Bulb tube service life prediction method and device, equipment and storage medium

Abstract

The invention provides a bulb tube service life prediction method and device, equipment and a storage medium, and relates to the technical field of medical equipment, and the method comprises the steps: obtaining a filament current prediction model, a failure current threshold of a current target bulb tube, and filament current data, obtaining target prediction current data based on the filament current data and the filament current prediction model, and obtaining the target prediction current data; and determining the current target bulb tube life according to the failure current threshold and the target prediction current data. Based on the current target bulb tube filament current data and the filament current prediction model, the filament current change trend in the current target bulb tube prediction time can be predicted, and the life of the bulb tube can be predicted in combination with the failure current threshold and the target prediction current data. The service life of the bulb tube is predicted through the filament current which is a key factor influencing the service life of the bulb tube, the prediction accuracy can be improved, equipment shutdown caused by sudden failure of the bulb tube is avoided, the operation reliability of the bulb tube is improved, and the operation cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of medical equipment technology, and in particular to a method, device, equipment and storage medium for predicting the life of a tube. Background Art

[0002] Computed Tomography (CT) is a technology widely used in medical imaging. It generates tomographic images of the scanned object by acquiring X-ray projection images from different angles and reconstructing them. One of the core components of a CT machine is the X-ray tube, which is responsible for generating high-energy X-rays to penetrate the human body and be received by the detector to obtain projection data. The X-ray tube is mainly composed of a cathode and an anode, and has a heating filament inside. When current passes through the filament, the filament heats up and emits electrons. The high-voltage electric field accelerates these electrons to the anode target, and the electrons interact with the target material to produce X-rays.

[0003] At present, CT equipment generally adopts the method of regular maintenance and replacement of tubes to ensure the normal operation of the equipment. This method not only increases the operating cost, but may also cause equipment downtime due to sudden tube failure. Therefore, a tube life prediction method is urgently needed to predict the life of the tube in advance. Summary of the invention

[0004] In view of the shortcomings of the related technologies mentioned above, the present application provides a tube life prediction method, device, equipment and storage medium to solve the above technical problems.

[0005] The present application provides a method for predicting the life of a tube, characterized by comprising: Obtain the filament current prediction model, the failure current threshold of the current target tube, and the filament current data; Based on the filament current data and the filament current prediction model, target predicted current data are obtained, wherein the target predicted current data includes a plurality of predicted filament currents of the current target bulb, the plurality of predicted filament currents are arranged in chronological order, and the target predicted current data are used to characterize a change trend of the filament current of the current target bulb within a prediction time; The current target tube life is determined according to the failure current threshold and the target predicted current data.

[0006] In one embodiment of the present application, obtaining a filament current prediction model includes: Obtaining a current sample data set and an initial prediction model; The initial prediction model is trained based on the current sample data set to obtain a filament current prediction model, and the filament current prediction model is used to predict the change trend of the filament current.

[0007] In one embodiment of the present application, obtaining a current sample data set includes: Acquire initial current data, the initial current data including filament current data corresponding to a plurality of sample bulbs, each of the filament current data including a plurality of historical filament currents sorted by time; Performing data standardization processing on the plurality of historical filament currents to obtain a plurality of standardized data; Sliding window processing is performed on the multiple standardized data to obtain multiple historical current sample data, and the multiple historical current sample data constitute the current sample data set.

[0008] In one embodiment of the present application, data standardization is performed on a plurality of the historical filament currents to obtain a plurality of standardized data, including: The filament current data are grouped according to a preset time period to obtain multiple groups of data to be calculated; Performing mean filtering processing on the historical filament current of each group of the data to be calculated to obtain a plurality of data to be processed; Data standardization is performed on each of the data to be processed to obtain a plurality of standardized data.

[0009] In one embodiment of the present application, the current sample data set includes historical current sample data corresponding to the sample tube, and training the initial prediction model based on the current sample data set includes: Inputting the historical current sample data into a first prediction model to obtain a first prediction result, wherein the first prediction model is used to capture the dependency relationship of the historical current sample data; Inputting the historical current sample data into a second prediction model to obtain a second prediction result, wherein the second prediction model is used to perform trend quantity prediction and residual quantity prediction respectively after data decomposition of the historical current sample data, and obtain the second prediction result based on the trend quantity prediction and the residual quantity prediction; Inputting the historical current sample data into a third prediction model to obtain a third prediction result, wherein the third prediction model is used to analyze data within a preset data length in the historical current sample data; Based on the self-attention mechanism, the first prediction result, the second prediction result, and the third prediction result are first concatenated to obtain model predicted current data.

