Storage equipment freshness identification method and device and storage equipment

Abstract

The invention discloses a storage equipment freshness identification method and device and storage equipment, and belongs to the field of storage equipment. The freshness identification method of the storage equipment comprises the following steps: acquiring real-time sensor data acquired by a sensor array at a current acquisition moment; performing wavelet packet decomposition on the real-time sensor data to obtain a first decomposition coefficient of a target frequency band; correcting the first decomposition coefficient based on an inter-board correction function, and obtaining target sensor data based on the corrected first decomposition coefficient; and on the basis of the target sensor data, freshness information corresponding to the target smell is obtained through prediction. According to the freshness identification method of the storage equipment, the mutual influence among the sensors can be fully considered, and the correction effect of the sensor array is remarkably improved, so that the precision and accuracy of freshness identification are improved.

Description

Technical Field

[0001] This application belongs to the field of storage equipment, and particularly relates to a method, apparatus and storage equipment for identifying the freshness of storage equipment. Background Technology

[0002] In storage devices and other fields, multiple electronic nose sensor arrays are often used for odor recognition. Different arrays exhibit variations in response characteristics, requiring consistency conversion to reduce these differences. Current technologies primarily involve individually calibrating each sensor. However, this method fails to consider the mutual influence between sensors and the inconsistency of this influence across different sensor arrays, affecting the calibration results and consequently impacting subsequent recognition performance. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method, apparatus, and storage device for freshness identification of storage equipment, which can fully consider the mutual influence between various sensors, realize correction based on the response values ​​between sensor arrays, improve the consistency of response values ​​between sensor arrays, and thus improve the accuracy and precision of freshness identification.

[0004] In a first aspect, this application provides a method for identifying the freshness of a storage device, the storage device including a sensor array, the method comprising:

[0005] Obtain the real-time sensor data collected by the sensor array at the current acquisition time;

[0006] The real-time sensor data is subjected to wavelet packet decomposition to obtain the first decomposition coefficients of the target frequency band; the target frequency band is determined based on the target odor used for freshness identification.

[0007] The first decomposition coefficients are corrected based on the inter-plate correction function, and the target sensor data is obtained based on the corrected first decomposition coefficients.

[0008] Based on the target sensor data, the freshness information corresponding to the target odor is predicted.

[0009] According to the freshness identification method for storage equipment in this application, wavelet packet decomposition is performed on the real-time sensor data collected by the sensor array to obtain decomposition coefficients, and the real-time sensor data is corrected based on the inter-plate correction function to obtain target sensor data, thereby realizing the correction based on the response values ​​between the sensor arrays and improving the consistency of the response values ​​between the sensor arrays; on this basis, freshness identification is performed based on the target sensor data of the corrected sensor array, thereby improving the accuracy and precision of freshness identification and enhancing the user experience.

[0010] According to one embodiment of this application, the step of correcting the first decomposition coefficients based on the inter-plate correction function and obtaining the target sensor data based on the corrected first decomposition coefficients includes:

[0011] The first decomposition coefficient is input into the inter-board correction function to obtain the corrected first decomposition coefficient output by the inter-board correction function;

[0012] The corrected first decomposition coefficients are reconstructed using wavelet packet inverse transform to obtain the target sensor data.

[0013] According to one embodiment of this application, before acquiring the real-time sensor data collected by the sensor array at the current acquisition time, the method further includes:

[0014] Calculate the first edge distribution difference between the historical sensor data set and the standard sensor data set corresponding to the sensor array; the historical sensor data set includes sensor data from multiple historical acquisition times before the storage device leaves the factory; the standard sensor data set is collected by a standard sensor array.

[0015] If the first edge distribution difference exceeds a first threshold, the initial inter-plate correction coefficient and the initial inter-plate correction function are built into the storage device; the initial inter-plate correction coefficient and the initial inter-plate correction function are calculated based on the historical sensor data set and the standard sensor data set.

[0016] According to one embodiment of this application, before acquiring the real-time sensor data collected by the sensor array at the current acquisition time, the method further includes:

[0017] Obtain the historical sensor data set of the sensor array at the multiple historical acquisition times and the standard sensor data set of the standard sensor array at the multiple acquisition times;

[0018] Wavelet packet decomposition is performed on the standard sensor data set and the historical sensor data set respectively to obtain the standard decomposition coefficients and the second decomposition coefficients of the target frequency band;

[0019] Based on the standard decomposition coefficient and the second decomposition coefficient, the initial inter-board correction coefficient is calculated;

[0020] Based on the standard decomposition coefficients and the initial inter-panel correction coefficients, an initial inter-panel correction function is determined; wherein, the initial inter-panel correction coefficients and the initial inter-panel correction function are deployed in the storage device as factory settings.

[0021] According to one embodiment of this application, determining the initial inter-board correction function based on the standard decomposition coefficients and the initial inter-board correction coefficients includes:

[0022] Based on the first upper and lower limit thresholds corresponding to the standard decomposition coefficients and the initial inter-board correction coefficients, the second upper and lower limit thresholds corresponding to the second decomposition coefficients are calculated.

[0023] The initial inter-plate correction function is constructed based on the second upper and lower thresholds and the second decomposition coefficients.

[0024] According to one embodiment of this application, after acquiring the real-time sensor data collected by the sensor array at the current acquisition time, the method further includes:

[0025] Calculate the second edge distribution difference between the first sensor data set corresponding to the sensor array and the standard sensor data set; the first sensor data set includes real-time sensor data at the current acquisition time and real-time sensor data at multiple acquisition times prior to the current acquisition time; the standard sensor data set is acquired by the standard sensor array.

[0026] If the difference in the second edge distribution exceeds the first threshold, the inter-plate correction coefficients of the sensor array and the parameters of the inter-plate correction function are updated based on the first sensor data set and the standard sensor data set.

[0027] According to one embodiment of this application, predicting the freshness information corresponding to the target odor based on the target sensor data includes:

[0028] The target sensor data is input into the freshness classification model to obtain the freshness information output by the freshness classification model;

[0029] The freshness classification model is trained using the standard sensor data as samples and the sample freshness category corresponding to the target odor corresponding to the standard sensor data as the sample label.

[0030] According to one embodiment of this application, predicting the freshness information corresponding to the target odor based on the target sensor data includes:

[0031] Principal component analysis and independent component analysis are performed on the target sensor data at the current acquisition time and the target sensor data at at least one historical acquisition time to obtain the independent component matrix;

[0032] Each independent component in the independent component matrix is ​​filtered, and the filtered independent component matrix is ​​reconstructed to obtain the processed target sensor data.

[0033] Based on the processed target sensor data, the freshness information corresponding to the target sensor data is predicted.

[0034] Secondly, this application provides a freshness identification device for a storage device, the storage device including a sensor array, the device comprising:

[0035] The first processing module is used to acquire real-time sensor data collected by the sensor array at the current acquisition time.

[0036] The second processing module is used to perform wavelet packet decomposition on the real-time sensor data to obtain the first decomposition coefficients of the target frequency band; the target frequency band is determined based on the target odor used for freshness identification.

[0037] The third processing module is used to correct the first decomposition coefficients based on the inter-board correction function, and to obtain the target sensor data based on the corrected first decomposition coefficients.

[0038] The fourth processing module is used to predict the freshness information corresponding to the target odor based on the target sensor data.

[0039] According to the freshness identification device for storage equipment of this application, the decomposition coefficients are obtained by wavelet packet decomposition of real-time sensor data collected by sensor array, and the target sensor data is obtained by correcting the real-time sensor data based on the inter-plate correction function, thereby realizing the correction based on the response values ​​between sensor arrays and improving the consistency of response values ​​between sensor arrays; on this basis, freshness identification is performed based on the target sensor data of the corrected sensor array, thereby improving the accuracy and precision of freshness identification and enhancing the user experience.

[0040] Thirdly, this application provides a storage device, comprising:

[0041] A sensor array, the sensor array including at least one odor sensor;

[0042] The freshness identification device for the storage equipment as described in the second aspect is electrically connected to the sensor array.

[0043] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the freshness identification method for storage devices as described in the first aspect above.

[0044] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the freshness identification method for storage devices as described in the first aspect above.

