A method, apparatus, device and medium for processing plasma profile data

Abstract

This invention discloses a method, apparatus, device, and medium for processing plasma profile data. The invention detects transient events in boundary radiation signals and constructs time exclusion intervals. By combining the sampling duration of each measurement point to determine the sampling time interval, measurement data whose sampling process overlaps with the transient event occurrence time are identified as interfered data and discarded. This ensures that subsequent data used in profile analysis reflects the steady-state profile characteristics during the transient event interval, avoiding distortion caused by transient disturbances in profile parameter extraction. Based on the magnetic axis flux and boundary flux in magnetic balance data, interpolation is used to uniformly map the geometric spatial positions of each valid data point to normalized flux coordinates. After filtering data points within a preset flux coordinate range, profile model fitting is automatically performed. This achieves automatic extraction of profile parameters from different diagnostic measurement data under a unified coordinate system without manual intervention.

Description

Technical Field

[0001] This invention relates to the technical field of tokamak plasma diagnostic data processing, specifically to a plasma profile data processing method, apparatus, equipment, and medium. Background Technology

[0002] During the operation of a tokamak magnetic confinement fusion device, when the plasma enters the high confinement mode (H mode), an edge transport barrier forms in its edge region, accompanied by the appearance of a platform structure. The evolution of parameters such as the height, width, center position, and gradient of the platform over time can reflect the confinement performance and edge stability of the device, and is an important basis for discharge operation analysis and related physics research.

[0003] Taking the HL-3 space station in China as an example, its experimental data is characterized by large data volume, numerous diagnostic channels, and dense time slices. To obtain the slab parameters during the H-mode, it is typically necessary to integrate multiple diagnostic data, including electron temperature and electron density profiles obtained from Thomson scattering (TS) diagnostics, ion temperature profiles obtained from charge-exchange recombination spectroscopy (CXRS) diagnostics, Ha / Da boundary radiation signals, gfile files obtained from EFIT equilibrium inversion, and H-mode time windows annotated by the experimenters. These data are not uniform in terms of time reference and spatial coordinates; only by organizing them into a unified time window and magnetic coordinate system can comparable and statistically significant slab parameter results be obtained.

[0004] Currently, most of the data processing is done manually by researchers: first, they download diagnostic data files from the big data platform or local directory; then, they manually define the start and end times of the H-mode based on the signal waveform; they identify the periods of edge localized mode (ELM) bursts based on experience; and then, they perform coordinate mapping, outlier removal, and curve fitting on the profile data of each time slice one by one. Piecewise functions or hyperbolic tangent functions are usually used for fitting, and the initial fitting values ​​and fitting intervals are mostly set manually or selected interactively on a graphical interface. Although some scripts can automate individual steps such as data download, file reading, coordinate mapping, ELM detection, or curve fitting, these scripts are independent of each other, lack a unified data organization method, and are difficult to connect into a complete data processing workflow.

[0005] The existing processing methods described above have the following shortcomings. First, during an ELM outbreak, the pedestal profile undergoes instantaneous collapse and recovery. If the data from this period is not excluded, the fitted pedestal parameters will not reflect the quasi-steady-state characteristics during the ELM interval. Commonly used single-threshold or single-derivative detection methods are poorly adaptable to baseline drift, low-amplitude spikes, spike tails, and noise spikes, resulting in unstable removal effects. Second, the fitting process relies on manually given initial values, boundaries, and fitting intervals. However, the profile scale and pedestal position vary significantly under different shot numbers, heating powers, and density platforms, making it difficult to ensure compatibility of fixed parameters. Manual selection also leads to poor repeatability of the processing results. Third, the diagnostic data differ in format, time base, spatial coordinates, and data quality. Manual splicing can easily cause inconsistencies in shot numbers, time slices, coordinates, and parameter versions. Fourth, a single HL-3 experiment generates a large number of H-mode shot numbers and time slices. Relying on manual interpretation of each image makes it difficult to complete batch statistical analysis, failing to provide stable data input for subsequent research.

[0006] In summary, in the relevant technologies, there are technical problems in the extraction of H-mode plasma profile parameters, such as the transient interference of edge local mode (ELM) and the lack of uniformity in the spatiotemporal reference of diagnostic data, which leads to insufficient accuracy in the extraction of effective steady-state plasma profile parameters. Summary of the Invention

[0007] The technical problem this invention aims to solve is that, in related technologies, the extraction of H-mode plasma profile parameters suffers from insufficient accuracy due to transient interference from edge localized modes (ELMs) and inconsistent spatiotemporal references in diagnostic data. The purpose is to provide a plasma profile data processing method, apparatus, device, and medium that solves the aforementioned technical problem.

[0008] This invention is achieved through the following technical solution:

[0009] In a first aspect, the present invention provides a method for processing plasma profile data, comprising:

[0010] Acquire raw measurement data, boundary radiation signals, and corresponding magnetic balance data from the profile diagnostic equipment; wherein, the raw measurement data includes time information and spatial location information of each measurement point;

[0011] Transient events are detected in the boundary radiation signal, and time exclusion intervals are constructed based on the detected transient events;

[0012] Based on the sampling duration of the original measurement data, determine the sampling time interval corresponding to each measurement point, and determine whether there is an overlap between each sampling time interval and the time exclusion interval. If there is an overlap, mark the corresponding original measurement data as interfered data and remove it.

