A device life prediction method, apparatus, device and storage medium

Abstract

The present disclosure relates to a device life prediction method, device, equipment and storage medium. The device life prediction method comprises: obtaining historical use data of a to-be-predicted device, the historical use data being historical data representing a device operation state of the to-be-predicted device; determining a current characteristic value of a target statistical characteristic of the historical use data; and predicting a remaining use life of the to-be-predicted device based on the current characteristic value and a failure threshold of the target statistical characteristic. The life prediction of each to-be-predicted device is based on its own historical use data, and the sensitivity of the target statistical characteristic to the degree of device degradation makes it more suitable for predicting the remaining use life, so that the prediction result is closer to the true result, the prediction deviation and the maintenance lag occurrence rate are reduced, the product yield is improved, and the production capacity loss is reduced.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent manufacturing, specifically to a device lifetime prediction method, apparatus, equipment, and storage medium. Background Technology

[0002] In recent years, with the accelerated development of computer, communication, and consumer electronics manufacturing (collectively known as 3C electronics), many new intelligent manufacturing models that can be replicated and promoted have gradually emerged. For key components of product manufacturing equipment, it is usually necessary to predict their remaining service life to guide their maintenance and use.

[0003] However, predicting the remaining lifespan of devices using related technologies has the problem of large deviations between the predicted and actual results, which will lead to delayed and ineffective device maintenance, low product yield, and high production capacity loss. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides a device lifetime prediction method, apparatus, equipment, and storage medium.

[0005] According to a first aspect of the present disclosure, a device lifetime prediction method is provided, the device lifetime prediction method comprising:

[0006] Obtain historical usage data of the device to be predicted, wherein the historical usage data is historical data characterizing the device's operating status;

[0007] Determine the current feature value of the target statistical feature of the historical usage data;

[0008] Based on the current feature value and the failure threshold of the target statistical feature, the remaining service life of the device to be predicted is predicted.

[0009] In some embodiments of this disclosure, determining the current feature value of the target statistical feature of the historical usage data includes:

[0010] Based on the historical usage data, a feature function is constructed, which is used to characterize the relationship between the number of operations of the device to be predicted and the feature value of the target statistical feature;

[0011] Based on the feature function and the current number of operations of the device to be predicted, the current feature value corresponding to the target statistical feature is determined.

[0012] In some embodiments of this disclosure, constructing the feature function based on the historical usage data includes:

[0013] The historical usage data is preprocessed to obtain a first data set, which includes multiple first data groups, and each first data group includes the target force data of a single operation of the device to be predicted.

[0014] Based on the target statistical characteristics and the first data set, determine the initial feature values, degradation parameters, and fluctuation parameters;

[0015] The feature function is constructed based on the initial feature value, the degradation parameter, and the fluctuation parameter.

[0016] In some embodiments of this disclosure, predicting the remaining lifetime of the device to be predicted based on the current feature value and the failure threshold of the target statistical feature includes:

[0017] Based on the current feature value, the failure threshold, the degradation parameter, and the fluctuation parameter, the probability density of remaining useful life is determined.

[0018] The remaining useful life is determined based on the probability density of the remaining useful life.

[0019] In some embodiments of this disclosure, the historical usage data includes multiple second data groups, each second data group including the raw force application data of a single operation of the device to be predicted, and the preprocessing of the historical usage data to obtain a first data set includes:

[0020] Each of the second data groups is resampled to obtain multiple third data groups, each of which contains the same number of data points;

[0021] Differentiate the data in each of the third data groups to obtain multiple fourth data groups;

[0022] The first data set is obtained by taking the absolute value of the data in each of the fourth data groups.

[0023] In some embodiments of this disclosure, determining the initial feature value, degradation parameter, and fluctuation parameter based on the target statistical features and the first data set includes:

[0024] Based on the first data set, determine the feature value of the target statistical feature corresponding to each of the first data groups;

[0025] The feature value corresponding to the first data group in the first group is determined as the initial feature value;

[0026] Based on the initial feature values ​​corresponding to each of the first data groups, the degradation parameter and the fluctuation parameter are determined.

[0027] In some embodiments of this disclosure, before determining the current feature value of the target statistical feature of the historical usage data, the device lifetime prediction method further includes:

[0028] Acquire sample data, which includes historical usage data of the failed devices;

[0029] Based on the sample data, the target statistical feature and the failure threshold of the target statistical feature are determined.

[0030] In some embodiments of this disclosure, determining the target statistical feature and the failure threshold of the target statistical feature based on the sample data includes:

[0031] The sample data is preprocessed to obtain a second data set, which includes multiple fifth data groups, each of which includes the target force data of a single operation of the failed device.

[0032] Based on the second data set, the feature values ​​of multiple statistical features corresponding to each of the fifth data groups are determined;

[0033] The target statistical feature is determined based on the feature values ​​of multiple statistical features corresponding to each of the fifth data groups;

[0034] The failure threshold is determined based on the target statistical characteristics.

[0035] In some embodiments of this disclosure, the sample data includes multiple sixth data groups, each of which includes the original force application data of a single operation of the failed device. The preprocessing of the sample data to obtain a second data set includes:

[0036] Each of the sixth data groups is resampled to obtain multiple seventh data groups, each of which contains the same number of data points.

[0037] Differentiate the data in each of the seventh data groups to obtain multiple eighth data groups;

[0038] The absolute value of each of the eighth data groups is taken to obtain the second data set.

[0039] In some embodiments of this disclosure, determining the target statistical feature based on the feature values ​​of multiple statistical features corresponding to each of the fifth data groups includes:

[0040] Based on the feature values ​​of multiple statistical features corresponding to each of the fifth data groups, determine multiple evaluation index values ​​corresponding to each of the statistical features;

[0041] The target statistical feature is determined based on the multiple evaluation index values ​​corresponding to each of the statistical features.

[0042] In some embodiments of this disclosure, the plurality of evaluation index values ​​include trend index values, monotonicity index values, and robustness index values. Determining the target statistical feature based on the plurality of evaluation index values ​​corresponding to each statistical feature includes:

[0043] Based on the trend index value, monotonicity index value, and robustness index value corresponding to each of the statistical characteristics, the mean of the trend index, the mean of the monotonicity index, and the mean of the robustness index are determined.

[0044] Based on the trend index value, monotonicity index value and robustness index value corresponding to each statistical feature, the total evaluation index value corresponding to each statistical feature is determined. The total evaluation index value is the sum of the trend index value, monotonicity index value and robustness index value corresponding to each statistical feature.

[0045] The statistical feature that is the largest sum of the evaluation indicators, where the trend indicator value, the monotonicity indicator value, and the robustness indicator value are all greater than the mean of the trend indicator, the mean of the monotonicity indicator, and the mean of the robustness indicator, is determined as the target statistical feature.

[0046] In some embodiments of this disclosure, the plurality of statistical features include any combination of a plurality of time-domain features and / or a plurality of frequency-domain features.

[0047] According to a second aspect of the present disclosure, a device lifetime prediction apparatus is provided, the device lifetime prediction apparatus comprising:

[0048] The first acquisition module is used to acquire historical usage data of the device to be predicted, wherein the historical usage data is historical data characterizing the device operation status of the device to be predicted.

[0049] The first determining module is used to determine the current feature value of the target statistical feature of the historical usage data;

[0050] The second determining module is used to predict the remaining service life of the device to be predicted based on the current feature value and the failure threshold of the target statistical feature.

[0051] According to a third aspect of the present disclosure, a device lifetime prediction apparatus is provided, the device lifetime prediction apparatus comprising:

[0052] processor;

[0053] Memory used to store processor-executable instructions;

[0054] The processor is configured as follows:

[0055] Obtain historical usage data of the device to be predicted, wherein the historical usage data is historical data characterizing the device's operating status;

[0056] Determine the current feature value of the target statistical feature of the historical usage data;

[0057] Based on the current feature value and the failure threshold of the target statistical feature, the remaining service life of the device to be predicted is predicted.