[0045] The above-described one or more technical solutions in the embodiments of this application have at least one of the following technical effects:

[0046] By performing wavelet packet decomposition on the real-time sensor data collected by the sensor array to obtain decomposition coefficients, and then correcting the real-time sensor data based on the inter-board correction function to obtain the target sensor data, the correction based on the response values ​​between the sensor arrays is realized, thereby improving the consistency of the response values ​​between the sensor arrays. On this basis, freshness identification is performed based on the corrected target sensor data of the sensor array, thereby improving the accuracy and precision of freshness identification and enhancing the user experience.

[0047] Furthermore, by calculating the second edge distribution difference between the first sensor data set collected by the sensor array and the standard sensor data set collected by the standard sensor array, the storage equipment after leaving the factory can be screened, and the storage equipment that needs to be corrected between boards can be updated, thereby improving the efficiency and accuracy of subsequent sensor array correction between boards, and thus improving the efficiency and accuracy of freshness identification.

[0048] Furthermore, by acquiring a first sensor data set including real-time sensor data from multiple acquisition times, and updating the initial inter-plate correction coefficients and initial inter-plate correction function based on the decomposition coefficients of the first sensor data set and the upper and lower thresholds of the decomposition coefficients, recursive updates of the initial inter-plate correction coefficients and initial inter-plate correction function can be achieved. Thus, based on the recursively updated initial inter-plate correction function, inter-plate correction of the sensor array can be performed, improving the accuracy of inter-plate correction of the sensor array, thereby improving the accuracy and precision of freshness prediction.

[0049] Furthermore, by correcting all wavelet packet decomposition coefficients of real-time sensor data and then reconstructing the corrected real-time sensor data decomposition coefficients using inverse wavelet packet transform, the original characteristics of the real signal at the target frequency point of the real-time sensor data are preserved, achieving the smoothness and integrity of the corrected decomposition coefficients. Thus, freshness identification is performed based on the corrected, smooth, and complete target sensor data, improving the accuracy of freshness identification.

[0050] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0051] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0052] Figure 1 This is one of the flowcharts illustrating the freshness identification method for storage equipment provided in this application embodiment;

[0053] Figure 2 This is a second schematic flowchart of the freshness identification method for storage equipment provided in the embodiments of this application;

[0054] Figure 3 This is the third flowchart illustrating the freshness identification method for storage equipment provided in this application embodiment;

[0055] Figure 4 This is the fourth flowchart illustrating the freshness identification method for storage equipment provided in this application embodiment;

[0056] Figure 5 This is a schematic diagram of the standard data set composition of the freshness identification method for storage equipment provided in the embodiments of this application;

[0057] Figure 6 This is one of the schematic diagrams showing the results of the freshness identification method for storage equipment provided in the embodiments of this application;

[0058] Figure 7 This is the second schematic diagram showing the result of the freshness identification method for storage equipment provided in the embodiments of this application;

[0059] Figure 8 This is a schematic diagram of the structure of the freshness identification device for storage equipment provided in the embodiments of this application;

[0060] Figure 9 This is a schematic diagram of the structure of the storage device provided in the embodiments of this application;

[0061] Figure 10 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0062] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0063] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0064] The following description, in conjunction with the accompanying drawings, details the freshness identification method, freshness identification device, electronic device, and readable storage medium for storage equipment provided in this application, through specific embodiments and application scenarios.

[0065] The freshness identification method for storage devices can be applied to the terminal, specifically executed by the hardware or software within the terminal.

[0066] The terminal includes, but is not limited to, storage devices, or portable communication devices such as mobile phones or tablets. It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer.

[0067] The storage devices mentioned in the embodiments of this application can be understood as refrigeration storage devices in a broad sense, including but not limited to refrigerators, freezers, display cases, beverage cabinets, wine cabinets, refrigerated display cases, and refrigerated vending machines. The storage devices have diverse structural forms and a wide range of applications.

[0068] The storage device can store a variety of items, including but not limited to: meat, fruits and vegetables, beverages, and daily necessities.

[0069] The freshness identification method for storage devices provided in this application can be executed by the storage device itself, such as by a microcontroller installed inside the storage device; in some embodiments, it can also be executed by a server that is communicatively connected to the storage device; or by a combination of a server and a microcontroller.

[0070] The following describes the freshness identification method for storage devices provided in this application, using storage devices as the implementing entity as an example.

[0071] like Figure 1 As shown, the freshness identification method for the storage device includes steps 110, 120, 130 and 140.

[0072] Step 110: Obtain the real-time sensor data collected by the sensor array at the current acquisition time;

[0073] In this step, the sensor array can be placed inside the storage device.

[0074] The sensor array may include one or more odor sensors. Of course, in other embodiments, temperature, humidity or airflow sensors may also be provided, which is not limited here.

[0075] The real-time sensor data at the current acquisition moment is the data corresponding to the target odor collected by the odor sensor on the sensor array at the current acquisition moment.

[0076] The target odor can be the odor used for freshness identification.

[0077] For example, the smell of meat, the smell of vegetables, etc.

[0078] It should be noted that the real-time sensor data is the data collected by the sensor array after the storage device leaves the factory.

[0079] In some embodiments, before acquiring the real-time sensor data at the current acquisition time collected by the sensor array, the method may further include:

[0080] Acquire historical sensor data sets of the sensor array at multiple historical acquisition times and standard sensor data sets of the standard sensor array at multiple acquisition times;

[0081] Wavelet packet decomposition was performed on the standard sensor dataset and the historical sensor dataset respectively to obtain the standard decomposition coefficients and the second decomposition coefficients of the target frequency band.

[0082] The initial inter-board correction coefficients are calculated based on the standard decomposition coefficients and the second decomposition coefficients.

[0083] Based on the standard decomposition coefficients and the initial inter-plate correction coefficients, the initial inter-plate correction function is determined; wherein, the initial inter-plate correction coefficients and the initial inter-plate correction function are deployed as the factory settings of the storage equipment.

[0084] In this embodiment, the historical sensor data set can be the data set collected by the sensor array before the storage device leaves the factory.

[0085] Historical data acquisition time can be the moment when the sensor array collects data before the storage device leaves the factory.

[0086] A standard sensor dataset can be a collection of data collected by a standard sensor array at multiple acquisition times.

[0087] A standard sensor array can be the sensor array selected before the storage device leaves the factory.

[0088] The standard sensor array selection method can be to screen by specifying, randomly selecting, or testing the target odor response capability, etc., and this application does not limit it.

[0089] Once selected, the standard sensor array is not changed and is not included in the storage device after shipment.

[0090] For example, there are sensor arrays numbered 1 to 9. Sensor array number 1 is selected as the standard sensor array. In subsequent processing, sensor array number 1 is always the standard sensor array and is not installed in the storage device after leaving the factory.

[0091] The sensor array used for collecting historical sensor data sets is a non-standard sensor array other than a standard sensor array.

[0092] Continuing with the example of sensors 1 to 9, sensor array 1 is the standard sensor array, while sensor arrays 2 to 9 are non-standard sensor arrays, which are installed in the storage device after leaving the factory.

[0093] It is understandable that the storage devices after leaving the factory contain non-standard sensor arrays other than the standard sensor array.

[0094] It should be noted that the number of odor sensors in a non-standard sensor array is the same as the number of odor sensors in a standard sensor array.

[0095] In some embodiments, both the historical sensor dataset and the standard sensor dataset can be datasets that have undergone freshness classification.

[0096] This freshness classification process can categorize data into either empty storage equipment data or freshness level data.

[0097] For example, a standard sensor dataset can include four types of data: storage equipment empty data, freshness grade data, near-freshness grade data, and spoilage grade data; of course, it can also include other types.

[0098] The types and data volume of each type in the historical sensor dataset should be consistent with those in the standard sensor dataset.

[0099] For example, such as Figure 5As shown, a standard sensor dataset can be a sample dataset with dimension N×d, where N is the number of samples and d is the number of sensors. Each sample contains d sensor response values. The first N1 rows store empty data samples from storage devices, N1+1 to N2 rows store freshness level data samples, N2+1 to N3 rows store near-freshness level data samples, and N3+1 to N rows store spoilage level data samples. Similarly, a historical sensor dataset is also a dataset with dimension N×d, with the first N1 rows storing empty data samples from storage devices, N1+1 to N2 rows store freshness level data samples, N2+1 to N3 rows store near-freshness level data samples, and N3+1 to N rows store spoilage level data samples. N, N1, N2, N3, and d are all positive integers.