[0013] For valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data.

[0014] Valid data points that meet the preset magnetic flux coordinate range are selected, and profile model fitting is performed on the valid data points to output plasma profile parameters.

[0015] Furthermore, the transient event detection of the boundary radiation signal includes:

[0016] The boundary radiation signal is smoothed, and the first derivative sequence of the smoothed signal is calculated. Local fluctuations that satisfy the sequence characteristics are identified as the first candidate transient events. The sequence characteristics are characterized by a rapid rise, reaching a peak, and then falling back.

[0017] Local maxima in the smoothed signal are identified, and local maxima with peak heights greater than a preset minimum peak height are identified as second candidate transient events; wherein the amplitude of the local maxima is greater than the signal value in a preset neighborhood.

[0018] The local background level and local fluctuation of the boundary radiation signal are calculated based on the sliding time window. A dynamic threshold that changes with time is constructed based on the local background level and local fluctuation. Signal segments whose amplitude exceeds the dynamic threshold are identified as the third candidate transient events.

[0019] The first, second, and third candidate transient events are fused, and the candidate transient events retained after fusion are used as the detected transient events.

[0020] Furthermore, the fusion of the first candidate transient event, the second candidate transient event, and the third candidate transient event includes:

[0021] The first candidate transient event, the second candidate transient event, and the third candidate transient event are merged into a candidate event set;

[0022] Perform temporal clustering on the candidate event set, grouping candidate transient events whose occurrence time interval is less than or equal to a preset time tolerance into the same event cluster;

[0023] For each event cluster, the number of detection methods from which the candidate transient events within the event cluster originate is counted, which is used as the support number for that event cluster.

[0024] Event clusters with a number of supports greater than or equal to a preset support threshold are retained, while event clusters with a number of supports less than the preset support threshold undergo morphological verification. Event clusters that pass the morphological verification and the directly retained event clusters are used together as detected transient events.

[0025] Furthermore, the step of constructing a time exclusion interval based on the detected transient events includes:

[0026] For each detected transient event, the rising slope of the transient event is determined based on the signal changes before the peak time, and the falling slope of the transient event is determined based on the signal changes after the peak time.

[0027] Transient events that only have a rising edge but no falling edge, whose rising or falling slope does not reach the corresponding preset slope threshold, or whose peak duration is less than the preset minimum duration are removed as noise spikes.

[0028] For transient events retained after removing noise spikes, determine the peak time, peak amplitude, and local background value before the peak time of the transient event;

[0029] Based on the local background value, the target amplitude level is determined according to the effective amplitude of the peak amplitude relative to the local background value and the first preset ratio.

[0030] The signal sequence is searched forward and backward from the peak time, and the time when the signal amplitude first crosses the target amplitude level is taken as the left boundary time and the right boundary time, respectively. The reference peak width is determined based on the left boundary time and the right boundary time.

[0031] Centered on the peak moment, the length of the second preset proportion of the reference peak width is extracted as the time exclusion interval corresponding to the transient event.

[0032] Furthermore, the step of determining the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, and determining whether each sampling time interval overlaps with the time exclusion interval, includes:

[0033] When the original measurement data record has a sampling start time and a sampling end time, the time interval between the sampling start time and the sampling end time is determined as the sampling time interval of the corresponding measurement point;

[0034] When the original measurement data records the nominal time and integration duration but not the sampling start and end times, the time interval centered on the nominal time and with the integration duration as the time length will be determined as the sampling time interval for the corresponding measurement point.

[0035] When the sampling time interval of any measurement point overlaps with any of the time exclusion intervals, the original measurement data corresponding to that measurement point is marked as interfered data.

[0036] When the original measurement data does not record the sampling start and end times and the integration duration, the original measurement data corresponding to the measurement points whose nominal times fall within any of the time exclusion intervals are marked as interfered data.

[0037] Furthermore, for the valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data, including:

[0038] Determine the two-dimensional flux grid, magnetic axis flux, and boundary flux in the magnetic balance data;

[0039] Based on the spatial location of each valid data point, bilinear interpolation is performed on the two-dimensional magnetic flux grid to obtain the poloidal magnetic flux value corresponding to each valid data point;

[0040] Based on the first difference between the pole magnetic flux and the magnetic axis flux, and the second difference between the boundary magnetic flux and the magnetic axis flux, the ratio of the first difference to the second difference is determined as the normalized magnetic flux coordinate corresponding to each valid data point.

[0041] Further, the process of selecting valid data points that meet the preset magnetic flux coordinate range and performing profile model fitting on the valid data points to output plasma profile parameters includes:

[0042] Retain valid data points whose normalized flux coordinates are within the preset flux coordinate range;

[0043] The initial value of the sump top value in the preset four-parameter model is estimated based on the statistics of the retained effective data points in the core magnetic flux region, and the initial value of the sump base value in the four-parameter model is estimated based on the statistics of the retained effective data points in the boundary magnetic flux region.