[0058] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein when instructions in the storage medium are executed by a processor of a device lifetime prediction device, the device lifetime prediction device is enabled to perform a device lifetime prediction method, the device lifetime prediction method comprising:

[0059] Obtain historical usage data of the device to be predicted, wherein the historical usage data is historical data characterizing the device's operating status;

[0060] Determine the current feature value of the target statistical feature of the historical usage data;

[0061] Based on the current feature value and the failure threshold of the target statistical feature, the remaining service life of the device to be predicted is predicted.

[0062] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: by acquiring historical usage data of the device to be predicted, determining the current characteristic value of the target statistical feature of the historical usage data, and then predicting the remaining service life of the device to be predicted based on the current characteristic value and the failure threshold of the target statistical feature, the prediction of device lifespan is realized. The lifespan prediction of each device to be predicted is based on its own historical usage data, and the sensitivity of the target statistical feature to the degree of device degradation makes it more suitable for predicting the remaining service life, making the prediction results closer to the actual results, reducing prediction deviation and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0063] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0064] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0065] Figure 1 This is a flowchart illustrating a device lifetime prediction method according to an exemplary embodiment.

[0066] Figure 2 This is a flowchart illustrating, according to an exemplary embodiment, the determination of the current feature value of a target statistical feature of historical usage data.

[0067] Figure 3 This is a flowchart illustrating the construction of a feature function based on historical usage data, according to an exemplary embodiment.

[0068] Figure 4 This is a flowchart illustrating, according to an exemplary embodiment, a method for predicting the remaining lifespan of a device based on a failure threshold of current eigenvalues ​​and target statistical characteristics.

[0069] Figure 5 This is a flowchart illustrating, according to an exemplary embodiment, the preprocessing of historical usage data to obtain a first data set.

[0070] Figure 6 It is a flowchart illustrating, according to an exemplary embodiment, the determination of initial feature values, degradation parameters, and fluctuation parameters based on target statistical features and a first data set.

[0071] Figure 7 This is a flowchart illustrating a device lifetime prediction method according to another exemplary embodiment.

[0072] Figure 8 This is a flowchart illustrating, according to an exemplary embodiment, the determination of target statistical features and failure thresholds of target statistical features based on sample data.

[0073] Figure 9 This is a flowchart illustrating, according to an exemplary embodiment, the preprocessing of sample data to obtain a second data set.

[0074] Figure 10 This is a flowchart illustrating, according to an exemplary embodiment, the determination of target statistical features based on feature values ​​of multiple statistical features corresponding to each fifth data group.

[0075] Figure 11 This is a flowchart illustrating, according to an exemplary embodiment, the determination of target statistical features based on multiple evaluation index values ​​corresponding to each statistical feature.

[0076] Figure 12 This is a comparison chart of the predicted and actual remaining useful life values ​​according to an exemplary embodiment.

[0077] Figure 13 This is a flowchart illustrating a device lifetime prediction method according to another exemplary embodiment.

[0078] Figure 14 This is a block diagram illustrating a device lifetime prediction apparatus according to an exemplary embodiment.

[0079] Figure 15 This is a block diagram of a device lifetime prediction apparatus according to another exemplary embodiment.

[0080] Figure 16 This is a block diagram illustrating a device lifetime prediction apparatus according to an exemplary embodiment.

[0081] In the picture:

[0082] 10-First acquisition module; 20-First determination module; 30-Second determination module; 40-Second acquisition module; 50-Third determination module; 101-Processing component; 102-Memory; 103-Power component; 104-Multimedia component; 105-Audio component; 106-Input / output interface; 107-Sensor component; 108-Communication component; 109-Processor. Detailed Implementation

[0083] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0084] With the rapid development of 3C product manufacturing, some replicable and scalable new intelligent manufacturing models have gradually emerged. These are advanced production models that achieve lean goals such as improved production quality and reliability, and enhanced production and energy efficiency. For example, electronic products such as mobile phones are mostly assembled from several tiny components fastened with screws. Nowadays, automatic screw fastening devices are widely used to achieve fully automated fastening operations. As a key component of automatic screw fastening devices, the remaining service life of the screwdriver bits needs to be predicted to guide maintenance personnel in scheduling maintenance work.

[0085] In related technologies, the actual number of operations performed on a device is typically counted, and the difference between the actual number of operations and the permitted number of operations stated in the device's specifications is used as a predicted remaining lifespan. However, predicting the remaining lifespan of a device using these technologies relies on empirical summaries of permitted operation counts. This fails to account for external random factors and individual differences that arise during actual use, leading to significant discrepancies between the predicted and actual results. Consequently, maintenance based on the predicted results is delayed and ineffective, resulting in low production yields and high capacity losses.

[0086] Based on this, an exemplary embodiment of this disclosure provides a device lifetime prediction method. By acquiring historical usage data of the device to be predicted, determining the current characteristic value of the target statistical feature of the historical usage data, and then predicting the remaining lifetime of the device based on the current characteristic value and the failure threshold of the target statistical feature, the device lifetime prediction is achieved. The lifetime prediction of each device is based on its own historical usage data, and the sensitivity of the target statistical feature to the degree of device degradation makes it more suitable for predicting remaining lifetime, resulting in prediction results that are closer to the actual results, reducing prediction bias and maintenance delay rates, improving product yield, and reducing production capacity loss.

[0087] In one exemplary embodiment, reference Figure 1 As shown, a device lifetime prediction method is provided, which includes:

[0088] S100. Obtain historical usage data of the device to be predicted. The historical usage data is historical data that characterizes the operating status of the device to be predicted.

[0089] In step S100, the device to be predicted can be, for example, a key component of a processing or assembly equipment. This device experiences performance degradation, such as wear, under high-frequency use. The historical usage data of the device to be predicted is at least a portion of the usage data from when the device was put into use until the current moment. This historical usage data can be all the usage data of the device, or it can be a portion of the data extracted from all the usage data through methods such as interval sampling. It can characterize the device's operating state. For example, the historical usage data can be force data characterizing the device's operating state, such as torque or pressure, during the use of the device. It can also be other high-frequency data reflecting the device's operating state, such as vibration, current, or sound, during the use of the device.

[0090] S200, Determine the current feature value of the target statistical feature of the historical usage data.

[0091] In step S200, the historical usage data possesses various statistical features that characterize its data and attribute features. The target statistical feature can be, for example, one of these features extracted from a pool of statistical features. Each statistical feature represents a data calculation method and reflects characteristics associated with the historical usage data. For instance, the target statistical feature can be any one of the statistical features such as mean, standard deviation, or root mean square. The target statistical feature can, for example, exhibit strong trend, monotonicity, and low volatility to be suitable for predicting remaining lifespan and to be highly sensitive to the degree of degradation of the device being predicted. The target statistical feature can be determined during the device lifespan prediction process or set in advance before device lifespan prediction.

[0092] The current characteristic value of the target statistical characteristic in historical usage data can be determined based on historical usage data and target statistical characteristics. The current characteristic value is the characteristic value of the target statistical characteristic at the current time or the current number of operations. The current characteristic value can be represented directly by numerical values ​​or indirectly by assigning values ​​to functions.

[0093] S300: Based on the current eigenvalues ​​and the failure threshold of the target statistical characteristics, predict the remaining service life of the device to be predicted.

[0094] In step S300, the target statistical feature has a corresponding failure threshold, representing the device failure after the feature value of the target statistical feature reaches the failure threshold. The failure threshold can be determined, for example, during the device lifetime prediction process, or it can be determined in advance before the device lifetime prediction. Based on the current feature value and the failure threshold, the remaining lifetime of the device to be predicted can be predicted. The remaining lifetime of the device to be predicted can be, for example, the remaining usage time or the remaining number of uses.

