X-ray tube and filament performance degradation prediction and failure discrimination method and system

Abstract

The invention relates to the technical field of X-ray tubes, and discloses an X-ray tube and a filament performance degradation prediction and failure discrimination method and system, and the method comprises the steps: obtaining the filament data of the X-ray tube, and constructing training sample data and test sample data; predicting the predicted current of the filament through a degradation prediction model, and predicting the predicted current distribution of the filament through an uncertainty prediction model, the uncertainty prediction model being constructed based on at least two degradation prediction models; calculating the uncertainty of the predicted current; calculating the failure range of the filament; in response to the degradation rate of the filament exceeding the failure range at a certain moment, determining the moment as a failure moment; and calculating the failure time of the filament based on the failure moment and the uncertainty. In the embodiment of the invention, based on the uncertainty, the reliability degree of the predicted current of the filament can be judged, the failure time of the filament is calculated according to the uncertainty, and the accuracy of predicting the failure time of the filament is further improved.

Description

Technical Field

[0001] The present application relates to the technical field of X-ray tubes, and in particular to a method and system for predicting performance degradation and failure identification of X-ray tubes and filaments. Background Art

[0002] Computed tomography (CT) is an important medical device in the diagnosis and treatment process of hospitals. The X-ray tube is the core component of CT, which directly affects the imaging quality and diagnostic accuracy of CT. Due to strict clinical working conditions, such as long-term exposure to high voltage, rotation and non-standard operation, the components of the X-ray tube, such as the anode target, bearings and cathode filament, are prone to failure, resulting in X-ray tube failure, which poses great risks during critical surgical and other treatment processes.

[0003] When using an X-ray tube, the cathode filament is heated to generate free electrons, which collide with the metal target to produce X-rays. Long-term use will cause the filament to evaporate and become thinner, resulting in filament failure and X-ray tube failure.

[0004] In response to the above problems, the relevant technology proposes to determine the filament failure by measuring parameters such as filament resistance and predicting the filament failure time through models, but the relevant methods lack test evaluation of the prediction results of the model output, and it is difficult to ensure the accuracy of the output data. Furthermore, there are many reasons for the failure of X-ray tubes, and different tubes have different degrees of influence on the degradation process, causes and speed of the filament due to differences in their own structures, service environments and other factors, resulting in fuzzy filament degradation characteristics. Therefore, the relevant technology cannot reasonably predict performance degradation and failure judgment for the problem of X-ray tube failure caused by filament degradation. Summary of the invention

[0005] The present application provides a method and system for predicting performance degradation and failure identification of an X-ray tube and a filament, which can at least be used to solve the problem of low accuracy in predicting the failure time of a filament in an X-ray tube.

[0006] In a first aspect, an embodiment of the present application provides a method for predicting performance degradation and failure identification of an X-ray tube filament, comprising the following steps: acquiring filament data of an X-ray tube, and constructing training sample data and test sample data; based on the training sample data, predicting a predicted current of the filament through a degradation prediction model, and predicting a predicted current distribution of the filament through an uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two of the degradation prediction models; calculating the uncertainty of the predicted current based on the predicted current distribution and the predicted current; calculating a failure range of the filament; in response to the degradation rate of the filament exceeding the failure range at a certain moment, determining the moment as a failure moment; and calculating the failure time of the filament based on the failure moment and the uncertainty.

[0007] The embodiment of the present application provides a method for predicting performance degradation and determining failure of an X-ray tube, comprising the following steps: executing the aforementioned method for predicting performance degradation and determining failure of an X-ray tube filament; in response to the degradation rate of the filament at any time not exceeding the failure range, excluding the failure cause of the X-ray tube as filament failure.

[0008] The embodiment of the present application provides an X-ray tube filament performance degradation prediction and failure discrimination system, comprising: an acquisition module, configured to acquire filament data of an X-ray tube, and construct training sample data and test sample data; a prediction module, configured to predict the predicted current of the filament through a degradation prediction model based on the training sample data, and predict the predicted current distribution of the filament through an uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two of the degradation prediction models; a calculation module, configured to calculate the uncertainty of the predicted current based on the predicted current distribution and the predicted current; calculate the failure range of the filament; a determination module, configured to determine that the moment is a failure moment in response to the degradation rate of the filament exceeding the failure range at a certain moment; and calculate the failure time of the filament based on the failure moment and the uncertainty.