[0010] In one embodiment of the present application, based on the filament current data and the filament current prediction model, the target predicted current data is obtained, including: Determine the filament current data as input data, and perform a data input step, wherein the data input step includes inputting the input data into the filament current prediction model to obtain initial predicted current data; Executing a data judgment step, the data judgment step comprising determining whether there is target data in the initial predicted current data, the target data being less than or equal to the failure current threshold; If the target data does not exist in the initial predicted current data, a data updating step is performed, wherein the data updating step includes performing a second splicing of the initial predicted current data and the input data to obtain spliced ​​data; determining the spliced ​​data as the input data, and performing the data input step again; If the target data exists in the initial predicted current data, the target predicted current data is determined based on all the initial predicted current data.

[0011] In one embodiment of the present application, obtaining the failure current threshold of the current target tube includes: Obtaining a target initial current of the current target tube; The product of the target initial current and a preset value is determined as the failure current threshold.

[0012] The current initial current and the historical initial current are weightedly summed to obtain the target initial current.

[0013] To achieve the above-mentioned purpose and other related purposes, the present application provides a device for predicting the life of a tube, comprising: A data acquisition module is used to obtain a filament current prediction model, a failure current threshold of a current target tube, and filament current data; A data prediction module, used for obtaining target predicted current data based on the filament current data and the filament current prediction model, wherein the target predicted current data includes a plurality of predicted filament currents of the current target bulb, the plurality of predicted filament currents are arranged in chronological order, and the target predicted current data is used for characterizing a change trend of the filament current of the current target bulb within a predicted time period; The life prediction module is used to determine the current target tube life according to the failure current threshold and the target predicted current data.

[0014] To achieve the above-mentioned purpose and other related purposes, the present application also provides an electronic device, including one or more processors; a memory for storing one or more programs, when the one or more programs are executed by the one or more processors, the electronic device executes one or more of the aforementioned tube life prediction methods.

[0015] To achieve the above-mentioned purpose and other related purposes, the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of a computer, the computer executes one or more of the aforementioned methods for predicting the life of the tube.

[0016] As described above, the present application provides a method, device, equipment and storage medium for predicting the life of a tube, which has the following beneficial effects: A method for predicting the life of a bulb in the present application, the method obtains a filament current prediction model, a failure current threshold of the current target bulb, and filament current data, obtains target predicted current data based on the filament current data and the filament current prediction model, and determines the life of the current target bulb according to the failure current threshold and the target predicted current data. Based on the filament current data of the current target bulb and the filament current prediction model, the filament current change trend of the current target bulb within the prediction time can be predicted, and the life of the bulb can be predicted in combination with the failure current threshold and the target predicted current data. By predicting the life of the bulb through the filament current, which is a key factor affecting the life of the bulb, the accuracy of the prediction can be improved, equipment shutdown caused by sudden failure of the bulb can be avoided, the reliability of the operation of the bulb can be improved, and the operating cost can be reduced.

[0017] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 is a flow chart of a tube life prediction method shown in an exemplary embodiment of the present application; Figure 2 is a schematic diagram of performing sliding window processing on multiple standardized data, shown in an exemplary embodiment of the present application; Figure 3 It is a structural block diagram of a tube life prediction device shown in an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0019] The following will describe the implementation methods of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustrating the present application, not for limiting the scope of protection of the present application.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application, and thus the drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed at will, and the component layout may also be more complicated.

[0021] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0022] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for predicting the life of a tube according to an exemplary embodiment of the present invention. Figure 1 It can be seen that the tube life prediction method can include: Step S110, obtaining a filament current prediction model, a failure current threshold of a current target tube, and filament current data.

[0023] In one embodiment of the present application, a filament current prediction model can be obtained, and the filament current prediction model can be used to predict the change trend of the filament current of the target tube within the prediction time. The failure current threshold and filament current data of the current target tube can also be obtained. The failure current threshold can be used to determine whether the filament of the current target tube has failed, which can be used as a basis for determining the life of the current target tube.

[0024] It should be noted that the filament current is the current required for heating the filament, and its magnitude directly affects the temperature of the filament and the number of electrons emitted, thereby affecting the generation of X-rays. The magnitude and fluctuation of the filament current is one of the key factors affecting the life of the tube.

[0025] In a possible implementation, the filament current data may include a plurality of current history data within a preset usage time of the current target tube. Initial history data within a preset usage time of the current target tube may be obtained, the initial history data may be divided into a plurality of groups according to a preset division period, and the mean of each group of initial history data may be determined as the current history data. The preset division period may be one day or 12 hours, which is not limited in the embodiment of the present application.

