Refrigerator fault prediction system and method based on big data and computing equipment

Abstract

The invention discloses a refrigerator fault prediction system and method based on big data and computing equipment, relates to the technical field of refrigerator big data testing, and solves the technical problem that periodic characteristics of different fault modes cannot be captured due to the fact that setting of a preset time interval lacks dynamic adaptability in the prior art. According to the method, whether components in the refrigerator break down or not is judged by constructing a refrigerator monitoring task; if yes, overhauling and recording to element maintenance log data; if not, continuing to judge; setting a plurality of statistical cycles, calculating the time from the current time to the next fault occurrence according to the component maintenance log data, and determining the initial probability of the fault occurrence based on the time of the next fault occurrence; counting the number of times of the same type of faults of the element and the adjacent time interval of the same type of faults, and constructing an increment factor to optimize the initial probability to obtain a final output probability; the prediction precision is improved.

Description

Technical Field

[0001] The present invention belongs to the field of refrigerator big data testing, involves automation monitoring technology, and specifically is a refrigerator fault prediction system, method and computing device based on big data. Background Art

[0002] Through big data analysis, the system can identify abnormal patterns during the operation of the refrigerator, thus issuing early warnings before a fault occurs, enabling users to have enough time to arrange for repairs or replacement of components, avoiding food spoilage and other potential problems; predictive maintenance reduces the need for emergency repairs caused by unexpected faults, thereby reducing maintenance costs; in addition, since the repair time can be planned, manufacturers and service providers can arrange resources more efficiently, further reducing costs; by detecting and repairing potential problems in a timely manner, the prediction system based on big data helps to extend the overall service life of the refrigerator, reduces the frequency of users replacing new devices, and saves resources; predictive maintenance reduces the equipment downtime caused by faults, thus ensuring the continuous and efficient operation of the refrigerator, contributing to reducing energy consumption and carbon emissions, and conforming to the current environmental protection trend.

[0003] The prior art (a patent application for an invention with a publication number of CN118691258A) discloses a refrigerator fault prediction processing method, device, medium and equipment, including: obtaining the time series of operation parameters of the refrigerator to be predicted within a preset time period; inputting the preprocessed time series of operation parameters into the refrigerator fault prediction model to obtain a prediction result; if the prediction result includes that the refrigerator to be predicted will have a fault within a future preset time period, generating a notification message and sending the notification message to the after-sales platform, so that the staff of the after-sales platform can confirm with the user of the refrigerator to be predicted, and when it is confirmed that the refrigerator to be predicted will have a fault within the future preset time period, notifying the repair personnel to come to repair; determining the actual operation situation of the refrigerator to be predicted within the future preset time period; if the actual operation situation is inconsistent with the prediction result, optimizing and adjusting the refrigerator fault prediction model;

[0004] In the prior art, the setting of the preset time interval lacks dynamic adaptability, and the fixed time window makes it impossible to capture the periodic characteristics of different fault modes. At the same time, the prior art does not integrate multi-dimensional environmental data, resulting in a single dimension of fault prediction, thus causing the technical problem of decreased prediction accuracy;

[0005] The present invention provides a refrigerator fault prediction system, method and computing device based on big data to solve the above technical problems. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art; for this purpose, the present invention proposes a refrigerator fault prediction system, method and computing device based on big data, which is used to solve the technical problems that the setting of the preset time interval in the prior art lacks dynamic adaptability and the fixed time window leads to the inability to capture the periodic characteristics of different fault modes.

[0007] To achieve the above object, the first aspect of the present invention provides a refrigerator fault prediction system based on big data, including: a data acquisition module, a fault prediction module and a probability optimization module;

[0008] Data acquisition module: Obtain the operation data of the refrigerator through a number of sensors; among them, the operation data includes: component maintenance log data, environmental variables;

[0009] Fault prediction module: Construct a refrigerator monitoring task to judge whether a component in the refrigerator fails; if so, perform maintenance and record it in the component maintenance log data; if not, continue to judge;

[0010] Set a number of statistical periods, calculate the time from the current time to the next fault occurrence according to the component maintenance log data, and determine the initial probability of the fault occurrence based on the time of the next fault occurrence;

[0011] Probability optimization module: Count the number of times the same type of fault occurs in the component and the adjacent time interval of the same type of fault, and construct an increment factor to optimize the initial probability to obtain the final output probability.

