A chamber leakage monitoring method for a medical cyclotron vacuum chamber

Abstract

The present application relates to the field of data cleaning technology, and specifically to a chamber leakage monitoring method for a medical cyclotron vacuum chamber, the method comprising: obtaining in real time the vacuum degree at each moment in the vacuum chamber when the medical cyclotron is running, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment; dividing all moments into multiple monitoring periods; determining the monitoring sensitivity of each monitoring period; determining the power influence index at any moment; determining the relative amplitude difference of each monitoring period; determining the magnetic field influence index at any moment; determining the metric distance between any moment and the remaining moments; using an anomaly detection algorithm to perform data cleaning on the vacuum degree at all moments, and combining with a deep learning model to monitor the chamber leakage of the vacuum chamber. The present application reduces the interference of redundant information in the data and improves the accuracy of chamber leakage monitoring of the vacuum chamber.

Description

Technical Field

[0001] The present application relates to the technical field of data cleaning, and in particular to a chamber leakage monitoring method for a medical cyclotron vacuum chamber. Background Art

[0002] A medical cyclotron is a device used to accelerate charged particles. It is widely used in the medical field for radiotherapy and positron emission tomography. It accelerates charged particles to high energy by using electromagnetic fields, and then bombards the target material to produce radioactive isotopes. Among them, the vacuum chamber is a crucial component of the device, which constitutes the environment for particle acceleration. The chamber of this vacuum chamber needs to maintain an extremely low pressure, close to a vacuum state, to ensure that the charged particles will not have unnecessary collisions with molecules in the air during the acceleration process, thereby reducing energy loss and beam scattering, and ensuring that the particles can be efficiently and accurately accelerated and focused.

[0003] Therefore, a high vacuum degree in the vacuum chamber is one of the important conditions to ensure the normal operation of the medical cyclotron, and the vacuum state of the vacuum chamber needs to be monitored and evaluated in real time during the operation of the accelerator. However, during the operation of the cyclotron, the electromagnetic field and mechanical vibration generated by the accelerator may have a certain impact on the vacuum degree monitored by the vacuum gauge, so the monitored vacuum degree needs to be cleaned. However, the existing data cleaning method fails to take into account the changing characteristics of the vacuum degree of the cyclotron at different operating stages, and the data cleaning results cannot meet the accuracy requirements of vacuum detection, thereby affecting the accuracy of chamber leak detection in the vacuum chamber. Summary of the invention

[0004] In order to solve the above technical problems, a chamber leakage monitoring method for a medical cyclotron vacuum chamber is provided to solve the existing problems.

[0005] The solution to the technical problem of the present application is to provide a chamber leakage monitoring method for a medical cyclotron vacuum chamber, comprising the following steps:

[0006] Real-time acquisition of the vacuum degree at each moment in the vacuum chamber of the medical cyclotron accelerator, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment;

[0007] Based on the changing trend of the operating power at all times, all times are divided into multiple monitoring periods; the discreteness of the rate of change of the vacuum degree in each monitoring period, as well as the correlation between the changing trend of the vacuum degree and the changing trend of the operating power are analyzed to obtain the monitoring sensitivity of each monitoring period;

[0008] Determine the power impact index at any moment according to the interval between any moment and different moments before it, and the change of the operating power at different moments, combined with the monitoring sensitivity;

[0009] Analyze the amplitude difference of the magnetic field strength at all times in different monitoring periods in the frequency domain to determine the relative amplitude difference of each monitoring period; calculate the magnetic field influence index at any time based on the change trend of the magnetic field strength at any time and the relative amplitude difference;

[0010] Determine the measurement distance between any moment and the remaining moments according to the difference in vacuum degree, operating power, magnetic field strength and the power influence index between any moment and the remaining moments, combined with the magnetic field influence index;

[0011] Based on the metric distance, an anomaly detection algorithm is used to clean the vacuum degree at all times, and the chamber leakage of the vacuum chamber is monitored in combination with a deep learning model.

[0012] Preferably, the step of dividing all time periods into a plurality of monitoring periods includes:

[0013] The operating power at each moment and its corresponding power are combined into a two-dimensional array; curve fitting is performed on the two-dimensional array at all moments and recorded as a power curve;

[0014] Obtain the extreme points and inflection points of the power curve; use all extreme points and inflection points as segmentation points, and divide all moments into multiple monitoring periods.