[0095] In this embodiment, by acquiring historical usage data of the device to be predicted and determining the current feature value of the target statistical feature of the historical usage data, the remaining service life of the device to be predicted is predicted based on the current feature value and the failure threshold of the target statistical feature, thus realizing the prediction of device life. The life prediction of each device to be predicted is based on its own historical usage data, and the sensitivity of the target statistical feature to the degree of device degradation makes it more suitable for predicting the remaining service life. It fully considers the external random factors and individual differences of the device to be predicted during actual use, making the prediction results closer to the actual results, reducing prediction bias and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0096] In some embodiments, reference Figure 2 As shown, the current feature values ​​of the target statistical features of historical usage data are determined, including:

[0097] S210. Based on historical usage data, construct a feature function. The feature function is used to characterize the relationship between the number of operations of the device to be predicted and the feature value of the target statistical feature.

[0098] In step S210, a feature function is constructed based on historical usage data. Where t represents the number of operations of the device to be predicted, and the characteristic function is... The relationship between the number of operations of the device to be predicted and the characteristic value of the target statistical feature is used as a degradation model for the device to be predicted.

[0099] S220. Based on the characteristic function and the current number of operations of the device to be predicted, determine the current characteristic value corresponding to the target statistical characteristic.

[0100] In step S220, the characteristic function can be used as a basis. And the current number of operations t1 of the device to be predicted, and substitute the current number of operations t1 into the characteristic function. In the process, determine the current feature value corresponding to the target statistical feature.

[0101] In this embodiment, a feature function is constructed based on historical usage data to characterize the relationship between the number of operations of the device under test and the feature value of the target statistical characteristic. The current feature value corresponding to the target statistical characteristic can be determined based on the feature function and the current number of operations of the device under test, thus realizing the determination of the current feature value and providing a basis for subsequent prediction of the remaining service life of the device under test. The construction of the feature function is based on the historical usage data of the device under test, and the current feature value of the target statistical characteristic is determined according to the current number of operations of the device under test. This fully considers external random factors and individual differences in the actual use of the device under test, making the determination of the current feature value consistent with the usage of the device under test and highly timely, thereby improving the prediction accuracy of the remaining service life of the device under test.

[0102] In some embodiments, reference Figure 3 As shown, based on historical usage data, a feature function is constructed, including:

[0103] S211. Preprocess the historical usage data to obtain a first data set. The first data set includes multiple first data groups, and each first data group includes the target force data of a single operation of the device to be predicted.

[0104] In step S211, preprocessing the historical usage data enhances the variation of different distribution characteristics of the historical usage data, resulting in a preprocessed first data set. The first data set includes multiple first data groups, each of which includes target force data for a single operation of the device to be predicted. For example, the number of first data groups can be the current number of operations of the device to be predicted. Each first data group includes multiple target force data for a single operation of the device to be predicted. The target force data can be, for example, the torque data of the screw when the bit is tightened onto the screw.

[0105] S212. Based on the target statistical characteristics and the first data set, determine the initial characteristic values, degradation parameters, and fluctuation parameters.

[0106] In step S212, initial feature values ​​are determined based on the target statistical features and the first data set. Degeneration parameter v and fluctuation parameter σ, initial eigenvalues The initial values ​​corresponding to the target statistical characteristics are given, and the degradation parameter v and the fluctuation parameter σ are the parameters describing the degradation rate and the fluctuation intensity, respectively.

[0107] S213. Construct the characteristic function based on the initial eigenvalues, degradation parameters, and fluctuation parameters.

[0108] In step S213, based on the initial eigenvalues The degradation parameter v and the fluctuation parameter σ can be used to construct the characteristic function. For example, the feature function can be The formula is determined as follows:

[0109]

[0110] Here, t and τ have the same practical meaning, both representing the number of operations of the device to be predicted, and B(t) is the standard Brownian motion formula. Once the current number of operations t1 of the device to be predicted is determined, it can be substituted into the characteristic function. In the process, determine the current feature value corresponding to the target statistical feature.

[0111] In this embodiment, by preprocessing the historical usage data of the device to be predicted, the fluctuations of different distribution characteristics of the historical usage data can be obtained by merging and enhancing the first dataset. Based on the target statistical features and the first dataset, the initial feature values, degradation parameters, and fluctuation parameters can be determined. A feature function is then constructed based on the initial feature values, degradation parameters, and fluctuation parameters, thus determining the feature function and providing a basis for subsequently determining the current feature value corresponding to the target statistical features. The parameters required to construct the feature function are determined based on the first dataset and the target statistical features, fully considering the external random factors and individual differences of the device during its actual use. This ensures that the determination of the current feature value conforms to the usage of the device and has strong timeliness, thereby improving the prediction accuracy of the remaining service life of the device.

[0112] In some embodiments, reference Figure 4 As shown, based on the current eigenvalues ​​and the failure threshold of the target statistical characteristics, the remaining lifetime of the device to be predicted is predicted, including:

[0113] S310. Based on the current eigenvalues, failure threshold, degradation parameters, and fluctuation parameters, determine the probability density of remaining useful life.

[0114] In step S310, based on the current eigenvalues, failure threshold, degradation parameters, and fluctuation parameters, the probability density of the remaining useful life can be determined according to stochastic process theory. For example, the probability density f of the remaining useful life can be... RUL (τ) is determined by the following formula:

[0115]

[0116] in, Let Γ be the characteristic function of the current eigenvalue, Γ be the failure threshold, and v and σ be the degradation parameter and fluctuation parameter, respectively.

[0117] S320. Determine the remaining useful life based on the probability density of the remaining useful life.

[0118] In step S320, the remaining lifetime of the device to be predicted can be determined based on the remaining lifetime probability density, where the remaining lifetime of the device to be predicted is represented by the number of jobs. For example, the expected value of the remaining lifetime probability density can be calculated, and the expected value can be used as the remaining lifetime of the device to be predicted.

[0119] In this embodiment, the remaining lifetime probability density can be determined based on the current eigenvalues, failure threshold, degradation parameters, and fluctuation parameters. The remaining lifetime of the device under test can be obtained by calculating the expectation of the remaining lifetime probability density, thus achieving device lifetime prediction. The construction of the remaining lifetime probability density, based on the current eigenvalues, failure threshold, degradation parameters, and fluctuation parameters, fully considers external random factors and individual differences in the actual use of the device under test. This makes the prediction results closer to the actual results, reduces prediction bias and maintenance lag rate, improves product yield, and reduces production capacity loss.

[0120] In some embodiments, historical usage data includes multiple second data sets, each second data set including raw force data from a single operation of the device to be predicted, with reference to... Figure 5 As shown, historical usage data is preprocessed to obtain the first dataset, which includes:

[0121] S211-1. Resample each second data group to obtain multiple third data groups, with each third data group containing the same number of data points.

[0122] In step S211-1, each second data group includes the original force data of a single operation of the device to be predicted. For example, each second data group may be the torque data of screws tightened each time by the same batch head. The number of data points in each second data group varies depending on the number of test points tightened each time. Each second data group is resampled. Resampling may include upsampling and downsampling. For example, second data groups with fewer than a preset number of data points may be upsampled, and second data groups with more than a preset number of data points may be downsampled, so that the number of data points in each second data group is the preset number, resulting in multiple third data groups. The preset number may be the average number of data points in each second data group; for example, the preset number is 1800.

[0123] S211-2. Differentiate the data in each third data group to obtain multiple fourth data groups.

[0124] In step S211-2, the data in each third data group are differentially processed, that is, each data in each third data group is subtracted from its previous data to obtain multiple fourth data groups corresponding to each third data group.

[0125] S211-3. Take the absolute value of the data in each fourth data group to obtain the first data set.

