Temperature prediction model training method, prediction method, device, equipment and medium

Abstract

The embodiment of the present invention discloses a training method, prediction method, device, equipment and medium of a temperature prediction model, wherein the training method includes: collecting temperature data of electronic devices and electrical characteristic data of electronic devices at different temperatures; segmenting the temperature data and the corresponding electrical characteristic data according to a preset time interval; for the temperature data and electrical characteristic data of each time period after segmentation, data collection is performed according to a preset window sliding step, and a feature matrix is ​​formed using all the collected temperature data and electrical characteristic data; the temperature data and electrical characteristic data in the first time period and the electrical characteristic data in the second time period are selected from the feature matrix as basic data, and the temperature data in the second time period is used as the data to be predicted; the temperature prediction model is trained using the basic data and the data to be predicted. By adopting the above technical solution, the problems of high difficulty in data analysis and poor temperature prediction accuracy are effectively solved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of predictive maintenance of electronic devices, and in particular, to a training method, prediction method, device, equipment and medium for a temperature prediction model. Background Art

[0002] With the innovation and development of social economy, electronic equipment (such as motors, etc.) has become more efficient and intelligent, but the maintenance of the equipment itself faces huge challenges. Traditional maintenance methods will reduce productivity, and unplanned production interruptions will cause some large manufacturers to lose a lot every year. However, due to the lack of understanding of maintenance methods in many companies, they still use traditional equipment maintenance methods in equipment maintenance, one is the post-failure maintenance method, and the other is a preventive maintenance method of periodic downtime maintenance.

[0003] Among them, predictive maintenance is a higher-level maintenance method developed from preventive maintenance. It is based on the equipment status and the prediction of the development trend of the equipment status. After information collection, processing and comprehensive analysis, the maintenance cycle and maintenance items are purposefully arranged, and "repairs are made when necessary and repairs must be made well". Compared with traditional maintenance methods, it has obvious advantages.

[0004] Traditional methods for electronic equipment temperature monitoring usually use two methods: one is physical modeling, and the other is data statistical analysis.

[0005] However, the common problem with physical modeling is that the principles are complex and cannot be accurately modeled. Although the statistical analysis method does not need to analyze the specific principles of electronic equipment, it still cannot obtain universal and accurate results. Summary of the invention

[0006] The embodiments of the present invention provide a training method, prediction method, device, equipment and medium for a temperature prediction model, which effectively solve the problems of difficult data analysis and insufficient prediction accuracy, and improve the accuracy of temperature prediction of electronic devices.

[0007] In a first aspect, the present invention provides a method for training a temperature prediction model, the method comprising:

[0008] Collecting temperature data of the electronic device and electrical characteristic data of the electronic device at different temperatures; wherein the temperature data and each electrical characteristic data are sampled at the same frequency;

[0009] Segmenting the temperature data and the corresponding electrical characteristic data according to preset time intervals;

[0010] For the temperature data and electrical characteristic data of each time period after segmentation, data extraction is performed according to the preset window sliding step size, and a characteristic matrix is ​​formed using all the extracted temperature data and electrical characteristic data;

[0011] Selecting temperature data and electrical characteristic data in a first time period and electrical characteristic data in a second time period from the characteristic matrix as basic data, and using temperature data in the second time period as data to be predicted;

[0012] Using the basic data and the data to be predicted as a training sample set to train a temperature prediction model, so that the temperature prediction model establishes an association relationship between temperature data and electrical characteristic data in different time periods;

[0013] The first time period and the second time period are arranged in chronological order.

[0014] Optionally, for the temperature data and electrical characteristic data of each time period after segmentation, data extraction is performed according to a preset window sliding step, and a characteristic matrix is ​​formed using all the extracted temperature data and electrical characteristic data, including:

[0015] For the temperature data and electrical characteristic data of each time period after segmentation, decompose them into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies respectively;

[0016] Normalizing the sub-temperature signal data and the sub-electrical characteristic signal data respectively;

[0017] For the normalized sub-temperature signal data and the normalized sub-electrical characteristic signal data in each time period, data extraction is performed according to a preset window sliding step, and a characteristic matrix is ​​formed using all the extracted temperature data and characteristic data.