[0009] An embodiment of the present application provides an X-ray tube performance degradation prediction and failure determination system, comprising: an X-ray tube filament performance degradation prediction and failure determination system as described above; and a second determination module, configured to exclude the failure cause of the X-ray tube as filament failure in response to the degradation rate of the filament at any time not exceeding the failure range.

[0010] An embodiment of the present application provides a computer-readable storage medium, which stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer runs a method for predicting and distinguishing performance degradation and failure of an X-ray tube filament and a method for predicting and distinguishing performance degradation and failure of an X-ray tube.

[0011] The technical solution disclosed in this application includes the following beneficial effects: In the method for predicting and distinguishing the performance degradation and failure of an X-ray tube filament disclosed in the embodiment of the present application, the predicted current of the current filament is obtained by predicting the degradation prediction model based on the training sample data. Since the predicted current is a single data, it is impossible to determine whether the prediction result of the degradation prediction model is accurate based on it. Therefore, on this basis, the embodiment of the present application constructs an uncertainty prediction model, performs multiple predictions based on different degradation prediction models, and forms a predicted current distribution based on the predicted current obtained by the multiple predictions. Therefore, the predicted current distribution can more comprehensively reflect the prediction results of the current filament current by the degradation prediction model in different states and at different training levels. Based on the predicted current distribution, it is possible to determine whether the predicted current obtained by the degradation prediction model is credible in turn, and obtain the uncertainty of the current filament prediction current. Secondly, the degradation prediction model can also be trained based on the predicted current distribution, thereby continuously improving the prediction ability of the degradation prediction model and further improving the accuracy of the degradation prediction model prediction. In addition, the failure time is corrected based on the above uncertainty, the failure time of the filament is calculated, and the final predicted failure time is adaptively adjusted according to the accuracy of the model prediction results. This can further optimize the failure time prediction process, improve the intelligence and automation of filament failure prediction, make the predicted filament failure time closer to the true value, and make the prediction result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.

[0013] In the attached picture: Figure 1 An exemplary flow chart of a method for predicting performance degradation and failure determination of an X-ray tube filament disclosed in some embodiments of the present application; Figure 2 A schematic diagram of the calculation uncertainty disclosed for some embodiments of the present application; Figure 3 A schematic diagram of determining the failure moment disclosed in some embodiments of the present application; Figure 4 A module diagram of an X-ray tube filament performance degradation prediction and failure determination system disclosed in some embodiments of the present application; FIG5( a ) is a schematic diagram of the original filament current disclosed in some embodiments of the present application; FIG5( b ) is a schematic diagram of the pre-processed filament current disclosed in some embodiments of the present application; Figure 6 A schematic diagram showing a comparison between a predicted current and an actual current predicted by a degradation prediction model disclosed in some embodiments of the present application; Figure 7 This is a schematic diagram showing the change of first-order difference data over time according to some embodiments of the present application. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0015] In order to facilitate understanding of the X-ray tube and filament performance degradation prediction and failure judgment method and system provided in the embodiments of the present application, the relevant technologies are first introduced in combination with application scenarios.

[0016] X-ray tubes are important components of medical equipment that use X-rays and computer technology to generate images to quickly and accurately diagnose various diseases and injuries. Due to strict clinical working conditions, such as long-term exposure to high voltage, rotation, and non-standard operation, X-ray tubes are prone to malfunction and failure, resulting in interruptions in use.

[0017] There are various reasons for the failure of X-ray tubes, such as filament failure. In the prior art, the method for determining the failure of X-ray tube filaments usually uses the measured filament resistance as a parameter and performs failure determination through a big data model. However, due to the diverse service environments of X-ray tubes, the working parameters change all the time, and the structural parameters of different X-ray tubes are different, resulting in the data of the big data model being messy and unable to measure the reliability of its determination results.

[0018] In view of this, the embodiments of the present application provide a method and system for predicting the performance degradation and failure judgment of an X-ray tube and a filament. Figure 1-Figure 7 Various embodiments disclosed in this application are described.

[0019] Figure 1 This is an exemplary flow chart of the method for predicting performance degradation and failure identification of an X-ray tube filament disclosed in some embodiments of the present application. Process 100 is executed by a system 400 for predicting performance degradation and failure identification of an X-ray tube filament.

[0020] Step 110, obtaining the filament data of the X-ray tube, and constructing training sample data and test sample data.

[0021] Filament data refers to monitoring data related to the filament in the X-ray tube.