[0012] Preferably, a refrigerator fault prediction system based on big data further includes: a remote monitoring module: used to construct a WiFi control module to remotely control the refrigerator, and report and display the operation data of the refrigerator to the user according to a preset period;

[0013] When constructing the WiFi module to remotely control the refrigerator, use the WiFi module of ESP32 or ESP8266, and cooperate with components such as the main control unit, temperature sensor, and relay to realize refrigerator control and environmental monitoring.

[0014] The present invention remotely controls the inside of the refrigerator through the WiFi module. The user can remotely view and adjust the temperature of each compartment of the refrigerator and change the operation mode of the refrigerator on the mobile phone App side, which is convenient for the user to manage the refrigerator when away from home; on the other hand, the refrigerator can regularly report iOT data such as the operation status through the WiFi module, which is convenient for professionals to remotely diagnose the operation status of the refrigerator.

[0015] Preferably, the construction of the refrigerator monitoring task includes:

[0016] Retrieve the operating data of the components, parse and integrate the operating data of the components to form the big field format, confirm the data type of the big field, split the number of data entries and content included in the big field, and at the same time extract the data timestamp corresponding to the big field, and traverse several operating data of the components in real time to form a monitoring task.

[0017] Preferably, determining whether a component in the refrigerator fails includes:

[0018] Retrieve the monitoring task and select execution parameters through the monitoring task; wherein, the execution parameters include: parameter one and parameter two;

[0019] Judge whether the component operating data field information of parameter one and parameter two in the execution parameters is consistent; if so, mark the corresponding component in the refrigerator as normal; if not, mark the corresponding component of the refrigerator as faulty.

[0020] It should be noted that parameter one is used to be passed to the data platform interface and directly obtain the component operating data field under the normal operating state of the refrigerator through the interface; parameter two is used to request the cloud according to the field of the corresponding timestamp and obtain the real-time component operating data field of the corresponding refrigerator returned by the cloud.

[0021] In the present invention, by using parameter two to obtain real-time component operating data from the cloud, the timeliness of the data can be ensured, so that the system can quickly respond to any abnormal state of the refrigerator components.

[0022] Preferably, calculating the time from the current time to the next failure according to the component maintenance log data includes:

[0023] Retrieve the statistical period T and the component maintenance log data of the faulty component within the statistical period; wherein, the component maintenance log data includes: the start time and end time of the component failure, the total number of times ZC of the corresponding component failure, the i-th type of failure type LS i , the number of times LS of different types of failures, the average adjacent time interval LT of different types of failures, and a preset maintenance factor WH; the value range of i is [1, n], and n is a positive integer;

[0024] The maintenance factor is obtained based on the maintenance method: when the maintenance method is to replace the original component, WH = z1, and when the maintenance method is to repair the original component, WH = z2; where 1 > z1 > z2 > 0.

[0025] Adopt the dynamic time warping algorithm and set range rules to automatically match the optimal historical interval; statistically record the changes in the temperature, relative humidity and frequency of the refrigerator within the optimal historical interval as ΔT, ΔH and ΔF respectively;

[0026] Through the formula Calculate the time T from the current time to the next failure sc ; where T 0 is the failure reference time, and the failure reference time is the mean value of the time intervals of all failures of the same component; θ1, θ2, and θ3 are the first, second, and third influencing factors respectively.

[0027] It should be noted that the adjacent time interval is the absolute value of the difference between the time when the component fails and the end time of the previous component failure; the influencing factor is determined according to the influencing factor value table; among them, the end time of the failure includes: the time when the component runs again after being repaired due to failure.

[0028] The present invention can automatically match the optimal historical interval through the dynamic time warping algorithm, and can more accurately find the historical data similar to the current failure situation, thereby improving the accuracy of failure prediction; by considering the changes in the temperature, relative humidity, and frequency of the refrigerator, these factors may be closely related to the occurrence of component failures, thereby enhancing the comprehensiveness and accuracy of model prediction.

[0029] Preferably, the dynamic time warping algorithm is adopted and a range rule is set to automatically match the optimal historical interval, including:

[0030] Mark the optimal historical interval as [A, B], and set the range rule as: A ∈ T, B ∈ (T - 1), and [A, B] includes the start time of component failure within the current statistical period T;

[0031] Retrieve the failure sequence within the statistical period and the environmental variable sequence in the optimal historical interval, where the environmental variable sequence includes: the temperature T of the refrigerator j1 , relative humidity H j2 and frequency F j3 corresponding time series; j1, j2, and j3 are the lengths of the time series;

[0032] Align the failure sequence and the environmental variable sequence through the dynamic time warping algorithm to obtain a distance function;

[0033] Calculate the distance corresponding to the interval through the distance function D = ∑ (i，j) cost(LS i , T j1 , H j2 , F j3 ); in the formula, t is the length of the failure sequence;

[0034] Take the interval with the minimum distance as the optimal historical interval.