[0018] Optionally, the temperature data and electrical characteristic data of each time period after segmentation are decomposed into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies, including:

[0019] The empirical mode decomposition (EMD) method is used to decompose the temperature data and electrical characteristic data of each time period after segmentation into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies.

[0020] Optionally, the temperature prediction model is a deep neural network model, and the deep neural network model includes:

[0021] The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence; wherein,

[0022] The first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of:

[0023] It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout;

[0024] The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout, and the fourth feature extraction module is used for temperature prediction data.

[0025] In a second aspect, an embodiment of the present invention further provides a temperature prediction method for an electronic device, the method comprising:

[0026] Using the collected temperature data of the electronic device in the first time period, the collected electrical characteristic data in the second time period as basic data;

[0027] Decomposing the basic data into sub-signal data of different frequencies, and normalizing the sub-signal data;

[0028] Based on the temperature prediction model, temperature prediction is performed according to the normalized sub-signal data to obtain temperature prediction data of the electronic device in a second time period;

[0029] The temperature prediction model is trained using the temperature prediction model training method provided by any embodiment of the present invention.

[0030] In a third aspect, an embodiment of the present invention further provides a training device for a temperature prediction model, the device comprising:

[0031] A data acquisition module, configured to acquire temperature data of the electronic device and electrical characteristic data of the electronic device at different temperatures;

[0032] A time segmentation module, configured to segment the temperature data and the corresponding electrical characteristic data according to preset time intervals;

[0033] The feature matrix forming module is configured to extract data according to a preset window sliding step for the temperature data and electrical feature data of each time period after segmentation, and form a feature matrix using all the extracted temperature data and electrical feature data;

[0034] a training sample determination module, configured to select temperature data and electrical characteristic data within a first time period, and electrical characteristic data within a second time period from the characteristic matrix as basic data, and use the temperature data within the second time period as data to be predicted;

[0035] A model training module is configured to use the basic data and the data to be predicted as a training sample set to train a temperature prediction model so that the temperature prediction model establishes an association relationship between temperature data and electrical characteristic data in different time periods;

[0036] The first time period and the second time period are arranged in chronological order.

[0037] Optionally, a feature matrix forming module specifically includes:

[0038] A frequency decomposition unit is configured to decompose the temperature data and electrical characteristic data of each time period after segmentation into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies;

[0039] A normalization unit, configured to normalize the sub-temperature signal data and the sub-electrical characteristic signal data respectively;

[0040] The characteristic matrix forming unit is configured to collect data according to a preset window sliding step for the normalized sub-temperature signal data and the normalized sub-electrical characteristic signal data in each time period, and form a characteristic matrix using all the collected temperature data and characteristic data.

[0041] Optionally, the frequency decomposition unit is configured to:

[0042] The empirical mode decomposition (EMD) method is used to decompose the temperature data and electrical characteristic data of each time period after segmentation into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies.

[0043] Optionally, the temperature prediction model is a deep neural network model, and the deep neural network model includes:

[0044] The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence; wherein,

[0045] The first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of:

[0046] It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout;

[0047] The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout, and the fourth feature extraction module is used to output temperature prediction data.

[0048] In a fourth aspect, an embodiment of the present invention further provides a temperature prediction device for an electronic device, the device comprising:

[0049] A basic data determination module is configured to use the collected temperature data of the electronic device in the first time period, the electrical characteristic data in the second time period as basic data;

[0050] A data transformation module, configured to decompose the basic data into sub-signal data of different frequencies and normalize the sub-signal data;

[0051] a temperature prediction module configured to perform temperature prediction according to the normalized sub-signal data based on a temperature prediction model to obtain temperature prediction data of the electronic device in a second time period;

[0052] The temperature prediction model is trained using the temperature model training method provided in any embodiment of the present invention.