[0044] Based on the position of the maximum gradient of the profile formed by the retained valid data points and the position corresponding to the preset amplitude ratio of the profile amplitude, the initial value of the platform center position in the four-parameter model is estimated by fusing the data.

[0045] The initial value of the platform width in the four-parameter model is estimated based on the gradient half-width of the profile.

[0046] Based on the range of magnetic flux coordinates and the range of profile amplitude of the retained valid data points, the fitting boundaries of each parameter are generated;

[0047] The initial values ​​of each parameter are perturbed to obtain multiple sets of initial parameters. The four-parameter model is called, and the retained effective data points are fitted based on each set of initial parameters and the fitting boundary. The best fitting result is selected from each fitting result according to the goodness of fit.

[0048] The apex top value, apex base value, apex center position, and apex width from the optimal fitting result are output as the plasma profile parameters.

[0049] In a second aspect, the present invention provides a plasma profile data processing apparatus, comprising:

[0050] Acquisition module: Acquires raw measurement data, boundary radiation signals, and corresponding magnetic balance data from the profile diagnostic equipment; wherein, the raw measurement data includes time information and spatial location information of each measurement point;

[0051] Detection module: performs transient event detection on boundary radiation signals and constructs time exclusion intervals based on the detected transient events;

[0052] Elimination module: Determine the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, determine whether there is an overlap between each sampling time interval and the time exclusion interval, and if there is an overlap, mark the corresponding original measurement data as interfered data and eliminate it;

[0053] Calculation module: For valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial location of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data.

[0054] Output module: Filter out valid data points that meet the preset magnetic flux coordinate range, perform profile model fitting on the valid data points, and output plasma profile parameters.

[0055] Thirdly, the present invention provides an electronic device, comprising: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, the instructions being executed by the one or more processors to cause the one or more processors to implement the method described above.

[0056] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0057] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0058] This invention detects transient events in boundary radiation signals and constructs time exclusion intervals. By combining the sampling duration of each measurement point to determine the sampling time interval, measurement data whose sampling process overlaps with the transient event occurrence time are identified as interfered data and discarded. This ensures that subsequent data used in profile analysis reflects the steady-state profile characteristics during the transient event interval, avoiding distortion caused by transient disturbances in profile parameter extraction. Furthermore, based on the magnetic axis flux and boundary flux in the magnetic balance data, interpolation is used to uniformly map the geometric spatial positions of each valid data point to normalized flux coordinates. After filtering data points within a preset flux coordinate range, profile model fitting is automatically performed. This achieves automatic extraction of profile parameters from different diagnostic measurement data under a unified coordinate system without manual intervention, ensuring the accuracy and repeatability of plasma profile parameter processing results. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0060] Figure 1 This is one of the flowcharts for a plasma profile data processing method provided in the embodiments of this specification;

[0061] Figure 2 This is a second flowchart of a plasma profile data processing method provided in the embodiments of this specification;

[0062] Figure 3 This is a schematic diagram of the electronic temperature profile fitting results provided in the embodiments of this specification;

[0063] Figure 4 This is a schematic diagram of the residual analysis of the electronic temperature profile fitting provided in the embodiments of this specification;

[0064] Figure 5 This is one of the schematic diagrams showing the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; wherein, the horizontal axis is the normalized magnetic flux coordinate; and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis;

[0065] Figure 6This is the second schematic diagram of the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; wherein, the horizontal axis is the normalized magnetic flux coordinate; and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis;

[0066] Figure 7 This is the third schematic diagram of the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; where the horizontal axis is the normalized magnetic flux coordinate; and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis.

[0067] Figure 8 This is the fourth schematic diagram of the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; where the horizontal axis is the normalized magnetic flux coordinate and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis;

[0068] Figure 9 This is the fifth schematic diagram of the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; where the horizontal axis is the normalized magnetic flux coordinate and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis;

[0069] Figure 10 This is the sixth schematic diagram of the fitting results of electron temperature profiles at different time slices provided in the embodiments of this specification; where the horizontal axis is the normalized magnetic flux coordinate; and the vertical axis is the electron temperature, which is the electron temperature measurement value obtained by TS diagnosis.

[0070] Figure 11 This is a schematic diagram of the electron temperature profile after removing zero values, provided in the embodiments of this specification.

[0071] Figure 12 This is a schematic diagram of the calculation results of the electronic temperature profile gradient provided in the embodiments of this specification;

[0072] Figure 13 A schematic diagram of the correction results for electron temperature profile data based on gradient threshold provided in the embodiments of this specification;

[0073] Figure 14 This is a comparative schematic diagram of ion temperature profiles across multiple time slices after removing zero values, provided in the embodiments of this specification. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are only for explaining this invention and are not intended to limit this invention.

[0075] like Figures 1-14 As shown, this embodiment provides a plasma profile data processing method. The execution entity of the method can be an electronic device with data processing capabilities, such as a server deployed on a laboratory intranet, a computing node in a computing cluster, or a personal computer. This electronic device interacts with the tokamak experimental big data platform (or local data cache directory) and a storage device storing the H-mode time window table.