[0022] In some embodiments, the filament data may include multiple parameters. Exemplarily, the filament data includes filament current, scanning frequency, focus, and tube voltage. In some embodiments, the filament data may be obtained in a variety of ways. For example, based on an existing medical information system, filament data corresponding to an X-ray tube replaced due to filament failure is collected.

[0023] Training sample data refers to sample data used for model training, and test sample data refers to sample data used to test the model prediction results.

[0024] In some embodiments, based on the filament data, the training sample data and the test sample data can be divided in a variety of ways. For example, according to a certain ratio, a part of the sample data is divided into the training sample data, and the remaining sample data is divided into the test sample data.

[0025] In some embodiments, constructing training sample data and test sample data by acquiring filament data of an X-ray tube also includes: acquiring filament data of N X-ray tubes; preprocessing the filament data; constructing training sample data based on the preprocessed filament data of N-1 X-ray tubes and the first sample data of the preprocessed filament data of the N-th X-ray tube; and constructing test sample data based on other sample data in the preprocessed filament data of the N-th X-ray tube except the first sample data.

[0026] In some embodiments, preprocessing the filament data includes the following steps: First, the filament data of the X-ray tube is obtained, and the mode of the scanning frequency, focus and tube voltage of each X-ray tube in the use process is calculated in the filament data. The mode of the scanning frequency, focus and tube voltage is used as a filtering condition to filter the filament data of each X-ray tube.

[0027] For example, the modes of the scanning frequency, focus and tube voltage of a certain X-ray tube are: scanning frequency 120Hz, focus is a small focus, and tube voltage 120kV. Then the filtering conditions are: scanning frequency is 120Hz, focus is a small focus, and tube voltage is 120kV. During the use of the X-ray tube, the filament data that meets the above filtering conditions is retained, and the rest of the filament data is removed.

[0028] The filament current in the filtered filament data is interpolated by an interpolation method, and the interpolated filament current is normalized.

[0029] As shown in FIG. 5( a ) and FIG. 5( b ), FIG. 5( a ) is the original filament current data without preprocessing, and FIG. 5( b ) is the filament current after interpolation and normalization.

[0030] Furthermore, the normalized filament current is constructed as sample data through sliding window slicing, and the sample data is divided into training sample data and test sample data.

[0031] Specifically, the interpolated and normalized filament current of each X-ray tube can be expressed as , express The filament current at time Indicates the life span of the filament.

[0032] In some embodiments, the sample data includes input samples and output samples, wherein the input samples refer to sample data input to the model, and the output samples refer to actual filament currents.

[0033] In some embodiments, the aforementioned model includes a degradation prediction model and an uncertainty prediction model. For more information about the degradation prediction model and the uncertainty prediction model, please refer to the following content.

[0034] Specifically, the input sample can be expressed as , the output sample can be expressed as That is, the input sample is Time has come Between moments, The filament current at the moment, the output sample is The actual measured filament current at that moment.

[0035] In some embodiments, Moments can include various time units, such as hours, days, weeks, months, etc. Specifically, The smaller the unit of time (that is, the shorter the interval between adjacent moments), the closer the sample data that can be obtained are, and the smaller the degree of data change; conversely, the more separated the sample data obtained are, the greater the degree of data change.

[0036] In the specific implementation process, if the sample data are too close or too separated, the degradation characteristics of the filament in the data sample will be blurred. Preferably, in order to ensure the validity of the sample data, The unit can be days.

[0037] In some embodiments, when When used to represent a period of time, it can be determined in a variety of ways For example, the filament current at a certain time in a day can be determined as the filament current of the day. For another example, the filament current at a certain time in a day can be determined as the filament current of the day.

[0038] In some embodiments, all data samples for each X-ray tube may be represented as: ; in, Represents each X-ray tube, Indicates the number of X-ray tubes, Indicates All data samples of an X-ray tube.

[0039] To ensure The time corresponding to each sample in is the time after the X-ray tube starts to be used, that is, The time must be greater than the initial time 1, so, , so we can get Start taking sample data of the X-ray tube at time . Among them, is the life span of the X-ray tube filament. Therefore, all data samples of each X-ray tube are Time to All input samples and output samples at time.

[0040] In some embodiments, in order to verify the effectiveness of the X-ray tube filament performance degradation prediction and failure judgment method, N-1 of the N X-ray tubes collected are used as training X-ray tubes, and the Nth X-ray tube is used as the X-ray tube to be tested. The training sample data includes all sample data of the N-1 X-ray tubes and the first sample data of the Nth X-ray tube, which can be expressed as ; The test sample data includes the Nth X-ray tube, except for the first sample data All other sample data except ,Right now .