[0015] Preferably, obtaining the monitoring sensitivity of each monitoring period includes:

[0016] Each moment and its corresponding vacuum degree are formed into a two-dimensional array; a curve fitting is performed on the two-dimensional array of all moments, recorded as a vacuum curve, and the derivative value of the vacuum curve at each moment is recorded as the vacuum change rate at each moment;

[0017] Calculating the discrete degree of the vacuum change rate at all times in each monitoring period;

[0018] The derivative value of the power curve at each moment is recorded as the power change rate at each moment;

[0019] Calculating the correlation between the vacuum change rate and the power change rate at all times in each monitoring period;

[0020] The ratio of the correlation degree to the discrete degree is used as the monitoring sensitivity of each monitoring period.

[0021] Preferably, the determining the power impact index at any moment includes:

[0022] Obtain the starting time of the previous monitoring period of any moment in the corresponding monitoring period, and record all the moments between the starting time and any moment as the local impact period of any moment;

[0023] Calculating the product of the operating power at each moment and the power change rate as the stage characterization value at each moment;

[0024] No. Power impact index at the moment The calculation method is: ,in, For the The local impact period of the moment The monitoring sensitivity of the monitoring period to which the moment belongs, For the The time and its local influence period The time interval between moments, For the The local impact period of the moment The stage characterization value at the moment, For the The number of all moments in the local impact period of a moment, is the normalization function.

[0025] Preferably, determining the relative amplitude difference in each monitoring period includes:

[0026] Perform frequency domain analysis on the magnetic field intensity at all times in each monitoring period to obtain the amplitude of each frequency component;

[0027] The difference between the amplitude of each frequency component in each monitoring period and the amplitude of the same frequency component in its adjacent monitoring period is recorded as the amplitude difference;

[0028] The average value of the amplitude differences of all frequency components in each monitoring period is taken as the relative amplitude difference of each monitoring period.

[0029] Preferably, the calculating the magnetic field influence index at any moment includes:

[0030] The magnetic field strengths at each moment and the corresponding magnetic field strengths are formed into a two-dimensional array; curve fitting is performed on the two-dimensional arrays at all moments, recorded as a magnetic field curve, and the derivative value of the magnetic field curve at each moment is recorded as the magnetic field change rate at each moment;

[0031] Calculate the product of the magnetic field strength at any moment and the magnetic field change rate, and record it as the magnetic field interference degree at any moment;

[0032] The product of the magnetic field interference degree at any moment and the relative amplitude difference of the monitoring period to which the any moment belongs is used as the magnetic field influence index at any moment.

[0033] Preferably, determining the metric distance between any one moment and the remaining moments comprises:

[0034] Determine the relative difference between any one moment and the other moments based on the difference in vacuum degree, operating power, and magnetic field strength between any one moment and the other moments;

[0035] No. Moment and Metric distance between moments The calculation method is: ,in, For the Moment and The mean value of the magnetic field influence index at time, For the The power impact index at the moment, For the The power impact index at the moment, For the Moment and The relative difference between moments.

[0036] Preferably, determining the relative difference between any one moment and the remaining moments comprises:

[0037] The vacuum degree, operating power and magnetic field strength at each moment are combined into the characteristic vector at each moment;

[0038] The difference between the feature vector at any moment and the feature vectors at other moments is used to form a difference vector;

[0039] The sum of all elements in the difference vector is taken as the relative difference between any one moment and the remaining moments.

[0040] Preferably, the data cleaning of the vacuum degree at all times includes:

[0041] Based on the measurement distance, anomaly detection is performed on the vacuum degree at all times through an anomaly detection algorithm, and abnormal vacuum degrees are eliminated; and the missing vacuum degrees are filled in through a missing value filling method.

[0042] Preferably, the monitoring of chamber leakage of the vacuum chamber includes:

[0043] The vacuum degree and operating power at all times after cleaning are used as inputs to the deep learning model to obtain the leakage probability of the vacuum chamber.

[0044] If the leakage probability is greater than a preset threshold, there is leakage in the vacuum chamber of the cyclotron; otherwise, there is no leakage in the vacuum chamber of the cyclotron.

[0045] This application has at least the following beneficial effects:

[0046] The present application divides all moments into multiple monitoring periods based on the changing trend of the operating power at all moments. The beneficial effect of the present application is that by analyzing the changing trend of the operating power of the vacuum pump when the vacuum pump is extracting the vacuum, the operation process of the vacuum pump is divided into different periods to reflect the different operation stages of the vacuum pump. The monitoring sensitivity of each monitoring period is determined according to the discreteness of the rate of change of the vacuum degree in each monitoring period and the correlation between the trend of the change of the vacuum degree and the trend of the change of the operating power. The beneficial effect of the present application is that the fluctuation of the rate of change of the vacuum degree detected by the vacuum gauge in the vacuum chamber at different operation stages and the correlation between the rate of change of the vacuum degree and the rate of change of the operating power are considered to reflect the timely capture of the vacuum gauge by the vacuum gauge of the vacuum degree change caused by the operation of the vacuum pump, and the sensitivity of the vacuum gauge to the monitoring of the vacuum extraction process of the vacuum pump in different operation stages is explained. The power influence index at any moment is determined according to the interval time between any moment and the different moments before it, and the change of the operating power at different moments, combined with the monitoring sensitivity. The beneficial effect of the present application is that the rate of change of the operating power is considered to reflect the operation state of the vacuum pump, so as to analyze the influence of the operating power at different moments on the vacuum degree monitoring. The magnetic field at all moments in different monitoring periods The amplitude difference of the intensity in the frequency domain is used to determine the relative amplitude difference of each monitoring period; the changing trend of the magnetic field intensity at any moment is analyzed, and the magnetic field influence index at any moment is determined in combination with the relative amplitude difference. The beneficial effect is that the adjustment frequency of the magnetic field by the cyclotron in the acceleration stage is taken into account. Secondly, through frequency domain analysis, the significance of the magnetic field fluctuation is taken into account to reflect the intensity of electromagnetic interference on the monitoring of vacuum degree; according to the difference in vacuum degree, operating power, magnetic field strength and the power influence index between any moment and the rest, combined with the magnetic field influence index, the influence index of the magnetic field at any moment and the rest is determined. The measurement distance between each moment has the beneficial effect of increasing the distance between the vacuum degrees at different moments by introducing the power influence index, thereby avoiding the temperature rise caused by the acceleration magnetic field change, which makes it impossible to distinguish the normal fluctuation of the vacuum degree caused by the temperature from the abnormal fluctuation caused by the change of the gas amount; based on the measurement distance, an abnormality detection algorithm is used to clean the vacuum degree at all moments, and the chamber leakage of the vacuum chamber is monitored in combination with the deep learning model. The beneficial effect is that by predicting the vacuum degree after cleaning, the interference of redundant information in the data is reduced, and the accuracy of chamber leakage monitoring of the vacuum chamber is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] A chamber leakage monitoring method for a medical cyclotron vacuum chamber of the present application is further described in detail below in conjunction with the accompanying drawings.

[0048] Figure 1 A flow chart of the steps of a chamber leakage monitoring method for a medical cyclotron vacuum chamber provided in an embodiment of the present application;

[0049] Figure 2 A flowchart of the steps of a method for obtaining the monitoring sensitivity of each monitoring period provided in an embodiment of the present application;

[0050] Figure 3 A flowchart of the steps of a method for obtaining a magnetic field influence index at any time provided in an embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the following is a further detailed description of a chamber leakage monitoring method for a medical cyclotron vacuum chamber proposed in the present application in conjunction with the accompanying drawings and implementation examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0053] See also Figure 1 , which shows a flow chart of the steps of a chamber leakage monitoring method for a medical cyclotron vacuum chamber provided by an embodiment of the present application, the method comprising the following steps:

[0054] Step 1: Real-time acquisition of the vacuum degree in the vacuum chamber at each moment when the medical cyclotron is in operation, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment.

[0055] The cyclotron is mainly composed of a magnetic field system, a vacuum system, a radio frequency system, an ion source system, a beam extraction system, a target system, and a cooling system. The core of a medical cyclotron includes a circular acceleration path located in a magnetic field, in which an ion source generates and emits charged particles, such as negative hydrogen ions. These particles move in a circle between two semicircular electrode boxes (D boxes), accelerated by the alternating radio frequency electric field, and accelerated outward along a spiral track in a constant magnetic field. As the speed of the particles increases, the radius of the particle's orbit increases, but the magnetic field guides them to always remain in the acceleration region. When the particles reach the required maximum energy, they are guided to the target material through the beam extraction system, and a nuclear reaction occurs to produce radioactive nuclides.

[0056] The vacuum system of a medical cyclotron accelerator includes: a vacuum chamber, a vacuum pump, a vacuum gauge, and a control section. The vacuum system needs to maintain the vacuum degree in the vacuum chamber to reduce the loss of the beam and the activation inside the accelerator, while providing an insulating environment for the high-frequency electric field to protect the precision components inside the accelerator from moisture and pollutants in the air, extend the service life of the equipment, and reduce maintenance costs. Therefore, in order to reduce the interference in the acceleration process of charged particles, it is necessary to monitor the gas pressure in the vacuum chamber in real time through a vacuum gauge to ensure that the required vacuum degree is reached and maintained in the vacuum chamber.

[0057] Based on the above analysis, during the operation of the accelerator, the vacuum degree in the vacuum chamber is monitored in real time by the vacuum gauge in the vacuum system of the medical cyclotron accelerator; in the medical cyclotron accelerator, the vacuum degree directly affects the motion stability and acceleration efficiency of the charged particles. The higher the vacuum degree, the lower the pressure in the vacuum chamber, the better the stability of the particle motion, and the higher the acceleration efficiency.