[0076] Specifically, the execution entity acquires raw measurement data from profiling diagnostic equipment such as TS and CXRS, Ha / Da boundary radiation signals, and magnetic equilibrium data derived from EFIT equilibrium inversion, according to the shot number, diagnostic channel, and time slice configuration, through the application programming interface or local cache of the big data platform. It also reads the H-mode time window table annotated by the experimenters or identified by the program. Then, the execution entity sequentially completes the time reference unification of each diagnostic data, transient event detection and interference data removal, magnetic coordinate mapping, and profiling model fitting locally, and outputs the corresponding plasma profiling parameters, fitting plots, and statistical reports.

[0077] The method may include:

[0078] Step S11: Obtain the raw measurement data, boundary radiation signal and corresponding magnetic balance data of the profile diagnostic equipment; wherein, the raw measurement data includes the time information and spatial location information of each measurement point.

[0079] In this embodiment, the executing entity can acquire profile diagnostic data, boundary radiation diagnostic data, and magnetic balance data corresponding to the discharge process to be analyzed, and establish the correspondence between different types of data based on the shot number and time information to form a data foundation for subsequent transient event detection, interference data removal, and magnetic coordinate mapping.

[0080] Specifically, the profile diagnostic data, as raw measurement data, is used to characterize the parameter distribution of plasma at different spatial locations. In one embodiment, the profile diagnostic device includes at least one of a Thomson scattering diagnostic device and a charge-exchange composite spectroscopy diagnostic device. The raw measurement data obtained by the Thomson scattering diagnostic device may include at least one of electron temperature and electron density, while the raw measurement data obtained by the charge-exchange composite spectroscopy diagnostic device may include ion temperature. For different profile parameters, the executing entity can read the measurement data output by the corresponding diagnostic device and organize it into a corresponding profile data set according to measurement time and spatial location.

[0081] In this embodiment, for each measurement point, the raw measurement data includes at least the measured value, time information, and spatial location information. The time information may include one or more of the following: sampling start time, sampling end time, nominal time, and integration duration, used to subsequently determine the sampling time interval corresponding to each measurement point and to determine whether the measurement process is affected by transient events. The spatial location information is used to characterize the geometric measurement position of the measurement point in the plasma cross-section.

[0082] The executing entity also reads the spatial position information of each measurement point. For profile diagnostic data represented by a two-dimensional cross-section, the spatial position information includes the radial and vertical positions of the measurement point within the plasma cross-section. The spatial position information of each measurement point is associated with and stored in conjunction with the corresponding measurement value and time information, so that the measurement value, measurement time, and geometric spatial position of the same measurement point form a corresponding data record.

[0083] In this embodiment, the executing entity acquires a boundary radiation signal belonging to the same discharge process as the original measurement data. The boundary radiation signal is a time-series signal that is continuous over time or acquired according to a preset sampling frequency, used to characterize the change in radiation intensity of the plasma boundary region over time. In one embodiment, the boundary radiation signal includes at least one of a Ha signal, a Da signal, or other radiation diagnostic signals capable of characterizing transient disturbances at the plasma boundary.

[0084] For the boundary radiation signal, the execution entity reads the signal sampling time and the signal amplitude corresponding to each sampling time, and constructs the boundary radiation signal sequence in chronological order.

[0085] In this embodiment, the executing entity can also acquire magnetic balance data corresponding to the original measurement data. In one embodiment, the magnetic balance data is EFIT magnetic balance reconstruction data obtained based on the current discharge process, including a two-dimensional magnetic flux grid, magnetic axis flux, boundary flux, and corresponding magnetic balance parameters.

[0086] For magnetic balance data with time information, the execution entity simultaneously reads the time information corresponding to the magnetic balance data and determines the magnetic balance data corresponding to the current measurement data based on the measurement time of the original measurement data, so that the spatial position of the original measurement data can be mapped to magnetic coordinates based on the magnetic balance result under the corresponding discharge state.

[0087] After acquiring the aforementioned data, the executing entity uses the shot number and time information as data association information to establish a correspondence between the original measurement data, boundary radiation signal, and magnetic balance data. Specifically, the time information in the original measurement data is used for comparison with the subsequently constructed time exclusion interval; the spatial location information is used to calculate the normalized magnetic flux coordinates in conjunction with the magnetic balance data; and the boundary radiation signal is used to detect transient events and construct the time exclusion interval. This allows the three types of data to form an interconnected data set according to the same discharge process.

[0088] In this embodiment, the executing entity can also read the time metadata carried in each diagnostic data and unify the time axis of each diagnostic data to a preset unified experimental time benchmark. The time metadata includes timestamps, time units, time zero-point definitions, and sampling frequencies. The unified experimental time benchmark is a second-level time axis with the unified trigger time or discharge start time of the current shot as the time zero point.

[0089] In cases where the time units of different data sources are inconsistent, the execution entity converts the time information of the corresponding diagnostic data according to a preset conversion relationship. In cases where the zero point of time or the time reference is inconsistent, the execution entity performs translation correction on the time axis of the corresponding diagnostic data based on the trigger delay recorded by the data acquisition system or the pre-configured time offset parameter, so that the time information of the original measurement data, boundary radiation signal, and magnetic balance data are all mapped to a unified experimental time axis.