[0126] In step S211-3, the absolute value of the data in each fourth data group is taken to obtain the first data group corresponding to each fourth data group. Multiple first data groups constitute the first data set, completing the conversion from the original applied force data to the target applied force data. It can be understood that the above conversion process applies to the data in each second data group without changing the number of data groups. Each first data group, second data group, third data group, and fourth data group corresponds to the data of each operation of the device to be predicted.

[0127] In this embodiment, each second data group is resampled, differencing, and averaged sequentially to obtain the third, fourth, and first data groups, thus determining the first data set and completing the conversion from the original force application data to the target force application data for a single operation. By resampling each second data group, deviations in the subsequent determination of current feature values ​​can be reduced. Through the calculation of differencing and averaging, the target force application data of bits with different wear levels can more easily reflect different distribution characteristics, improving the prediction accuracy of the remaining service life of the device to be predicted.

[0128] In some embodiments, reference Figure 6 As shown, based on the target statistical characteristics and the first dataset, the initial feature values, degradation parameters, and fluctuation parameters are determined, including:

[0129] S212-1. Based on the first data set, determine the feature values ​​of the target statistical features corresponding to each first data group.

[0130] In step S212-1, based on the first data set, the feature values ​​of the target statistical features corresponding to each first data group are determined, forming a feature value sequence with a sequence length equal to the number of first data groups.

[0131] S212-2, Determine the feature value corresponding to the first data group of the first group as the initial feature value.

[0132] In step S212-2, the first feature value in the feature value sequence corresponding to the first data group is determined as the initial feature value. Initial eigenvalues ​​were achieved The determination.

[0133] S212-3. Based on the characteristic values ​​corresponding to each first data group, determine the degradation parameter and fluctuation parameter.

[0134] In step S212-3, the degradation parameter v and fluctuation parameter σ are determined based on the eigenvalues, i.e., the eigenvalue sequences, corresponding to each first data group. For example, the degradation parameter can be estimated using maximum likelihood estimation. and the estimated value of the fluctuation parameter And the estimated value of the degradation parameter and the estimated value of the fluctuation parameter These are respectively used as the degradation parameter v and the fluctuation parameter σ. The estimated values ​​of the degradation parameters can be... The formula is determined as follows:

[0135]

[0136] And the estimated value of the fluctuation parameter The formula is determined as follows:

[0137]

[0138] in, This represents the difference sequence obtained by differencing the eigenvalues ​​(i.e., the eigenvalue sequences) corresponding to each first data group. Its sequence length is N, and Δt represents the acquisition interval of the eigenvalues ​​corresponding to each first data group. Since each first data group corresponds to the target force data of the device to be predicted for each operation, Δt = 1. for any element in, i.e.

[0139] In this embodiment, by determining the feature values ​​of the target statistical features corresponding to each first data group, the feature values ​​corresponding to the first data group can be determined as initial feature values. Furthermore, degradation parameters and fluctuation parameters are determined based on the feature values ​​corresponding to each first data group, thus realizing the determination of initial feature values, degradation parameters, and fluctuation parameters. This provides a basis for subsequent construction of feature functions and determination of current feature values. The initial feature values, degradation parameters, and fluctuation parameters are all determined based on the first data set, fully considering external random factors and individual differences in the actual use of the device under test. This ensures that the determination of current feature values ​​conforms to the usage of the device under test and has strong timeliness, thereby improving the prediction accuracy of the remaining service life of the device under test.

[0140] In some embodiments, reference Figure 7 As shown, before determining the current feature values ​​of the target statistical features of historical usage data, the device lifetime prediction method further includes:

[0141] S400. Obtain sample data, which includes historical usage data of the failed device.

[0142] In step S400, historical usage data of the failed device is acquired. The failed device can be, for example, a device of the same type and model as the device to be predicted, so that the service life of the failed device can reflect the service life of the device to be predicted to a certain extent. The historical usage data of the failed device includes at least a portion of its usage data from when it was put into use until its failure. This historical usage data can be all the usage data of the failed device, or it can be a portion of the data extracted from all the usage data through methods such as interval sampling, and it can characterize the device's operating state. For example, the historical usage data can be force data that characterizes the device's operating state during the use of the failed device, such as torque or pressure, or it can be other high-frequency data that reflects the device's operating state during the use of the device to be predicted, such as vibration, current, or sound. The type of historical usage data of the failed device is the same as the type of historical usage data of the device to be predicted.

[0143] S500. Based on sample data, determine the target statistical characteristics and the failure threshold of the target statistical characteristics.

[0144] In step S500, the target statistical characteristics and failure thresholds of the target statistical characteristics suitable for lifetime prediction can be determined based on the sample data, providing a basis for subsequently determining the current characteristic value of the target statistical characteristics and the failure thresholds based on the current characteristic value and the target statistical characteristics, and predicting the remaining lifetime of the device to be predicted.

[0145] In this embodiment, by acquiring sample data including historical usage data of failed devices, target statistical characteristics and failure thresholds for these characteristics can be determined based on the sample data. This provides definite target statistical characteristics and failure thresholds for subsequent lifetime prediction. The target statistical characteristics and failure thresholds determined based on the sample data of failed devices enable the target statistical characteristics to be applicable to the prediction of the remaining lifetime of the device to be predicted, and ensure that the failure thresholds characterize the threshold values ​​of the target statistical characteristics corresponding to the failure of the same device. This results in prediction results that are closer to the actual failure situation of the same device, thus making the prediction results closer to the true results, reducing prediction bias and maintenance lag rates, improving product yield, and reducing production capacity loss.

[0146] In some embodiments, reference Figure 8 As shown, based on sample data, the target statistical features and their failure thresholds are determined, including:

[0147] S510. Preprocess the sample data to obtain a second data set. The second data set includes multiple fifth data groups, and each fifth data group includes the target force data of a single operation of the failed device.

[0148] In step S510, the sample data is preprocessed to enhance the fluctuations of different distribution characteristics of the sample data, resulting in a preprocessed second data set. The second data set includes multiple fifth data groups, each of which includes target force data for a single operation of the failed device. For example, the number of fifth data groups may be the total number of operations performed on the failed device. Each fifth data group includes multiple target force data for a single operation of the failed device, such as the torque data of the screw when the screw is tightened by a screwdriver bit.

[0149] S520. Based on the second data set, determine the characteristic values ​​of multiple statistical features corresponding to each fifth data group.

[0150] In step S520, based on the second data set, the feature values ​​of multiple statistical features corresponding to each fifth data group are determined, forming a sequence of feature values ​​for multiple different statistical features, each with a sequence length equal to the number of fifth data groups. Each statistical feature can characterize a data calculation method and reflect certain data characteristics. For example, feature values ​​such as mean, standard deviation, and root mean square can be determined for each fifth data group.

[0151] S530. Based on the feature values ​​of multiple statistical features corresponding to each fifth data group, determine the target statistical feature.

[0152] In step S530, based on the feature values ​​of multiple statistical features corresponding to each fifth data group, i.e., the feature value sequence of multiple different statistical features, the target statistical feature can be extracted from multiple statistical features, thus realizing the determination of the target statistical feature.

[0153] S540. Determine the failure threshold based on the target statistical characteristics.

[0154] In step S540, based on the target statistical characteristics, the characteristic values ​​of the target statistical characteristics corresponding to each fifth data group can be determined, and the characteristic value of the target statistical characteristics corresponding to the last fifth data group, i.e. the last operation before failure, is determined as the failure threshold of the target statistical characteristics, thus realizing the determination of the failure threshold.