[0035] In the present invention, DTW dynamically bends the time axis to find the optimal matching path between two sequences. Even if their lengths are different or there are local feature offsets, DTW can still identify similar patterns, increasing the recognition accuracy and reducing the possibility of misjudgment.

[0036] Preferably, determining the initial probability of a fault occurring based on the time of the next fault occurrence includes:

[0037] Retrieving the time from the current time to the occurrence of the next fault;

[0038] Through the formula Calculate the initial probability of a fault occurring at the corresponding time.

[0039] Preferably, constructing an incremental factor to optimize the initial probability to obtain the final output probability includes:

[0040] Retrieve the number of times of the same type of fault and denote it as LS ix , and denote the adjacent time interval of the i-th type of the same type of fault as LT ix and the initial probability P 0 ; where the value range of x is [1, m], and m is a positive integer;

[0041] Through the formula Calculate the incremental factor; where b1 is the weight coefficient of the number of faults, b2 is the weight coefficient of the time interval, both b1 and b2 are greater than 0, and b1 + b2 = 1;

[0042] Through the formula Calculate the final output probability.

[0043] It should be noted that the weight coefficient of the number of faults represents the contribution of a single fault to the risk, which is set according to the importance of the component where the fault occurs in the refrigerator. The greater the importance, the greater the weight coefficient of the number of faults; the setting of the weight coefficient of the time interval: the shorter the time interval, the greater the probability of the next risk occurrence.

[0044] To achieve the above object, the second aspect of the present invention provides a refrigerator fault prediction method based on big data, including:

[0045] Obtain the operation data of the refrigerator through a number of sensors; where the operation data includes: component maintenance log data, environmental variables;

[0046] Construct a refrigerator monitoring task to determine whether a component in the refrigerator has a fault; if so, perform maintenance and record it in the component maintenance log data; if not, continue to judge;

[0047] Set several statistical periods, calculate the time from the current time to the next failure based on the component maintenance log data, and determine the initial probability of failure based on the time of the next failure;

[0048] Count the number of times the same type of failure occurs in the component and the adjacent time intervals of the same type of failure, and construct an incremental factor to optimize the initial probability to obtain the final output probability.

[0049] To achieve the above object, the third aspect of the present invention provides a refrigerator fault prediction calculation device based on big data, which is characterized by including: a memory and a processor, wherein an executable code is stored in the memory, and when the processor executes the executable code, it implements a refrigerator fault prediction system provided by the first aspect.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] 1. The present invention can timely discover and respond to the fault conditions of the components in the refrigerator by obtaining the operation data of the refrigerator in real time through sensors, which helps to reduce the downtime caused by faults and improve the operation efficiency of the refrigerator; the fault prediction mathematical model constructed based on the component maintenance log data can predict the time range from the current time to the next failure, which is beneficial for the maintenance team to perform preventive maintenance before the failure occurs and avoid the impact of sudden failures on the operation of the refrigerator; by counting the number of times the same type of failure occurs in the component and the adjacent time intervals, an incremental factor can be constructed to optimize the fault prediction model, which is beneficial for the maintenance strategy to be more accurate and can formulate personalized maintenance plans according to the fault characteristics of different components.

[0052] 2. The present invention can reduce the unplanned downtime caused by sudden failures through real-time monitoring and fault prediction, thereby reducing the maintenance cost. Through fault prediction, the maintenance team can prepare the required parts and tools in advance and shorten the maintenance time; at the same time, accurate fault location can also reduce unnecessary inspections and maintenance work and improve work efficiency. Preventive maintenance and timely fault handling help to extend the service life of the refrigerator and its components and reduce the frequency and cost of replacing equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0054] Figure 1 It is a schematic diagram of the module relationship included in the present invention;

[0055] Figure 2 Schematic diagram of the specific steps for fault prediction of the present invention;

[0056] Figure 3 Schematic diagram of the specific steps for probability prediction and optimization of the present invention;

[0057] Figure 4 Schematic diagram of the process for fault prediction of the present invention. Specific embodiments

[0058] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0059] Please refer to Figure 1 , an embodiment of the first aspect of the present invention provides a refrigerator fault prediction system based on big data, including: a data acquisition module, a fault prediction module, and a probability optimization module;