[0090] After unifying the time reference, the execution entity uses the unified experimental time axis as a benchmark to extract data for each diagnostic channel at the corresponding time according to the target analysis time slice. For continuously sampled signals, the execution entity uses nearest neighbor matching or linear interpolation to obtain the signal value at the target time. For low sampling rate or discrete measurement data, the execution entity selects the data point with the smallest time difference from the target time, and this time difference is less than a preset tolerance, as the data corresponding to the target time.

[0091] Furthermore, after acquiring the raw measurement data, the executing entity can check the integrity of the data records. If measurement values, time information, or spatial location information are missing, or if the corresponding data is invalid, NaN, or infinite and cannot be used in subsequent calculations, the corresponding data record is marked as invalid data and will not be used as input for subsequent valid data processing. Data records for which valid measurement values ​​and corresponding time and spatial location information can be obtained are retained as raw measurement data to be processed.

[0092] Step S12: Perform transient event detection on the boundary radiation signal and construct a time exclusion interval based on the detected transient events.

[0093] Step S13: Determine the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, and determine whether there is an overlap between each sampling time interval and the time exclusion interval. If there is an overlap, mark the corresponding original measurement data as interfered data and remove it.

[0094] Step S14: For valid data points in the original measurement data that are not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data.

[0095] Step S15: Select valid data points that meet the preset magnetic flux coordinate range, and perform profile model fitting on the valid data points to output plasma profile parameters.

[0096] In some implementations, the transient event detection of the boundary radiation signal includes:

[0097] Step S121: Smooth the boundary radiation signal, calculate the first derivative sequence of the smoothed signal, and determine the local fluctuations that satisfy the sequence characteristics as the first candidate transient events; wherein, the sequence characteristics are represented by a rapid rise, reaching a peak and then falling back.

[0098] In this embodiment, the executing entity first performs smoothing processing on the acquired boundary radiation signal. The boundary radiation signal is a Ha signal or a Da signal. Smoothing processing is used to suppress high-frequency noise in the boundary radiation signal, so that the change trend of the smoothed signal can be used for subsequent first-order derivative calculations. The window width of the smoothing processing can be preset according to the sampling rate of the boundary radiation signal and the time scale of the transient event to be detected.

[0099] After smoothing, the executing entity can calculate the first derivative sequence of the smoothed signal along the time axis.

[0100] Specifically, for two temporally adjacent sampling points, the signal amplitude difference between the latter and the former is calculated, and the ratio of this signal amplitude difference to the time interval between the two sampling points is taken as the corresponding first derivative value. A positive first derivative indicates that the boundary radiation signal is in an upward state, while a negative first derivative indicates that it is in a downward state.

[0101] In this embodiment, the executing entity searches for local fluctuations that satisfy the sequence characteristics in the first derivative sequence.

[0102] Specifically, the sequential characteristics include a rapid rise in the boundary radiation signal, reaching a local peak, and then falling back after the local peak. The executing entity can identify the local fluctuations that satisfy the sequential characteristics as the first candidate transient event and record the peak time and peak amplitude of the first candidate transient event.

[0103] When no local fluctuations satisfying the sequential characteristics are found, an empty first candidate transient event set is generated.

[0104] Step S122: Determine the local maxima in the smoothed signal, and identify the local maxima with a peak height greater than the preset minimum peak height as the second candidate transient event; wherein the amplitude of the local maxima is greater than the signal value in the preset neighborhood.

[0105] In this embodiment, the executing entity can sequentially read each sampling point in the smoothed signal and compare the signal amplitude of each sampling point with the signal amplitudes in the preset neighborhood before and after that sampling point. When the signal amplitude of a sampling point is greater than the signal amplitudes of other sampling points in the preset neighborhood, that sampling point is determined as a local maximum point.

[0106] In this embodiment, the executing entity can further determine the peak height of the local maximum point relative to the local background level and compare the peak height with a preset minimum peak height. When the peak height is greater than the preset minimum peak height, the local fluctuation corresponding to the local maximum point is determined as a second candidate transient event.

[0107] In this embodiment, the executing entity can also examine the signal change trends on both sides of the local maximum point. When there is a signal rise before the local maximum point and a signal fall after the local maximum point, the corresponding second candidate transient event is retained, and its peak time, peak amplitude, and peak width are recorded. When no local maximum point that meets the conditions is detected, an empty set of second candidate transient events is generated.

[0108] Step S123: Calculate the local background level and local fluctuation amount of the boundary radiation signal based on the sliding time window, construct a dynamic threshold that changes with time based on the local background level and local fluctuation amount, and determine the signal segment whose amplitude exceeds the dynamic threshold as the third candidate transient event.

[0109] In this embodiment, the executing entity can set a sliding time window along the time axis and calculate the local background level and local fluctuation of the boundary radiation signal within each sliding time window.

[0110] Specifically, the arithmetic mean of the signal amplitude within the sliding time window can be used as the local background level, and the standard deviation of the signal amplitude within the sliding time window can be used as the local fluctuation.