[0100] Wavelet Packet Decomposition (WPD) is an extension of wavelet transform; wavelet transform is a mathematical technique used for hierarchical decomposition of signals.

[0101] Wavelet packet decomposition can adaptively select the optimal wavelet basis function based on the characteristics of the signal, and finely divide the time-frequency plane.

[0102] Wavelet basis functions are mathematical tools used to describe the local features of a signal; wavelet transform uses wavelet basis functions to decompose and reconstruct signals.

[0103] For example, when the sensor data is decomposed at the first scale using wavelet basis functions, a set of low-frequency coefficients and a set of high-frequency coefficients with the same bandwidth can be obtained; when the data is decomposed at the second scale, the set of low-frequency coefficients is also divided into a set of low-frequency coefficients and a set of high-frequency coefficients with the same bandwidth, and similarly, the set of high-frequency coefficients is also divided into low-frequency and high-frequency parts; and so on, a binary tree of wavelet packet decomposition can be generated.

[0104] The target frequency band can be the range of frequency points that have a response result obtained by wavelet packet decomposition of the target odor.

[0105] For example, such as Figure 7 As shown, the wavelet packet decomposition coefficients of a vinegar sample collected by a gas sensor at different frequency points were tested. The test results mainly focused on the low frequency band, while other frequency bands were almost zero. Therefore, the subsequent analysis of the decomposition coefficients of the sensor data of the vinegar odor was selected as the low frequency coefficient set, which is the target frequency band.

[0106] Standard decomposition coefficients can be the decomposition coefficients of the target frequency band based on wavelet packet decomposition of a standard sensor dataset.

[0107] The second decomposition coefficient can be the decomposition coefficient of the target frequency band based on wavelet packet decomposition of historical sensor data sets.

[0108] The initial inter-plate correction coefficient can be a coefficient calculated based on the standard sensor array for non-standard sensors before the storage equipment leaves the factory, and is used to correct the non-standard sensor array after the storage equipment leaves the factory.

[0109] The initial inter-panel correction function can be constructed based on the initial inter-panel correction coefficients before the storage equipment leaves the factory.

[0110] The initial inter-plate correction coefficient and the initial inter-plate correction function are deployed in the storage equipment as factory settings.

[0111] Understandably, before the storage device leaves the factory, the initial inter-panel correction coefficients and initial inter-panel correction functions based on the standard sensor array can be calculated for all non-standard sensor arrays other than the standard sensor array, and these can be deployed into the storage device as factory settings.

[0112] In some embodiments, before acquiring the real-time sensor data at the current acquisition time collected by the sensor array, the method may further include:

[0113] Calculate the first marginal distribution difference between the historical sensor data set and the standard sensor data set corresponding to the sensor array;

[0114] If the difference in the first edge distribution exceeds a first threshold, the initial inter-plate correction coefficient and the initial inter-plate correction function are built into the storage device;

[0115] In this embodiment, the historical sensor data set can be the data set collected by the sensor array before the storage device leaves the factory.

[0116] The historical sensor dataset includes sensor data collected at multiple historical moments before the storage device leaves the factory.

[0117] A standard sensor dataset can be acquired by a standard sensor array.

[0118] The initial inter-plate correction coefficients and initial inter-plate correction functions can be calculated based on historical sensor datasets and standard sensor datasets.

[0119] Marginal distributions refer to the probability distributions in probability theory and statistics that contain only a subset of the variables in a multidimensional random variable.

[0120] The first marginal distribution difference can be the marginal distribution of two datasets, historical sensor data set and standard sensor data set, on the same dimension; it is used to measure the consistency between historical sensor data set and standard sensor data set.

[0121] In actual implementation, the first edge distribution difference can be achieved based on a variety of algorithms.

[0122] For example, one of the following methods can be used to calculate the difference in marginal distributions between two datasets: the Kolmogorov-Smirnov test or the Wasserstein distance.

[0123] In some embodiments, the consistency of the marginal distributions of two datasets can also be assessed by calculating the Pearson correlation coefficient between the historical sensor dataset and the standard sensor dataset.

[0124] For example, such as Figure 2 As shown, the Pearson correlation coefficient between the historical sensor data set and the standard sensor data set is calculated. If the Pearson correlation coefficient between the variables corresponding to the two data sets is close to 1, the marginal distributions of the two data sets are considered to be highly similar, that is, the two data sets are similar. If the Pearson correlation coefficient between the variables corresponding to the two data sets is close to 0 or -1, the two data sets are considered to be dissimilar.

[0125] like Figure 6 As shown, the first threshold can be a threshold used to measure the difference in marginal distribution between historical sensor data sets and standard sensor data sets.

[0126] In actual implementation, the first threshold can be determined by experimental methods.

[0127] It is understandable that when the difference between the first edge distribution of the historical sensor data set collected by the sensor array and the standard sensor data set collected by the standard sensor array exceeds the first threshold, the sensor array can be considered to be dissimilar to the standard sensor array. The initial inter-board correction coefficient and the parameters of the initial inter-board correction function based on the standard sensor array can be calculated only for the sensor array that is dissimilar to the standard sensor array, and these can be deployed into the storage device as the factory settings.

[0128] This involves screening non-standard sensor arrays other than standard sensor arrays, calculating the initial inter-board correction coefficients and initial inter-board correction functions based on the standard sensor array for qualified non-standard sensor arrays, and deploying them as factory settings into the storage device.

[0129] It should be noted that when the first edge distribution difference is greater than the first threshold, the parameters of the initial inter-plate correction coefficient and the initial inter-plate correction function are used when the inter-plate correction is performed for the first time after the storage device is powered on after leaving the factory; when the first edge distribution difference is less than the first threshold, the initial inter-plate correction coefficient and the initial inter-plate correction function of the sensor array based on the standard sensor array are not calculated, and the inter-plate correction within the storage device is not performed during the first period of use after the storage device equipped with the sensor array is powered on after leaving the factory.

[0130] According to the freshness identification method for storage devices provided in this application, by calculating the first edge distribution difference between the historical sensor data set and the standard sensor data set, the sensor array that needs to be corrected between boards when the storage device is powered on for the first time after leaving the factory is selected, and the parameters of the initial board correction coefficient and the initial board correction function are calculated before leaving the factory. This can reduce unnecessary calculation processes, improve the efficiency of board correction of the sensor array, and thus improve the efficiency of freshness identification.

[0131] In some embodiments, determining the initial inter-board correction function based on the standard decomposition coefficients and the initial inter-board correction coefficients may include:

[0132] Based on the first upper and lower thresholds corresponding to the standard decomposition coefficients and the initial inter-board correction coefficients, the second upper and lower thresholds corresponding to the second decomposition coefficients are calculated.

[0133] An initial inter-plate correction function is constructed based on the second upper and lower thresholds and the second decomposition coefficient.

[0134] In this embodiment, the first upper and lower threshold values ​​can be the deviation range of the standard decomposition coefficients.

[0135] The second upper and lower thresholds can be the deviation range of the second decomposition coefficients.

[0136] In some embodiments, the first upper and lower thresholds can be calculated from the mean and standard deviation of the standard decomposition coefficients.

[0137] For example, the calculation method for the first upper and lower thresholds can be expressed as:

[0138]

[0139] Where, α i1 It is the upper threshold of the i-th decomposition frequency point in the standard sensor data set, α. i2 It is the corresponding lower threshold, c i It is the mean of the decomposition coefficients at the i-th decomposition frequency point in the standard sensor data set, θ i It is the standard deviation of the decomposition coefficients at the i-th decomposition frequency point in the standard sensor data set; where i is a positive integer.

[0140] In this embodiment, the average value of the decomposition coefficients at the i-th decomposition frequency point can be the average value of the decomposition coefficients of each sensor in the standard sensor array at the i-th decomposition frequency point.

[0141] It is understandable that each sensor has its corresponding decomposition coefficient at each decomposition frequency point.