[0041] Due to the diverse usage scenarios of X-ray tubes, in different operations, the tube voltage scanning frequency and focus parameters used in X-ray tubes are also changing all the time. Due to the differences in their service environments, they also have different effects on the loss of the filaments of the X-ray tubes. Therefore, when predicting the failure time of the filament, the data should be controlled under the same service conditions as much as possible to avoid the influence of interference data on the prediction results. The embodiment of the present application determines the filtering conditions by the mode of the scanning frequency, focus and tube voltage, and filters the filament data, so as to obtain as much data as possible, and at the same time removes the filament current corresponding to different usage parameters, thereby improving the purity of the data, avoiding interference data from reducing the accuracy of the prediction results, and facilitating analysis and processing. At the same time, reducing the amount of data is also conducive to reducing the amount of calculation and improving calculation efficiency. At the same time, normalizing, interpolating and sliding window slicing the data can further summarize and screen the data, and summarize the huge and complex data into sample data that can be quantified and calculated, simplifying the calculation while ensuring the accuracy of the data. In addition, since different X-ray tubes have different structures and processing methods, each X-ray tube has its own characteristics. It is difficult to accurately control the impact of the characteristics of the X-ray tube to be tested on the accuracy of the prediction results by only using the remaining X-ray tubes as sample data. Therefore, by classifying the first sample data of the Nth X-ray tube (the X-ray tube to be tested) into the training sample data, the training sample data can be further made more real and reliable based on the characteristics of the X-ray tube itself and combined with universal data, thereby improving the accuracy of the model prediction results.

[0042] Step 120 , based on the training sample data, predict the predicted current of the filament using the degradation prediction model, and predict the predicted current distribution of the filament using the uncertainty prediction model.

[0043] The predicted current refers to the filament current predicted by the degradation prediction model. The predicted current distribution refers to the collection of serialized predicted current data.

[0044] The degradation prediction model is a model for predicting the filament current of the X-ray tube. In some embodiments, the degradation prediction model is a machine learning model, such as a convolutional neural network, a recurrent neural network, a fully connected neural network, an attention mechanism, etc.

[0045] In some embodiments, the input of the degradation prediction model includes input samples in the training sample data , the output includes the predicted current at time t .

[0046] In some embodiments, the degradation prediction model is expressed as: ,in, Representation Model The parameters can be preset by staff in this field.

[0047] In some embodiments, the degradation prediction model can be trained based on a large number of first training samples with first labels through various feasible training methods, such as step-down method.

[0048] The first training sample can be an input sample , the first label can be the output sample .

[0049] In some embodiments, the uncertainty prediction model is a model for predicting the filament current distribution of the X-ray tube. In some embodiments, the uncertainty prediction model is a machine learning model, such as a convolutional neural network, a recurrent neural network, a fully connected neural network, an attention mechanism, etc.

[0050] In some embodiments, the input of the uncertainty prediction model includes the input sample , the output includes the predicted current distribution at time t.

[0051] The training method of the uncertainty prediction model is the same as that of the degradation prediction model and will not be described in detail here.

[0052] For more information on uncertainty prediction model construction, see Figure 2 Detailed description.

[0053] Step 130 , calculating uncertainty of the predicted current based on the predicted current distribution and the predicted current.

[0054] For more information about step 130, see Figure 2 Related content.

[0055] Step 140, calculating the failure range of the filament.

[0056] The failure range refers to the degradation rate of the filament exceeding the allowable range of the standard value. In some embodiments, the failure range of the filament can be determined based on the training sample data in a variety of ways. For example, it can be preset based on the experience of the staff in the field.

[0057] In some embodiments, the failure range can be obtained by calculation. First, based on the training sample data, the first-order difference data of the predicted current distribution at two adjacent moments are calculated.

[0058] Specifically, each input sample in the training data samples is input into the uncertainty prediction model to obtain the predicted current distribution corresponding to each input sample, and the first-order difference of the predicted current distribution of the input samples at every two adjacent moments is calculated.

[0059] The first-order difference data refers to a data set of the first-order difference of the predicted current distribution of the input samples at every two adjacent moments.