[0026] For example, the starting point of the preset use time may be the start time of the current target tube, and the length of the preset use time may be 30 days; the end point of the preset use time may be the current time, and the length of the preset use time may be 14 days. This embodiment of the application is not limited to this.

[0027] In an exemplary embodiment, the process of obtaining the failure current threshold of the current target tube in step S110 may include obtaining the target initial current of the current target tube; and determining the product of the target initial current and a preset value as the failure current threshold.

[0028] In one embodiment of the present application, the target initial current may be the average value of the filament current when the current target bulb is used on the first day. The target initial current may also be the filament current when the current target bulb is used for the first time.

[0029] In another exemplary embodiment, the process of obtaining the target initial current of the current target tube may further include obtaining the current initial current of the current target tube and the historical initial current of the historical target tube; performing weighted summation on the current initial current and the historical initial current to obtain the target initial current. The historical initial current of the historical target tube may be the historical initial current of any historical target tube, or may be the average of all historical initial currents of all historical target tubes.

[0030] It should be noted that the historical initial current of the historical target tube may be the initial value of the filament current of other tubes in the historical database. The current initial current may be the average value of the filament current of the current target tube when it was used on the first day. The current initial current may also be the filament current of the current target tube when it was used for the first time.

[0031] For example, the weight of the current initial current may be 70%, and the weight of the historical initial current may be 30%. , the historical initial current can be , target initial current .

[0032] In one embodiment of the present application, the product of the target initial current and the preset value can be determined as the failure current threshold. The target initial current can be used to characterize the current state when the current target tube starts to be used, and the failure current threshold can be used to characterize the critical condition for the normal operation of the current target tube.

[0033] Exemplarily, the preset value may be 5%.

[0034] It should be noted that in the embodiment of the present application, the product of the target initial current and the preset value is determined as the failure current threshold, which ensures the standardization of the failure current thresholds of the filaments of different bulbs, effectively reflects the specific characteristics of each bulb, and thus can improve the accuracy of the current target life prediction in the subsequent steps. At the same time, the setting of a fixed ratio simplifies the calculation process, making failure prediction more convenient and reliable, which helps to carry out maintenance and replacement in advance, reduce equipment downtime, and improve overall operating efficiency.

[0035] In another exemplary embodiment, the process of acquiring the filament current prediction model in step S110 may include step S111 and step S112.

[0036] Step S111, obtaining a current sample data set and an initial prediction model.

[0037] In one embodiment of the present application, a current sample data set and an initial prediction model may be obtained. The current sample data set may be used to train the initial prediction model.

[0038] In a possible implementation, the process of obtaining a current sample data set may include: obtaining initial current data, the initial current data including filament current data corresponding to a plurality of sample bulbs, each filament current data including a plurality of historical filament currents sorted by time; performing data standardization processing on the plurality of historical filament currents to obtain a plurality of standardized data; performing sliding window processing on the plurality of standardized data to obtain a plurality of historical current sample data, wherein the plurality of historical current sample data constitute a current sample data set.

[0039] In another possible implementation, the process of performing data standardization processing on multiple historical filament currents to obtain multiple standardized data may include: grouping the filament current data according to a preset time period to obtain multiple groups of data to be calculated; performing mean filtering processing on the historical filament current of each group of data to be calculated to obtain multiple data to be processed; performing data standardization processing on each data to be processed to obtain multiple standardized data.

[0040] Exemplarily, the current sample data set may be obtained by taking the following steps.

[0041] Step 1: Data Collection Multiple initial current data are collected from the data of several sample tubes, and each initial current data can be full-life data, that is, the data is collected from the installation of the sample tube until it is damaged. Assuming that there are four sample tubes, the collected initial current data can be divided into the first data sample, the second data sample, the third data sample and the fourth data sample. Each data sample represents the current change of a tube in a specific period: the first data sample contains data in the initial stage, the second data sample contains mid-term data, the third data sample contains late data, and the fourth data sample contains data close to damage. This data segmentation can comprehensively record and analyze the changing characteristics of the tube filament current over time, thereby providing a scientific basis for predicting the failure time of the tube. Observe whether the collected data meets the above standards, and eliminate those that do not meet the standards, and analyze the reasons.