[0142] In some embodiments, the second upper and lower thresholds can be calculated from the first upper and lower thresholds and the initial inter-plate correction coefficient.

[0143] For example, the calculation method for the second upper and lower thresholds can be expressed as:

[0144]

[0145] Where, β i1 β is the upper threshold of the decomposition coefficients at the i-th decomposition frequency point of the historical sensor data set. i2 It is the lower limit threshold of the decomposition coefficients at the i-th decomposition frequency point of the historical sensor data set, α. i1 It is the upper threshold of the i-th decomposition frequency point in the standard sensor data set, α i2 It is the lower limit threshold of the i-th decomposition frequency point in the standard sensor data set, m i is the initial inter-plate correction coefficient; where i is a positive integer.

[0146] In some embodiments, the initial inter-board correction coefficients can be calculated based on the standard decomposition coefficients and the second decomposition coefficients.

[0147] For example, such as Figure 3 As shown, the initial inter-plate correction coefficient can be calculated as follows:

[0148]

[0149] Where, m i It is the initial inter-plate correction factor, w i c is the mean of the decomposition coefficients at the i-th decomposition frequency point of the historical sensor data set. i Let be the mean of the decomposition coefficients at the i-th decomposition frequency point in the standard sensor data set; where i is a positive integer.

[0150] According to the freshness identification method of storage equipment provided in the embodiments of this application, the inter-plate correction coefficients calculated from the standard sensor data set and the historical sensor data set can reveal the inter-plate differences of different sensor arrays. For different sample data, the inter-plate differences can be removed by reconstruction using the initial inter-plate correction coefficients.

[0151] In some embodiments, the initial inter-plate correction function can be constructed based on the second upper and lower thresholds and the decomposition coefficients of the analyzed sample data.

[0152] For example, the initial inter-plate correction function can be constructed as a correction function based on the second upper and lower thresholds and the decomposition coefficients of the analyzed sample data. Specifically, it can be expressed as:

[0153]

[0154] Among them, w ij w is the decomposition coefficient of the j-th sample at the i-th frequency point of the analyzed sample data.* ij It is the correction value of the decomposition coefficient of the j-th sample at the i-th frequency point of the analyzed sample data, max i β is the maximum value among the decomposition coefficients of the analyzed sample data at the i-th wavelet packet decomposition frequency point. i1 β is the upper threshold of the decomposition coefficients at the i-th decomposition frequency point of the historical sensor data set. i2 It is the lower limit threshold of the decomposition coefficient at the i-th decomposition frequency point of the historical sensor data set; where i and j are both positive integers.

[0155] In actual execution, the sample data being analyzed can be real-time sensor data collected by the sensor array. Wavelet packet decomposition can be performed on the real-time sensor data to obtain the decomposition coefficients to be corrected. These decomposition coefficients are then input into the initial inter-board correction function to obtain the corrected decomposition coefficients of the real-time sensor data.

[0156] According to the freshness identification method for storage devices provided in this application, an initial inter-panel correction coefficient is calculated based on a standard sensor data set and a historical sensor data set, and an initial inter-panel correction function is constructed. The initial inter-panel correction coefficient and the constructed initial inter-panel correction function are then deployed as factory values ​​within the storage device. This facilitates timely inter-panel correction of real-time sensor data after the storage device is powered on for the first time after leaving the factory, and also facilitates subsequent updates to the initial inter-panel correction coefficient and the initial inter-panel correction function, thereby improving the inter-panel correction efficiency of the sensor array.

[0157] In some embodiments, after acquiring the real-time sensor data collected by the sensor array at the current acquisition time, the method may further include:

[0158] Calculate the second marginal distribution difference between the first sensor data set corresponding to the sensor array and the standard sensor data set;

[0159] If the difference in the second edge distribution exceeds the first threshold, the inter-plate correction coefficients of the sensor array and the parameters of the inter-plate correction function are updated based on the first sensor data set and the standard sensor data set.

[0160] In this embodiment, the first sensor data set is a collection of real-time sensor data collected at multiple acquisition times within a certain period of time after the storage device leaves the factory.

[0161] The first sensor data set may include real-time sensor data at the current acquisition time and real-time sensor data at multiple acquisition times prior to the current acquisition time.

[0162] The standard sensor dataset is collected by a standard sensor array.

[0163] In some embodiments, the first sensor data set can be a data set of real-time sensor data from multiple acquisition times that has been processed for freshness classification.

[0164] The types and data volume of each type in the first sensor dataset should be consistent with the types and data volume of each type in the standard sensor dataset.

[0165] For example, such as Figure 5 As shown, the standard sensor dataset can be a sample dataset with dimension N×d, where N is the number of samples and d is the number of sensors. Each sample contains d sensor response values. The first N1 rows store empty data samples from the storage device, N1+1 to N2 rows store freshness level data samples, N2+1 to N3 rows store near-freshness level data samples, and N3+1 to N rows store spoilage level data samples. Similarly, the first sensor dataset can also be a dataset with dimension N×d, with the first N1 rows storing empty data samples from the storage device, N1+1 to N2 rows store freshness level data samples, N2+1 to N3 rows store near-freshness level data samples, and N3+1 to N rows store spoilage level data samples. N, N1, N2, N3, and d are all positive integers.

[0166] The second edge distribution difference can be the edge distribution of two datasets, the first sensor dataset and the standard sensor dataset, on the same dimension; it is used to measure the consistency between the first sensor dataset and the standard sensor dataset.

[0167] In actual implementation, the difference in the second marginal distribution can be calculated in the same way as the difference in the first marginal distribution.

[0168] For example, such as Figure 2 As shown, the Pearson correlation coefficient between the first sensor data set and the standard sensor data set is calculated. If the Pearson correlation coefficient between the variables corresponding to the two data sets is close to 1, the marginal distributions of the two data sets are considered to be highly similar, that is, the two data sets are similar. If the Pearson correlation coefficient between the variables corresponding to the two data sets is close to 0 or -1, the two data sets are considered to be dissimilar.

[0169] like Figure 6 As shown, the first threshold can be a threshold that measures the difference in marginal distribution between the first sensor data set or the historical sensor data set and the standard sensor data set.

[0170] In actual implementation, the first threshold can be determined by experimental methods.

[0171] It is understandable that when the difference between the second edge distribution of the first sensor data set collected by the sensor array and the standard sensor data set collected by the standard sensor array exceeds the first threshold, the sensor array can be considered to be dissimilar to the standard sensor array. Therefore, it is necessary to update the inter-board correction coefficients and the parameters of the inter-board correction function of the sensor array based on the first sensor data set and the standard sensor data set.

[0172] It should be noted that when the first edge distribution difference between the historical sensor data set collected by the sensor array and the standard sensor data set is calculated, and the first edge distribution difference exceeds the first threshold, when the second edge distribution difference between the first sensor data set and the standard sensor data set is calculated for the first time after the storage device leaves the factory, and the second edge distribution difference exceeds the first threshold, the inter-panel correction coefficient and the parameters of the inter-panel correction function of the sensor array are updated, that is, the initial inter-panel correction coefficient and the initial inter-panel correction function deployed in the storage device as the factory settings of the storage device are updated.

[0173] When the difference between the first edge distribution of the historical sensor data set collected by the sensor array and the standard sensor data set is calculated, and the difference between the first edge distribution does not exceed the first threshold, when the difference between the first edge distribution and the standard sensor data set is calculated for the first time after the storage device leaves the factory, and the difference between the second edge distribution exceeds the first threshold, the parameters of the inter-plate correction coefficient and the inter-plate correction function of the sensor array can be updated by calculating the initial inter-plate correction coefficient and the initial inter-plate correction function in the storage device based on the first sensor data set and the standard sensor data set.

[0174] According to the freshness identification method for storage devices provided in this application, by calculating the second edge distribution difference between the first sensor data set collected by the sensor array and the standard sensor data set collected by the standard sensor array, the storage devices after leaving the factory can be screened, and the storage devices that need to be corrected between boards can be updated, thereby improving the efficiency and accuracy of subsequent sensor array correction between boards, and thus improving the efficiency and accuracy of freshness identification.