[0053] In a fifth aspect, an embodiment of the present invention further provides a computing device, including:

[0054] A memory storing executable program code;

[0055] a processor coupled to the memory;

[0056] The processor calls the executable program code stored in the memory to execute the training method of the temperature prediction model provided by any embodiment of the present invention.

[0057] In a sixth aspect, an embodiment of the present invention further provides a computing device, including:

[0058] A memory storing executable program code;

[0059] a processor coupled to the memory;

[0060] The processor calls the executable program code stored in the memory to execute the temperature prediction method of the electronic device provided by any embodiment of the present invention.

[0061] In a seventh aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the training method of the temperature prediction model provided by any embodiment of the present invention.

[0062] In an eighth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the temperature prediction method for an electronic device provided by any embodiment of the present invention.

[0063] The technical solution provided by the embodiment of the present invention predicts the temperature of electronic devices based on a neural network method, which can effectively solve the problems of difficulty in analyzing existing temperature prediction data and insufficient prediction accuracy. By using the historical temperature data of electronic devices for model training, there is no need to perform complex modeling analysis, and the prediction of the temperature of electronic devices at future moments is achieved, solving the problem of modeling difficulties. In addition, by using electrical signal characteristic data that can characterize the operating state of electronic devices for temperature prediction, the accuracy of electronic device temperature prediction is further improved.

[0064] The innovative features of the embodiments of the present invention include:

[0065] 1. The method based on neural network can effectively solve the problems of difficult data analysis and insufficient prediction accuracy. In the training process of the neural network model, in addition to using historical temperature data as training samples, electrical characteristic data at future times are also used as training samples, which effectively improves the model's prediction accuracy for temperature. For situations where temperature data is not easy to measure, or in some cases where the return speed of the temperature sensor is different and there is a certain delay in the return value, the use of the trained neural network model can achieve accurate prediction of temperature data at future times, which is one of the innovative points of the embodiment of the present invention.

[0066] 2. Before inputting the training samples into the neural network model, the temperature data and electrical characteristic data are decomposed by empirical mode EMD, so that the data can be decomposed into different high and low frequency parts, so that the distribution of different frequencies can be predicted separately, which can make the model prediction results more accurate. In addition, the data after EMD decomposition is normalized to further improve the effect of model prediction, which is one of the innovative points of the embodiment of the present invention.

[0067] 3. The first three feature extraction modules connected in sequence in the neural network model are respectively composed of: a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN and a partial discarding neuron unit Dropout, which are used to effectively extract features from the training sample set in sequence. This design method can effectively improve the prediction accuracy of temperature data and is one of the innovations of the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0069] Figure 1aA flow chart of a method for training a temperature prediction model provided in Embodiment 1 of the present invention;

[0070] Figure 1b A structural diagram of a deep neural network model provided in Example 1 of the present invention;

[0071] Figure 2 A flow chart of a method for predicting temperature of an electronic device provided in Embodiment 2 of the present invention;

[0072] Figure 3 A structural block diagram of a temperature prediction model training device provided in Embodiment 3 of the present invention;

[0073] Figure 4 A structural block diagram of a temperature prediction device for an electronic device provided in a fourth embodiment of the present invention;

[0074] Figure 5 It is a structural diagram of a computing device provided in Embodiment 5 of the present invention. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] It should be noted that the terms "including" and "having" and any variations thereof in the embodiments of the present invention and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0077] The embodiments of the present invention disclose a method for training a temperature prediction model and a method for performing temperature prediction on an electronic device using the trained temperature prediction model. The details are described below.

[0078] Embodiment 1

[0079] Figure 1a This is a flow chart of a method for training a temperature prediction model provided in the first embodiment of the present invention. The temperature prediction model can be applied to different components, especially to the scenario of temperature prediction of traction motors. The method can be executed by a training device for the temperature prediction model, which can be implemented in software and / or hardware. Figure 1a, the method of this embodiment specifically includes:

[0080] 110. Collect temperature data of electronic devices and electrical characteristic data of electronic devices at different temperatures.