[0060] In some embodiments, the first-order difference data can be calculated based on formula (1): (1) in, for The predicted current distribution at time, for Predicted current distribution at time instant. Time is The moment before the moment. and It can be predicted by the uncertainty prediction model.

[0061] Next, the first-order difference data is filtered Filtering is performed to obtain a reference degradation rate of the filament.

[0062] The reference degradation rate refers to a set of degradation rates of the filaments obtained by filtering based on the training sample data.

[0063] In some embodiments, the degradation rate It can be calculated based on formula (2): (2) in, is the filtering coefficient, which can be determined based on experience by those skilled in the art.

[0064] The filter is a filter circuit composed of a capacitor, an inductor and a resistor. In some embodiments, the filter used in this embodiment can be of various types, such as an infinite impulse response filter.

[0065] Then, the degradation mean value and degradation standard deviation of the reference degradation rate are calculated. The degradation mean value refers to the average value of the reference degradation rate, and the degradation standard deviation refers to the standard deviation of the reference degradation rate.

[0066] The failure range is calculated based on the degradation mean, degradation standard deviation and failure threshold coefficient.

[0067] In some embodiments, the failure range can be expressed as ,in, is the degradation mean value, is the fault threshold coefficient, is the degradation standard deviation. Among them, the fault threshold coefficient can be preset according to requirements.

[0068] like Figure 7As shown, the first-order difference data that has not been filtered by the filter presents a scattered distribution, and it is difficult to capture its corresponding change law, while the data that has been filtered by the filter presents a linear distribution with obvious and clear change characteristics.

[0069] By training sample data, the reference degradation rate of the filament can be calculated based on a large amount of data, so as to widely obtain the variation characteristics of the filament degradation rate, and use it as a reference, and then use the filter to process and filter the reference degradation rate data of the filament, so that the variation characteristics of the reference degradation rate of the filament are clearer and more intuitive. Furthermore, the failure range of the filament is determined according to the mean value, standard deviation and fault threshold coefficient of the reference degradation rate of the filament, with a large amount of actual data as support. At the same time, considering the possible differences in filaments in different X-ray tubes, the three data of mean value, standard deviation and fault threshold coefficient are introduced to calculate and correct the failure range of the filament, so that the calculated failure range is more accurate and has universal applicability.

[0070] Step 150: In response to the degradation rate of the filament exceeding the failure range at a certain moment, determine the moment as a failure moment.

[0071] The failure moment refers to the point in time when the filament fails.

[0072] In some embodiments, in response to the degradation rate of the filament not exceeding the failure range at any time, the main failure cause of the X-ray tube is excluded as filament failure.

[0073] In some embodiments, the degradation rate corresponding to each moment can be calculated based on the test sample data, and the failure moment can be determined based on the degradation rate. For more information about step 150, see Figure 3 Detailed description.

[0074] Step 160 , calculating the failure time of the filament based on the failure moment and the uncertainty.

[0075] The failure time refers to the time when the filament fails after the correction calculation of the filament failure moment.

[0076] In some embodiments, the failure time can be calculated by formula (3): (3) in, is the failure time, is the failure time, is the safety margin, is the uncertainty.

[0077] In some embodiments, the safety margin is a parameter used to calibrate the failure time.

[0078] In some embodiments, the safety margin can be preset according to workers in this field.

[0079] In a specific implementation process, the failure time can be used as the final maintenance time, and maintenance personnel can repair or replace the filament of the X-ray tube before the failure time arrives to avoid filament failure and cause failure of the X-ray tube. For example, maintenance personnel can inspect the filament in the X-ray tube one week before the failure time arrives.

[0080] like Figure 6 As shown, the actual value of the filament current after preprocessing is highly consistent with the predicted current data output by the degradation prediction model, which further illustrates that after the degradation prediction model is trained by the above method, it is possible to obtain a prediction result close to the true value, and the prediction accuracy is greatly improved.

[0081] Obtaining filament data, processing the filament data, and constructing the processed filament data as sample data for model training can improve the purity of the data, simplify the calculation process, and further improve the accuracy and precision of the model measurement. In addition, based on the filament data, the model predicts its current and current distribution, and the reliability of the predicted current is evaluated according to the predicted current distribution, which can further determine whether the prediction result is accurate, and in turn train the model based on the reliability of the model prediction result, improve the model prediction ability, obtain better prediction results, and improve the intelligence and automation of the filament failure prediction method. By calculating the degradation rate of the filament and determining the filament failure based on the degradation rate, the change characteristics of the degradation rate during the filament degradation process can be effectively obtained, and based on the change characteristics, the failure process of the filament can be intuitively expressed, thereby improving the accuracy of the determination of the filament failure time, avoiding the failure of the X-ray tube due to the filament failure, and causing medical accidents.

