] This invention provides a refrigerator fault location method based on feature similarity.

[0026] See Figure 1 The method includes the following steps S110 to S140:

[0027] S110. Obtain the data collected by each sensor uploaded by the refrigerator within the current unit of time.

[0028] In real-world scenarios, after collecting data, the various sensors in the refrigerator upload the data to a big data platform in the cloud for storage. Therefore, the collected data can be obtained from the big data platform in the cloud.

[0029] The unit of time can be in the form of a day, that is, step S110 is executed once every day at 24:00 to obtain the data collected by each sensor for that day.

[0030] The sensors inside the refrigerator may include temperature sensors, humidity sensors, etc., to collect temperature, humidity and other status data inside the refrigerator.

[0031] Understandably, since there is a connection between the refrigerator and the big data platform in the cloud, the refrigerator uploads the collected data to the big data platform according to the Internet of Things (IoT) communication protocol. Therefore, the collected data can also be called IoT data.

[0032] The data collected by each sensor is a data sequence arranged in chronological order.

[0033] S120. Extract the feature data corresponding to the current unit time from the data collected by each sensor;

[0034] In other words, feature extraction is performed on the data collected by various sensors to obtain feature data, which can be called the feature data corresponding to the current unit of time.

[0035] In one embodiment, S120 may specifically include A1 to A3:

[0036] A1. If the collected data is categorical, then one-hot encoding is used to convert the categorical collected data into numerical collected data, and the numerical collected data is used as feature data; wherein, the categorical collected data is state data.

[0037] For example, categorical data such as compressor status, damper status, fan status, and heater status need to be converted into numerical data. Specifically, one-hot encoding is used to convert this data into numerical data, achieving sparse encoding of categorical data from low to high dimensions. The resulting numerical data is then used as the corresponding feature data.

[0038] One-hot encoding, also known as one-bit valid encoding, uses a multi-bit state register to encode multiple states. Each state has its own independent register bit, and at any given time, only one bit is valid.

[0039] A2. Based on the data collected by the temperature sensor, generate temperature difference feature data and temperature change rate feature data, wherein the temperature difference feature data is the temperature difference between each collected data. Before calculating the temperature difference, subtract the refrigerator compartment set temperature from each collected data.

[0040] In other words, for temperature data, the corresponding feature data includes two categories: temperature difference feature data and temperature change rate feature data. Temperature difference feature data refers to the temperature difference between various temperatures in the collected data. Temperature change rate feature data refers to the change in temperature relative to a baseline value at each time point.

[0041] Specifically, because different refrigerators may operate under different fault conditions, their compartment temperatures exhibit different distributions. To ensure the universality of the feature data, relative values ​​of compartment temperature differences are used to avoid the influence of individual refrigerators on the features. Since different refrigerators have different set values ​​for compartment temperature, the set temperature of each refrigerator is subtracted from each collected data point to obtain individual depersonalized collected data. Then, the differences between the depersonalized collected data are calculated to obtain the temperature difference feature data.

[0042] Specifically, the characteristic data of the rate of temperature change can be determined using the first calculation formula, which is: Δε t =ε t -ε o ;

[0043] Where, Δε t ε represents the characteristic data of the rate of temperature change at time t within the current unit of time. t ε represents the data collected at time t within the current unit of time; O As the baseline value; if Δε t With Δε t1 If the directions are the same, then ε o Unchanged; if Δε t With Δε t-1 If the directions are opposite, then the second calculation formula is used for ε. o Update; the second calculation formula is ε′ o =aε o -(1-a)ε t-1 ,ε′ o is the updated baseline value used to calculate the temperature change rate characteristic data at time t+1; a is the forgetting factor.

[0044] In this context, "same direction" means both values ​​are positive or both are negative. "Opposite direction" means one value is positive and the other is negative. For time t+1, after updating the baseline value, the updated baseline value is input into the first calculation formula, and then the collected data at time t+1 is input into the first calculation formula to obtain the temperature change rate characteristic data at time t+1.