[0042] Step 2: Mean filtering The sampling interval of the CT tube filament current is about a few seconds, and the amount of data collected is large. However, due to the influence of the working mode, the filament current will have a large jump in amplitude in a short time and at a high frequency (during the entire life cycle, the filament current shows an overall attenuation trend). For this type of data with large jumps, the prediction accuracy may be very low. In fact, it is not necessary to predict the remaining life of the tube filament based on the prediction of the filament current with an accuracy of seconds or hours. Therefore, it can be processed by taking the average on a daily basis, and the prediction is made in units of days. The collected initial current data is grouped by day (preset time period) to obtain multiple groups of data to be calculated, and then the weighted average of each group of data to be calculated is taken to obtain multiple data to be processed.

[0043] For example, suppose the data for a day is (that is, the data to be calculated), the average current value of the day after processing (that is, the data to be processed) is: (Formula 1) Step 3: Data standardization Data standardization is an important step in data preprocessing. Data standardization can eliminate dimensionality effects, speed up convergence, improve model performance, and avoid numerical instability, thus preparing for subsequent model training and testing.

[0044] For example, standardization is to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1. Assume that the data to be processed is , the standardized data is , the formula is: (Formula 2) in, is the mean of all the data to be processed, is the standard deviation of all the data to be processed. Through this process, all the data to be processed will be adjusted to the same scale, ready for subsequent model training and testing.

[0045] Step 4: Sliding Window Processing Sliding window processing is performed on multiple standardized data. The core idea of ​​sliding window processing is to slide a fixed-size window on the data sequence and gradually process subsets of the data sequence. Figure 2 As shown in Figure 1, it is a schematic diagram of sliding window processing for multiple standardized data. Figure 2 It can be seen that given a length of Data series (that is, multiple standardized data), select Length window and Length window, select sliding step , the sliding window operation will construct multiple historical sequences and label sequences. The calculation formula for the number of samples obtained after sliding the window is as follows: (Formula 3) For example, The input sequence is , No. The tag sequence is In this way, sliding window processing converts the original data into multiple samples suitable for model training and prediction, enabling the model to learn the temporal dependencies and trends in the data sequence.

[0046] Step 5: Divide the training set and test set Any initial current data is divided, and a small part of the front part and other initial current data are used as a training set for training the model, and the rest of the data is used as a test set to evaluate the generalization performance of the model.

[0047] For example, the specific method of dividing the training set and the test set can be: first take a set of initial current data, merge a small part of the front part with all other initial current data as the training set for training the model; then, take the remaining data of this set of initial current data as the test set to evaluate the generalization performance of the model. This division method ensures that the model can fully learn the characteristics of each tube during the training process, and can accurately evaluate the performance of the model on unseen data during the test.

[0048] Step S112: training the initial prediction model based on the current sample data set to obtain a filament current prediction model.

[0049] Among them, the filament current prediction model is used to predict the changing trend of the filament current.

[0050] In one embodiment of the present application, the current sample data set may include historical current sample data corresponding to the sample bulb, and the initial prediction model may be trained based on the current sample data set to obtain a filament current prediction model. The initial prediction model may include a first prediction model, a second prediction model, and a third prediction model, and based on a self-attention mechanism, the first prediction result output by the first prediction model, the second prediction result output by the second prediction model, and the third prediction result output by the third prediction model may be first spliced ​​to obtain model predicted current data. The loss function of the initial prediction model may be a mean-square error (MSE), and the loss function may quantify the difference between the current data predicted by the initial prediction model and the actual current.

[0051] Exemplarily, the formula for mean square error is as follows: (Formula 4) in, is the number of samples, It is The true value of the samples, It is The predicted value of a sample.

[0052] MSE measures the difference between the model prediction and the actual result by calculating the square of all prediction errors and taking the average. This formula can help minimize large errors when training the model and improve the overall prediction effect.

[0053] In one embodiment, the process of step S112 of training the initial prediction model based on the current sample data set to obtain the filament current prediction model may include steps S1121 to S1124.

[0054] Step S1121: input historical current sample data into a first prediction model to obtain a first prediction result.

[0055] The first prediction model can be used to capture the dependency relationship of historical current sample data.

[0056] Exemplarily, the first prediction model may be an LSTM (Long Short-Term Memory) model, which can effectively capture long-term dependencies in sequence data. The first prediction model may also be a Gated Recurrent Unit (GRU) or a Transformer model.

[0057] Step S1122, input the historical current sample data into the second prediction model to obtain a second prediction result. The second prediction model is used to perform trend quantity prediction and residual quantity prediction on the historical current sample data after data decomposition, and obtain the second prediction result based on the trend quantity prediction and the residual quantity prediction.

