A CT equipment abnormality prediction method and system based on Internet of Things data

By using IoT data and machine learning models in CT devices, building features and predicting abnormal phenomena, the problem of CT device failure prediction in the prior art is solved, and more efficient anomaly prediction performance is achieved.

CN116364263BActive Publication Date: 2024-11-15SICHUAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211619820.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-12-15
Publication Date
2024-11-15
Estimated Expiration
2042-12-15

AI Technical Summary

Technical Problem

Existing IoT systems have not yet effectively utilized medical IoT monitoring data to predict CT device anomalies, and there is a lack of suitable machine learning models to predict medical device failures.

Method used

Using a combination of sliding window algorithms based on IoT data and machine learning models (such as Naive Bayes, K-nearest neighbors, and decision trees), data preprocessing and building features are carried out by inputting the operation data of CT equipment collected by the medical IoT. , to predict abnormal phenomena in CT equipment, such as ignition inside the tube.

Benefits of technology

Accurate prediction of CT device abnormalities is realized, prediction performance is improved, and application prospects are available in CT device monitoring.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116364263B_ABST
    Figure CN116364263B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of medical equipment maintenance technology, and specifically relates to a method and system for predicting CT equipment abnormalities based on Internet of Things data. The method of the present invention comprises the following steps: Step 1, inputting CT equipment operation data collected by the medical Internet of Things; Step 2, performing data preprocessing on the CT equipment operation data, and constructing features using a sliding window algorithm; Step 3, inputting the features into a machine learning model to obtain a prediction result of an abnormality of the CT equipment; wherein the machine learning model is selected from a naive Bayes model, a K-nearest neighbor model or a decision tree model. The present invention also constructs a system that applies the above-mentioned CT equipment abnormality prediction method. Compared with other machine learning methods in the prior art, the method of the present invention has better prediction performance, and therefore has a good application prospect in CT equipment abnormality monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of medical equipment maintenance, and in particular relates to a method and system for predicting abnormalities of CT equipment based on Internet of Things data. Background Art

[0002] At present, various medical devices have been widely used in various links of medical services, including disease diagnosis, patient status monitoring and rehabilitation treatment. In particular, large digital radiology equipment, such as computed tomography (CT), can obtain clear cross-sectional images of internal organs through X-rays, which is essential for medical institutions to treat patients. However, CT equipment embedded with complex operating systems is susceptible to various types of damage during operation. Abnormal phenomena such as unexpected component failure and system interruption during equipment operation have long plagued hospitals. Equipment abnormalities may lead to low-quality radiological images, unexpected delays in patient care, expensive maintenance services, and even serious patient accidents. According to a survey by the Joint Commission (TJC), safety accidents such as premature death, serious injury and disability accidents are closely related to medical equipment failures. It is reported that there were 176 medical device-related incidents in the United States between 2004 and 2011, accounting for 2.9% of the total 6093 incidents collected from 8 hospitals. Therefore, hospitals and medical institutions must ensure the high reliability of medical equipment to avoid operational interruptions and ensure patient safety.

[0003] The working principle of CT equipment is as follows: during scanning, a fan-shaped X-ray beam emitted from the CT tube passes through the patient to multiple digital detectors, which receive the X-rays and convert them into medical images. X-ray tubes operating at high voltage require a high vacuum environment inside the tube. When the required vacuum environment is destroyed and a conductive bridge is formed between the cathode and impurities, spark discharge occurs. Sparking not only leads to low-quality radiographic images, but is also closely related to unstable CT performance or tube rupture. Sparking can cause artifacts, which appear as approximately parallel and equidistant stripe patterns or "horizontal" low-density stripes in the image, which can reduce image quality and affect clinical diagnosis. In addition, although various abnormalities of X-ray tubes have been discussed in the literature, sparking in the tube is generally considered to be the most typical and dangerous early sign of CT equipment failure. Therefore, sparking deserves attention because it is closely related to image quality and CT failure.