[0155] In this embodiment, by preprocessing the historical usage data of failed devices, the fluctuations of different distribution characteristics of the second dataset's merged and enhanced sample data can be obtained. Based on the second dataset, the feature values ​​of multiple statistical features corresponding to each fifth data group can be determined. Furthermore, based on the feature values ​​of these multiple statistical features corresponding to each fifth data group, the target statistical features and failure thresholds are determined, thus realizing the extraction of target statistical features and the determination of failure thresholds. The target statistical features extracted from multiple statistical features are suitable for predicting the remaining service life of the devices to be predicted, and the failure threshold can characterize the threshold value of the target statistical feature corresponding to the failure of the same device. The prediction results are closer to the actual failure situation of the same device, thereby making the prediction results closer to the true results, reducing prediction bias and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0156] In some embodiments, reference Figure 9 As shown, the sample data includes multiple sixth data groups, each containing the original force data of a single operation of the failed device. Preprocessing the sample data yields a second data set, including:

[0157] S511. Resample each sixth data group to obtain multiple seventh data groups, with each seventh data group containing the same number of data points.

[0158] In step S511, each sixth data group includes the original force data of a single operation of the failed device. For example, each sixth data group may be the torque data of the screws tightened each time by the same batch head. The number of data points in each sixth data group varies depending on the test point used each time the screws are tightened. Each sixth data group is resampled. Resampling may include upsampling and downsampling. For example, sixth data groups with fewer than a preset number of data points may be upsampled, while sixth data groups with more than a preset number of data points may be downsampled, so that the number of data points in each sixth data group is the preset number, resulting in multiple seventh data groups. The preset number may be, for example, the average number of data points in each sixth data group.

[0159] S512. Differentiate the data in each seventh data group to obtain multiple eighth data groups.

[0160] In step S512, the data in each seventh data group are differentially processed, that is, each data in each seventh data group is subtracted from its previous data to obtain multiple eighth data groups corresponding to each seventh data group.

[0161] S513. Take the absolute value of the data in each of the eighth data groups to obtain the second data set.

[0162] In step S513, the absolute value of the data in each eighth data group is taken to obtain the fifth data group corresponding to each eighth data group. Multiple fifth data groups constitute the second data set, completing the conversion from the original force application data to the target force application data. It can be understood that the above conversion process applies to the data in each sixth data group without changing the number of data groups. Each fifth, sixth, seventh, and eighth data group corresponds to the data of each operation of the failed device.

[0163] In this embodiment, each of the sixth data groups was resampled, differencing, and averaging sequentially to obtain the seventh, eighth, and fifth data groups, thus determining the second data set and completing the conversion from the original force application data to the target force application data for a single operation. By resampling each of the sixth data groups, deviations in the subsequent target statistical feature determination process can be reduced. Through the calculation of differencing and averaging, the target force application data of bits with different wear levels can more easily reflect different distribution characteristics, improving the prediction accuracy of the remaining service life of the device to be predicted.

[0164] In some embodiments, reference Figure 10 As shown, based on the feature values ​​of multiple statistical features corresponding to each fifth data group, the target statistical features are determined, including:

[0165] S531. Based on the feature values ​​of multiple statistical features corresponding to each fifth data group, determine the multiple evaluation index values ​​corresponding to each statistical feature.

[0166] In step S531, based on the feature values ​​of multiple statistical features corresponding to each fifth data group, multiple evaluation index values ​​corresponding to each statistical feature are determined. Each evaluation index value is used to characterize whether each statistical feature is applicable to the prediction of the remaining service life of the device to be predicted.

[0167] S532. Determine the target statistical characteristics based on multiple evaluation index values ​​corresponding to each statistical characteristic.

[0168] In step S532, based on the multiple evaluation index values ​​corresponding to each statistical feature, the target statistical feature that is most suitable for predicting the remaining service life of the device to be predicted is extracted from the multiple statistical features.

[0169] In this embodiment, based on the feature values ​​of multiple statistical features corresponding to each fifth data group, multiple evaluation index values ​​are determined for each statistical feature. These evaluation index values ​​characterize whether each statistical feature is applicable to predicting the remaining useful life of the device under test. Furthermore, a target statistical feature is extracted from the multiple statistical features based on the evaluation index values, thus determining the target statistical feature and providing a basis for determining the current feature value. The target statistical feature extracted from the multiple statistical features is applicable to predicting the remaining useful life of the device under test, and the prediction result is closer to the actual failure situation of the same device. This makes the prediction result closer to the true result, reducing prediction bias and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0170] In some embodiments, the multiple evaluation index values ​​include trend index values, monotonicity index values, and robustness index values, which are used to evaluate the trend, monotonicity, and robustness of the characteristic values ​​of each statistical feature, respectively. For example, the trend index value Tred(H) can be determined by the following formula:

[0171]

[0172] The monotonicity index value Mon(H) is determined by the following formula:

[0173]

[0174] The robustness index value Rob(H) is determined by the following formula:

[0175]

[0176] Where K is the length of any statistical feature H, that is, the number of feature values ​​of that statistical feature corresponding to each fifth data group, h k and t k These are the k-th eigenvalue and its sequential number among the eigenvalues ​​of this statistical feature. and These represent the median of the eigenvalues ​​and the median number of eigenvalues ​​for this statistical feature, respectively; dH is the difference sequence of the eigenvalues ​​for this statistical feature; No.of represents the counting operator; H... R This represents the smoothed residual after smoothing the eigenvalues ​​of the statistical feature. It is understandable that different statistical features have corresponding trend index values ​​(Tred(H), monotonicity index values ​​(Mon(H), and robustness index values ​​(Rob(H))).

[0177] refer to Figure 11 As shown, the target statistical characteristics are determined based on multiple evaluation index values ​​corresponding to each statistical characteristic, including:

[0178] S532-1. Based on the trend index value, monotonicity index value, and robustness index value corresponding to each statistical characteristic, determine the mean of the trend index, the mean of the monotonicity index, and the mean of the robustness index.

[0179] In step S532-1, the average values ​​of the trend index, monotonicity index, and robustness index corresponding to each statistical feature are respectively taken as the trend index mean, monotonicity index mean, and robustness index mean. For example, if there are 5 statistical features, namely H1, H2, H3, H4, and H5, then there are also 5 corresponding trend indices, namely Trend(H1), Trend(H2), Trend(H3), Trend(H4), and Trend(H5). The average value of the 5 trend indices is taken and used as the trend index mean. Similarly, the monotonicity index mean and robustness index mean can be obtained.

[0180] S532-2. Based on the trend index value, monotonicity index value and robustness index value corresponding to each statistical characteristic, determine the total value of the evaluation index corresponding to each statistical characteristic. The total value of the evaluation index is the sum of the trend index value, monotonicity index value and robustness index value corresponding to each statistical characteristic.

[0181] In step S532-2, the trend index value, monotonicity index value and robustness index value corresponding to each statistical feature are summed to obtain the total evaluation index value corresponding to each statistical feature.

[0182] S532-3. The statistical characteristic that has the largest total value of the evaluation index and is greater than the mean of the trend index, the mean of the monotonic index, and the mean of the robust index, respectively, is determined as the target statistical characteristic.

[0183] In step S523-3, if the trend indicator value corresponding to any of the multiple statistical features is less than the mean of the trend indicator, or the corresponding monotonicity indicator value is less than the mean of the monotonicity indicator, or the corresponding robustness indicator value is less than the mean of the robustness indicator, then that statistical feature is excluded. The statistical feature representing the total value of the evaluation indicators among the remaining multiple statistical features after exclusion is then determined as the target statistical feature.

[0184] By extracting target statistical features from multiple different statistical characteristics using the method described above, the extracted target statistical features exhibit stronger trends, better monotonicity, and less fluctuation compared to other statistical features, making them more suitable for predicting the remaining lifespan of the device to be predicted. For example... Figure 12As shown, the predicted value of remaining service life determined by the above method is compared with the actual value. The horizontal axis of the figure is the current number of operations, and the vertical axis is the number of operations with remaining service life. The predicted value of remaining service life is very close to the actual value, which can reduce prediction deviation and maintenance delay rate, improve product yield, and reduce production capacity loss.