[0175] In some embodiments, when the difference in the second edge distribution exceeds a first threshold, updating the inter-plate correction coefficients of the sensor array and the parameters of the inter-plate correction function based on the first sensor data set and the standard sensor data set may include:

[0176] Wavelet packet decomposition was performed on the standard sensor data set and the first sensor data set respectively to obtain the standard decomposition coefficients and the third decomposition coefficients of the target frequency band.

[0177] Update the initial inter-board correction coefficients based on the standard decomposition coefficients and the third decomposition coefficients;

[0178] The initial inter-board correction function is updated based on the standard decomposition coefficients and the updated initial inter-board correction coefficients.

[0179] In this embodiment, the third decomposition coefficient can be the decomposition coefficient of the target frequency band of the first sensor data set based on wavelet packet decomposition.

[0180] Updating the initial inter-board correction coefficients can be understood as updating the parameters of the initial inter-board correction coefficient calculation formula based on the standard decomposition coefficients and the third decomposition coefficients.

[0181] For example, the updated initial inter-panel correction factor can be calculated based on the parameters of the formula for calculating the initial inter-panel correction factor before the updated storage equipment leaves the factory, which can be specifically expressed as:

[0182]

[0183] Where, m * i It is the updated initial inter-board correction factor, v i c is the mean of the decomposition coefficients at the i-th decomposition frequency point of the first sensor data set. i Let be the mean of the decomposition coefficients at the i-th decomposition frequency point in the standard sensor data set; where i is a positive integer.

[0184] It should be noted that when the sensor array is used for the first inter-board calibration after the storage device is powered on, the initial inter-board calibration coefficient before the update is the factory setting value deployed in the storage device before it leaves the factory; when the sensor array is used for inter-board calibration after the storage device is powered on, the initial inter-board calibration coefficient before the update is obtained based on the initial inter-board calibration coefficient of the previous update.

[0185] In some embodiments, updating the initial inter-board correction function based on the standard decomposition coefficients and the updated initial inter-board correction coefficients may include:

[0186] Based on the first upper and lower limit thresholds corresponding to the standard decomposition coefficients and the updated initial inter-board correction coefficients, the third upper and lower limit thresholds corresponding to the third decomposition coefficients are calculated.

[0187] The initial inter-board correction function is updated based on the third upper and lower thresholds and the third decomposition coefficients.

[0188] In this embodiment, the third upper and lower limit thresholds can be the deviation range of the third decomposition coefficient.

[0189] In actual implementation, the third upper and lower limit thresholds can be obtained by using the calculation formula based on the first upper and lower limit thresholds and the updated initial inter-plate correction coefficient; the calculation formula can be the one built into the storage device after leaving the factory when calculating the second upper and lower limit thresholds based on the first upper and lower limit thresholds and the initial inter-plate correction coefficient.

[0190] For example, the calculation method for the third upper and lower thresholds can be expressed as:

[0191]

[0192] Where, γ i1 γ is the upper limit threshold of the decomposition coefficients at the i-th decomposition frequency point of the first sensor data set. i2 It is the lower limit threshold of the decomposition coefficients at the i-th decomposition frequency point of the first sensor data set, α. i1 It is the upper threshold of the i-th decomposition frequency point in the standard sensor data set, α. i2 It is the lower limit threshold of the i-th decomposition frequency point in the standard sensor data set, m * i is the updated initial inter-board correction coefficient; where i is a positive integer.

[0193] Updating the initial inter-panel correction function can be done by updating the upper and lower thresholds of the decomposition coefficients in the initial inter-panel correction function built into the storage device, based on the updated initial inter-panel correction coefficients.

[0194] For example, when the storage device is powered on for the first time after leaving the factory and undergoing inter-board calibration, the updated initial inter-board calibration function can be an inter-board calibration function that replaces the second upper and lower limit thresholds in the initial inter-board calibration function with the third upper and lower limit thresholds.

[0195] Specifically, it can be expressed as:

[0196]

[0197] Among them, w ij w is the decomposition coefficient of the j-th sample at the i-th frequency point of the analyzed sample data. * ij It is the correction value of the decomposition coefficient of the j-th sample at the i-th frequency point of the analyzed sample data, max i γ is the maximum value among the decomposition coefficients of the analyzed sample data at the i-th wavelet packet decomposition frequency point. i1 γ is the upper limit threshold of the decomposition coefficients at the i-th decomposition frequency point of the first sensor data set. i2 It is the lower limit threshold of the decomposition coefficient at the i-th decomposition frequency point of the first sensor data set; where i and j are both positive integers.

[0198] In actual execution, the sample data being analyzed can be real-time sensor data collected by the sensor array. Wavelet packet decomposition can be performed on the real-time sensor data to obtain the decomposition coefficients to be corrected. These decomposition coefficients to be corrected are then input into the updated initial inter-board correction function to obtain the corrected decomposition coefficients of the real-time sensor data.

[0199] Understandably, the initial inter-plate correction function can be updated as the first set of sensor data collected over a certain period of time is updated.

[0200] In actual implementation, the first sensor data set update can be composed of real-time sensor data collected within a fixed period of time, or it can be based on receiving new real-time sensor data, discarding old real-time sensor data, and maintaining a fixed data capacity.

[0201] For example, a set of real-time sensor data is collected every 20 or 30 minutes to form a new first sensor data set;

[0202] For example, the first sensor data set receives real-time sensor data at the current acquisition time while discarding real-time sensor data from previous acquisition times.

[0203] Of course, other methods of updating the first sensor data set are also possible, and this application does not limit them here.

[0204] It should be noted that since this method constructs the correction function within the measurement period of the sample measurement time window, determining the size of the sample measurement time window is very important, as the size of the sample measurement time window will affect the determination of the upper and lower limits of the decomposition coefficient threshold. As the sample measurement time window moves forward, a new correction function is generated, thereby enabling recursive correction.

[0205] According to the freshness identification method for storage devices provided in this application, a first sensor data set including real-time sensor data at multiple acquisition times is obtained. Based on the decomposition coefficients of the first sensor data set and the upper and lower thresholds of the decomposition coefficients, the initial inter-plate correction coefficients and the initial inter-plate correction function are updated. This enables recursive updating of the initial inter-plate correction coefficients and the initial inter-plate correction function. Therefore, based on the recursively updated initial inter-plate correction function, inter-plate correction of the sensor array can be performed, improving the accuracy of inter-plate correction of the sensor array and thus improving the precision and accuracy of freshness prediction.

[0206] Step 120: Perform wavelet packet decomposition on the real-time sensor data to obtain the first decomposition coefficients of the target frequency band;

[0207] The target frequency band is determined based on the target odor used for freshness identification;

[0208] In this step, the first decomposition coefficients are the decomposition coefficients of the target frequency band based on wavelet packet decomposition of real-time sensor data.

[0209] The target frequency band can be determined based on the target odor used for freshness identification.

[0210] In actual execution, the target frequency band can be the range of frequency points that have a response result obtained by wavelet packet decomposition of the target odor.

[0211] According to the freshness identification method for storage equipment provided in the embodiments of this application, by performing wavelet packet decomposition on real-time sensor data, the decomposition coefficients of the target frequency band corresponding to the real-time sensor data can be obtained. Based on these decomposition coefficients, the inter-plate correction function can be updated, and then the inter-plate correction of the real-time sensor data can be realized based on the corrected correction function, thereby improving the accuracy of freshness identification based on the corrected real-time sensor data.

[0212] Step 130: Correct the first decomposition coefficients based on the inter-plate correction function, and obtain the target sensor data based on the corrected first decomposition coefficients;

[0213] In this step, the target sensor data is used for freshness identification.

[0214] The target sensor data can be calculated based on the first decomposition coefficients after inter-plate correction.

[0215] In some embodiments, such as Figure 4 As shown, step 130 may include:

[0216] Input the first decomposition coefficients into the inter-board correction function, and obtain the corrected first decomposition coefficients output by the inter-board correction function;

[0217] The corrected first decomposition coefficients are reconstructed using wavelet packet inverse transform to obtain the target sensor data.

[0218] In this embodiment, the wavelet packet inverse transform is used to reconstruct the original signal from each sub-band signal.