[0045] It is evident that different feature extraction methods are used for different types of data.

[0046] A3. High-dimensional transformation of feature data is performed through polynomial transformation to obtain nonlinear feature data.

[0047] Understandably, feature data can be obtained through A1 to A2 above. To further obtain high-dimensional feature data, a polynomial transformation is used to perform a high-dimensional transformation on the feature data, resulting in high-dimensional feature data that is non-linear. For example, a polynomial of degree 2 can be used to perform the high-dimensional transformation of the feature data. To avoid overfitting, only the cross terms are taken as high-dimensional feature data.

[0048] Thus, we obtain the feature data corresponding to different collected data, and these are high-dimensional feature data.

[0049] In one embodiment, before performing the step "extracting the feature data corresponding to the current unit time from the collected data" in step S120, the method may further include B1 to B3:

[0050] B1. Count the number of missing values ​​in the data collected by each sensor;

[0051] B2. If the percentage of missing values ​​in the data collected by a sensor exceeds a preset percentage, the data collected by that sensor will be deleted.

[0052] B3. If the proportion of missing values ​​in a sensor does not exceed the preset proportion, the missing values ​​are filled. Specifically, for missing values ​​in numerical data, the mean is used to fill them; for missing values ​​in categorical data, the mode is used to fill them.

[0053] In other words, before feature extraction, it is first determined whether there are missing values ​​in the collected data, and then the number of missing values ​​determines whether to retain the collected data.

[0054] For example, if the collected data is a temperature sequence, and temperature values ​​are missing at certain points in the sequence, the system checks if the percentage of missing temperature values ​​reaches 60%. If it does, the temperature sequence is deleted, and no further feature extraction is performed on it. If the percentage of missing temperature values ​​is less than 60%, the temperature sequence can be retained, but the missing values ​​need to be filled in. Since temperature values ​​are numerical data, the mean can be used for filling; for example, the temperature value at the missing location can be the average of several surrounding temperature values.

[0055] For example, if the number of missing values ​​in a state sequence is relatively small, the state sequence can be retained. Specifically, the mode can be used to fill in the missing values.

[0056] The mode refers to the value that has a clear central tendency in the statistical distribution. It represents the general level of the data and is also the value that appears most frequently in a set of data. In other words, the state that appears most frequently in the state sequence is selected to fill the missing state.

[0057] Furthermore, before extracting the feature data corresponding to the current unit time from the collected data and after filling in missing values, the process may further include the following steps C1 to C3:

[0058] C1. Sort the data collected by each sensor in ascending order to obtain a data sequence;

[0059] C2. Calculate the difference between the data at three-quarters position and the data at one-quarter position in the data sequence, and determine the outlier distribution range based on the difference;

[0060] C3. Remove data that falls within the outlier distribution range;

[0061] The outlier distribution intervals include [-∞, Q1-3*IQR] and [Q3+3*IQR, +∞], where Q1 is the data at the quarter position, Q3 is the data at the three-quarter position, and IQR is the difference.

[0062] For example, given a temperature sequence, the temperature values ​​are arranged in ascending order to obtain a data sequence with numerical order. Then, the temperature values ​​at three-quarters and one-quarter positions are taken. If the one-quarter and / or three-quarter positions are not integers, they can be determined by rounding up or down. The difference between the temperature values ​​at the three-quarters and one-quarter positions is calculated, and based on this difference, an outlier distribution range is determined. Temperature values ​​within this range are removed as outliers, thus achieving outlier handling.

[0063] The intervals [-∞, Q1-3*IQR] and [Q3+3*IQR, +∞] are not the ranges where most temperature values ​​are located, but rather the ranges where individual temperatures are located. This method can be used to identify individual special temperature values ​​and then remove them as outliers.

[0064] The above method can be simply referred to as the quartile method.

[0065] S130. Based on the characteristic data corresponding to each unit time within the time window where the current unit time is located and the standard characteristic data obtained in advance through big data statistics, determine whether the refrigerator has a malfunction.