[0081] Among them, for electronic devices, various types of sensors can be used to collect the electrical characteristic data and temperature data corresponding to the electronic devices. For example, for the train traction system, the sensor coverage is already very comprehensive, and the current and voltage data of each branch in the system are measured and recorded by corresponding professional sensors, which will generate massive amounts of data every moment.

[0082] The embodiment of the present invention uses a neural network model to predict temperature data, and the training and testing of the model are based on the actual collected data. The sensor records the data during the operation of the system in a time series. The collected data is stored in tdms and csv (Comma-Separated Values) files according to different sampling frequencies. In this embodiment, the data stored in the above files can be processed uniformly, for example, the temperature data and electrical characteristic data can be sampled at the same sampling frequency, such as 10Hz. The data after uniform processing can be used in the subsequent data segmentation process.

[0083] The electrical characteristic column data may include: network current, instantaneous value of U-phase current, instantaneous value of W-phase current, effective value of U-phase current, effective value of W-phase current and operating speed of electronic devices, such as motor speed.

[0084] 120. Segment the temperature data and the corresponding electrical characteristic data according to preset time intervals.

[0085] Among them, when the adjacent time intervals exceed the preset time interval, they are considered to be data of different time periods. In this embodiment, after the temperature data and the corresponding electrical characteristic data are segmented according to the preset time intervals, the time segment information of the temperature data and the corresponding electrical characteristic data can be obtained. The time segment information can be used for data extraction through the sliding window below. Through the sliding window method, multiple temperature data and electrical characteristic data of the same time period can be extracted as sample data for model training.

[0086] 130. For the temperature data and electrical characteristic data of each time period after segmentation, data extraction is performed according to a preset window sliding step, and a characteristic matrix is ​​formed using all the extracted temperature data and electrical characteristic data.

[0087] In this embodiment, in order to obtain more sample points, the preset window sliding step size may be 1, corresponding to 0.1 seconds, and the sliding window size may be 160 steps. The window sliding step size and window size may be designed according to actual conditions, and this embodiment does not specifically limit this.

[0088] In this embodiment, for the temperature data and electrical characteristic data of the same time period, they can be extracted by using a sliding window. For all the temperature data and electrical characteristic data extracted in different time periods, they can be combined to form a feature matrix. The elements in the feature matrix include temperature signal data and electrical characteristic signal data at different time points intercepted by the sliding window. The elements in the same row of the feature matrix represent the data at the same time point, for example, the first row represents the electrical characteristic column data and temperature data collected in 0.1 seconds; the elements in the same column represent the same type of data, for example, the first column is current, the second column is voltage, etc.

[0089] Furthermore, in order to make the model prediction results more accurate, the temperature data and electrical characteristic data of each time period after segmentation can be decomposed into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies. Separate predictions of distributions of different frequencies can make the model prediction results more accurate, especially in areas with large high-frequency components. Among them, the frequency decomposition can preferably adopt the EMD (Empirical Mode Decomposition) method, which can decompose the data into high and low frequency parts, so as to separately predict the distributions of different frequencies, which can make the model prediction results more accurate, especially in areas with large high-frequency components. Accurate prediction of temperature can be achieved.

[0090] After frequency decomposition, the sub-temperature signal data and the sub-electrical characteristic signal data can be normalized respectively to effectively improve the effect of model prediction.

[0091] For the normalized sub-temperature signal data and the normalized sub-electrical characteristic signal data in each time period, data extraction is performed according to the preset window sliding step, and a feature matrix is ​​formed using all the extracted temperature data and characteristic data. The elements in the feature matrix include the sub-temperature signal data and the sub-electrical characteristic signal data at different time points intercepted by the sliding window.