[0219] In practice, the inverse wavelet packet transform can be the inverse operation of the wavelet packet decomposition process. That is, the corresponding synthesis filter is applied to each sub-band signal, and the sub-band signals are gradually merged until the original signal is reconstructed.

[0220] In some embodiments, the wavelet packet inverse transform reconstruction of the corrected first decomposition coefficients can be performed through a single-step merging operation.

[0221] For example, the single-step merging operation in the inverse wavelet packet transform process of the corrected first decomposition coefficients can be expressed as:

[0222]

[0223] in, It is the subband coefficient of the (j+1)th layer. is the wavelet packet coefficient sequence of the k-th subband obtained from the j-th level decomposition, and h is a low-pass filter; where j and k are both positive integers.

[0224] In actual implementation, the low-pass filter is defined based on the selected wavelet basis function. These filters are usually symmetrical and satisfy the perfect reconstruction condition, ensuring that the inverse transform can accurately recover the original signal.

[0225] After the wavelet packet decomposition coefficients of the real-time sensor data acquired by the sensor array are corrected, the corrected real-time sensor data can be obtained by reconstruction.

[0226] According to the freshness identification method for storage devices provided in this application, all wavelet packet decomposition coefficients of real-time sensor data are corrected, and then the corrected real-time sensor data decomposition coefficients are reconstructed by wavelet packet inverse transform. This preserves the original characteristics of the real signal at the target frequency point of the real-time sensor data, and realizes the smoothness and integrity of the corrected decomposition coefficient data. Thus, freshness identification is performed based on the corrected smooth and complete target sensor data, thereby improving the accuracy of freshness identification.

[0227] Step 140: Based on the target sensor data, predict the freshness information corresponding to the target odor.

[0228] In this step, the freshness information is information related to the freshness category of the stored objects in the storage device.

[0229] For example, freshness information could include whether the ingredients are fresh or spoiled.

[0230] For example, freshness information can include: fresh, not very fresh, or spoiled.

[0231] In actual implementation, a correlation table between candidate target odors and freshness information can be pre-built. In subsequent applications, the correlation table can be searched based on the target data to obtain the freshness category corresponding to the target odor.

[0232] In some embodiments, predicting the freshness information corresponding to the target odor based on target sensor data may include:

[0233] The target sensor data is input into the freshness classification model to obtain the freshness information output by the freshness classification model.

[0234] In this embodiment, the freshness classification model is trained using standard sensor data as samples and the sample freshness category corresponding to the target odor corresponding to the standard sensor data as the sample label.

[0235] The sample freshness category corresponding to the target odor in the standard sensor data can be manually labeled and classified during the standard sensor data acquisition process. The specific category can be determined according to the user's actual needs.

[0236] For example, if the freshness categories that users actually need include: empty storage equipment, fresh, slightly fresh, and spoiled, then the sample freshness categories need to include empty storage equipment, fresh, slightly fresh, and spoiled in order to train a freshness classification model that meets users' actual needs.

[0237] It should be noted that the parameters of the freshness classification model are fixed after training, and are used to classify the input target sensor data in subsequent tests.

[0238] In practical applications, establishing a freshness classification model corresponding to a target odor can enable freshness prediction for multiple target odors and can simultaneously process large amounts of data on multiple target gases, thereby achieving efficient freshness prediction. The freshness classification model based on machine learning algorithms has good adaptive learning capabilities and can automatically adjust its parameters and structure as the user's usage time increases, thereby improving the accuracy and precision of freshness prediction.

[0239] In some embodiments, predicting the freshness information corresponding to the target odor based on target sensor data may include:

[0240] Principal component analysis and independent component analysis are performed on the target sensor data at the current acquisition time and the target sensor data at at least one historical acquisition time to obtain the independent component matrix;

[0241] Each independent component in the independent component matrix is ​​filtered, and the filtered independent component matrix is ​​reconstructed to obtain the processed target sensor data.

[0242] Based on the processed target sensor data, the freshness information corresponding to the target sensor data is predicted.

[0243] In this embodiment, the target sensor data at the current acquisition time is obtained by performing inter-board calibration on the real-time sensor data at the current acquisition time collected by the sensor array after the storage device leaves the factory.

[0244] The target sensor data at the historical acquisition time is obtained by performing inter-board correction on the real-time sensor data at the historical acquisition time before the current acquisition time collected by the sensor array after the storage device leaves the factory.

[0245] In actual execution, the target sensor data at the current acquisition time and the target sensor data at at least one historical acquisition time can constitute a target sensor data set for subsequent processing.

[0246] Principal component analysis is an algorithm used for dimensionality reduction processing of target sensor datasets.

[0247] Principal component analysis transforms a target sensor dataset into a new set of variables (i.e., principal components) through linear transformation. These new variables are linear combinations of the original variables and are uncorrelated with each other.

[0248] The main goal of principal component analysis is to find the main features in the target sensor dataset and represent these features in a new way.

[0249] Independent component analysis is an algorithm used for signal processing and data analysis.

[0250] The main goal of independent component analysis is to find a linear transformation, or unmixing matrix, that makes the transformed signal components as statistically independent as possible.

[0251] The independent component matrix is ​​a matrix composed of the independent components obtained from principal component analysis and independent component analysis of the original data.

[0252] Independent components can be understood as components that are statistically independent.

[0253] It is understandable that the degree to which each independent component is affected by different factors may vary. For example, different independent components may be affected to different degrees by factors such as changes in temperature and humidity and changes in airflow.

[0254] The processed target sensor dataset can be used for freshness identification.

[0255] In some embodiments, a data matrix can be constructed from the target sensor data set at the current acquisition time and the target sensor data set at at least one historical acquisition time. Principal component analysis is performed on the data matrix to obtain the principal component matrix. Then, independent component analysis is performed on the data matrix based on the principal component matrix to obtain the independent component matrix.

[0256] In this embodiment, the principal component matrix can be obtained from the standardized data matrix and the principal component transformation matrix.

[0257] The specific method for obtaining the principal component matrix is ​​explained below.

[0258] In some embodiments, performing principal component analysis on the data matrix to obtain the principal component matrix may include:

[0259] Standardize the data matrix.

[0260] In practice, the standardization of the data matrix can be achieved through mean centering or variance scaling.

[0261] Data matrix standardization can prevent the over-capturing of certain large-value features, ensuring that the data size is within the same range and that different variables have the same scale, which facilitates subsequent principal component analysis.

[0262] Solve for the covariance matrix of the standardized data matrix to obtain the eigenvalues ​​and eigenvectors of the covariance matrix. Then, select the principal components and their corresponding eigenvectors to obtain the principal component transformation matrix.

[0263] Based on the data matrix and the principal component transformation matrix, the principal component matrix is ​​obtained.

[0264] In this embodiment, the principal component transformation matrix may include k principal components, each of which is a k eigenvector of the covariance matrix corresponding to the data matrix.

[0265] The principal component transformation matrix can be expressed as:

[0266]

[0267] Wherein, the columns of P are the eigenvectors corresponding to the first k largest eigenvalues ​​in the eigenvector group after solving the covariance matrix, v i (j) Let i represent the j-th element of the i-th eigenvector, k be the number of principal components, and n be the number of features in the data matrix, i.e., the number of sensors; where i, j, k, and n are all positive integers.

[0268] In some embodiments, the value of k can be determined by experimentally observing whether the curve of the data matrix after processing the data matrix is ​​smooth.

[0269] After obtaining the principal component transformation matrix, the principal component matrix is ​​calculated based on the following formula:

[0270] The calculation method is as follows:

[0271] Z = X·P

[0272] Where Z is an m×k principal component matrix, m is the number of samples in the data matrix, and k is the number of principal components selected. Each row of Z... i Let X represent the coordinates of the i-th sample in the principal component space, X be the standardized data matrix, and P be the principal component transformation matrix, where m and k are both positive integers.

[0273] After obtaining the principal component matrix, independent component processing is performed based on the principal component matrix to obtain the independent component matrix.

[0274] The independent component matrix can be obtained from the principal component matrix and the unmixing matrix.

[0275] The calculation method can be expressed as:

[0276] U = W·Z

[0277] Where U is the independent component matrix, W is the unmixing matrix, and Z is the principal component matrix, where the unmixing matrix W is obtained after performing independent component analysis on the data matrix.