[0185] In this embodiment, the target statistical feature is determined by defining the trend index value, monotonicity index value, and robustness index value as being greater than the mean of the trend index value, the mean of the monotonicity index value, and the mean of the robustness index value, respectively, and having the largest total value of the evaluation index. This achieves the extraction of the target statistical feature from multiple statistical features, making the target statistical feature more suitable for predicting the remaining service life of the device to be predicted. The prediction result is closer to the actual failure situation of the same device, thereby making the prediction result closer to the true result, reducing prediction deviation and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0186] In some embodiments, statistical features may include any combination of multiple time-domain features, or statistical features may include any combination of multiple frequency-domain features, or statistical features may include any combination of multiple time-domain features and multiple frequency-domain features.

[0187] Time-domain features may include, for example, the mean, standard deviation, and root mean square (RMS), while frequency-domain features may include, for example, spectral centroid and spectral kurtosis. Standard deviation σ x The calculation formula is as follows:

[0188]

[0189] Root mean square x rms The calculation formula is as follows:

[0190]

[0191] The formula for calculating the spectral centroid f1 is as follows:

[0192]

[0193] The formula for calculating spectral kurtosis f2 is as follows:

[0194]

[0195] Where x(n) is the value of the nth data in each fifth data group, s(l) is the frequency domain data after Fourier transform of the absolute value sequence data, and l∈{1,…,L}, L is the sequence length after Fourier transform, and p(l) is the frequency value corresponding to s(l).

[0196] In this embodiment, by setting multiple statistical features, target statistical features can be extracted from a sufficient number of statistical features. This allows for the analysis of device operation data from multiple different dimensions, making the extracted target statistical features more suitable for predicting the remaining service life of the device to be predicted. The prediction results are closer to the actual failure situation of the same device, thus making the prediction results closer to the real results, reducing prediction deviation and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0197] In some embodiments, the device to be predicted includes a screwdriver bit for a screw-locking device.

[0198] The device to be predicted can be, for example, the bit of a screw-fastening device such as an automatic screw-fastening machine. The bit is used to apply the torque output by the motor of the screw-fastening device to the screw to complete the screw fastening operation. When the device to be predicted is the bit of a screw-fastening device, each first data group includes the real-time torque data of the screw when the bit performs a single screw fastening operation.

[0199] In this embodiment, the device to be predicted includes the screwdriver bit of the screw fastening device. The remaining service life of the screwdriver bit can be predicted by the above-mentioned device life prediction method, thereby guiding maintenance personnel to arrange maintenance work in advance, reducing the frequency of poor fastening such as stripping, improving the stability of screw fastening of the product, and thus improving the product yield.

[0200] In one exemplary embodiment, reference Figure 13 As shown, a device lifetime prediction method is provided, which includes:

[0201] S1. Obtain sample data;

[0202] S2. Resample each sixth data group to obtain multiple seventh data groups;

[0203] S3. Differentiate the data in each seventh data group to obtain multiple eighth data groups;

[0204] S4. Take the absolute value of the data in each of the eighth data groups to obtain the second data set;

[0205] S5. Based on the second data set, determine the feature values ​​of multiple statistical features corresponding to each fifth data group;

[0206] S6. Based on the feature values ​​of multiple statistical features corresponding to each fifth data group, determine the multiple evaluation index values ​​corresponding to each statistical feature;

[0207] S7. Based on the trend index value, monotonicity index value, and robustness index value corresponding to each statistical characteristic, determine the mean of the trend index, the mean of the monotonicity index, and the mean of the robustness index.

[0208] S8. Based on the trend index value, monotonic index value and robust index value corresponding to each statistical characteristic, determine the total value of the evaluation index corresponding to each statistical characteristic;

[0209] S9. The statistical characteristic with the largest total value of the evaluation index and the largest value of the trend index, monotonicity index, and robustness index, respectively, is determined as the target statistical characteristic.

[0210] S10. Determine the failure threshold based on the target statistical characteristics;

[0211] S11. Obtain historical usage data of the device to be predicted;

[0212] S12. Resample each second data set to obtain multiple third data sets;

[0213] S13. Difference the data in each third data group to obtain multiple fourth data groups;

[0214] S14. Take the absolute value of the data in each fourth data group to obtain the first data set;

[0215] S15. Based on the first data set, determine the feature values ​​of the target statistical features corresponding to each first data group;

[0216] S16. Determine the feature value corresponding to the first data group of the first group as the initial feature value;

[0217] S17. Determine the degradation parameter and fluctuation parameter based on the characteristic values ​​corresponding to each first data group;

[0218] S18. Construct the characteristic function based on the initial eigenvalues, degradation parameters, and fluctuation parameters;

[0219] S19. Based on the characteristic function and the current number of operations of the device to be predicted, determine the current characteristic value corresponding to the target statistical characteristic;

[0220] S20. Based on the current eigenvalues, failure thresholds, degradation parameters, and fluctuation parameters, determine the probability density of remaining useful life;

[0221] S21. Determine the remaining useful life based on the probability density of the remaining useful life.

[0222] In this embodiment, historical usage data of the device to be predicted is acquired, and the current feature value corresponding to the target statistical feature is determined based on the historical usage data and the target statistical feature. Then, the remaining service life of the device to be predicted is predicted based on the current feature value and the failure threshold of the target statistical feature, thus realizing the prediction of device life. The life prediction of each device to be predicted is based on its own historical usage data, and the sensitivity of the target statistical feature to the degree of device degradation makes it more suitable for predicting the remaining service life. It fully considers the external random factors and individual differences of the device to be predicted in the actual use process, making the prediction results closer to the actual results, reducing prediction deviation and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0223] In one exemplary embodiment, reference Figure 14 As shown, a device lifetime prediction apparatus is provided. The device lifetime prediction apparatus includes a first acquisition module 10, a first determination module 20, and a second determination module 30. The first acquisition module 10 is used to acquire historical usage data of the device to be predicted, whereby the historical usage data is historical data characterizing the operating status of the device to be predicted. The first determination module 20 is used to determine the current feature value of the target statistical feature of the historical usage data. The second determination module 30 is used to predict the remaining lifetime of the device to be predicted based on the current feature value and the failure threshold of the target statistical feature.

[0224] In this embodiment, the historical usage data of the device to be predicted is acquired by the first acquisition module 10, and the current feature value of the target statistical feature of the historical usage data is determined by the first determination module 20. Then, the remaining service life of the device to be predicted is predicted by the second determination module 30 based on the current feature value and the failure threshold of the target statistical feature, thus realizing the prediction of device life. The life prediction of each device to be predicted is based on its own historical usage data, and the sensitivity of the target statistical feature to the degree of device degradation makes it more suitable for predicting the remaining service life. It fully considers the external random factors and individual differences of the device to be predicted in the actual use process, so that the prediction results are closer to the actual results, reducing prediction deviation and maintenance lag rate, improving product yield, and reducing production capacity loss.

[0225] In one embodiment, the first determining module 20 is further configured to: construct a feature function based on historical usage data, wherein the feature function is used to characterize the relationship between the number of operations of the device to be predicted and the feature value of the target statistical feature; and determine the current feature value corresponding to the target statistical feature based on the feature function and the current number of operations of the device to be predicted.

[0226] In one embodiment, the first determining module 20 is further configured to: preprocess historical usage data to obtain a first data set, the first data set including multiple first data groups, each first data group including target force data of a single operation of the device to be predicted; determine initial feature values, degradation parameters and fluctuation parameters based on target statistical features and the first data set; and construct a feature function based on the initial feature values, degradation parameters and fluctuation parameters.

[0227] In one embodiment, the first determining module 20 is further configured to: resample each second data group to obtain multiple third data groups, wherein the number of data in each third data group is the same; differentiate the data in each third data group to obtain multiple fourth data groups; and take the absolute value of the data in each fourth data group to obtain a first data set.

[0228] In one embodiment, the first determining module 20 is further configured to: determine the feature value of the target statistical feature corresponding to each first data group based on the first data set; determine the feature value corresponding to the first group of first data groups as the initial feature value; and determine the degradation parameter and fluctuation parameter based on the feature value corresponding to each first data group.