[0092] 140. Select temperature data and electrical characteristic data within a first time period and electrical characteristic data within a second time period from the characteristic matrix as basic data, and use the temperature data within the second time period as data to be predicted.

[0093] In addition, historical data of other related variables may also be added to the basic data.

[0094] Among them, the first time period and the second time period are arranged in chronological order, for example, the first time period can be the data collected in the first 16 seconds, and the second time period can be the data collected in the last 16 seconds. The electrical characteristic data and temperature data of the first time period can be used as historical data, and the electrical characteristic data of the second time period can be used as data at a future time. In this embodiment, in addition to using historical data when training the model, the electrical characteristic data that is easy to measure at a future time is also used as part of the training sample for predicting the temperature data at a future time. This embodiment is arranged in this way. On the one hand, for some situations where temperature data is not easy to measure, or in some cases where the return speed of the temperature sensor is different and there is a certain delay in the return value, the prediction of the temperature of the electronic device at a future time is realized, solving the problem of modeling difficulties. On the other hand, this embodiment is arranged in this way. Compared with the method of using only historical data as training samples, by also using electrical characteristic data that is easy to measure at a future time as training samples, the prediction accuracy of the model can be effectively improved.

[0095] In this embodiment, some data can also be selected from the feature matrix, such as the data collected in the first 16 seconds, to test the effect of the trained temperature prediction model. The test data used for effect testing and the training sample data used for model training can also be randomly selected from the data collected in different time periods after segmentation, and this embodiment does not specifically limit this.

[0096] 150. The basic data and the data to be predicted are used as training sample sets to train the temperature prediction model, so that the temperature prediction model can establish the correlation between the temperature data and the electrical characteristic data in different time periods.

[0097] Among them, the temperature prediction model is a deep neural network model. Figure 1b A structural diagram of a deep neural network model provided in the first embodiment of the present invention, such as Figure 1b As shown, the deep neural network model includes:

[0098] The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence, such as Figure 1b As shown, by continuously performing feature extraction, the dimensions of the input data are reduced to 128, 64, and 32 respectively, until the temperature prediction data is finally output.

[0099] Among them, the first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of:

[0100] It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout;

[0101] The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout. The fourth feature extraction module is used to output the temperature prediction data for the second time period.

[0102] In this embodiment, the basic data and the data to be predicted are training sample sets, which are input into the temperature prediction model for training. During the training process, the parameters of the neural network model are updated so that the value of the loss function of the neural network model is gradually reduced until the value of the loss function reaches convergence, and the last updated parameters are used as the parameters of the neural network model. After the training of the neural network model is completed, the correlation between the temperature data and the electrical characteristic data in different time periods is established.

[0103] The technical solution provided by the embodiment of the present invention predicts the temperature of electronic devices based on a neural network method, which can effectively solve the problems of difficulty in analyzing existing temperature prediction data and insufficient prediction accuracy. By using the historical temperature data of electronic devices for model training, there is no need to perform complex modeling analysis, and the prediction of the temperature of electronic devices at future moments is achieved, solving the problem of modeling difficulties. In addition, by using electrical signal characteristic data that can characterize the operating state of electronic devices for temperature prediction, the accuracy of electronic device temperature prediction is further improved.

[0104] After completing the training of the temperature prediction model, you can use the trained model to predict the temperature of electronic devices at future times. For details, please refer to Figure 2 .

[0105] Embodiment 2

[0106] Figure 2 This is a flow chart of a method for predicting the temperature of an electronic device provided in the second embodiment of the present invention. The method can be applied to the application scenario of predicting the temperature of an electronic device, and can be typically applied to the prediction process of the temperature of a traction converter chassis and a motor. The method can be executed by an electronic device temperature prediction device, and the device can be implemented by software and / or hardware. Figure 2 As shown, the method includes:

[0107] 210. Use the collected temperature data and electrical characteristic data of the electronic device in the first time period and the collected electrical characteristic data in the second time period as basic data.