[0278] In some embodiments, the independent components in the independent component matrix are filtered, and the filtered independent component matrix is ​​reconstructed to obtain a processed target sensor data set, which may include:

[0279] Calculate the correlation coefficient between each independent component and each data point in the data matrix;

[0280] The independent components with correlation coefficients greater than the first threshold are updated to 0, resulting in the updated independent component matrix.

[0281] The updated independent component matrix is ​​inversely transformed to obtain the processed target sensor data set.

[0282] In this embodiment, the correlation coefficient is used to characterize the degree to which temperature, humidity, or airflow affects the system.

[0283] The first threshold is the correlation coefficient value that can distinguish between target data and non-target data.

[0284] The first threshold is used to filter components in the independent component matrix that are highly correlated with the temperature and humidity sensor but lowly correlated with the gas sensor.

[0285] The first threshold can be determined by observing whether the curve of the data matrix after processing is smooth.

[0286] The correlation coefficients between each independent component and each data point in the data matrix are obtained by the formula.

[0287] The calculation method is as follows:

[0288]

[0289] Where R is the correlation coefficient matrix, i and j represent rows and columns respectively, and n is the number of sensors; where i, j and n are all positive integers.

[0290] Based on the principal component transformation matrix and unmixing matrix corresponding to the data matrix, the updated independent component matrix can be inversely transformed to obtain the processed target sensor data set.

[0291] In some embodiments, the calculation process of inverse transformation of the updated independent component matrix can be represented as follows:

[0292] X = W -1 ·U·P -1

[0293] Where X is the processed target sensor data set, W is the unmixing matrix corresponding to the processed target sensor data set, P is the unmixing matrix corresponding to the processed target sensor data set, and U is the updated independent component matrix.

[0294] According to the freshness identification method for storage devices provided in this application, principal component analysis and independent component analysis are performed on the target sensor data set at the current and historical acquisition times. Based on the processed independent component matrix, each independent component is filtered, which can reduce the dimensionality of the target sensor data set and obtain the target sensor data set that excludes the influence of temperature, humidity and airflow in the mixed signal, thereby achieving gas sensor interference suppression. On this basis, freshness identification is performed based on more accurate gas sensor data, which helps to improve the freshness identification results and achieve more accurate freshness monitoring.

[0295] Through multiple experiments, the inventors verified that the processing method in this embodiment suppresses interference such as temperature, humidity, and airflow in the target sensor data collected by the sensor array. Compared with the case without this processing, the accuracy of freshness identification results is significantly improved by 14.4075% when based on the target sensor data for freshness identification.

[0296] In this embodiment, by further processing the target sensor data at the current acquisition time to suppress interference such as temperature, humidity, and airflow, the accuracy and precision of predicting freshness category based on real-time sensor data can be significantly improved, further enhancing the user experience.

[0297] During the research and development process, the inventors discovered that in related technologies, when calibrating a sensor array, each sensor in the array is calibrated individually. This method does not consider the mutual influence between the sensors in the sensor array, nor the inconsistency of such mutual influence across different sensor arrays, which affects the calibration results and thus the subsequent recognition performance.

[0298] In this application, by acquiring real-time sensor data collected by a sensor array, wavelet packet decomposition is performed on the real-time sensor data to obtain the first decomposition coefficients of the target frequency band of the real-time sensor data. Then, the first decomposition coefficients of the real-time sensor data are corrected based on the inter-board correction function to obtain the corrected target sensor data. This achieves correction based on the response values ​​between sensor arrays, improving the consistency of response values ​​between sensor arrays. On this basis, freshness recognition is performed based on the corrected target sensor data of the sensor array, improving the accuracy and precision of freshness recognition and enhancing the user experience.

[0299] According to the freshness identification method for storage devices provided in this application, wavelet packet decomposition is performed on real-time sensor data collected by sensor arrays to obtain decomposition coefficients, and the real-time sensor data is corrected based on the inter-plate correction function to obtain target sensor data. This achieves correction based on the response values ​​between sensor arrays, improving the consistency of response values ​​between sensor arrays. On this basis, freshness identification is performed based on the corrected target sensor data of the sensor array, improving the accuracy and precision of freshness identification and enhancing the user experience.

[0300] The freshness identification method for storage equipment provided in this application can be executed by a freshness identification device for the storage equipment. This application uses the example of a freshness identification device for the storage equipment executing the freshness identification method to illustrate the freshness identification device for the storage equipment provided in this application.

[0301] This application also provides a freshness identification device for storage equipment.

[0302] like Figure 8 As shown, the freshness identification device of the storage equipment includes a sensor array and comprises a first processing module 810, a second processing module 820, a third processing module 830, and a fourth processing module 840.

[0303] The first processing module 810 is used to acquire real-time sensor data at the current acquisition time collected by the sensor array;

[0304] The second processing module 820 is used to perform wavelet packet decomposition on the real-time sensor data to obtain the first decomposition coefficients of the target frequency band; the target frequency band is determined based on the target odor used for freshness identification.

[0305] The third processing module 830 is used to correct the first decomposition coefficients based on the inter-board correction function, and to obtain the target sensor data based on the corrected first decomposition coefficients.

[0306] The fourth processing module 840 is used to predict the freshness information corresponding to the target odor based on the target sensor data.

[0307] According to the freshness identification device for storage equipment provided in the embodiments of this application, the decomposition coefficients are obtained by performing wavelet packet decomposition on the real-time sensor data collected by the sensor array, and the target sensor data is obtained by correcting the real-time sensor data based on the inter-plate correction function. This realizes the correction based on the response values ​​between the sensor arrays, thereby improving the consistency of the response values ​​between the sensor arrays. On this basis, freshness identification is performed based on the target sensor data of the corrected sensor array, thereby improving the accuracy and precision of freshness identification and enhancing the user experience.

[0308] In some embodiments, the third processing module 830 may also be used for:

[0309] Input the first decomposition coefficients into the inter-board correction function, and obtain the corrected first decomposition coefficients output by the inter-board correction function;

[0310] The corrected first decomposition coefficients are reconstructed using wavelet packet inverse transform to obtain the target sensor data.

[0311] In some embodiments, the fourth processing module 840 can also be used for:

[0312] Input the target sensor data into the freshness classification model and obtain the freshness information output by the freshness classification model;

[0313] The freshness classification model is trained using standard sensor data as samples and the freshness category of the sample corresponding to the target odor of the standard sensor data as the sample label.

[0314] In some embodiments, the fourth processing module 840 can also be used for:

[0315] Principal component analysis and independent component analysis are performed on the target sensor data at the current acquisition time and the target sensor data at at least one historical acquisition time to obtain the independent component matrix;

[0316] Each independent component in the independent component matrix is ​​filtered, and the filtered independent component matrix is ​​reconstructed to obtain the processed target sensor data.

[0317] Based on the processed target sensor data, the freshness information corresponding to the target sensor data is predicted.

[0318] In some embodiments, the device may further include a fifth processing module for:

[0319] Before acquiring the real-time sensor data at the current acquisition time from the sensor array, the first marginal distribution difference between the historical sensor data set corresponding to the sensor array and the standard sensor data set is calculated; the historical sensor data set includes sensor data from multiple historical acquisition times before the storage device leaves the factory; the standard sensor data set is acquired by the standard sensor array.

[0320] When the difference in the first edge distribution exceeds a first threshold, the initial inter-plate correction coefficient and the initial inter-plate correction function are built into the storage device; the initial inter-plate correction coefficient and the initial inter-plate correction function are calculated based on the historical sensor data set and the standard sensor data set.

[0321] In some embodiments, the device may further include a sixth processing module for:

[0322] Before acquiring the real-time sensor data at the current acquisition time collected by the sensor array, acquire the historical sensor data set of the sensor array at multiple historical acquisition times and the standard sensor data set of the standard sensor array at multiple acquisition times.

[0323] Wavelet packet decomposition was performed on the standard sensor dataset and the historical sensor dataset respectively to obtain the standard decomposition coefficients and the second decomposition coefficients of the target frequency band.

[0324] The initial inter-board correction coefficients are calculated based on the standard decomposition coefficients and the second decomposition coefficients.