[0060] Data acquisition module: Obtain the operation data of the refrigerator through a number of sensors; among them, the operation data includes: component maintenance log data, environmental variables;

[0061] Fault prediction module: Construct a refrigerator monitoring task to determine whether a component in the refrigerator fails; if yes, perform maintenance and record it in the component maintenance log data; if not, continue to judge;

[0062] Set a number of statistical periods, calculate the time from the current time to the next fault occurrence based on the component maintenance log data, and determine the initial probability of failure based on the time of the next fault occurrence;

[0063] Probability optimization module: Count the number of times of the same type of failure of the component and the adjacent time intervals of the same type of failure, and construct an increment factor to optimize the initial probability to obtain the final output probability;

[0064] Remote monitoring module: Used to construct a WiFi module to remotely control the refrigerator, and report and display the operation data of the refrigerator to the user according to the user-defined period.

[0065] Please refer to Figure 2 , the specific steps of fault prediction,

[0066] Retrieve the statistical period T, and retrieve the component maintenance log data of the failed component within the statistical period; among them, the component maintenance log data includes: the start time and end time of the component failure, the total number of times ZC of the corresponding component failure, and the i-th type of failure type LSi The number LS of different types of faults, the average adjacent time interval LT of different types of faults, and a preset maintenance factor WH; the value range of i is [1, n], and n is a positive integer;

[0067] Adopt the dynamic time warping algorithm and set range rules to automatically match the optimal historical interval; mark the optimal historical interval as [A, B], and set the range rules as: A ∈ T, B ∈ (T - 1), and [A, B] includes the start time of component failure within the current statistical period T;

[0068] Retrieve the fault sequence within the statistical period and the environmental variable sequence in the optimal historical interval. Among them, the environmental variable sequence includes: the temperature T of the refrigerator j1 , the relative humidity H j2 and the frequency F j3 The corresponding time series; j1, j2, and j3 are the lengths of the time series;

[0069] Align the fault sequence and the environmental variable sequence through the dynamic time warping algorithm to obtain a distance function;

[0070] Through the distance function D = Σ (i，j) cost(LS i , T j1 , H j2 , F j3 ) Calculate the distance corresponding to the interval; in the formula, t is the length of the fault sequence;

[0071] Take the interval with the minimum distance as the optimal historical interval;

[0072] Statistically record the changes in the temperature, relative humidity, and frequency of the refrigerator within the optimal historical interval as ΔT, ΔH, and ΔF respectively;

[0073] Through the formula Calculate the time T from the current time to the next fault occurrence; sc ; where, T 0 is the fault reference time, and the fault reference time is the average value of the time intervals of all faults of the same component; θ1, θ2, and θ3 are the first, second, and third influencing factors respectively.

[0074] For example, the steps to predict the time and corresponding probability of the next component failure of a refrigerator BX are as follows. There is a refrigerator component Y. Set the statistical period as two months and retrieve the component maintenance log data of the failed component within the statistical period; among them, the component maintenance log data includes: the start time and end time of component failure within two months,

[0075] The total number of times the corresponding component fails is 4 times, and the first type of fault type LS1 For the deformation of component Y, it occurred 1 time; the second type of fault is LS 2 For the noise generated by component Y, it occurred 1 time; the third type of fault is LS 3 For the reduced operating efficiency of component Y, it occurred 2 times;

[0076] The order of fault occurrence is as follows: First, event one is the deformation of component Y. Second, event two is a reduction in operating efficiency occurring once. Third, event three is the generation of noise by component Y. Finally, event four is another reduction in operating efficiency;

[0077] The maintenance method for event one is to replace the original component. The maintenance method for event two is to repair the original component. The maintenance method for event three is to repair the original component. The maintenance method for event four is to replace the original component. Among them, the maintenance method of replacing the original component is assigned WH = 0.15, and the maintenance method of repairing the original component is assigned WH = 0.05

[0078] The time interval between event one and event two is 15 days. The time interval between event two and event three is 1 day. The time interval between event three and event four is 32 days. The average adjacent time interval for different types of faults is 16 days;

[0079] Using the dynamic time warping algorithm and setting range rules to automatically match the optimal historical interval of 48 days. The difference ΔT between the maximum temperature and the minimum temperature in the refrigerator within 48 days is 5°C. The difference ΔH between the maximum relative humidity and the minimum relative humidity is 3%. The difference ΔF between the maximum frequency and the minimum frequency is 15 Hz;

[0080] Components in the refrigerator that are not sensitive to temperature difference changes are marked as Y1 - type components. Components that are not sensitive to humidity changes are marked as Y2 - type components. Components that are not sensitive to frequency changes are marked as Y3 - type components. The judgment criterion for insensitivity is that when the environmental variable changes, the component does not fail;

[0081] Table 1 Influence factor value table

[0082]

[0083] Since the refrigerator component Y is a component sensitive to temperature difference changes, the influence factor is obtained according to the influence factor value table (as shown in Table 1). Among them, the fault reference time T 0 is 16 days;

[0084] Then through the formula

[0085]

[0086] The calculated time from the current time to the next failure is 11.5. By rounding, it is obtained that the component Y will fail on the 12th day from the current time.