[0066] In other words, this method compares the characteristic data of each unit time within the current time window with the standard characteristic data to determine the difference between the characteristic data and the standard characteristic data within the time window, thereby determining whether the characteristic data for each unit time is normal. The standard characteristic data refers to the characteristic data corresponding to the data collected by each sensor inside the refrigerator under normal operating conditions. This standard characteristic data can be obtained by performing big data statistics and feature extraction on the collected data under normal operating conditions through a big data platform. The big data platform can determine the standard characteristic data offline.

[0067] Instead of simply comparing the characteristic data of the current unit of time with the standard characteristic data, this method compares the characteristic data of each unit of time within the current time window with the standard characteristic data separately, which can improve the accuracy of fault diagnosis.

[0068] In one embodiment, S130 may specifically include:

[0069] S131. Calculate the weighted Euclidean distance between the feature data corresponding to each unit time within the time window where the current unit time is located and the standard feature data;

[0070] S132. If the weighted Euclidean distance corresponding to N consecutive units of time within the time window is greater than the preset distance, then it is determined that the refrigerator has a fault; where N is an integer greater than 1 and less than or equal to the length of the time window.

[0071] For example, if the time window is 4 days long and the current unit of time is today, then the time window includes today and the three days prior to today. For each of these 4 days, the weighted Euclidean distance between the corresponding feature data and the standard feature data is calculated, resulting in 4 distance values. If the distance values ​​for three consecutive days are all greater than a preset distance, it indicates that the refrigerator is malfunctioning. The feature data for the three days prior to today can be obtained during the fault location process for the previous three days.

[0072] The greater the weighted Euclidean distance, the further the feature data deviates from the standard feature data, and therefore the higher the probability of failure. The weighted Euclidean distance can be calculated using the following formula:

[0073]

[0074] In the formula, Dis refers to the weighted Euclidean distance, S is the covariance matrix, and x i For each unit of time within the time window, u represents the i-th data point in the feature data. iLet represent the i-th data in the standard feature data, T be the inverted sign, and n be the number of data in the feature data.

[0075] S140. If it is determined that the refrigerator has a fault, the fault type of the refrigerator is determined based on the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault obtained through big data statistics.

[0076] Understandably, after determining that the refrigerator is faulty via S130, it is necessary to determine what type of fault the refrigerator is experiencing.

[0077] In one embodiment, S140 may specifically include:

[0078] S141. Calculate the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault.

[0079] S142. Calculate the similarity distance corresponding to the feature similarity for each type of fault using the third calculation formula, wherein the third calculation formula is: d sim = 1 - (sim(a,b)) / min(len) a ,len b ); where d sim Let sim(a,b) be the similarity distance, len be the feature similarity, and len be the similarity distance. a len represents the length of the feature data corresponding to the current unit of time. b This refers to the characteristic data corresponding to this type of fault;

[0080] S143. If the similarity distance corresponding to a type of fault is less than a preset threshold, then the refrigerator has that type of fault.

[0081] The process of determining the characteristic data corresponding to each type of fault is similar to the process of determining the characteristic data for the current unit of time. Specifically, the fault type can be determined through after-sales maintenance records. Big data analysis is then performed on the data collected by various sensors when that type of fault occurs. Feature extraction is then performed on the collected data obtained from the big data analysis to obtain the characteristic data for each type of fault. Different types of faults correspond to different characteristic data.

[0082] Feature similarity can be determined using a recursive method based on the longest common subsequence.

[0083] Understandably, this process calculates the feature similarity between the feature data corresponding to the current time unit and the feature data corresponding to each type of fault. After calculating the feature similarity, the similarity distance is calculated, with one similarity distance corresponding to each type of fault. The smaller the similarity distance, the greater the probability that this type of fault has occurred in the current time unit. If the similarity distance is less than a preset threshold, it indicates that the refrigerator has a fault of the corresponding type. For example, the preset threshold is set to 0.2.