[0325] Based on the standard decomposition coefficients and the initial inter-plate correction coefficients, the initial inter-plate correction function is determined; wherein, the initial inter-plate correction coefficients and the initial inter-plate correction function are deployed as the factory settings of the storage equipment.

[0326] In some embodiments, the sixth processing module can also be used for:

[0327] Based on the first upper and lower thresholds corresponding to the standard decomposition coefficients and the initial inter-board correction coefficients, the second upper and lower thresholds corresponding to the second decomposition coefficients are calculated.

[0328] An initial inter-plate correction function is constructed based on the second upper and lower thresholds and the second decomposition coefficient.

[0329] In some embodiments, the device may further include a seventh processing module for:

[0330] After acquiring the real-time sensor data collected by the sensor array at the current acquisition time, the second edge distribution difference between the first sensor data set corresponding to the sensor array and the standard sensor data set is calculated; the first sensor data set includes the real-time sensor data at the current acquisition time and the real-time sensor data at multiple acquisition times prior to the current acquisition time; the standard sensor data set is collected by the standard sensor array.

[0331] If the difference in the second edge distribution exceeds the first threshold, the inter-plate correction coefficients of the sensor array and the parameters of the inter-plate correction function are updated based on the first sensor data set and the standard sensor data set.

[0332] The freshness identification device for storage devices in this application embodiment can be the storage device itself, or a component within the storage device, such as an integrated circuit or a chip.

[0333] The freshness identification device for storage equipment in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0334] The freshness identification device for storage equipment provided in this application embodiment can achieve… Figures 1 to 7 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0335] This application also provides a storage device.

[0336] like Figure 9 As shown, the storage device includes a housing 910, a sensor array 920, and a freshness identification device for the storage device as described in any of the above embodiments.

[0337] In this embodiment, the storage device can be understood as a refrigeration storage device in a broad sense, including but not limited to refrigerators, freezers, display cases, beverage cabinets, wine cabinets, refrigerated display cases, and refrigerated vending machines. The storage devices have diverse structural forms and a wide range of applications.

[0338] The sensor array 920 is housed within the housing 910.

[0339] The sensor array 920 includes at least one odor sensor.

[0340] The freshness identification device for the storage equipment is electrically connected to the sensor array 920.

[0341] The freshness identification device for storage equipment is used to perform the steps of the freshness identification method for storage equipment as described in any of the above embodiments, which will not be repeated here.

[0342] According to the storage device provided in the embodiments of this application, the decomposition coefficients are obtained by performing wavelet packet decomposition on the real-time sensor data collected by the sensor array, and the target sensor data is obtained by correcting the real-time sensor data based on the inter-plate correction function. This realizes the correction based on the response values ​​between the sensor arrays and improves the consistency of the response values ​​between the sensor arrays. On this basis, freshness recognition is performed based on the target sensor data of the corrected sensor array, which improves the accuracy and precision of freshness recognition and enhances the user experience.

[0343] In some embodiments, such as Figure 10 As shown, this application embodiment also provides an electronic device 1000, including a processor 1001, a memory 1002, and a computer program stored on the memory 1002 and executable on the processor 1001. When the program is executed by the processor 1001, it implements the various processes of the above-described freshness detection method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0344] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0345] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described freshness identification method embodiment for storage devices and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0346] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0347] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for identifying the freshness of storage devices.

[0348] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0349] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described freshness identification method embodiment for storage devices, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0350] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0351] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0352] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0353] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0354] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0355] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A freshness identification method of a storage device, characterized by, The storage device comprises a sensor array, and the method comprises: obtaining real-time sensor data collected by the sensor array at a current collection time point; wavelet packet decomposing the real-time sensor data to obtain first decomposition coefficients of a target frequency band; the target frequency band is determined based on a target odor used for freshness identification; correcting the first decomposition coefficients based on an inter-plate correction function, and obtaining target sensor data based on the corrected first decomposition coefficients; based on the target sensor data, predicting freshness information corresponding to the target odor.

2. The method of claim 1, wherein The method further comprises: calculating a first edge distribution difference between a historical sensor data set corresponding to the sensor array and a standard sensor data set; the historical sensor data set comprises sensor data at a plurality of historical collection time points before the storage device is shipped; the standard sensor data set is collected by a standard sensor array; in a case where the first edge distribution difference exceeds a first threshold value, embedding an initial inter-plate correction coefficient and an initial inter-plate correction function in the storage device; the initial inter-plate correction coefficient and the initial inter-plate correction function are calculated based on the historical sensor data set and the standard sensor data set.

3. The method of claim 1, wherein The method further comprises: obtaining a historical sensor data set of the sensor array at a plurality of historical collection time points and a standard sensor data set of a standard sensor array at a plurality of collection time points; wavelet packet decomposing the standard sensor data set and the historical sensor data set respectively to obtain standard decomposition coefficients and second decomposition coefficients of the target frequency band; 4. The freshness identification method of the storage apparatus according to any one of claims 1 to 3, characterized by, calculating an initial inter-plate correction coefficient based on the standard decomposition coefficients and the second decomposition coefficients; determining an initial inter-plate correction function based on the standard decomposition coefficients and the initial inter-plate correction coefficient; wherein the initial inter-plate correction coefficient and the initial inter-plate correction function are deployed in the storage device as factory setting values of the storage device. The method further comprises: calculating a second upper and lower limit threshold value corresponding to the second decomposition coefficients based on a first upper and lower limit threshold value corresponding to the standard decomposition coefficients and the initial inter-plate correction coefficient; constructing the initial inter-plate correction function based on the second upper and lower limit threshold value and the second decomposition coefficients.

5. The freshness identification method of the storage apparatus according to claim 4, characterized by, ​ ​ ​ 6. The freshness identification method of the storage apparatus according to any one of claims 1 to 3, characterized by, After the real-time sensor data at a current collection time point collected by the sensor array is acquired, the method further comprises: calculating a second edge distribution difference between a first sensor data set corresponding to the sensor array and a standard sensor data set; the first sensor data set comprises the real-time sensor data at the current collection time point and real-time sensor data at a plurality of collection time points before the current collection time point; the standard sensor data set is collected by a standard sensor array; in a case where the second edge distribution difference exceeds a first threshold, updating an inter-plate correction coefficient of the sensor array and a parameter of the inter-plate correction function based on the first sensor data set and the standard sensor data set.

7. The method of claim 1-3, wherein The freshness information corresponding to the target odor is predicted based on the target sensor data, comprising: inputting the target sensor data into a freshness classification model to obtain the freshness information output by the freshness classification model; The freshness classification model is trained based on standard sensor data as samples and sample freshness categories corresponding to target odors corresponding to the standard sensor data as sample labels.

8. The method of claim 1-3, wherein The freshness information corresponding to the target sensor data is predicted based on the target sensor data, comprising: performing principal component analysis and independent component analysis on the target sensor data at the current collection time point and the target sensor data at at least one historical collection time point to obtain an independent component matrix; screening each independent component in the independent component matrix and reconstructing the screened independent component matrix to obtain processed target sensor data; based on the processed target sensor data, the freshness information corresponding to the target sensor data is predicted.

9. A freshness identification device for a storage apparatus, characterized by comprising: The storage device comprises a sensor array, and the device comprises: a first processing module for acquiring real-time sensor data collected by the sensor array at a current collection time point; a second processing module for wavelet packet decomposition of the real-time sensor data to obtain first decomposition coefficients of a target frequency band; the target frequency band is determined based on a target odor used for freshness identification; a third processing module for correcting the first decomposition coefficients based on an inter-plate correction function and obtaining target sensor data based on the corrected first decomposition coefficients; a fourth processing module for predicting freshness information corresponding to the target odor based on the target sensor data.

10. A storage device, characterized by comprising: a sensor array comprising at least one odor sensor; The freshness identification device of the storage device of claim 9, the freshness identification device of the storage device is electrically connected with the sensor array. 11.A non-transitory computer-readable storage medium having stored thereon a computer program. The computer program is executed by the processor to realize the freshness identification method of the storage device of any one of claims 1-8.

12. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to realize the freshness identification method of the storage device of any one of claims 1-8.

Images