[0058] In one embodiment of the present application, historical current sample data may be input into a second prediction model to obtain a second prediction result. The second prediction result obtained based on trend quantity prediction and residual quantity prediction is more comprehensive in feature extraction.

[0059] Exemplarily, the second prediction model can be a DLinear model. The DLinear model can decompose the historical current sample data into trend quantities and residual quantities. This decomposition helps the model better understand the different components in the sequence. The trend quantity can capture the long-term trend of change, while the residual quantity can capture short-term fluctuations or noise. The decomposed trend quantity and residual quantity are modeled independently. This allows the model to learn the characteristics of each component without interfering with each other, which helps to capture the different characteristics of the sequence more accurately.

[0060] It should be noted that the DLinear (Decomposition-Linear) model is a time series prediction method that decomposes data into trend and residual quantities and then performs linear processing on these components for prediction. The specific formula is as follows: ① Data decomposition: Assume that the input historical current sample data is ,Will Decomposed into trend volume and the residual .

[0061] (Formula 5) ②Trend volume prediction: Perform linear prediction and set the linear model weight to , biased to .

[0062] (Formula 6) ③ Residual prediction: Perform linear prediction and set the linear model weight to , biased to .

[0063] (Formula 7) ④ Combined prediction: Combine the prediction results of trend quantity and residual quantity to obtain the final second prediction result.

[0064] (Formula 8) Step S1123: input the historical current sample data into the third prediction model to obtain a third prediction result.

[0065] The third prediction model is used to analyze data within a preset data length in the historical current sample data.

[0066] In one embodiment of the present application, considering that the degree of data degradation increases in the later period, a third prediction model is introduced to pay more attention to the later period data.

[0067] Exemplarily, the third prediction model may be a Lookback model. The preset data length may be a Lookback value, and the Lookback value may indicate how many time steps of input information the Lookback model needs to consider when predicting the output.

[0068] Step S1124: Based on the self-attention mechanism, the first prediction result, the second prediction result, and the third prediction result are first concatenated to obtain model predicted current data.

[0069] In one embodiment of the present application, the self-attention mechanism is a method used in deep learning to capture the dependencies between elements in an input sequence. Self-Attention generates a weighted sum by calculating the correlation between each element in the input sequence and all other elements. The specific formula is as follows: ① Input representation: Assume that the input sequence is ,in is the first elements.

[0070] ②Linear transformation: transform the input sequence Mapping to query vector , key vector Sum value vector .

[0071] (Formula 9) in, is the weight matrix that needs to be learned.

[0072] ③Calculate attention weight: Calculate query vector through dot product and key vector similarity, and divided by the scaling factor To stabilize the gradient, use The function converts similarity into attention weights.

[0073] (Formula 10) in, is the dimension of the key vector. is the key vector The transpose of .

[0074] ④Calculation output: value vector Perform weighted sum to get the final output.

[0075] (Formula 11) The Self-Attention mechanism can generate a weighted sum by calculating the correlation between each element in the input sequence and all other elements, thereby capturing the dependencies between elements in the sequence.

[0076] In an exemplary embodiment of the present application, the first prediction model may be an LSTM model, the second prediction model may be a DLinear model, and the third prediction model may be a Lookback model. The process of obtaining the model prediction current data based on the historical current sample data by the initial prediction model may include: Input historical current sample data into the LSTM model to generate the first prediction result .

[0077] (Formula 12) in, It is the historical current sample data.

[0078] The DLinear model decomposes historical current sample data into trend quantity and residual quantity, and improves the prediction accuracy of the model by independently processing these two parts of features.

[0079] (Formula 13) (Formula 14) (Formula 15) (Formula 16) in, is the second prediction result.

[0080] Use the Lookback model to analyze the changes in the data in the later period and generate the third prediction results .

[0081] (Formula 17) The prediction results generated by the LSTM model, DLinear model and Lookback model are spliced ​​together, and the attention weight of each part is calculated using the self-attention mechanism to generate the model prediction current data .

[0082] (Formula 18) Through the above steps and methods, the model can effectively predict the future change trend of CT tube filament current and help make scientific decisions on maintenance and replacement.

[0083] In addition, it should be noted that when training the initial prediction model, the loss function of the initial prediction model can be mean-square error (MSE), and the gradient descent algorithm can be used to iteratively adjust the initial prediction model parameters to minimize the value of the loss function.

[0084] Step S120, obtaining target predicted current data based on the filament current data and the filament current prediction model.