[0082] Figure 2 Schematic diagram of the calculation uncertainty disclosed for some embodiments of the present application.

[0083] In some embodiments, the uncertainty prediction model 220 is constructed based on the degradation prediction model 230 .

[0084] Specifically, based on the training sample data 210, M degradation prediction models are constructed through different random seeds, which are recorded as .in, The above M degradation prediction models are combined to form an uncertainty prediction model 220 .

[0085] like Figure 2As shown, M degradation prediction models 230 - 1 , 230 - 2 . . . 230 -M are constructed respectively through different random seeds 1 -M, and M predicted currents are obtained based on 230 - 1 , 230 - 2 . . . 230 -M to form a predicted current distribution 250 .

[0086] In some embodiments, the M degradation prediction models are arranged in parallel.

[0087] In some embodiments, the training sample data 210 is input into the uncertainty prediction model 220 to obtain the predicted current distribution 250 .

[0088] In some embodiments, the predicted current distribution 250 includes the predicted currents 240 predicted by M degradation prediction models.

[0089] Uncertainty 260 is a parameter used to measure the reliability of the predicted current. In some embodiments, uncertainty 260 can be determined based on a variety of methods. For example, uncertainty can be the difference between the predicted current 240 and the average value of the predicted current distribution 250.

[0090] In some embodiments, uncertainty 260 may also be determined based on a standard deviation of predicted current 240 and predicted current distribution 250 .

[0091] In some embodiments, the smaller the uncertainty 260 between the predicted current and the predicted current distribution is, the smaller the difference between the predicted current 240 and the predicted current distribution 250 is, and the predicted current 240 is closer to the true value. Conversely, the larger the uncertainty 260 between the predicted current 240 and the predicted current distribution 250 is, the more the predicted current 240 deviates from the predicted current distribution 250 in terms of value, which means that the predicted result is further away from the true value and the reliability is lower.

[0092] In some embodiments, the uncertainty It can be calculated by formula (4): (4) (5) By constructing an uncertainty prediction model through multiple degradation prediction models obtained by training with different random seeds, the output of the uncertainty model can cover as many prediction results as possible in the degradation prediction model. On this basis, the predicted current distribution is obtained. Through the distributed prediction results, the comprehensive conclusions presented by the prediction results of the degradation prediction model on a large amount of data can be intuitively reflected. Then, the standard deviation between the predicted current distribution and the predicted current predicted by the degradation prediction model is calculated to obtain the uncertainty, so that the uncertainty can quantitatively reflect the accuracy of the degradation prediction model prediction, help model training, and improve the computing power of the model.

[0093] Figure 3 A schematic diagram of determining the failure moment disclosed in some embodiments of the present application.

[0094] In some embodiments, the X-ray tube filament performance degradation prediction and failure determination system 400 can update the test sample data based on the queue algorithm; input the updated test sample data into the uncertainty prediction model 220 to obtain the predicted current distribution; calculate the degradation rate at the current moment based on the predicted current distribution of the test sample data corresponding to the current moment and the predicted current distribution of the test sample data corresponding to the previous moment; in response to the degradation rate exceeding the failure range, determine the current moment as the failure moment.

[0095] A queue algorithm refers to a first-in, first-out linear data structure. In an embodiment of the present application, updating the test sample data based on the queue algorithm means that, in a sequence of test sample data, a single test sample data enters from the head of the queue and exits from the tail of the queue. For example, a number set is {1, 2, 3}, and now the number 4 is added, then 4 is added to the head of the queue, and the number 3 at the tail of the queue is removed, that is, the number set becomes {4, 1, 2}.

[0096] For example, Figure 3 As shown, the first sample data in the test sample data is Input sample corresponding to time Input to the uncertainty prediction model 220 to obtain the predicted current distribution I1, and add the predicted current distribution I1 to Input sample corresponding to time The first place of the original last sample data is removed , forming an updated Input sample corresponding to time .

[0097] Then update the Input sample corresponding to time The predicted current distribution I2 is input into the uncertainty prediction model 220, and the degradation rate 3-1 of the predicted current distribution I1 and the predicted current distribution I2 is calculated according to formulas (1) and (2).