[0229] In one embodiment, the second determining module 30 is further configured to: determine the remaining useful life probability density based on the current feature value, failure threshold, degradation parameter and fluctuation parameter; and determine the remaining useful life based on the remaining useful life probability density.

[0230] In some embodiments, reference Figure 15 As shown, the device lifetime prediction device also includes a second acquisition module 40 and a third determination module 50. The second acquisition module 40 is used to acquire sample data, which includes historical usage data of failed devices. The third determination module 50 is used to determine target statistical characteristics and failure thresholds of target statistical characteristics based on the sample data.

[0231] In this embodiment, the second acquisition module 40 acquires sample data including historical usage data of failed devices. The third determination module 50 then determines the target statistical characteristics and failure thresholds of these characteristics based on the sample data, providing definite target statistical characteristics and failure thresholds for subsequent lifetime prediction. The target statistical characteristics and failure thresholds determined based on the sample data of failed devices enable the target statistical characteristics to be applicable to the prediction of the remaining lifetime of the device to be predicted, and the failure thresholds to characterize the threshold values ​​of the target statistical characteristics corresponding to the failure of the same device. This results in prediction results that are closer to the actual failure situation of the same device, thus making the prediction results closer to the true results, reducing prediction bias and maintenance lag rates, improving product yield, and reducing production capacity loss.

[0232] In one embodiment, the third determining module 50 is further configured to: preprocess the sample data to obtain a second data set, the second data set including multiple fifth data groups, each fifth data group including target force data of a single operation of the failed device; determine the feature values ​​of multiple statistical features corresponding to each fifth data group based on the second data set; determine the target statistical feature based on the feature values ​​of multiple statistical features corresponding to each fifth data group; and determine the failure threshold based on the target statistical feature.

[0233] In one embodiment, the third determining module 50 is further configured to: resample each sixth data group to obtain multiple seventh data groups, wherein the number of data in each seventh data group is the same; differentiate the data in each seventh data group to obtain multiple eighth data groups; and take the absolute value of the data in each eighth data group to obtain a second data set.

[0234] In one embodiment, the third determining module 50 is further configured to: determine multiple evaluation index values ​​corresponding to each statistical feature based on the feature values ​​of multiple statistical features corresponding to each fifth data group; and determine the target statistical feature based on the multiple evaluation index values ​​corresponding to each statistical feature.

[0235] In one embodiment, the third determining module 50 is further configured to: determine the mean of the trend indicator, the mean of the monotonic indicator, and the mean of the robust indicator based on the trend indicator value, the monotonic indicator value, and the robust indicator value corresponding to each statistical feature; determine the total value of the evaluation indicators corresponding to each statistical feature based on the trend indicator value, the monotonic indicator value, and the robust indicator value corresponding to each statistical feature, wherein the total value of the evaluation indicators is the sum of the trend indicator value, the monotonic indicator value, and the robust indicator value corresponding to each statistical feature; and determine the statistical feature whose trend indicator value, the monotonic indicator value, and the robust indicator value are respectively greater than the mean of the trend indicator, the mean of the monotonic indicator, and the mean of the robust indicator, and whose total value of the evaluation indicators is the largest, as the target statistical feature.

[0236] In one exemplary embodiment, a device lifetime prediction apparatus is provided, with reference to Figure 16 As shown, the device lifetime prediction device may include one or more of the following components: processing component 101, memory 102, power component 103, multimedia component 104, audio component 105, input / output (I / O) interface 106, sensor component 107, and communication component 108.

[0237] Processing component 101 typically controls the overall operation of the device lifetime prediction device, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 101 may include one or more processors 109 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 101 may include one or more modules to facilitate interaction between processing component 101 and other components. For example, processing component 101 may include a multimedia module to facilitate interaction between multimedia component 104 and processing component 101.

[0238] Memory 102 is configured to store various types of data to support the operation of the device lifetime prediction device. Examples of such data include instructions for any application or method operating on the device lifetime prediction device, contact data, phone book data, messages, pictures, videos, etc. Memory 102 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0239] Power assembly 103 provides power to various components of the device lifetime prediction device. Power assembly 103 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device lifetime prediction device.

[0240] Multimedia component 104 includes a screen that provides an output interface between the device lifetime prediction device and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 104 includes a front-facing camera and / or a rear-facing camera. When the device lifetime prediction device is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0241] Audio component 105 is configured to output and / or input audio signals. For example, audio component 105 includes a microphone (MIC) configured to receive external audio signals when the device lifetime prediction device is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 102 or transmitted via communication component 108. In some embodiments, audio component 105 also includes a speaker for outputting audio signals.

[0242] I / O interface 106 provides an interface between processing component 101 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0243] Sensor assembly 107 includes one or more sensors for providing a state assessment of various aspects of the device lifetime prediction device. For example, sensor assembly 107 may detect the on / off state of the device lifetime prediction device, the relative positioning of components such as the device's display and keypad, changes in the position of the device lifetime prediction device or a component thereof, the presence or absence of user contact with the device, the device's orientation or acceleration / deceleration, and temperature changes. Sensor assembly 107 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 107 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 107 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0244] Communication component 108 is configured to facilitate wired or wireless communication between the device lifetime prediction device and other devices. The device can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 108 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 108 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0245] In an exemplary embodiment, the device lifetime prediction device may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the device lifetime prediction method described above applied to the device lifetime prediction device.

[0246] In one exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 102 including instructions, which can be executed by a processor 109 of the device lifetime prediction device to complete the device lifetime prediction method applied to the device lifetime prediction device described above. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the instructions in the storage medium are executed by the processor of the device lifetime prediction device, the device lifetime prediction device is able to perform the device lifetime prediction method shown in the above embodiments.

[0247] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0248] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A device lifetime prediction method, characterized by, The device life prediction method comprises: obtaining historical use data of a device to be predicted, the historical use data being historical data representing a device operation state of the device to be predicted; obtaining sample data, the sample data comprising historical use data of a failed device, the historical use data of the failed device being at least part of use data of the failed device from when the failed device was put into use to a failure time, the historical use data of the failed device comprising force data and / or high-frequency data capable of representing a device operation state, the force data comprising at least one of torque data and pressure data, and the high-frequency data comprising at least one of vibration data, current data, and sound data; preprocessing the sample data to obtain a plurality of data groups, and determining feature values of a plurality of statistical features corresponding to each of the data groups, each of the data groups comprising target force data of a single operation of the failed device; determining a target statistical feature based on the feature values of the plurality of statistical features corresponding to each of the data groups; determining a failure threshold of the target statistical feature based on the target statistical feature; determining a current feature value of the target statistical feature of the historical use data of the device to be predicted; predicting a remaining use life of the device to be predicted based on the current feature value and the failure threshold; wherein the device to be predicted comprises a driver bit, the driver bit being used to lock a screw, the current feature value being determined based on a current operation number of the driver bit, and the current operation number of the driver bit being a number of times of locking a screw by the driver bit.

2. The device lifetime prediction method according to claim 1, characterized by, The determining of the current feature value of the target statistical feature of the historical use data of the device to be predicted comprises: constructing a feature function based on the historical use data, the feature function being used to represent a relationship between an operation number of the device to be predicted and a feature value of the target statistical feature; determining a current feature value corresponding to the target statistical feature based on the feature function and a current operation number of the device to be predicted.

3. The device lifetime prediction method according to claim 2, characterized by, The constructing of the feature function based on the historical use data comprises: preprocessing the historical use data to obtain a first data set, the first data set comprising a plurality of first data groups, each of the first data groups comprising target force data of a single operation of the device to be predicted; determining an initial feature value, a degradation parameter, and a fluctuation parameter based on the target statistical feature and the first data set; constructing the feature function based on the initial feature value, the degradation parameter, and the fluctuation parameter.