[0108] Among them, the electrical characteristic data is the electrical signal characteristic data that characterizes the operating status of the electronic device, such as network current, U-phase current instantaneous value, W-phase current instantaneous value, U-phase current effective value, W-phase current effective value and operating speed light.

[0109] In this embodiment, the first time period and the second time period are arranged in chronological order. The temperature data and electrical characteristic data of the first time period can be used as historical data, and the historical data and the electrical characteristic data that can be easily measured in the second time period (future time) are used as input data of the model and input into the trained temperature prediction model. The model can input the temperature data at the future time to achieve accurate prediction of the temperature of the electronic device.

[0110] 220. Decompose the basic data into sub-signal data of different frequencies, and normalize the sub-signal data.

[0111] In this embodiment, the empirical mode decomposition method can be used to decompose the basic data into sub-signal data of different frequencies, specifically, into sub-signal data of different high and low frequencies. In the model prediction process, the distribution of different frequencies can be predicted separately, which can make the model prediction results more accurate, especially in areas with large high-frequency components to achieve accurate temperature prediction. In addition, after frequency decomposition, the decomposed sub-signal data can be normalized to effectively improve the effect of model prediction.

[0112] 230. Based on the temperature prediction model, perform temperature prediction according to the normalized sub-signal data to obtain temperature prediction data of the electronic device in the second time period.

[0113] The temperature prediction model is a deep neural network model, which includes:

[0114] The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence; wherein,

[0115] The first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of:

[0116] It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout;

[0117] The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout. The fourth feature extraction module is used to output the temperature prediction data for the second time period.

[0118] In this embodiment, the training method of the temperature prediction model can refer to the description of the above embodiment and will not be repeated here.

[0119] The technical solution provided in this embodiment, after the temperature prediction model training is completed, uses historical temperature data and electrical signal characteristic data characterizing the operating status of electronic devices as input data of the model, and can obtain the temperature data of electronic devices at future times, thereby achieving accurate prediction of temperature data.

[0120] Embodiment 3

[0121] Figure 3 A structural block diagram of a temperature prediction model training device provided in Embodiment 3 of the present invention, such as Figure 3 As shown, the device includes: a data acquisition module 310, a time segmentation module 320, a feature matrix formation module 330, a training sample determination module 340 and a model training module 350; wherein,

[0122] The data acquisition module 310 is configured to acquire temperature data of the electronic device and electrical characteristic data of the electronic device at different temperatures;

[0123] The time segmentation module 320 is configured to segment the temperature data and the corresponding electrical characteristic data according to preset time intervals;

[0124] The feature matrix forming module 330 is configured to extract data according to a preset window sliding step for the temperature data and electrical feature data of each time period after segmentation, and form a feature matrix using all the extracted temperature data and electrical feature data;

[0125] The training sample determination module 340 is configured to select the temperature data and the electrical characteristic data in the first time period and the electrical characteristic data in the second time period from the characteristic matrix as basic data, and use the temperature data in the second time period as data to be predicted;

[0126] The model training module 350 is configured to use the basic data and the data to be predicted as a training sample set to train the temperature prediction model so that the temperature prediction model establishes an association relationship between the temperature data and the electrical characteristic data in different time periods;

[0127] The first time period and the second time period are arranged in chronological order.

[0128] Optionally, the feature matrix forming module 330 specifically includes:

[0129] A frequency decomposition unit is configured to decompose the temperature data and electrical characteristic data of each time period after segmentation into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies;

[0130] A normalization unit, configured to normalize the sub-temperature signal data and the sub-electrical characteristic signal data respectively;

[0131] The feature matrix forming unit is configured to extract data according to a preset window sliding step for the normalized sub-temperature signal data and the normalized sub-electrical feature signal data in each time period, and form a feature matrix using all the extracted temperature data and feature data.

[0132] Optionally, the frequency decomposition unit is configured to:

[0133] The empirical mode decomposition (EMD) method is used to decompose the temperature data and electrical characteristic data of each time period after segmentation into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies.