[0098] Determine whether the degradation rate 3-1 exceeds the failure range. If not, continue to repeat the above steps for calculation; if it exceeds the failure range, The time is determined as the failure time 330 .

[0099] In some embodiments, in response to the degradation rate of the filament not exceeding the failure range at any time, the main failure cause of the X-ray tube is excluded as filament failure.

[0100] By calculating the degradation rate of the filament, the changing trend of the filament degradation rate at each moment can be dynamically obtained. Whether the filament has failed can be determined based on whether the degradation rate of the filament exceeds the failure range, which can further improve the accuracy of the filament failure determination and avoid the sudden failure or damage of the filament before maintenance and replacement to the greatest extent, resulting in medical equipment failure and interruption of treatment. Furthermore, in response to the degradation rate not exceeding the degradation rate, the influence of the filament failure on the X-ray tube can be ruled out, which further helps maintenance personnel to identify the cause of the failure, reduces the economic loss of uniformly replacing parts, extends the service life of the X-ray tube, and reduces its maintenance cost.

[0101] Figure 4 A module diagram of an X-ray tube filament performance degradation prediction and failure identification system disclosed in some embodiments of the present application.

[0102] In some embodiments, the X-ray tube filament performance degradation prediction and failure judgment system 400 includes an acquisition module 410, a prediction module 420, a calculation module 430 and a determination module 440. For detailed description of the relevant functions of each module of the X-ray tube filament performance degradation prediction and failure judgment system 400, please refer to Figures 1 to 3 Related content.

[0103] The acquisition module 410 is a module for acquiring filament data. In some embodiments, the acquisition module 410 is configured to acquire filament data of an X-ray tube and construct training sample data and test sample data.

[0104] The prediction module 420 is a module for predicting the degradation characteristics of the filament. In some embodiments, the prediction module 420 is configured to predict the predicted current of the filament through the degradation prediction model based on the training sample data, and predict the predicted current distribution of the filament through the uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two degradation prediction models.

[0105] The calculation module 430 is a module for implementing a calculation function. In some embodiments, the calculation module 430 is configured to calculate the uncertainty of the predicted current based on the predicted current distribution and the predicted current.

[0106] In some embodiments, the calculation module 430 is configured to calculate the failure range of the filament.

[0107] The determination module 440 is a module for determining the filament failure time.

[0108] In some embodiments, the determination module 440 is configured to determine a moment as a failure moment in response to a degradation rate of the filament at a certain moment exceeding a failure range.

[0109] In some embodiments, the determination module 440 is configured to calculate the failure time of the filament based on the failure moment and the uncertainty.

[0110] In some embodiments, the X-ray tube performance degradation prediction and failure determination system includes all modules in the above-mentioned X-ray tube filament performance degradation prediction and failure determination system 400.

[0111] The X-ray tube performance degradation prediction and failure determination system also includes a second determination module.

[0112] In some embodiments, the second determination module is used to determine the failure cause of the X-ray tube. In some embodiments, the second determination module is configured to exclude the failure cause of the X-ray tube as filament failure in response to the degradation rate of the filament at any time not exceeding the failure range.

[0113] It should be noted that the above description of the X-ray tube filament performance degradation prediction and failure judgment system 400, the X-ray tube performance degradation prediction and failure judgment system and its modules is only for the convenience of description, and cannot limit this specification to the scope of the embodiments. It is understandable that for those skilled in the art, after understanding the principle of the system, it is possible to arbitrarily combine the modules or form a subsystem to connect with other modules without deviating from this principle.

[0114] In some embodiments, Figure 4 The acquisition module 410, prediction module 420, calculation module 430 and determination module 440 disclosed in the specification, as well as the second determination module in the X-ray tube performance degradation prediction and failure judgment system can be different modules in one system, or one module can realize the functions of two or more modules mentioned above. For example, each module can share a storage module, or each module can have its own storage module. Such variations are all within the protection scope of this specification.

[0115] The above embodiments of the present application focus on describing the differences between the various embodiments. As long as the different optimization features between the various embodiments are not contradictory, they can be combined to form a better embodiment. Considering the simplicity of the text, they will not be repeated here.

[0116] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for predicting performance degradation and failure determination of an X-ray tube filament, characterized in that: The following steps are involved: Obtain the filament data of the X-ray tube and construct training sample data and test sample data; Based on the training sample data, predicting the predicted current of the filament by a degradation prediction model, and predicting the predicted current distribution of the filament by an uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two of the degradation prediction models; Calculating uncertainty of the predicted current based on the predicted current distribution and the predicted current; Calculate the failure range of the filament; In response to the degradation rate of the filament exceeding the failure range at a certain moment, determining the moment as a failure moment; The failure time of the filament is calculated based on the failure moment and the uncertainty.