[0085] The target predicted current data may include multiple predicted filament currents of the current target tube, the multiple predicted filament currents are arranged in chronological order, and the target predicted current data is used to characterize the change trend of the filament current of the current target tube within the predicted time.

[0086] In one embodiment of the present application, the filament current data may be input into the filament current prediction model to obtain the target predicted current data. The filament current prediction model is the same as the initial prediction model, and both may include a first prediction model, a second prediction model, a third prediction model, and a self-attention mechanism.

[0087] In an exemplary embodiment, the process of obtaining the target predicted current data based on the filament current data and the filament current prediction model in step S120 may include steps S121 to S124.

[0088] Step S121, determining the filament current data as input data, and executing a data input step, the data input step including inputting the input data into a filament current prediction model to obtain initial predicted current data.

[0089] In one embodiment of the present application, when current prediction is performed based on the filament current data and the filament current prediction model, the filament current data may be first determined as input data, and a data input step may be performed. The initial predicted current data may characterize the current trend of the current target tube within a period of time in the future, the length of the initial predicted current data may be the length of the prediction time, and the length of the initial predicted current data may be greater than 0 and less than or equal to the length of the input data.

[0090] For example, if a certain data in the initial predicted current data represents a predicted current value for a certain day, the initial predicted current can represent the current trend of the current target tube in the next few days.

[0091] Step S122, executing a data judgment step, the data judgment step includes determining whether there is target data in the initial predicted current data, and the target data is less than or equal to the failure current threshold.

[0092] In one embodiment of the present application, after the initial predicted current data is obtained, a data judgment step may be performed to determine whether there is target data less than or equal to the failure current threshold in the initial predicted current data.

[0093] Step S123, if the failure current threshold does not exist in the initial predicted current data, a data updating step is executed, the data updating step includes performing a second splicing of the initial predicted current data and the input data to obtain spliced ​​data; determining the spliced ​​data as input data, and executing the data input step again.

[0094] In one embodiment of the present application, if there is no target data in the initial predicted current data, the failure time of the current target tube has not been predicted, and the filament current of the current target tube needs to be predicted again.

[0095] It should be noted that the initial predicted current data and the input data are concatenated for the second time and the current prediction is performed again, so that the consistency of the prediction sequence can be maintained.

[0096] Step S124: if target data exists in the initial predicted current data, target predicted current data is determined based on all the initial predicted current data.

[0097] In one embodiment of the present application, when there is a failure current threshold in the initial predicted current data, it can be determined that the remaining life of the current target tube has been predicted, and then all predicted current data can be spliced ​​in time sequence to obtain target predicted current data. Each data in the target predicted current data can represent a predicted current at a predicted time point.

[0098] For example, after the initial prediction model training is completed and the filament current prediction model is obtained, the prediction of the filament current of the CT tube can be carried out. A recursive prediction method is adopted. For a new tube, data can be collected within 30 days from its use as the input of the filament current prediction model. , to predict the filament current in the later period of time , and reuse the predicted data and historical real data as input to the model again , and then predict the data for a period of time in the future, and repeat this process to reach the required target prediction length. As shown below: (Formula 19) Among them, p is the model prediction length, . p can be equal to .

[0099] Step S130, determining the current target tube life according to the failure current threshold and the target predicted current data.

[0100] In one embodiment of the present application, target data less than or equal to the failure current threshold may be searched in the target predicted current data, and the current target tube life may be determined based on the ranking of the target data in the target predicted current data.

[0101] For example, each data in the target predicted current data represents the predicted current for a certain day, and the ranking of the target data in the target predicted current data may be 50, and it can be determined that the current target tube life is 50 days.

[0102] It should be noted that the embodiment of the present application constructs a filament current prediction model with a self-attention mechanism. The model integrates LSTM, DLinear, Lookback and Self-Attention to achieve high-precision prediction of the filament current of the CT tube. The LSTM model solves the problem of long-term dependence, the DLinear model comprehensively extracts data features, the Lookback model focuses on late data degradation, and the self-attention mechanism optimizes the weights of each part. By integrating these advantages, the filament current prediction model significantly improves the prediction accuracy and robustness, provides a scientific basis for equipment maintenance, reduces downtime, and improves operating efficiency.