[0134] Optionally, the temperature prediction model is a deep neural network model, and the deep neural network model includes:

[0135] The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence; wherein,

[0136] The first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of:

[0137] It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout;

[0138] The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout, and the fourth feature extraction module is used to output temperature prediction data.

[0139] The training device for the temperature prediction model provided in the embodiment of the present invention can execute the training method for the temperature prediction model provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in the above embodiment, please refer to the training method for the temperature prediction model provided in any embodiment of the present invention.

[0140] Embodiment 4

[0141] Figure 4 A structural block diagram of a temperature prediction device for an electronic device provided in a fourth embodiment of the present invention, such as Figure 4 As shown, the device includes: a basic data determination module 410, a data transformation module 420 and a temperature prediction module 430; wherein,

[0142] The basic data determination module 410 is configured to use the temperature data, electrical characteristic data of the electronic device collected in the first time period, and the electrical characteristic data in the second time period as basic data;

[0143] A data transformation module 420 is configured to decompose the basic data into sub-signal data of different frequencies and normalize the sub-signal data;

[0144] The temperature prediction module 430 is configured to perform temperature prediction according to the normalized sub-signal data based on the temperature prediction model to obtain temperature prediction data of the electronic device in the second time period;

[0145] The temperature prediction model is trained using the temperature model training method provided in any embodiment of the present invention.

[0146] The temperature prediction device of the electronic device provided in the embodiment of the present invention can execute the temperature prediction method of the electronic device provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in the above embodiment, please refer to the temperature prediction method of the electronic device provided in any embodiment of the present invention.

[0147] Embodiment 5

[0148] See also Figure 5 , Figure 5 Schematic diagram of a computing device provided by Embodiment 5 of the present invention. Figure 5 As shown, the computing device may include:

[0149] A memory 701 storing executable program codes;

[0150] a processor 702 coupled to the memory 701;

[0151] The processor 702 calls the executable program code stored in the memory 701 to execute the training method of the temperature prediction model provided by any embodiment of the present invention.

[0152] The embodiment of the present invention also provides a schematic diagram of the structure of a computing device. The computing device may include:

[0153] A memory storing executable program code;

[0154] a processor coupled to the memory;

[0155] The processor calls the executable program code stored in the memory to execute the temperature prediction method of the electronic device provided by any embodiment of the present invention.

[0156] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the training method of a temperature prediction model provided by any embodiment of the present invention.

[0157] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the temperature prediction method of an electronic device provided by any embodiment of the present invention.

[0158] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the above-mentioned processes does not mean the necessary order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0159] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0160] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0161] If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several requests for a computer device (which can be a personal computer, a server or a network device, etc., specifically a processor in a computer device) to perform some or all of the steps of the above methods of various embodiments of the present invention.

[0162] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0163] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0164] Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description of the embodiment, or can be changed accordingly and located in one or more devices different from the embodiment. The modules in the above embodiment can be combined into one module, or can be further divided into multiple sub-modules.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for training a temperature prediction model, applied to scenarios where the return speed of a temperature sensor is different or the return value is delayed, characterized in that: include: Collecting temperature data of the electronic device and electrical characteristic data of the electronic device at different temperatures; wherein the temperature data and each electrical characteristic data are sampled at the same frequency; Segmenting the temperature data and the corresponding electrical characteristic data according to preset time intervals; For the temperature data and electrical characteristic data of each time period after segmentation, data extraction is performed according to the preset window sliding step size, and a characteristic matrix is ​​formed using all the extracted temperature data and electrical characteristic data; Selecting temperature data and electrical characteristic data in a first time period and electrical characteristic data in a second time period from the characteristic matrix as basic data, and using temperature data in the second time period as data to be predicted; Using the basic data and the data to be predicted as a training sample set to train a temperature prediction model, so that the temperature prediction model establishes an association relationship between temperature data and electrical characteristic data in different time periods; The first time period and the second time period are arranged in chronological order.