[0111] In this embodiment, the executing entity can determine the dynamic threshold corresponding to each sampling time based on the local background level, local fluctuation amount, and preset threshold coefficient. Specifically, the dynamic threshold is determined by the sum of the products of the local background level at the corresponding sampling time, the local fluctuation amount, and the preset threshold coefficient.

[0112] The dynamic threshold is expressed as:

[0113] ;

[0114] Where T(t) is the dynamic threshold at time t, B(t) is the local background level at time t, σ(t) is the local fluctuation at time t, and k is the preset threshold coefficient.

[0115] Since the local background level and local fluctuation amount change with the position of the sliding time window, the dynamic threshold also changes with time.

[0116] In this embodiment, the amplitude of the boundary radiation signal can be compared point by point with the dynamic threshold at the corresponding time. When the amplitude of the boundary radiation signal exceeds the dynamic threshold, the corresponding signal sampling point is marked as an over-threshold sampling point, and temporally continuous over-threshold sampling points are merged into candidate signal segments.

[0117] In this embodiment, the peak time and peak amplitude of each candidate signal segment can be determined, and the candidate signal segment that meets the preset duration or slope conditions is determined as the third candidate transient event.

[0118] Step S124: Fuse the first candidate transient event, the second candidate transient event, and the third candidate transient event, and use the candidate transient event retained after fusion as the detected transient event.

[0119] In this embodiment, the executing entity fuses the candidate transient events obtained by the three detection methods, and uses the candidate transient events retained after fusion as the detected transient events, outputting the corresponding peak time, peak amplitude, and source of the detection method.

[0120] In some implementations, the fusion of the first candidate transient event, the second candidate transient event, and the third candidate transient event includes:

[0121] Step S1241: Merge the first candidate transient event, the second candidate transient event, and the third candidate transient event into a candidate event set.

[0122] In this embodiment, each candidate transient event set includes several candidate event records, and each candidate event record includes at least the peak time, peak amplitude, and source method identifier. The peak time is the occurrence time of the candidate transient event on a unified experimental time axis, the peak amplitude is the maximum signal amplitude in the corresponding signal segment, and the source method identifier indicates which detection method—derivative detection, local maximum detection, or dynamic threshold detection—was used to obtain the candidate transient event.

[0123] In this embodiment, the executing entity merges all candidate event records from the first, second, and third candidate transient event sets into a unified candidate event set. Candidate transient events obtained by any detection method are written into this candidate event set, without pre-screening or weighting the candidate transient events based on the source method.

[0124] The merged candidate event records retain the original source method identifiers. When multiple detection methods obtain candidate transient events at the same peak time, the executing entity can associate the corresponding candidate event records and record the source method identifiers corresponding to each candidate transient event. These source method identifiers are used for subsequent statistical analysis of the number of detection method types.

[0125] When any candidate transient event set is empty, the executing entity can merge the candidate event records from the remaining candidate transient event sets into a single candidate event set. When all three candidate transient event sets are empty, an empty candidate event set is output.

[0126] Step S1242: Perform time clustering on the candidate event set, and group candidate transient events whose time intervals at the time of occurrence are less than or equal to the preset time tolerance into the same event cluster.

[0127] In this embodiment, the executing entity clusters the execution times of the sorted candidate events according to a preset time tolerance. Candidate transient events whose occurrence time interval is less than or equal to the preset time tolerance are grouped into the same event cluster. Candidate transient events whose occurrence time interval is greater than the preset time tolerance are grouped into different event clusters.

[0128] In this embodiment, the preset time tolerance is used to identify candidate transient events obtained by different detection methods for the same local signal fluctuations. After time clustering is completed, each event cluster includes at least one candidate event record.

[0129] Step S1243: For each event cluster, count the number of detection methods from which the candidate transient events included in the event cluster originate, and use this count as the support number of the event cluster.

[0130] In this embodiment, the executing entity determines the number of deduplicated source method types as the support number for the corresponding event cluster. The support number is used to characterize the number of detection methods that obtain the candidate transient events corresponding to that event cluster.

[0131] In this embodiment, when an event cluster includes candidate transient events obtained by derivative detection, local maximum detection, and dynamic threshold detection simultaneously, the support number of the event cluster is 3. When an event cluster includes candidate transient events obtained by two of these detection methods, the support number of the event cluster is 2. When an event cluster includes candidate transient events obtained by only one detection method, the support number of the event cluster is 1.

[0132] Step S1244: Retain event clusters with a number of supports greater than or equal to a preset support threshold, perform morphological verification on event clusters with a number of supports less than the preset support threshold, and use the event clusters that pass the morphological verification and the directly retained event clusters together as detected transient events.

[0133] In this embodiment, when the number of supports for an event cluster is greater than or equal to the preset support threshold, the event cluster is directly retained. When the number of supports for an event cluster is less than the preset support threshold, the boundary radiation signal segment corresponding to the event cluster is read, and the boundary radiation signal segment is morphologically verified.

[0134] In this embodiment, the executing entity can determine the dynamic signal-to-noise ratio based on the peak amplitude, local background level, and local fluctuation of the boundary radiation signal segment. The rising slope is determined based on the signal changes before the peak moment. The falling slope is determined based on the signal changes after the peak moment. The peak width is determined based on the signal changes on both sides of the peak.