[0103] In summary, the scheme of the embodiment of the present application obtains the filament current prediction model, the failure current threshold of the current target tube, and the filament current data, obtains the target predicted current data based on the filament current data and the filament current prediction model, and determines the life of the current target tube according to the failure current threshold and the target predicted current data. Based on the filament current data of the current target tube and the filament current prediction model, the filament current change trend of the current target tube within the prediction time can be predicted, and the life of the tube can be predicted in combination with the failure current threshold and the target predicted current data. By predicting the life of the tube through the filament current, which is a key factor affecting the life of the tube, the accuracy of the prediction can be improved, equipment shutdown caused by sudden failure of the tube can be avoided, the reliability of the tube operation can be improved, and the operating cost can be reduced.

[0104] Figure 3 FIG. 1 is a block diagram of a device for predicting the life of a tube according to an exemplary embodiment of the present application. Figure 3 As shown, the exemplary tube life prediction device 300 includes: The data acquisition module 310 is used to acquire the filament current prediction model, the failure current threshold of the current target tube, and the filament current data.

[0105] The data prediction module 320 is used to obtain target predicted current data based on the filament current data and the filament current prediction model. The target predicted current data includes multiple predicted filament currents of the current target tube. The multiple predicted filament currents are arranged in chronological order. The target predicted current data is used to characterize the change trend of the filament current of the current target tube within the prediction time.

[0106] The life prediction module 330 is used to determine the current target tube life according to the failure current threshold and the target predicted current data.

[0107] In a possible implementation, the data acquisition module may also be used to: Obtaining a current sample data set and an initial prediction model; The initial prediction model is trained based on the current sample data set to obtain a filament current prediction model, which is used to predict the change trend of the filament current.

[0108] In a possible implementation, the data acquisition module may also be used to: Acquire initial current data, the initial current data including filament current data corresponding to a plurality of sample bulbs, each filament current data including a plurality of historical filament currents sorted by time; Performing data standardization processing on multiple historical filament currents to obtain multiple standardized data; Sliding window processing is performed on the multiple standardized data to obtain multiple historical current sample data, and the multiple historical current sample data constitute a current sample data set.

[0109] In a possible implementation, the data acquisition module may also be used to: The filament current data are grouped according to a preset time period to obtain multiple groups of data to be calculated; Performing mean filtering on the historical filament current of each group of data to be calculated to obtain a plurality of data to be processed; Data standardization is performed on each piece of data to be processed to obtain a plurality of standardized data.

[0110] In a possible implementation, the current sample data set includes historical current sample data corresponding to the sample tube, and the data acquisition module may also be used to: Inputting historical current sample data into a first prediction model to obtain a first prediction result, wherein the first prediction model is used to capture the dependency relationship of the historical current sample data; Inputting the historical current sample data into the second prediction model to obtain a second prediction result, the second prediction model is used to perform trend quantity prediction and residual quantity prediction respectively after data decomposition of the historical current sample data, and obtain the second prediction result based on the trend quantity prediction and the residual quantity prediction; Inputting the historical current sample data into a third prediction model to obtain a third prediction result, wherein the third prediction model is used to analyze data within a preset data length in the historical current sample data; Based on the self-attention mechanism, the first prediction result, the second prediction result, and the third prediction result are first concatenated to obtain the model predicted current data.

[0111] In a possible implementation, the data prediction module may also be used to: Determine the filament current data as input data, and perform a data input step, the data input step including inputting the input data into a filament current prediction model to obtain initial predicted current data; Executing a data judgment step, the data judgment step includes determining whether there is target data in the initial predicted current data, and the target data is less than or equal to a failure current threshold; If the target data does not exist in the initial predicted current data, a data updating step is performed, the data updating step including performing a second splicing of the initial predicted current data and the input data to obtain spliced ​​data; determining the spliced ​​data as the input data, and performing the data input step again; If target data exists in the initial predicted current data, the target predicted current data is determined based on all the initial predicted current data.

[0112] In a possible implementation, the data acquisition module may also be used to: Get the target initial current of the current target tube; The product of the target initial current and the preset value is determined as the failure current threshold.

[0113] In a possible implementation manner, the data acquisition module may also be used to: Obtain the current initial current of the current target tube and the historical initial current of the historical target tube; The current initial current and the historical initial current are weightedly summed to obtain the target initial current.

[0114] It should be noted that the tube life prediction device provided in the above embodiment and the tube life prediction method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiment, and will not be repeated here. In practical applications, the tube life prediction device provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the system into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0115] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by one or more processors, the electronic device implements the tube life prediction method provided in the above-mentioned embodiments.

[0116] Another aspect of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor of a computer, the computer executes the tube life prediction method provided in the above embodiments. The computer-readable storage medium may be included in the electronic device described in the above embodiments, or may exist independently without being assembled into the electronic device.