[0084] Understandably, refrigerators can have a variety of malfunctions.

[0085] At this point, it can be determined whether the refrigerator is malfunctioning, and if so, what type of malfunction it is.

[0086] It is understood that the method provided by the embodiments of the present invention can efficiently and accurately judge the fault information of networked refrigerators, reduce the time of after-sales maintenance personnel to troubleshoot faults, improve the quality of after-sales service, and has good stability and practicality.

[0087] Secondly, embodiments of the present invention provide a refrigerator fault location device based on feature similarity, see [link to relevant documentation]. Figure 2 The device includes:

[0088] The data acquisition module is used to acquire the data collected by the various sensors uploaded by the refrigerator within the current unit of time.

[0089] The feature extraction module is used to extract the feature data corresponding to the current unit time from the data collected by each sensor;

[0090] The fault determination module is used to determine whether the refrigerator has a fault based on the characteristic data corresponding to each unit time within the time window where the current unit time is located and the standard characteristic data obtained in advance through big data statistics.

[0091] The type determination module is used to determine the fault type of the refrigerator if it is determined that the refrigerator has a fault, based on the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault obtained through big data statistics.

[0092] In one embodiment, the apparatus further includes:

[0093] The missing value processing module is used to process missing values ​​before the feature extraction module extracts the feature data corresponding to the current unit time from the collected data.

[0094] The missing value handling module includes:

[0095] The first statistical unit is used to count the number of missing values ​​in the data collected by each sensor;

[0096] The first deletion unit is used to delete the sensor's data if the proportion of missing values ​​in the sensor's collected data exceeds a preset proportion.

[0097] The first filling unit is used to fill missing values ​​if the proportion of missing values ​​in a sensor does not exceed a preset proportion; specifically, for missing values ​​in numerical data, the mean is used for filling; for missing values ​​in categorical data, the mode is used for filling.

[0098] In one embodiment, the apparatus further includes:

[0099] An anomaly handling module is used to handle anomalies before the feature extraction module extracts the feature data corresponding to the current unit time from the collected data and after the missing value handling module performs missing value handling.

[0100] The exception handling module includes:

[0101] The first sorting unit is used to sort the data collected by each sensor in ascending order to obtain a data sequence;

[0102] An interval determination unit is used to calculate the difference between the data at three-quarters position and the data at one-quarter position in the data sequence, and to determine the outlier distribution interval based on the difference;

[0103] An anomaly removal unit is used to remove data located within the anomaly distribution range.

[0104] The outlier distribution intervals include [-∞, Q1-3*IQR] and [Q3+3*IQR, +∞], where Q1 is the data at the quarter position, Q3 is the data at the three-quarter position, and IQR is the difference.

[0105] In one embodiment, the feature extraction module includes:

[0106] The first encoding unit is used to convert the categorical collected data into numerical collected data using one-hot encoding if the collected data is categorical collected data, and to use the numerical collected data as feature data; wherein the categorical collected data is state data.

[0107] The feature calculation unit is used to generate temperature difference feature data and temperature change rate feature data based on the data collected by the temperature sensor. The temperature difference feature data is the temperature difference between each piece of collected data. Before calculating the temperature difference, the set temperature of the refrigerator compartment is subtracted from each piece of collected data.

[0108] The dimension transformation unit is used to perform high-dimensional transformation on feature data through polynomial transformation to obtain nonlinear feature data.

[0109] Furthermore, the feature calculation unit is used to determine the temperature change rate feature data using a first calculation formula, which is: Δε t =ε t -ε o ;

[0110] Where, Δε t ε represents the characteristic data of the rate of temperature change at time t within the current unit of time. t ε represents the data collected at time t within the current unit of time; o As the baseline value; if Δε t With Δε t-1 If the directions are the same, then ε o Unchanged; if Δε t With Δε t-1 If the directions are opposite, then the second calculation formula is used for ε. o Update; the second calculation formula is ε′ o =aε o -(1-a)ε t-1 ,ε′ o is the updated baseline value used to calculate the temperature change rate characteristic data at time t+1; a is the forgetting factor.