[0135] In this embodiment, the executing entity compares the dynamic signal-to-noise ratio, rising slope, falling slope, and peak width with corresponding preset verification conditions. When the dynamic signal-to-noise ratio meets the preset signal-to-noise ratio condition, the rising slope meets the preset rising slope condition, the falling slope meets the preset falling slope condition, and the peak width is within the preset peak width range, the event cluster is determined to pass the morphological verification and is retained. When any verification condition is not met, the event cluster is not retained.

[0136] In some implementations, constructing the time exclusion interval based on the detected transient event includes:

[0137] Step S125: For each detected transient event, determine the rising slope of the transient event based on the signal changes before the peak time, and determine the falling slope of the transient event based on the signal changes after the peak time.

[0138] In this embodiment, the executing entity sequentially reads the boundary radiation signal segments corresponding to transient events, and divides the corresponding signal segments into pre-peak signal segments and post-peak signal segments according to the peak time of each transient event. The pre-peak signal segment is used to characterize the process of the boundary radiation signal rising from the local background to the peak value, and the post-peak signal segment is used to characterize the process of the boundary radiation signal falling back from the peak value.

[0139] In one implementation, the executing entity selects a preset time period before the peak moment, reads the signal amplitude and time information corresponding to the start and end positions of the preset time period, and determines the rising slope based on the signal increment and time increment within the preset time period. After the peak moment, the executing entity selects a preset time period and determines the falling slope based on the signal decrease and time increment within the preset time period.

[0140] In another implementation, the executing entity reads the first derivative to the left of the peak moment of the transient event and determines the largest positive derivative therein as the rising slope. The executing entity reads the first derivative to the right of the peak moment and determines the largest negative derivative therein as the falling slope.

[0141] Step S126: Transient events that only have a rising edge but no falling edge, whose rising or falling slope does not reach the corresponding preset slope threshold, or whose peak duration is less than the preset minimum duration are removed as noise spikes.

[0142] In this embodiment, the execution entity reads the rising slope and falling slope obtained in step S125 for each transient event, and determines whether the transient event forms a complete peak structure by combining the signal changes on both sides of the peak.

[0143] Specifically, when there is an upward process that meets the preset upward slope condition before the peak moment, but no corresponding downward process after the peak moment, the executing entity determines the transient event as a noise spike. Correspondingly, when there is a signal decline after the peak moment, but the decline slope does not meet the preset decline slope condition, the executing entity also determines the transient event as a noise spike.

[0144] The executing entity also compares the rising and falling slopes of transient events with corresponding preset slope thresholds. Transient events whose rising slope does not reach the preset rising slope threshold or whose falling slope does not reach the preset falling slope threshold are not considered as valid transient events for subsequent time exclusion interval construction.

[0145] For transient events with both rising and falling phases, the executing entity further determines the peak duration based on the time information of the corresponding signal segment and compares the peak duration with a preset minimum duration. When the rising and falling phases occur only within one or a few sampling points, resulting in a peak duration less than the preset minimum duration, the executing entity classifies the transient event as a noise spike and discards it.

[0146] Step S127: For the transient event retained after removing noise spikes, determine the peak time, peak amplitude, and local background value before the peak time of the transient event.

[0147] In this embodiment, the execution entity sequentially reads the transient events retained in step S126 and reads the boundary radiation signal segments corresponding to each transient event. For each retained transient event, the execution entity determines its peak time and the peak amplitude corresponding to that peak time, and determines the local background value corresponding to that transient event from the signal before the peak time. The local background value is used as a baseline for determining the effective peak amplitude and peak width boundary.

[0148] Step S128: Based on the local background value, determine the target amplitude level according to the effective amplitude of the peak amplitude relative to the local background value and the first preset ratio.

[0149] In this embodiment, the executing entity compares the peak amplitude obtained in step S127 with the local background value, and determines the portion of the peak amplitude that is higher than the local background value as the effective amplitude of the transient event. Then, the executing entity determines the amplitude position for searching the peak width boundary from the effective amplitude according to a first preset ratio, and determines the corresponding target amplitude level using the local background value as a baseline.

[0150] In this embodiment, the first preset ratio is set to 70% to 80%. That is, the target amplitude level is located within the effective amplitude range of the change from the local background value to the peak amplitude, and corresponds to the 70% to 80% position of the effective amplitude.

[0151] Step S129: Search the signal sequence forward and backward from the peak time respectively, and take the time when the signal amplitude first crosses the target amplitude level as the left boundary time and the right boundary time respectively, and determine the reference peak width based on the left boundary time and the right boundary time.

[0152] In this embodiment, the executing entity uses the peak time obtained in step S127 as the search starting point and reads the target amplitude level determined in step S128.

[0153] The execution unit first reads signal sampling points sequentially along the time axis from the peak time to earlier times, and compares the amplitude of each signal sampling point with the target amplitude level. When the boundary radiation signal first crosses the target amplitude level from the peak side, the execution unit records the corresponding sampling time and determines this sampling time as the left boundary time.

[0154] Then, the execution entity sequentially reads signal sampling points along the time axis from the peak time to later times. When the signal after the peak crosses the target amplitude level again for the first time from the peak side, the execution entity records the corresponding sampling time and determines this sampling time as the right boundary time.