[0004] To date, maintenance strategies including corrective maintenance, preventive maintenance, and predictive maintenance have been widely used in various fields such as mechanical engineering, nuclear engineering, and management science, greatly improving the management level of these systems. However, the application of maintenance models to medical equipment such as CT equipment has not been thoroughly addressed. Usually, most medical institutions perform equipment maintenance in accordance with the manufacturer's recommendations. Manufacturers will develop maintenance plans and provide maintenance guidance for equipment. This routine maintenance scheduling does improve the reliability of medical equipment and reduce the risk of failure to a certain extent, but fails to predict and avoid abnormal or sudden failures.

[0005] With the development of various monitoring tools and technologies over the past few decades, it has been found that combining preventive maintenance with monitoring data and data analysis technology will be an appropriate way to predict equipment abnormalities. The Internet of Things (IoT) has gradually become a key technology for monitoring the real-time status of target equipment by integrating the status information of machine components through the Internet. It is worth mentioning that the Internet of Medical Things (IoMT), which obtains real-time healthcare data from wearable devices and sensors, has received widespread attention. At present, the development of the Internet of Things is still in its early stages, and most of the existing IoT systems focus on improving the level of diagnosis related to the human body rather than medical devices.

[0006] At present, the use of monitoring data from the Medical Internet of Things for the prediction of medical device anomalies has not yet been realized. The difficulty lies mainly in the fact that the Medical Internet of Things can obtain a large amount of data related to the operation of medical equipment, and people are not clear about what specific data should be used to predict medical device anomalies. In addition, the development of machine learning models has also made the current selection of prediction models diverse. However, the applicability of different models to specific prediction tasks also needs further research. People still do not know what kind of model is suitable for the prediction of medical device anomalies. Summary of the invention

[0007] Based on the problems of the prior art, the present invention provides a method and system for predicting abnormalities of CT equipment based on Internet of Things data, aiming to predict abnormalities of CT equipment by utilizing monitoring data of the medical Internet of Things combined with machine learning methods.

[0008] A method for predicting abnormality of CT equipment based on Internet of Things data comprises the following steps:

[0009] Step 1: Input the CT equipment operation data collected by the medical Internet of Things;

[0010] Step 2, preprocessing the CT device operation data and constructing features using a sliding window algorithm;

[0011] Step 3, inputting the features into a machine learning model to obtain a prediction result of abnormality of the CT device;

[0012] Wherein, the machine learning model is selected from a naive Bayes model, a K-nearest neighbor model or a decision tree model.

[0013] Preferably, in step 1, the CT equipment operation data is collected several times a day, and the following data is collected each time: oil temperature, anode voltage, cumulative tube scanning time, cumulative power consumption and daily ignition times.

[0014] Preferably, the features obtained in step 2 include at least one of the following parameters: AOT Max 、AAV Max ,I Max , IDLE, AOT dMax 、AAV dMax ,I dMax Arcing w ;

[0015] Among them, the definition of each parameter is:

[0016] Define a time period T, which means that the duration T is used as a time period, and several consecutive time periods T constitute the time window of the sliding window algorithm;

[0017] AOT Max is the maximum value of the average oil temperature in T among several consecutive time periods T of the time window, where the average oil temperature in T is the result of averaging the oil temperatures in the time period T;

[0018] AAV Max is the maximum value of the average anode voltage within T in a plurality of consecutive time periods T of the time window, wherein the average anode voltage within T is the result obtained by averaging the anode voltage within the time period T;

[0019] I Max is the maximum value of the average current in T among several consecutive time periods T of the time window, wherein the average current in T is calculated by the average anode voltage in T, the total scanning time in the time period T and the total consumed power in the time period T;

[0020] IDLE is defined as follows: among several consecutive time periods T in the time window, if at least one time period T does not collect CT equipment operation data, it is marked as 0, otherwise it is marked as 1;

[0021] AOT dMax The definition of AOT is: the average oil temperature in T is first-order differentiated to obtain the average oil temperature slope in T, and the maximum value of the average oil temperature slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average oil temperature slope in T exceeds the preset threshold, then AOT dMax The value is 1, otherwise it is 0;

[0022] AAV dMax The definition of AAV is as follows: the average anode voltage in T is first-order differentiated to obtain the average anode voltage slope in T, and the maximum value of the average anode voltage slope in T in several consecutive time periods T in the time window is taken. If the maximum value of the average anode voltage slope in T exceeds a preset threshold, then AAV dMax The value is 1, otherwise it is 0;