[0111] In one embodiment, the fault determination module includes:

[0112] The distance calculation unit is used to calculate the weighted Euclidean distance between the feature data corresponding to each unit time within the time window of the current unit time and the standard feature data;

[0113] The fault determination unit is used to determine that the refrigerator has a fault if the weighted Euclidean distance corresponding to N consecutive unit times within the time window is greater than a preset distance; wherein N is an integer greater than 1 and less than or equal to the length of the time window.

[0114] In one embodiment, the type determination module includes:

[0115] The first calculation unit is used to calculate the feature similarity between the feature data corresponding to the current unit time and the feature data corresponding to each type of fault.

[0116] The second calculation unit is used to calculate the similarity distance corresponding to the feature similarity for each type of fault using a third calculation formula, wherein the third calculation formula is: d sim = 1 - (sim(a,b) / min(len)a ,len b ); where d sim Let sim(a,b) be the similarity distance, len be the feature similarity, and len be the similarity distance. a len represents the length of the feature data corresponding to the current unit of time. b This refers to the characteristic data corresponding to this type of fault;

[0117] The type determination unit is used to determine if the similarity distance corresponding to a type of fault is less than a preset threshold, in which case the refrigerator has that type of fault.

[0118] It is understood that explanations, examples, and beneficial effects of the relevant content in the apparatus provided in the embodiments of the present invention can be referred to the relevant content in the first aspect, and will not be repeated here.

[0119] Thirdly, embodiments of the present invention provide a computer-readable medium storing computer instructions, which, when executed by a processor, cause the processor to perform the method provided in the first aspect.

[0120] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0121] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0122] Examples of storage media used to provide program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0123] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0124] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0125] It is understood that explanations, specific implementation methods, beneficial effects, examples, etc. of the contents in the computer-readable medium provided in the embodiments of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0126] Fourthly, one embodiment of this specification provides a computing device including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method of any embodiment of the specification.

[0127] It is understood that explanations, specific implementation methods, beneficial effects, examples, etc. of the computing device provided in the embodiments of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0128] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the apparatus embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0129] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, widgets, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium.

[0130] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A refrigerator fault location method based on feature similarity, characterized by, The method comprises the following steps: acquiring collection data of each sensor uploaded by a refrigerator in a current unit time; extracting feature data corresponding to the current unit time from the collection data of each sensor; determining whether the refrigerator has a fault according to feature data corresponding to each unit time in a time window in which the current unit time is located and standard feature data obtained through big data statistics in advance; if it is determined that the refrigerator has a fault, determining a fault type of the refrigerator according to a feature similarity between the feature data corresponding to the current unit time and feature data corresponding to each type of fault obtained through big data statistics; the step of extracting the feature data corresponding to the current unit time from the collection data of each sensor comprises the following steps: if the collection data is category type collection data, converting the category type collection data into numerical type collection data by using one-hot encoding, and taking the numerical type collection data as the feature data; wherein the category type collection data is state data; generating temperature difference feature data and temperature change rate feature data for the collection data of the temperature sensor, wherein the temperature difference feature data is a temperature difference value between each collection data, and each collection data is subtracted by a set temperature of a compartment of the refrigerator before the temperature difference value is calculated; performing high-dimensional transformation on the feature data by using a polynomial transformation method to obtain nonlinear feature data; The temperature change rate characteristic data is determined using a first calculation formula: Δε t = ε t - ε o ; wherein Δε t is the temperature change rate characteristic data at time t in the current unit time, ε t is the collection data at time t in the current unit time; Δε o is a reference value; if Δε t is in the same direction as Δε t-1 , then ε o is unchanged; if Δε t is in the opposite direction of Δε t-1 , then the second calculation formula is used to update ε o ; the second calculation formula is ε' o =aε o -(1-a)ε t-1 , ε' o is the updated reference value, used to calculate the temperature change rate characteristic data at time t+1; a is a forgetting factor.