[0087] Please refer to Figure 3 , the specific steps of probability prediction and optimization.

[0088] Retrieve the time from the current time to the next failure;

[0089] Through the formula Calculate the initial probability of failure at the corresponding time;

[0090] Record the number of failures of the same type as LS ix , and record the adjacent time interval of the i-th type of the same type of failure as LT ix and the initial probability P 0 ; where the value range of x is [1, m], and m is a positive integer;

[0091] Through the formula Calculate the incremental factor; where, b1 is the weight coefficient of the number of failures, b2 is the weight coefficient of the time interval, both b1 and b2 are greater than 0, and b1 + b2 = 1;

[0092] Through the formula Calculate the final output probability.

[0093] For example, retrieve the time of 12 days from the current time to the next failure,

[0094] Through the formula Calculate that the initial probability of failure at the corresponding time is 0.7;

[0095] The number of failures of the same type is 2 times, the adjacent time interval of the second type of the same type of failure is 33 days and the initial probability is 0.7; the weight coefficient of the number of failures b1 = 0.6, and the weight coefficient of the time interval b2 is 0.4

[0096] Through the formula Calculate that the incremental factor is 1.21;

[0097] Through the formula Calculate that the final output probability is 0.617, and the probability of failure on the 12th day is corrected to 61.7% through the incremental factor.

[0098] Please refer to Figure 4 , the second aspect embodiment of the present invention provides a refrigerator failure prediction method based on big data, including:

[0099] Obtain the operation data of the refrigerator through a number of sensors; among them, the operation data includes: component maintenance log data, environmental variables;

[0100] Construct a refrigerator monitoring task to determine whether a component in the refrigerator has failed; if so, perform maintenance and record it in the component maintenance log data; if not, continue to judge;

[0101] Set a number of statistical periods, calculate the time from the current time to the next failure based on the component maintenance log data, and determine the initial probability of failure based on the time of the next failure;

[0102] Count the number of times the same type of failure occurs in the component and the adjacent time intervals of the same type of failure, and construct an increment factor to optimize the initial probability to obtain the final output probability.

[0103] An embodiment of the third aspect of the present invention provides a refrigerator fault prediction calculation device based on big data, including: a memory and a processor. An executable code is stored in the memory. When the processor executes the executable code in the memory, it implements a refrigerator fault prediction system provided in the first aspect.

[0104] Some of the data in the above formula is calculated by removing the dimension and taking its value. The formula is obtained by software simulation of a large amount of collected data to obtain a formula closest to the actual situation; the preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained by simulation of a large amount of data.

[0105] The working principle of the present invention: Obtain the operation data of the refrigerator through a number of sensors; construct a refrigerator monitoring task to determine whether a component in the refrigerator has failed; if so, perform maintenance and record it in the component maintenance log data; if not, continue to judge; set a number of statistical periods, calculate the time from the current time to the next failure based on the component maintenance log data, and determine the initial probability of failure based on the time of the next failure; count the number of times the same type of failure occurs in the component and the adjacent time intervals of the same type of failure, and construct an increment factor to optimize the initial probability to obtain the final output probability.

[0106] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A refrigerator failure prediction system based on big data, characterized in that: include: Data acquisition module, fault prediction module and probability optimization module; Data acquisition module: obtains the operating data of the refrigerator through a number of sensors; the operating data includes: component maintenance log data and environmental variables; Fault prediction module: Build a refrigerator monitoring task to determine whether a component in the refrigerator has a fault; if so, perform maintenance and record it in the component maintenance log data; if not, continue to judge; Set several statistical cycles, calculate the time from the current time to the next fault occurrence based on the component maintenance log data, and determine the initial probability of a fault based on the time of the next fault occurrence; Probability optimization module: Count the number of times the same type of failure occurs in the component and the adjacent time intervals of the same type of failure, and construct incremental factors to optimize the initial probability to obtain the final output probability.

2. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: Also includes : Remote monitoring module: used to build a WiFi control module to remotely control the refrigerator, and report the refrigerator's operating data according to a preset cycle and display it to the user.

3. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: The refrigerator monitoring task is constructed, including: Retrieve component operation data, parse and integrate the component operation data into a big field format, confirm the data type of the big field, split the number and content of data contained in the big field, extract the data timestamp corresponding to the big field, and traverse several operation data of the component in real time to form a monitoring task.

4. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: The step of determining whether a component in the refrigerator fails comprises: Retrieve the monitoring task, and select execution parameters through the monitoring task; wherein the execution parameters include: parameter one and parameter two; Determine whether the component operation data field information of parameter one and parameter two in the execution parameters are consistent; if yes, mark the corresponding component in the refrigerator as normal; if not, mark the corresponding component in the refrigerator as faulty.

5. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: The calculating the time from the current time to the next fault occurrence according to the component maintenance log data includes: Retrieve the component maintenance log data of the statistical period T and the failed components within the statistical period; the component maintenance log data includes: the start time and end time of the component failure, the total number of failures of the corresponding component ZC, the i-th type of failure LS i , the number of different types of faults LS, the average adjacent time intervals of different types of faults LT, and the preset maintenance factor WH; the value range of i is [1, n], and n is a positive integer; The dynamic time warping algorithm is used and the range rule is set to automatically match the optimal historical interval; the changes in the refrigerator's temperature, relative humidity and frequency within the optimal historical interval are counted and recorded as ΔT, ΔH and ΔF respectively; By formula Calculate the time T from the current time to the next fault occurrence sc ; Wherein, T0 is the fault reference time, which is the mean of the time intervals between all faults of the same component; θ1, θ2 and θ3 are influencing factors one, two and three respectively.

6. The refrigerator failure prediction system based on big data according to claim 5, characterized in that: The use of a dynamic time warping algorithm and setting range rules to automatically match the optimal historical interval includes: The optimal historical interval is marked as [A, B], and the range rule is set as: A∈T, B∈(T-1), and [A, B] contains the start time of component failure within the current statistical period T; Retrieve the fault sequence within the statistical period and the environmental variable sequence in the optimal historical interval, where the environmental variable sequence includes: the temperature T of the refrigerator j1 , relative humidity H j2 and frequency F j3 The corresponding time series; j1, j2, j3 are the lengths of the time series; The distance function is obtained by aligning the fault sequence and the environmental variable sequence through the dynamic time warping algorithm; Through the distance function D = ∑ (i，j) cost(LS i , T j1 , H j2 , F j3 ) calculates the distance corresponding to the interval; where t is the length of the fault sequence; The interval with the smallest distance is taken as the optimal historical interval.

7. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: The determining of the initial probability of a failure based on the time when the next failure occurs comprises: The initial probability of a failure occurring at a corresponding time is obtained according to a mapping relationship between the time from the current time to the next failure occurrence and the initial probability of a failure occurring at a corresponding time.

8. The refrigerator failure prediction system based on big data according to claim 1, characterized in that: The construction of the incremental factor optimizes the initial probability to obtain the final output probability, including: The number of times the same type of fault is retrieved is recorded as LS ix , and the adjacent time interval of the i-th type of fault is recorded as LT ix And the initial probability P0; where the value range of x is [1, m], and m is a positive integer; By formula The incremental factor is calculated; wherein b1 is the weight coefficient of the number of failures, b2 is the weight coefficient of the time interval, b1 and b2 are both greater than 0, and b1+b2=1; By formula The final output probability is calculated.

9. A refrigerator failure prediction method based on big data, adapted to the refrigerator failure prediction system based on big data described in claims 1-8, characterized in that: include: The operation data of the refrigerator is obtained through a number of sensors; wherein the operation data includes: component maintenance log data and environmental variables; Construct a refrigerator monitoring task to determine whether a component in the refrigerator is faulty; if so, perform maintenance and record it in the component maintenance log data; if not, continue to determine; Set several statistical cycles, calculate the time from the current time to the next fault occurrence based on the component maintenance log data, and determine the initial probability of a fault based on the time of the next fault occurrence; The number of times the same type of fault occurs in the component and the adjacent time intervals of the same type of fault are counted, and the incremental factor is constructed to optimize the initial probability to obtain the final output probability.

10. A refrigerator failure prediction computing device based on big data, characterized in that: include: A memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, a refrigerator fault prediction system based on big data as described in claims 1-8 is implemented.

Images