2. The method according to claim 1, characterized in that: For the temperature data and electrical characteristic data of each time period after segmentation, data extraction is performed according to the preset window sliding step size, and a feature matrix is ​​formed using all the extracted temperature data and electrical characteristic data, including: For the temperature data and electrical characteristic data of each time period after segmentation, decompose them into sub-temperature signal data and sub-electrical characteristic signal data of different frequencies respectively; Normalizing the sub-temperature signal data and the sub-electrical characteristic signal data respectively; For the normalized sub-temperature signal data and the normalized sub-electrical characteristic signal data in each time period, data extraction is performed according to a preset window sliding step, and a characteristic matrix is ​​formed using all the extracted temperature data and characteristic data.

3. The method according to claim 1, characterized in that The temperature prediction model is a deep neural network model, and the deep neural network model includes: The first feature extraction module, the second feature extraction module, the third feature extraction module and the fourth feature extraction module connected in sequence are used to extract features from the training sample set in sequence; wherein, The first feature extraction module, the second feature extraction module, and the third feature extraction module are respectively composed of: It consists of a fully connected unit FC, a linear rectified linear unit ReLU, a batch normalization unit BN, and a partially discarded neuron unit Dropout; The fourth feature extraction module is composed of a fully connected unit FC, a batch normalization unit BN and a partial discard neuron unit Dropout, and the fourth feature extraction module is used to output temperature prediction data.

4. A temperature prediction method for an electronic device, applied to a scenario where the return speed of a temperature sensor is different or the return value is delayed, characterized in that: include: Using the collected temperature data of the electronic device in the first time period, the collected electrical characteristic data in the second time period as basic data; Decomposing the basic data into sub-signal data of different frequencies, and normalizing the sub-signal data; Based on the temperature prediction model, temperature prediction is performed according to the normalized sub-signal data to obtain temperature prediction data of the electronic device in a second time period; The temperature prediction model is trained using the temperature prediction model training method as described in any one of claims 1 to 3.

5. A training device for a temperature prediction model, applied to scenarios where the return speed of a temperature sensor is different or the return value is delayed, characterized in that: include: A data acquisition module, configured to acquire temperature data of the electronic device and electrical characteristic data of the electronic device at different temperatures; A time segmentation module, configured to segment the temperature data and the corresponding electrical characteristic data according to preset time intervals; The feature matrix forming module is configured to extract data according to a preset window sliding step for the temperature data and electrical feature data of each time period after segmentation, and form a feature matrix using all the extracted temperature data and electrical feature data; a training sample determination module, configured to select temperature data and electrical characteristic data within a first time period, and electrical characteristic data within a second time period from the characteristic matrix as basic data, and use the temperature data within the second time period as data to be predicted; A model training module is configured to use the basic data and the data to be predicted as a training sample set to train a temperature prediction model so that the temperature prediction model establishes an association relationship between temperature data and electrical characteristic data in different time periods; The first time period and the second time period are arranged in chronological order.

6. A temperature prediction device for an electronic device, applied to a scenario where the return speed of a temperature sensor is different or the return value is delayed, characterized in that: include: A basic data determination module is configured to use the collected temperature data of the electronic device in the first time period, the electrical characteristic data in the second time period as basic data; A data transformation module, configured to decompose the basic data into sub-signal data of different frequencies and normalize the sub-signal data; a temperature prediction module configured to perform temperature prediction according to the normalized sub-signal data based on a temperature prediction model to obtain temperature data of the electronic device in a second time period; The temperature prediction model is trained using the temperature prediction model training method as described in any one of claims 1 to 3.

7. A computing device, characterized in that The computing device comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the training method of the temperature prediction model as described in any one of claims 1-3.

8. A computing device, characterized in that The computing device comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the temperature prediction method for an electronic device as claimed in claim 4.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the training method of the temperature prediction model as described in any one of claims 1 to 3 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the temperature prediction method of an electronic device as claimed in claim 4 is implemented.

Images