2. The method for predicting performance degradation and failure determination of an X-ray tube filament according to claim 1, characterized in that: The obtaining of the filament data of the X-ray tube and the construction of the training sample data and the test sample data comprises: Acquiring the filament data of N X-ray tubes; Preprocessing the filament data; constructing training sample data based on the preprocessed filament data of N-1 X-ray tubes and first sample data of the preprocessed filament data of the Nth X-ray tube; Test sample data are constructed based on other sample data except the first sample data in the preprocessed filament data of the Nth X-ray tube.

3. The method for predicting performance degradation and failure determination of an X-ray tube filament according to claim 1, characterized in that: Based on the training sample data, the predicted current of the filament is predicted by a degradation prediction model, and the predicted current distribution of the filament is predicted by an uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two of the degradation prediction models and includes: Based on the training sample data, at least two degradation prediction models are constructed by using different random seeds; Combining the degradation prediction models to form the uncertainty prediction model; The training sample data is input into the uncertainty prediction model to obtain the predicted current distribution.

4. The method for predicting performance degradation and failure determination of an X-ray tube filament according to claim 1, characterized in that: The calculating the uncertainty of the predicted current based on the predicted current distribution and the predicted current includes: The uncertainty is determined based on a standard deviation of the predicted current and the predicted current distribution.

5. The method for predicting performance degradation and failure determination of an X-ray tube filament according to claim 1, characterized in that: The calculating the failure range of the filament comprises: Based on the training sample data, calculating first-order difference data of predicted current distribution corresponding to two adjacent moments; Filtering the first-order difference data through a filter to obtain a reference degradation rate of the filament; Calculating a degradation mean value and a degradation standard deviation of the reference degradation rate; The failure range is calculated based on the degradation mean value, the degradation standard deviation and a failure threshold coefficient.

6. The method for predicting performance degradation and failure determination of an X-ray tube filament according to claim 1, characterized in that: In response to the degradation rate of the filament exceeding the failure range at a certain moment, determining the moment as the failure moment includes: Based on a queue algorithm, updating the test sample data; Inputting the updated test sample data into the uncertainty prediction model to obtain the predicted current distribution; Calculating the degradation rate at the current moment based on the predicted current distribution of the test sample data corresponding to the current moment and the predicted current distribution of the test sample data corresponding to the previous moment; In response to the degradation rate exceeding the failure range, the current time is determined as the failure time.

7. A method for predicting performance degradation and failure determination of an X-ray tube, characterized in that: The following steps are involved: Executing the X-ray tube filament performance degradation prediction and failure judgment method as described in any one of claims 1 to 6; In response to the degradation rate of the filament at any time not exceeding the failure range, the failure cause of the X-ray tube is ruled out as filament failure.

8. An X-ray tube filament performance degradation prediction and failure judgment system, characterized in that: include: An acquisition module is configured to acquire filament data of an X-ray tube and construct training sample data and test sample data; A prediction module, configured to predict the predicted current of the filament by a degradation prediction model based on the training sample data, and predict the predicted current distribution of the filament by an uncertainty prediction model, wherein the uncertainty prediction model is constructed based on at least two of the degradation prediction models; a calculation module, configured to calculate the uncertainty of the predicted current based on the predicted current distribution and the predicted current; Calculate the failure range of the filament; A determination module, configured to determine a moment as a failure moment in response to a degradation rate of the filament exceeding the failure range at a certain moment; The failure time of the filament is calculated based on the failure moment and the uncertainty.

9. An X-ray tube performance degradation prediction and failure judgment system, characterized in that: include: An X-ray tube filament performance degradation prediction and failure judgment system as claimed in claim 8; The second determination module is configured to exclude the failure of the X-ray tube as a result of the failure of the filament in response to the degradation rate of the filament at any time not exceeding the failure range.

10. A computer-readable storage medium, characterized in that: The storage medium stores computer instructions. When the computer reads the computer instructions in the storage medium, the computer runs the X-ray tube filament performance degradation prediction and failure judgment method as described in any one of claims 1 to 6, or the X-ray tube performance degradation prediction and failure judgment method as described in claim 7.

Images