[0023] I dMax The definition of is: the average current slope in T is obtained by performing a first-order difference on the average current in T, and the maximum value of the average current slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average current slope in T exceeds a preset threshold, I dMax The value is 1, otherwise it is 0;

[0024] Arcing w The definition of is: when there is at least one time period T in the time window where sparking is observed in the tube from the CT device, the value is 1, otherwise it is 0;

[0025] The preprocessing includes calculating the device current I according to the following steps:

[0026] Step a, performing first-order difference on the accumulated tube scanning time and accumulated power consumption of the data collected by the medical Internet of Things, and obtaining the tube scanning time DTST in each time period T and the power consumption DCE in each time period T;

[0027] Step b, calculate the device current using the following formula

[0028]

[0029] Where AAV is the average anode voltage within T.

[0030] Preferably, the duration T is between 1 hour and 1 day.

[0031] Preferably, the features obtained in step 2 include AOT Max 、AAV Max ,I Max , IDLE, AOT dMax 、AAV dMax ,I dMax and Arcing w .

[0032] Preferably, the machine learning model is selected from a naive Bayes model, and the features include I dMax and AOT Max ;

[0033] Or, the machine learning model is selected from a K-nearest neighbor model, and the features include AOT Max and AAV Max ;

[0034] Or, the machine learning model is selected from a decision tree model, and the features include AOT Max and AAV Max .

[0035] Preferably, the machine learning model is selected from a Naive Bayes model.

[0036] Preferably, in step 3, the abnormality is sparking in the tube.

[0037] The present invention also provides a CT equipment abnormality prediction system based on Internet of Things data, comprising:

[0038] An input module is used to input the CT equipment operation data collected by the medical Internet of Things;

[0039] A feature construction module, used for performing data preprocessing on the CT device operation data and constructing features using a sliding window algorithm;

[0040] A prediction module, used for inputting the features into a machine learning model to obtain a prediction result of abnormality of the CT device, wherein the machine learning model is selected from a naive Bayes model, a K-nearest neighbor model or a decision tree model;

[0041] An output module is used to output the prediction result.

[0042] The present invention also provides a computer-readable storage medium on which is stored a computer program for implementing the above-mentioned CT equipment abnormality prediction method based on Internet of Things data.

[0043] The present invention optimizes the feature construction method and the selection of model types, and realizes a method and system for accurately predicting abnormalities of CT equipment by using monitoring data of the medical Internet of Things combined with machine learning methods. Experimental results show that the accuracy of the method and system of the present invention in predicting abnormalities of CT equipment is higher than that of other existing time series classification models, and has a good application prospect.

[0044] Obviously, according to the above contents of the present invention, in accordance with common technical knowledge and customary means in the art, without departing from the above basic technical ideas of the present invention, other various forms of modification, replacement or change may be made.

[0045] The above contents of the present invention are further described in detail below through specific implementation methods in the form of embodiments. However, this should not be understood as the scope of the above subject matter of the present invention being limited to the following examples. All technologies realized based on the above contents of the present invention belong to the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of the framework for building two training and test data sets in Experimental Example 1;

[0047] Figure 2 are the average ROC and AUC values ​​of the three models in Experimental Example 1. DETAILED DESCRIPTION

[0048] It should be noted that the algorithms of data collection, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structure, circuit connection, etc. not specifically described can all be implemented through the contents disclosed in the prior art.

[0049] Example 1 CT equipment abnormality prediction method and system based on Internet of Things data

[0050] The system of this embodiment includes:

[0051] An input module is used to input the CT equipment operation data collected by the medical Internet of Things;

[0052] A feature construction module, used for performing data preprocessing on the CT device operation data and constructing features using a sliding window algorithm;

[0053] A prediction module, used to input the features into a machine learning model to obtain a prediction result of abnormality of the CT device. The machine learning model specifically selected in this embodiment is a decision tree model, a logistic regression model, a support vector machine model, a naive Bayes model or a K-nearest neighbor model;

[0054] An output module is used to output the prediction result.