[0117] Another aspect of the present application also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the tube life prediction method provided in each of the above embodiments.

[0118] In the embodiments of the present application, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. "Including" and "comprising" mentioned throughout the specification and claims are open-ended terms and should be interpreted as "including but not limited to".

[0119] The above embodiments are merely illustrative of the principles and effects of the present application, and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A method for predicting tube life, characterized in that: include: Obtain the filament current prediction model, the failure current threshold of the current target tube, and the filament current data; Based on the filament current data and the filament current prediction model, target predicted current data are obtained, wherein the target predicted current data includes a plurality of predicted filament currents of the current target bulb, the plurality of predicted filament currents are arranged in chronological order, and the target predicted current data are used to characterize a change trend of the filament current of the current target bulb within a prediction time; The current target tube life is determined according to the failure current threshold and the target predicted current data.

2. The method for predicting tube life according to claim 1, characterized in that: Get the filament current prediction model, including: Obtaining a current sample data set and an initial prediction model; The initial prediction model is trained based on the current sample data set to obtain a filament current prediction model, and the filament current prediction model is used to predict the change trend of the filament current.

3. The method for predicting tube life according to claim 2, characterized in that: Get the current sample data set, including: Acquire initial current data, the initial current data including filament current data corresponding to a plurality of sample bulbs, each of the filament current data including a plurality of historical filament currents sorted by time; Performing data standardization processing on the plurality of historical filament currents to obtain a plurality of standardized data; Sliding window processing is performed on the multiple standardized data to obtain multiple historical current sample data, and the multiple historical current sample data constitute the current sample data set.

4. The method for predicting tube life according to claim 3, characterized in that: Performing data standardization processing on the plurality of historical filament currents to obtain a plurality of standardized data, including: The filament current data are grouped according to a preset time period to obtain multiple groups of data to be calculated; Performing mean filtering processing on the historical filament current of each group of the data to be calculated to obtain a plurality of data to be processed; Data standardization is performed on each of the data to be processed to obtain a plurality of standardized data.

5. The method for predicting the life of a tube according to any one of claims 2 to 4, characterized in that: The current sample data set includes historical current sample data corresponding to the sample tube, and training the initial prediction model based on the current sample data set includes: Inputting the historical current sample data into a first prediction model to obtain a first prediction result, wherein the first prediction model is used to capture the dependency relationship of the historical current sample data; Inputting the historical current sample data into a second prediction model to obtain a second prediction result, wherein the second prediction model is used to perform trend quantity prediction and residual quantity prediction respectively after data decomposition of the historical current sample data, and obtain the second prediction result based on the trend quantity prediction and the residual quantity prediction; Inputting the historical current sample data into a third prediction model to obtain a third prediction result, wherein the third prediction model is used to analyze data within a preset data length in the historical current sample data; Based on the self-attention mechanism, the first prediction result, the second prediction result, and the third prediction result are first concatenated to obtain model predicted current data.

6. The method for predicting tube life according to claim 1, characterized in that: Based on the filament current data and the filament current prediction model, target predicted current data is obtained, including: Determine the filament current data as input data, and perform a data input step, wherein the data input step includes inputting the input data into the filament current prediction model to obtain initial predicted current data; Executing a data judgment step, the data judgment step comprising determining whether there is target data in the initial predicted current data, the target data being less than or equal to the failure current threshold; If the target data does not exist in the initial predicted current data, a data updating step is performed, wherein the data updating step includes performing a second splicing of the initial predicted current data and the input data to obtain spliced ​​data; determining the spliced ​​data as the input data, and performing the data input step again; If the target data exists in the initial predicted current data, the target predicted current data is determined based on all the initial predicted current data.

7. The method for predicting tube life according to claim 1, characterized in that: Get the failure current threshold of the current target tube, including: Obtaining a target initial current of the current target tube; The product of the target initial current and a preset value is determined as the failure current threshold.

8. A device for predicting the life of a tube, characterized in that: include: A data acquisition module is used to obtain a filament current prediction model, a failure current threshold of a current target tube, and filament current data; A data prediction module, used for obtaining target predicted current data based on the filament current data and the filament current prediction model, wherein the target predicted current data includes a plurality of predicted filament currents of the current target bulb, the plurality of predicted filament currents are arranged in chronological order, and the target predicted current data is used for characterizing a change trend of the filament current of the current target bulb within a predicted time period; The life prediction module is used to determine the current target tube life according to the failure current threshold and the target predicted current data.

9. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to execute the tube life prediction method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the tube life prediction method according to any one of claims 1 to 7.

Images