[0058] The vacuum pump in the vacuum system of the medical cyclotron will also affect the vacuum degree in the vacuum chamber. The higher the power of the vacuum pump, the faster the gas in the vacuum chamber is pumped out. Therefore, the operating power of the vacuum pump during the operation of the accelerator is collected in real time.

[0059] In a medical cyclotron, the electromagnetic field and mechanical vibration generated during the operation of the accelerator may also interfere with the data measured by the vacuum gauge. In order to reduce the interference of the accelerator operation process on the vacuum degree detection, the magnetic field strength during the medical cyclotron process is collected in real time.

[0060] The collected data were normalized to remove the dimension in the data, and the vacuum degree in the vacuum chamber at each moment when the medical cyclotron was running, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment were obtained.

[0061] Preferably, in this embodiment, the maximum and minimum value normalization method is used for normalization processing, wherein the maximum and minimum value normalization method is a well-known technology and will not be described in detail here.

[0062] Thus, the vacuum degree in the vacuum chamber at each moment when the medical cyclotron is in operation, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment are obtained.

[0063] Step 2: Based on the changing trend of the operating power at all times, divide all times into multiple monitoring periods; analyze the discreteness of the rate of change of the vacuum degree in each monitoring period, as well as the correlation between the changing trend of the vacuum degree and the changing trend of the operating power, to obtain the monitoring sensitivity of each monitoring period.

[0064] The normal operation of medical cyclotrons requires a high vacuum environment to prevent charged particles from colliding with other atoms during the acceleration process and losing energy. Secondly, the vacuum pump plays a vital role in the cyclotron. The vacuum pump extracts the vacuum in the vacuum chamber in real time to maintain the required high vacuum environment in the vacuum chamber.

[0065] During the operation of the medical cyclotron, the operating power of the vacuum pump is not constant. When the accelerator is started, the vacuum system will operate at a higher power to quickly reduce the gas pressure in the vacuum chamber and quickly remove air and other gas molecules. Once the required vacuum degree is reached, the power of the vacuum pump will be reduced accordingly and enter the maintenance mode, working continuously at a lower power to ensure the stability of the vacuum degree.

[0066] Further, the step flow chart of the method for obtaining the monitoring sensitivity of each monitoring period provided in the embodiment of the present application is as follows: Figure 2 shown.

[0067] First, the change of vacuum degree in the vacuum chamber can be combined with the change characteristics of the operating power of the vacuum pump, and all moments can be divided into different monitoring periods for analysis, specifically:

[0068] The operating power at each moment and its corresponding power are combined into a two-dimensional array; curve fitting is performed on the two-dimensional array at all moments and recorded as a power curve;

[0069] Obtain extreme points and inflection points of the power curve; use all extreme points and inflection points as segmentation points to divide all moments into multiple monitoring periods;

[0070] Preferably, in this embodiment, the least squares method is used for curve fitting, wherein the least squares method is a well-known technology and will not be described in detail here; secondly, the extreme points and inflection points are obtained by using the derivative method, wherein the calculation of the derivative method is a well-known technology and will not be described in detail here.

[0071] It should be noted that the inflection point can reflect the change in the speed of change of the operating power of the vacuum pump at a certain moment, and it is divided into different monitoring periods to reflect the different operating stages of the vacuum pump.

[0072] Furthermore, during the operation of the medical cyclotron, the air in the vacuum chamber is pumped out by a vacuum pump. Therefore, for the vacuum degree in different monitoring periods, under ideal conditions, the change in vacuum degree is similar to the change in the operating power of the vacuum pump. The operating power of the vacuum pump directly reflects the operating status of the vacuum pump, and the vacuum gauges installed at different positions in the vacuum chamber have a certain delay in responding to the vacuum level in the chamber. In order to ensure the accuracy of abnormal vacuum detection in different states of the vacuum chamber, it is necessary to evaluate the monitoring sensitivity generated by the relative position of the vacuum pump and the vacuum gauge.

[0073] Based on the above analysis, the monitoring sensitivity is determined by analyzing the correlation between the operating power and the rate of change of vacuum degree, specifically:

[0074] Each moment and its corresponding vacuum degree are formed into a two-dimensional array; a curve fitting is performed on the two-dimensional array of all moments, recorded as a vacuum curve, and the derivative value of the vacuum curve at each moment is recorded as the vacuum change rate at each moment;

[0075] Preferably, in this embodiment, the least square method is used for curve fitting, wherein the least square method is a well-known technology and will not be described in detail here.

[0076] Calculating the discrete degree of the vacuum change rate at all times in each monitoring period;

[0077] Preferably, in this embodiment, the variance of the vacuum change rate at all times in each monitoring period is calculated.