[0055] The specific steps of the method for predicting abnormalities of CT equipment using this system are as follows:

[0056] Step 1, input the CT equipment operation data collected by the medical Internet of Things; these parameters include the average oil temperature within time T (AOT), the average anode voltage within time T (AAV), the average cumulative tube scan time within time T (ATST) and the average cumulative energy consumption within time T (ACE).

[0057] Step 2, preprocessing the CT device operation data and constructing features using a sliding window algorithm;

[0058] The features obtained include: AOT Max 、AAV Max ,I Max , IDLE, AOT dMax 、AAV dMax ,I dMax and Arcingw ;

[0059] Among them, the definition of each parameter is:

[0060] Define a time period T, which means that the duration T is used as a time period, and several consecutive time periods T constitute the time window of the sliding window algorithm;

[0061] AOT Max is the maximum value of the average oil temperature in T among several consecutive time periods T of the time window, where the average oil temperature in T is the result of averaging the oil temperatures in the time period T;

[0062] AAV Max is the maximum value of the average anode voltage within T in a plurality of consecutive time periods T of the time window, wherein the average anode voltage within T is the result obtained by averaging the anode voltage within the time period T;

[0063] I Max is the maximum value of the average current in T among several consecutive time periods T of the time window, wherein the average current in T is calculated by the average anode voltage in T, the total scanning time in the time period T and the total consumed power in the time period T;

[0064] IDLE is defined as follows: among several consecutive time periods T in the time window, if at least one time period T does not collect CT equipment operation data, it is marked as 0, otherwise it is marked as 1;

[0065] AOT dMax The definition of AOT is: the average oil temperature in T is first-order differentiated to obtain the average oil temperature slope in T, and the maximum value of the average oil temperature slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average oil temperature slope in T exceeds the preset threshold, then AOT dMax The value is 1, otherwise it is 0; for the average oil temperature slope within T, the preset threshold can be reasonably selected according to the CT equipment, for example, 5-15K (degrees Kelvin) can be selected, and 5K is selected in this embodiment.

[0066] AAV dMax The definition of AAV is as follows: the average anode voltage in T is first-order differentiated to obtain the average anode voltage slope in T, and the maximum value of the average anode voltage slope in T in several consecutive time periods T in the time window is taken. If the maximum value of the average anode voltage slope in T exceeds a preset threshold, then AAV dMax The value is 1, otherwise it is 0; for the average anode voltage within T, the preset threshold value can be reasonably selected according to the CT equipment, for example, 40-60kV can be selected, and 40kV is selected in this embodiment.

[0067] IdMax The definition of is: the average current slope in T is obtained by performing a first-order difference on the average current in T, and the maximum value of the average current slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average current slope in T exceeds a preset threshold, I dMax The value is 1, otherwise it is 0; for the average current in T, the preset threshold value can be reasonably selected according to the CT equipment, for example, 0.180-0.252A can be selected, and 0.180A is selected in this embodiment.

[0068] Arcing w The definition of is: when there is at least one time period T in the time window where sparking is observed in the tube from the CT device, the value is 1, otherwise it is 0;

[0069] The preprocessing includes calculating the device current I according to the following steps:

[0070] Step a, performing first-order difference on the accumulated tube scanning time and accumulated power consumption of the data collected by the medical Internet of Things, and obtaining the tube scanning time DTST in each time period T and the power consumption DCE in each time period T;

[0071] Step b, calculate the device current using the following formula

[0072]

[0073] Where AAV is the average anode voltage within T.

[0074] In this embodiment, the value of the duration T is 1 day.

[0075] Step 3: Input the features into a machine learning model to obtain a prediction result of abnormality of the CT device.

[0076] The technical solution of the present invention is further described below through experimental examples. The steps or parameters not described in the experimental examples can be set with reference to the contents of Example 1.

[0077] Experimental Example 1 Prediction Performance Investigation

[0078] 1. Experimental Methods

[0079] 1. Sample data

[0080] Real-time CT equipment status data continuously monitored by IoMT of West China Hospital is used to predict abnormalities of CT equipment. The dataset contains the operating status data of 3 CT devices (CT1, CT2, CT3). The features of the dataset include: oil temperature (OT), anode voltage (AV), cumulative tube scan time (TST), and cumulative electrical energy consumption (CE). The occurrence of sparking in the tube is closely related to image quality and CT failure and is regarded as a label. Due to failures in the IoMT system, missing data occasionally appear in the dataset. There are 33 observations in the sparking class, which is significantly less than the 733 in the non-sparking class, and the dataset is unbalanced. This will be considered in model development.