2. The method of claim 1, wherein, before the feature data corresponding to the current unit time is extracted from the collection data of each sensor, the method further comprises the following steps: counting a number of missing values in the collection data of each sensor; if a proportion of the number of missing values in the collection data of a sensor exceeds a preset proportion, deleting the collection data of the sensor; if the proportion of the number of missing values in the collection data of a sensor does not exceed the preset proportion, filling the missing values; wherein the missing values in the numerical type collection data are filled by using a mean value, and the missing values in the category type collection data are filled by using a mode.

3. The method of claim 2, wherein, before the feature data corresponding to the current unit time is extracted from the collection data of each sensor and after the missing values are filled, the method further comprises the following steps: sorting the collection data of each sensor in ascending order to obtain a data sequence; calculating a difference value between data at a three-quarter position and data at a one-quarter position in the data sequence, and determining an abnormal value distribution interval according to the difference value; eliminating data located in the abnormal value distribution interval; wherein the abnormal value distribution interval comprises [-∞, Q1-3*IQR] and [Q3+3*IQR, +∞], Q1 is the data at the one-quarter position, Q3 is the data at the three-quarter position, and IQR is the difference value.

4. The method of claim 1, wherein, the step of determining whether the refrigerator has a fault according to the feature data corresponding to each unit time in the time window in which the current unit time is located and the standard feature data obtained through big data statistics in advance comprises the following steps: calculating a weighted Euclidean distance between the feature data corresponding to each unit time in the time window in which the current unit time is located and the standard feature data, respectively. If the weighted Euclidean distances corresponding to the N continuous unit times in the time window are all greater than the preset distance, it is determined that the refrigerator has a fault; wherein N is an integer greater than 1 and less than or equal to the length of the time window. 5.A refrigerator fault location device based on feature similarity, characterized by, The method comprises: a data acquisition module configured to acquire collection data of each sensor uploaded by the refrigerator in a current unit time; a feature extraction module configured to extract feature data corresponding to the current unit time from the collection data of each sensor; a fault determination module configured to determine whether the refrigerator has a fault according to the feature data corresponding to each unit time in a time window in which the current unit time is located and standard feature data obtained by big data statistics in advance; a type determination module configured to determine a fault type of the refrigerator according to a feature similarity between the feature data corresponding to the current unit time and feature data corresponding to each type of fault obtained by big data statistics, if it is determined that the refrigerator has a fault. The feature extraction module comprises: a first encoding unit configured to convert category-type collection data into numerical-type collection data by using a one-hot encoding method, if the collection data is category-type collection data, and to take the numerical-type collection data as feature data; wherein the category-type collection data is state data; a feature calculation unit configured to generate temperature difference feature data and temperature change rate feature data for collection data of a temperature sensor, wherein the temperature difference feature data is a temperature difference value between each collection data, and each collection data is subtracted by a set temperature of a compartment of the refrigerator before the temperature difference value is calculated; a dimension transformation unit configured to perform high-dimensional transformation on feature data by using a polynomial transformation method to obtain nonlinear feature data. wherein the feature calculation unit is configured to determine the temperature change rate feature data using a first calculation formula, the first calculation formula being: Δε t = ε t - ε o ; wherein Δε t is the temperature change rate characteristic data at time t in the current unit time, ε t is the collection data at time t in the current unit time; Δε o is a reference value; if Δε t and Δε t-1 are in the same direction, then ε o is unchanged; if Δε t and Δε t-1 are in opposite directions, then ε o is updated using a second calculation formula; the second calculation formula is ε' o =aε o -(1-a)ε t-1 , ε' o is an updated reference value, used to calculate the temperature change rate characteristic data at time t+1; and a is a forgetting factor.

6. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-4.

7. A computing device comprising a memory and a processor, wherein the memory stores executable code, and the processor executes the executable code to implement the method of any one of claims 1-4.

Images