[0078] The derivative value of the power curve at each moment is recorded as the power change rate at each moment;

[0079] Calculating the correlation between the vacuum change rate and the power change rate at all times in each monitoring period;

[0080] Preferably, in this embodiment, the degree of correlation is measured by calculating the Pearson correlation coefficient of the vacuum change rate and the power change rate at all times in each monitoring period, wherein the calculation of the Pearson correlation coefficient is a well-known technique and will not be described in detail herein.

[0081] The ratio of the correlation degree to the discrete degree is used as the monitoring sensitivity of each monitoring period;

[0082] It should be noted that, the smaller the discreteness, the smaller the difference in the rate of change of the vacuum degree in the corresponding monitoring period, and the vacuum degree shows a monotonic change, because each monitoring period is divided according to the monotonicity of the operating power and the inflection point. Therefore, the smaller the discreteness, the higher the sensitivity of the vacuum gauge to the vacuum degree monitoring during the vacuum pump's vacuum extraction process; secondly, the greater the correlation, the higher the similarity between the rate of change of the operating power and the rate of change of the vacuum degree, and the greater the resulting monitoring sensitivity, which means that the vacuum gauge in different operating stages can timely capture the vacuum degree changes caused by the operation of the vacuum pump, that is, the vacuum gauge has a higher sensitivity to monitoring the vacuum pump's vacuum extraction process.

[0083] At this point, the monitoring sensitivity of each monitoring period is obtained.

[0084] Step 3, according to the interval time between any moment and the different moments before it, and the change of the operating power at different moments, combined with the monitoring sensitivity, determine the power impact index of any moment; analyze the amplitude difference of the magnetic field strength at all moments in different monitoring time periods in the frequency domain, and determine the relative amplitude difference of each monitoring time period; based on the change trend of the magnetic field strength at any moment, combined with the relative amplitude difference, calculate the magnetic field impact index of any moment.

[0085] Furthermore, during the operation of the cyclotron, as the accelerating magnetic field and the vacuum pump change, the vacuum degree in the vacuum chamber will also change to a certain extent. The vacuum degree detected by the vacuum gauge will be mainly affected by the vacuum pump power, the cyclotron acceleration magnetic field, etc. Therefore, when performing data cleaning on the monitored vacuum degree by measuring the distance between the vacuum degrees at different times, it is necessary to consider the impact of the accelerating magnetic field and the probability of the vacuum pump on the vacuum degree monitoring.

[0086] Secondly, in addition to the change in the operating power of the vacuum pump, for different monitoring periods, the change in vacuum degree has a certain delay relationship with the change in operating power. The operating power in different monitoring periods is different, and its impact on the vacuum degree is also different. The impact of the operation of the vacuum pump at each moment is related to the operating power of the vacuum pump at the previous moment. Therefore, the power impact index is determined by the change in operating power at each moment and all moments in the previous monitoring period, which is:

[0087] Calculating the product of the operating power at each moment and the power change rate as the stage characterization value at each moment;

[0088] It should be noted that the operating power at each moment reflects the working state of the vacuum pump at a certain moment, that is, the working intensity of the vacuum pump in extracting vacuum; the power change rate reflects the changing speed of the operating power, for example, whether the operating power is gradually increasing or decreasing, or remains constant, indicating the operating trend and efficiency of the vacuum pump; the stage characterization value can reflect the operating stage of the vacuum pump at a certain moment, for example, the larger the stage characterization value, the more the vacuum pump is in a stage of rapid vacuum extraction, and the smaller the stage characterization value, the more the vacuum pump is in a stable operation stage.

[0089] Obtain the starting time of the previous monitoring period of any moment in the corresponding monitoring period, and record all the moments between the starting time and any moment as the local impact period of any moment;

[0090] The calculation method of the power impact index at any moment is: ,in, For the The power impact index at the moment, For the The local impact period of the moment The monitoring sensitivity of the monitoring period to which the moment belongs, For the The time and its local influence period The time interval between moments, For the The local impact period of the moment The stage characterization value at the moment, For the The number of all moments in the local impact period of a moment, is a normalization function. In this embodiment, a sigmoid function is used for normalization processing. The sigmoid function is a well-known technology and will not be described in detail here.

[0091] It should be noted that The closer the time interval is, the greater the monitoring sensitivity of the monitoring period. The larger the When monitoring the vacuum degree at all times, the greater the impact of the vacuum pump's operating stage, the more The longer the time interval is, the smaller the monitoring sensitivity of the monitoring period is. The smaller the value, the greater the delay in the change of vacuum degree when the operating power of the vacuum pump changes. The greater the impact of the moment, therefore, when When smaller or larger, The larger the power influence index is, the greater the influence of the vacuum pump operation on the vacuum degree monitoring at the corresponding moment is.