[0081] 2. Construction of training and test data sets

[0082] A 5-fold cross validation was used for each parameter combination to compare the performance metrics. The data was divided into five datasets, one of which was used as a test set and the rest as training sets.

[0083] Since the positive examples mainly come from CT1, this experimental example proposes two frameworks to construct training and test datasets. Figure 1 As shown, framework 1 uses data from three devices and splits it into training and test sets, while framework 2 uses data from device CT1 only, splits it into training and test sets, and then adds positive examples from the training sets of the other two devices. Random oversampling is used to increase the size of the minority class.

[0084] 3. Multivariate Time Series Classification Model

[0085] The machine learning classification models selected in this experiment include decision tree (DT), support vector machine (SVM), logistic regression (LR), naive Bayes (NB) and K-nearest neighbor (KNN). By comparing the performance of these models and selecting the best model, the results of the best model performance are obtained. The best model is then compared with several state-of-the-art time series classification models reported in the existing literature, including BOSS (Data Mining Knowledge Discovery. 2015; 29(6): 1505-1530.), CIF (In: 2020 IEEE International Conference on Big Data (Big Data). IEEE; 2020. p. 188-195.), DrCIF (Machine Learning. 2021; 110: 3211-3243.), TDE (In: ECML PKDD 2020: Machine Learning and Knowledge Discovery in Databases. Springer; 2021. p. 660-676.) and DTW (Data Mining and Knowledge Discovery. 2017; 31(1): 1-31.).

[0086] 4. Performance evaluation

[0087] Accuracy (Acc), recall (Rec), precision (Pre), and F1-score (F1) are used as performance indicators to measure the classification model. Acc is calculated by formula (2) and reflects the overall classification ability. Rec is calculated by formula (3), which is the proportion of true anomalies predicted as anomalies. Pre is calculated by formula (4), which is the portion of predicted true anomalies. F1 is calculated by formula (5), which is a performance indicator that considers both Rec and Pre. This experimental example also draws the receiver operating characteristic (ROC) curve and calculates the area under the curve (AUC) to evaluate the performance of the classification model.

[0088]

[0089]

[0090]

[0091]

[0092] TP is true positive, TN is true negative, FP is false positive, and FN is false negative.

[0093] 2. Experimental Results

[0094] Table 4 shows the performance indicators of the five models under the two frameworks with the optimal parameters. It can be seen that the NB model has higher Acc, Pre, and AUC, indicating that the false alarm rate of this model is lower. On the contrary, the KNN model has higher Rec and F1, indicating that the model has higher abnormal alarm capability. In addition, the DT model also performs well. The ROC curves and AUC values ​​of the three models under Framework 2 are shown in Figure 4. Figure 2 It is observed that the NB model with the highest (0.88) AUC value has the best overall classification performance.

[0095] Table 4 Performance indicators of classification models

[0096]

[0097]

[0098] The two most important features determined by the three models under Framework 2 are shown in Table 5. The results show that AOT Max and AAV Max is the most important feature in the DT and KNN models, while I dMax and is the model AOT Max The most important features in NB. This shows that the original features OT and AV are very important in predicting anomalies.

[0099] Table 5 The two most important features identified by the three models under Framework 2

[0100]

[0101] In order to compare the above three models with other existing models, each model is applied to the data set under Framework 1 and Framework 2, and the performance indicators with the best performance are shown in Table 6. It can be seen that the model proposed in this application has the best performance among all models.

[0102] Table 6 Best performance indicators of the model

[0103]

[0104]

[0105] Through the above embodiments and experimental examples, it can be seen that the present invention realizes a system and method for accurately predicting CT equipment abnormalities using medical Internet of Things monitoring data by optimizing the selection of models and the construction of features. The method of the present invention has better prediction performance than other machine learning methods in the prior art, and thus has a good application prospect in CT equipment abnormality monitoring.