4. The device lifetime prediction method according to claim 3, characterized by, The predicting of the remaining use life of the device to be predicted based on the current feature value and the failure threshold comprises: determining a remaining use life probability density based on the current feature value, the failure threshold, the degradation parameter, and the fluctuation parameter; determining the remaining use life based on the remaining use life probability density.

5. The device lifetime prediction method according to claim 3, characterized by, The historical use data comprises a plurality of second data groups, each of the second data groups comprising original force data of a single operation of the device to be predicted, and the preprocessing of the historical use data to obtain a first data set comprises: Resample each of the second data sets to obtain a plurality of third data sets, the number of data in each of the third data sets being the same; Differencing the data in each of the third data sets to obtain a plurality of fourth data sets; Taking absolute values of the data in each of the fourth data sets to obtain the first data set.

6. The device lifetime prediction method according to claim 3, characterized by, The initial characteristic value, the degradation parameter and the fluctuation parameter are determined based on the target statistical characteristic and the first data set, comprising: Determine the characteristic value of the target statistical characteristic corresponding to each of the first data sets based on the first data set; Determine the initial characteristic value as the characteristic value corresponding to the first group of first data sets; Determine the degradation parameter and the fluctuation parameter based on the characteristic value corresponding to each of the first data sets.

7. The device lifetime prediction method according to any one of claims 1 to 6, characterized by, The sample data is preprocessed to obtain a plurality of data sets, and the characteristic value of a plurality of statistical characteristics corresponding to each of the data sets is determined, comprising: The sample data is preprocessed to obtain a second data set, the second data set comprising a plurality of fifth data sets, each of the fifth data sets comprising target force data of a single operation of the failed device; based on the second data set, the characteristic value of a plurality of statistical characteristics corresponding to each of the fifth data sets is determined.

8. The device lifetime prediction method according to claim 7, characterized by, The sample data comprises a plurality of sixth data sets, each of the sixth data sets comprising original force data of a single operation of the failed device, and the preprocessing of the sample data to obtain a second data set comprises: Resample each of the sixth data sets to obtain a plurality of seventh data sets, the number of data in each of the seventh data sets being the same; Differencing the data in each of the seventh data sets to obtain a plurality of eighth data sets; Taking absolute values of the data in each of the eighth data sets to obtain the second data set.

9. The device lifetime prediction method according to claim 7, characterized by, The target statistical characteristic is determined based on the characteristic value of a plurality of statistical characteristics corresponding to each of the data sets, comprising: Determine a plurality of evaluation index values corresponding to each of the statistical characteristics based on the characteristic value of a plurality of statistical characteristics corresponding to each of the fifth data sets; Determine the target statistical characteristic based on a plurality of evaluation index values corresponding to each of the statistical characteristics.

10. The device lifetime prediction method according to claim 9, characterized by, The plurality of evaluation index values comprises a trend index value, a monotonicity index value and a robustness index value, and the target statistical characteristic is determined based on a plurality of evaluation index values corresponding to each of the statistical characteristics, comprising: Determine a trend index mean value, a monotonicity index mean value and a robustness index mean value based on the trend index value, the monotonicity index value and the robustness index value corresponding to each of the statistical characteristics; Determine an evaluation index total value corresponding to each of the statistical characteristics based on the trend index value, the monotonicity index value and the robustness index value corresponding to each of the statistical characteristics, the evaluation index total value being the sum of the trend index value, the monotonicity index value and the robustness index value corresponding to each of the statistical characteristics; The statistical feature with the maximum evaluation index total value and greater trend index value, monotonicity index value and robustness index value than the trend index mean value, monotonicity index mean value and robustness index mean value is determined as the target statistical feature.

11. The device lifetime prediction method according to claim 7, wherein The plurality of statistical features include a combination of any of a plurality of time domain features and / or a plurality of frequency domain features.

12. A device lifetime prediction apparatus characterized by comprising: The device life prediction apparatus includes: A first obtaining module, configured to obtain historical usage data of a device to be predicted, the historical usage data being historical data representing a device operation state of the device to be predicted; A first determining module, configured to determine a current feature value of a target statistical feature of the historical usage data of the device to be predicted; A second determining module, configured to predict a remaining service life of the device to be predicted based on the current feature value and a failure threshold of the target statistical feature; A second obtaining module, configured to obtain sample data, the sample data including historical usage data of a failed device, the historical usage data of the failed device being at least part of usage data of the failed device from a time when the failed device is put into use to a time when the failed device fails, the historical usage data of the failed device including force data and / or high-frequency data that can represent a device operation state, the force data including at least one of torque data and pressure data, and the high-frequency data including at least one of vibration data, current data and sound data; A third determining module, configured to pre-process the sample data to obtain a plurality of data groups, and determine feature values of a plurality of statistical features corresponding to each of the data groups, each of the data groups including target force data of a single operation of the failed device; determine the target statistical feature based on the feature values of the plurality of statistical features corresponding to each of the data groups; and determine the failure threshold of the target statistical feature based on the target statistical feature; The device to be predicted includes a batch head, the batch head is used to lock a screw, the current feature value is determined based on a current operation number of the batch head, and the current operation number of the batch head is a number of times of locking the screw by the batch head.

13. A device lifetime prediction apparatus characterized by comprising: The device life prediction apparatus includes: A processor; A memory for storing processor-executable instructions; The processor is configured to: obtain historical usage data of a device to be predicted, the historical usage data being historical data representing a device operation state of the device to be predicted; obtain sample data, the sample data including historical usage data of a failed device, the historical usage data of the failed device being at least part of usage data of the failed device from a time when the failed device is put into use to a time when the failed device fails, the historical usage data of the failed device including force data and / or high-frequency data that can represent a device operation state, the force data including at least one of torque data and pressure data, and the high-frequency data including at least one of vibration data, current data and sound data; Preprocess the sample data to obtain a plurality of data groups, and determine the characteristic values of a plurality of statistical characteristics corresponding to each data group, each data group comprising target force data of a single operation of the failed device; Determine a target statistical characteristic based on the characteristic values of the plurality of statistical characteristics corresponding to each data group; Determine the failure threshold of the target statistical characteristic based on the target statistical characteristic; Determine the current characteristic value of the target statistical characteristic of the historical use data of the device to be predicted; Predict the remaining service life of the device to be predicted based on the current characteristic value and the failure threshold; Wherein the device to be predicted comprises a batch head, the batch head is used to lock the screw, the current characteristic value is determined based on the current operation times of the batch head, and the current operation times of the batch head is the number of times of locking the screw.

14. A non-transitory computer-readable storage medium, comprising: When the instructions in the storage medium are executed by the processor of the device life prediction equipment, the device life prediction equipment can perform a device life prediction method, the device life prediction method comprises: Obtain the historical use data of the device to be predicted, which is the historical data representing the device operation state of the device to be predicted; Obtain sample data, which includes the historical use data of the failed device, the historical use data of the failed device is at least part of the use data from the time when the failed device is put into use to the time when it fails, and the historical use data of the failed device includes force data and / or high frequency data that can represent the device operation state, the force data includes at least one of torque data and pressure data, and the high frequency data includes at least one of vibration data, current data and sound data; Preprocess the sample data to obtain a plurality of data groups, and determine the characteristic values of a plurality of statistical characteristics corresponding to each data group, each data group comprising target force data of a single operation of the failed device; Determine a target statistical characteristic based on the characteristic values of the plurality of statistical characteristics corresponding to each data group; Determine the failure threshold of the target statistical characteristic based on the target statistical characteristic; Determine the current characteristic value of the target statistical characteristic of the historical use data of the device to be predicted; Predict the remaining service life of the device to be predicted based on the current characteristic value and the failure threshold; Wherein the device to be predicted comprises a batch head, the batch head is used to lock the screw, the current characteristic value is determined based on the current operation times of the batch head, and the current operation times of the batch head is the number of times of locking the screw.

Images