[0092] Furthermore, in addition to being affected by the operating power of the vacuum pump at different operating stages, the vacuum degree monitored by the vacuum gauge is also affected by the accelerating magnetic field during the operation of the accelerator, where the interaction between the high-frequency electric field and the magnetic field during the acceleration stage of the ions will generate heat, which may cause the temperature in the vacuum chamber to rise. The increase in temperature will cause the activity of the residual gas molecules to increase, thereby causing fluctuations in the vacuum degree, which are caused by the activity of the gas molecules, not by changes in the amount of gas. Therefore, when measuring the distance between vacuum degrees, the normal fluctuations in vacuum degree caused by temperature and the abnormal fluctuations caused by changes in the amount of gas should be distinguished as much as possible.

[0093] Secondly, the electromagnetic interference generated by the cyclotron during operation may also affect the vacuum degree monitored by the vacuum gauge, causing noise interference in the vacuum degree detection process; in the acceleration stage, in order to keep the radius of the particle's cyclotron orbit unchanged, the size and frequency of the magnetic field need to be adjusted. This change in magnetic field strength may interfere with the electronic components inside the vacuum gauge, causing deviations in vacuum degree monitoring.

[0094] Based on the above analysis, the magnetic field influence index is determined by analyzing the magnetic field strength. The step flow chart of the method for obtaining the magnetic field influence index at any time provided in the embodiment of the present application is as follows: Figure 3 As shown, specifically including:

[0095] The magnetic field strengths at each moment and the corresponding magnetic field strengths are formed into a two-dimensional array; curve fitting is performed on the two-dimensional arrays at all moments, recorded as a magnetic field curve, and the derivative value of the magnetic field curve at each moment is recorded as the magnetic field change rate at each moment;

[0096] Perform frequency domain analysis on the magnetic field intensity at all times in each monitoring period to obtain the amplitude of each frequency component;

[0097] Preferably, in this embodiment, fast Fourier transform is used to perform frequency domain analysis, wherein fast Fourier transform is a well-known technology and will not be described in detail here.

[0098] The difference between the amplitude of each frequency component in each monitoring period and the amplitude of the same frequency component in the previous monitoring period is recorded as the amplitude difference;

[0099] Preferably, in this embodiment, the absolute value of the difference between the amplitude of each frequency component in each monitoring period and the amplitude of the same frequency component in the previous monitoring period is recorded as the amplitude difference;

[0100] Taking the mean of the amplitude differences of all frequency components in each monitoring period as the relative amplitude difference of each monitoring period;

[0101] Calculate the product of the magnetic field strength at any moment and the magnetic field change rate, and record it as the magnetic field interference degree at any moment;

[0102] The product of the magnetic field interference degree at any moment and the relative amplitude difference of the monitoring period to which the any moment belongs is taken as the magnetic field influence index at any moment;

[0103] It should be noted that the greater the magnetic field change rate, the more frequent the adjustment of the magnetic field by the cyclotron during the acceleration stage, and the greater the magnetic field interference, which means that the vacuum degree monitoring at this time will be subject to stronger electromagnetic interference; secondly, the greater the relative amplitude difference, the more significant the fluctuation of the magnetic field strength in different monitoring time periods, and the electromagnetic changes will cause the vacuum degree monitoring to be more disturbed. The greater the obtained magnetic field influence index, the greater the degree of vacuum degree monitoring is affected by the magnetic field, and the more likely the vacuum degree monitoring will deviate.

[0104] At this point, the power impact index and the magnetic field impact index at any moment are obtained.

[0105] Step 4: Determine the measurement distance between any moment and the remaining moments according to the differences in vacuum degree, operating power, magnetic field strength and the power influence index between any moment and the remaining moments, combined with the magnetic field influence index; based on the measurement distance, use an anomaly detection algorithm to perform data cleaning on the vacuum degree at all moments, and monitor the chamber leakage of the vacuum chamber in combination with a deep learning model.

[0106] Furthermore, based on the power influence index and the magnetic field influence index, the distances between different moments are measured, the abnormal vacuum degrees at all moments are eliminated, and the monitored vacuum degrees are cleaned, specifically:

[0107] The vacuum degree, operating power and magnetic field strength at each moment are combined into the characteristic vector at each moment;

[0108] The difference between the feature vector at any moment and the feature vectors at other moments is used to form a difference vector;

[0109] Preferably, in this embodiment, the square sum of the differences between the feature vector at any moment and the feature vectors at other moments is used to form a difference vector. It should be noted that, for ease of understanding, it is assumed that The eigenvector at time is , No. The eigenvector at time is , then the difference vector is .