Claims

1. A CT equipment abnormality prediction method based on IoT data, It is characterized in that The steps include: Step 1, inputting the CT equipment operation data collected by the medical Internet of Things; the CT equipment operation data is collected several times a day, and the following data is collected each time: oil temperature, anode voltage, cumulative tube scanning time, cumulative power consumption and daily ignition times; Step 2, preprocessing the CT device operation data and constructing features using a sliding window algorithm; Step 3, inputting the features into a machine learning model to obtain a prediction result of abnormality of the CT device; in, The machine learning model is selected from the Naive Bayes model, and the features include I dMax and AOT Max ; Or, the machine learning model is selected from a K-nearest neighbor model, and the features include AOT Max and AAV Max ; Or, the machine learning model is selected from a decision tree model, and the features include AOT Max and AAV Max; AOT Max is the maximum value of the average oil temperature within T in a number of consecutive time periods T of the time window, where the average oil temperature within T is the result of averaging the oil temperatures within the time period T; AAV Max is the maximum value of the average anode voltage within T in a plurality of consecutive time periods T of the time window, wherein the average anode voltage within T is the result obtained by averaging the anode voltage within the time period T; I dMax The definition of is: the average current slope in T is obtained by performing a first-order difference on the average current in T, and the maximum value of the average current slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average current slope in T exceeds a preset threshold, I dMax The value is 1, otherwise it is 0; The preprocessing includes calculating the device current I according to the following steps: Step a, performing first-order difference on the accumulated tube scanning time and accumulated power consumption of the data collected by the medical Internet of Things, and obtaining the tube scanning time DTST in each time period T and the power consumption DCE in each time period T; Step b, calculate the device current using the following formula Where AAV is the average anode voltage within T.

2. According to the method for predicting abnormality of CT equipment based on Internet of Things data as claimed in claim 1, Features: The value of T is 1 hour to 1 day.

3. According to the method for predicting abnormality of CT equipment based on Internet of Things data as claimed in claim 1, Features: The machine learning model is selected from a Naive Bayes model.

4. According to the method for predicting abnormality of CT equipment based on Internet of Things data as claimed in claim 1, Features: In step 3, the abnormality is sparking in the tube.

5. A CT equipment abnormality prediction system based on IoT data, It is characterized in that include: An input module is used to input the CT equipment operation data collected by the medical Internet of Things; the CT equipment operation data is collected several times a day, and the following data is collected each time: oil temperature, anode voltage, cumulative tube scanning time, cumulative power consumption and daily ignition times; A feature construction module, used for performing data preprocessing on the CT device operation data and constructing features using a sliding window algorithm; A prediction module, used to input the features into a machine learning model to obtain a prediction result of abnormality of the CT device; An output module, used for outputting the prediction result; in, The machine learning model is selected from the Naive Bayes model, and the features include I dMax and AOT Max ; Or, the machine learning model is selected from a K-nearest neighbor model, and the features include AOT Max and AAV Max ; Or, the machine learning model is selected from a decision tree model, and the features include AOT Max and AAV Max; AOT Max is the maximum value of the average oil temperature within T in a number of consecutive time periods T of the time window, where the average oil temperature within T is the result of averaging the oil temperatures within the time period T; AAV Max is the maximum value of the average anode voltage within T in a plurality of consecutive time periods T of the time window, wherein the average anode voltage within T is the result obtained by averaging the anode voltage within the time period T; I dMax The definition of is: the average current slope in T is obtained by performing a first-order difference on the average current in T, and the maximum value of the average current slope in T in several consecutive time periods T of the time window is taken. If the maximum value of the average current slope in T exceeds a preset threshold, I dMax The value is 1, otherwise it is 0; The preprocessing includes calculating the device current I according to the following steps: Step a, performing first-order difference on the accumulated tube scanning time and accumulated power consumption of the data collected by the medical Internet of Things, and obtaining the tube scanning time DTST in each time period T and the power consumption DCE in each time period T; Step b, calculate the device current using the following formula Where AAV is the average anode voltage within T.

6. A computer-readable storage medium, Features: A computer program for implementing the method for predicting abnormalities of CT equipment based on Internet of Things data as described in any one of claims 1 to 4 is stored thereon.