[0110] The sum of all elements in the difference vector is taken as the relative difference between any one moment and the other moments;

[0111] The calculation method of the metric distance between any one moment and the other moments is: ,in, For the Moment and The measured distance between moments, For the Moment and The mean value of the magnetic field influence index at time, For the The power impact index at the moment, For the The power impact index at the moment, For the Moment and The relative difference between moments.

[0112] It should be noted that The larger the value is, the more significant the influence of magnetic field interference is between the two moments, and there may be deviation in the monitoring data. Then the relative difference between the two moments should be larger, thereby increasing the difference between the two moments. The larger the relative difference, the different effects of the operating power of the vacuum pump on the two moments are. By introducing the influence of the operating power of the vacuum pump, the temperature rise caused by the change in the accelerated magnetic field can be used to distinguish the normal fluctuation of the vacuum degree caused by the temperature from the abnormal fluctuation caused by the change in the gas volume, increase the measurement distance of the vacuum degree with abnormal fluctuations, and improve the accuracy of subsequent analysis. The larger the relative difference, the greater the difference in vacuum degree, operating power, and magnetic field strength between the two moments, and the larger the obtained measurement distance, indicating the difference between the two moments, so as to reflect the degree of state difference between the two moments, so that the vacuum degree at all moments can be cleaned more accurately according to the measurement distance between different moments in the future, thereby eliminating abnormal vacuum degree data.

[0113] Based on the measurement distance, anomaly detection is performed on the vacuum degree at all times through an anomaly detection algorithm, and abnormal vacuum degrees are eliminated; and the missing vacuum degrees are filled in completely through a missing value filling method to obtain the vacuum degrees at all times after cleaning;

[0114] Preferably, in this embodiment, the LOF algorithm (Local Outlier Factor) is used for anomaly detection, wherein the LOF algorithm is a well-known technology and will not be described in detail here; secondly, the missing vacuum degree is filled by the cubic spline interpolation method, wherein the cubic spline interpolation method is a well-known technology and will not be described in detail here.

[0115] Furthermore, the vacuum degree after cleaning is used to monitor the leakage of the vacuum chamber using a deep learning model, specifically:

[0116] The vacuum degree monitored in historical periods, the operating power of the vacuum pump, and the specification parameters of the vacuum chamber of the cyclotron, including parameters such as size, chamber material, and vacuum pump configuration, are selected. The data in each historical period are manually scored to evaluate the probability of leakage in the corresponding vacuum chamber in each historical period. The vacuum degree and corresponding leakage probability of all historical periods are used as training sets.

[0117] It should be noted that the values ​​of manual scoring are 0.01 and 0.02 respectively. , 0.09, 1, contains a total of 100 probability values, corresponding to 100 vacuum chamber leakage probabilities.

[0118] The deep learning model is trained through the training set to obtain a trained deep learning model;

[0119] Preferably, in this embodiment, the neural network model is trained, wherein the neural network model is a well-known technology and will not be described in detail here; secondly, the loss function of the neural network model is cross entropy.

[0120] The vacuum degree, operating power and vacuum chamber specification parameters of the cyclotron at all times after cleaning are used as inputs to the trained deep learning model to obtain the leakage probability of the vacuum chamber;

[0121] If the leakage probability is greater than a preset threshold, there is leakage in the vacuum chamber of the cyclotron; if the leakage probability is less than or equal to the preset threshold, there is no leakage in the vacuum chamber of the cyclotron.

[0122] Preferably, in this embodiment, the preset threshold value is 0.6. As other implementation modes, the implementer can set it according to the actual situation.

[0123] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0124] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the present application. It should be pointed out that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application without departing from the content of the technical solution of the present application, shall fall within the protection scope of the technical solution of the present application.

Claims

1. A chamber leakage monitoring method for a medical cyclotron vacuum chamber, characterized in that: The method comprises the following steps: Real-time acquisition of the vacuum degree at each moment in the vacuum chamber of the medical cyclotron accelerator, the operating power of the vacuum pump at each moment, and the magnetic field strength at each moment; Based on the changing trend of the operating power at all times, all times are divided into multiple monitoring periods; the discreteness of the rate of change of the vacuum degree in each monitoring period, as well as the correlation between the changing trend of the vacuum degree and the changing trend of the operating power are analyzed to obtain the monitoring sensitivity of each monitoring period; Determine the power impact index at any moment according to the interval between any moment and different moments before it, and the change of the operating power at different moments, combined with the monitoring sensitivity; Analyze the amplitude difference of the magnetic field strength at all times in different monitoring periods in the frequency domain to determine the relative amplitude difference of each monitoring period; calculate the magnetic field influence index at any time based on the change trend of the magnetic field strength at any time and the relative amplitude difference; Determine the measurement distance between any moment and the remaining moments according to the difference in vacuum degree, operating power, magnetic field strength and the power influence index between any moment and the remaining moments, combined with the magnetic field influence index; Based on the metric distance, an anomaly detection algorithm is used to clean the vacuum degree at all times, and the chamber leakage of the vacuum chamber is monitored in combination with a deep learning model.

2. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: The method divides all time periods into multiple monitoring periods, including: The operating power at each moment and its corresponding power are combined into a two-dimensional array; curve fitting is performed on the two-dimensional array at all moments and recorded as a power curve; Obtain the extreme points and inflection points of the power curve; use all extreme points and inflection points as segmentation points, and divide all moments into multiple monitoring periods.

3. A chamber leakage monitoring method for a medical cyclotron vacuum chamber as claimed in claim 2, characterized in that: The obtaining of the monitoring sensitivity of each monitoring period includes: Each moment and its corresponding vacuum degree are formed into a two-dimensional array; a curve fitting is performed on the two-dimensional array of all moments, recorded as a vacuum curve, and the derivative value of the vacuum curve at each moment is recorded as the vacuum change rate at each moment; Calculating the discrete degree of the vacuum change rate at all times in each monitoring period; The derivative value of the power curve at each moment is recorded as the power change rate at each moment; Calculating the correlation between the vacuum change rate and the power change rate at all times in each monitoring period; The ratio of the correlation degree to the discrete degree is used as the monitoring sensitivity of each monitoring period.

4. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 3, characterized in that: The determining of the power impact index at any moment includes: Obtain the starting time of the previous monitoring period of any moment in the corresponding monitoring period, and record all the moments between the starting time and the any moment as the local impact period of the any moment; Calculating the product of the operating power at each moment and the power change rate as the stage characterization value at each moment; No. Power impact index at the moment The calculation method is: ,in, For the The local impact period of the moment The monitoring sensitivity of the monitoring period to which the moment belongs, For the The time and its local influence period The time interval between moments, For the The local impact period of the moment The stage characterization value at the moment, For the The number of all moments in the local impact period of a moment, is the normalization function.

5. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: Determining the relative amplitude difference in each monitoring period includes: Perform frequency domain analysis on the magnetic field intensity at all times in each monitoring period to obtain the amplitude of each frequency component; The difference between the amplitude of each frequency component in each monitoring period and the amplitude of the same frequency component in its adjacent monitoring period is recorded as the amplitude difference; The average value of the amplitude differences of all frequency components in each monitoring period is taken as the relative amplitude difference of each monitoring period.

6. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: The calculating the magnetic field influence index at any moment includes: The magnetic field strengths at each moment and the corresponding magnetic field strengths are formed into a two-dimensional array; curve fitting is performed on the two-dimensional arrays at all moments, recorded as a magnetic field curve, and the derivative value of the magnetic field curve at each moment is recorded as the magnetic field change rate at each moment; Calculate the product of the magnetic field strength at any moment and the magnetic field change rate, and record it as the magnetic field interference degree at any moment; The product of the magnetic field interference degree at any moment and the relative amplitude difference of the monitoring period to which the any moment belongs is used as the magnetic field influence index at any moment.

7. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: Determining the metric distance between any one moment and the remaining moments includes: Determine the relative difference between any one moment and the other moments based on the difference in vacuum degree, operating power, and magnetic field strength between any one moment and the other moments; No. Moment and Metric distance between moments The calculation method is: ,in, For the Moment and The mean value of the magnetic field influence index at time, For the The power impact index at the moment, For the The power impact index at the moment, For the Moment and The relative difference between moments.

8. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 7, characterized in that: Determining the relative difference between any one moment and the remaining moments includes: The vacuum degree, operating power and magnetic field strength at each moment are combined into the characteristic vector at each moment; The difference between the feature vector at any moment and the feature vectors at other moments is used to form a difference vector; The sum of all elements in the difference vector is taken as the relative difference between any one moment and the remaining moments.

9. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: The data cleaning of the vacuum degree at all times includes: Based on the measurement distance, anomaly detection is performed on the vacuum degree at all times through an anomaly detection algorithm, and abnormal vacuum degrees are eliminated; and the missing vacuum degrees are filled in through a missing value filling method.

10. The chamber leakage monitoring method of a medical cyclotron vacuum chamber according to claim 1, characterized in that: The monitoring of chamber leakage of the vacuum chamber comprises: The vacuum degree and operating power at all times after cleaning are used as inputs to the deep learning model to obtain the leakage probability of the vacuum chamber. If the leakage probability is greater than a preset threshold, there is leakage in the vacuum chamber of the cyclotron; otherwise, there is no leakage in the vacuum chamber of the cyclotron.

Images