[0155] The executing entity determines the reference peak width of the transient event based on the time length between the left and right boundary moments. Therefore, the reference peak width corresponds to the duration of the boundary radiation signal between the left and right crossing points at the target amplitude level.

[0156] Step S1210: Using the peak moment as the center, extract the length of the second preset ratio of the reference peak width as the time exclusion interval corresponding to the transient event.

[0157] In this embodiment, the executing entity takes the peak moment as the center, multiplies the reference peak width determined in step S129 by a second preset ratio to obtain the exclusion width.

[0158] The exclusion width is expressed as:

[0159] ;

[0160] In the formula, The exclusion width is the length of the time exclusion interval. The reference peak width determined in step S129; This is the second preset ratio.

[0161] The executing entity can extend the exclusion width forward by half and backward by half from the peak moment, and use this time interval as the time exclusion interval corresponding to the transient event.

[0162] The second preset ratio ranges from 0 to 1, and this ratio determines the proportion of the final exclusion interval relative to the reference peak width. When the second preset ratio is 1, the time exclusion interval completely overlaps with the reference peak width. When the second preset ratio is less than 1, the time exclusion interval is a narrowed interval near the center of the reference peak width, used to exclude only the core period of the most intense disturbance in the ELM event.

[0163] In this embodiment, the executing entity executes steps S127 to S1210 one by one for all transient events retained in step S126 to obtain the time exclusion interval corresponding to each transient event. The union of all time exclusion intervals constitutes the exclusion interval set for subsequent effective time slice screening.

[0164] In some implementations, determining the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, and determining whether each sampling time interval overlaps with the time exclusion interval, includes:

[0165] Step S131: When the original measurement data record has a sampling start time and a sampling end time, the time interval between the sampling start time and the sampling end time is determined as the sampling time interval of the corresponding measurement point.

[0166] In this embodiment, a measurement point in the raw measurement data corresponds to the measurement result obtained by the profile diagnostic device during a single sampling or integration process. In addition to the measured value, this measurement point also has time information characterizing the time when the measurement occurred.

[0167] Since a single measurement by the profile diagnostic equipment does not correspond to a single instant, when the original measurement data record contains the sampling start time and sampling end time, the execution entity reads the sampling start time and sampling end time corresponding to the measurement point and determines the time range constrained by the two as the sampling time interval of the measurement point.

[0168] The sampling time interval is represented as follows:

[0169] ;

[0170] in, Let i be the sampling time interval corresponding to the i-th measurement point. This is the sampling start time corresponding to this measurement point. This represents the sampling end time corresponding to this measurement point.

[0171] In this embodiment, the executing entity performs the above processing on each measurement point according to the order of the measurement points in the original measurement data, so that each measurement point with a sampling start time and a sampling end time is associated with a corresponding sampling time interval. The sampling time interval is used for subsequent time range overlap determination with the time exclusion interval.

[0172] Step S132: When the original measurement data record has a nominal time and integration duration but no sampling start and end time, the time interval centered on the nominal time and with the integration duration as the time length is determined as the sampling time interval of the corresponding measurement point.

[0173] In this embodiment, when the original measurement data does not directly record the sampling start time and sampling end time, but records the nominal time and integration duration of the measurement point, the execution entity reads the nominal time and integration duration corresponding to that measurement point. Here, the nominal time represents the center time corresponding to this integration measurement, and the integration duration represents the actual duration of this measurement.

[0174] The executing entity uses the nominal time as the center of the sampling time interval, and determines the time range before and after the nominal time according to the integration duration, thereby obtaining the sampling time interval corresponding to the measurement point.

[0175] The sampling time interval is represented as follows:

[0176] ;

[0177] in, Let i be the sampling time interval corresponding to the i-th measurement point. This is the nominal time corresponding to the measurement point. This is the integration time corresponding to this measurement point.

[0178] In this embodiment, for the same original set of measurement data, the executing entity determines the sampling time interval based on the actual time information recorded at each measurement point. Measurement points that record the sampling start time and sampling end time are processed according to step S131; measurement points that do not record the sampling start and end times but record the nominal time and integration duration are processed according to step S132. After processing, the executing entity obtains the correspondence between each measurement point and its sampling time interval.

[0179] Step S133: When the sampling time interval of any measurement point overlaps with any of the time exclusion intervals, the original measurement data corresponding to that measurement point is marked as interfered data.

[0180] In this embodiment, the executing entity reads the sampling time intervals of each measurement point determined in step S131 or step S132, and reads the time exclusion intervals constructed based on transient events. For each measurement point, the executing entity compares the sampling time interval corresponding to that measurement point with each time exclusion interval in sequence.

[0181] The overlap determination is expressed as follows:

[0182] ;

[0183] in, Let i be the sampling time interval corresponding to the i-th measurement point. To exclude the j-th time interval, This indicates that the sampling time interval and the time exclusion interval overlap. When the above relationship holds, the executing entity marks the original measurement data corresponding to the measurement point as interfered data.

[0184] When the sampling time interval of a measurement point shares a common time range with any of its exclusion time intervals, the executing entity determines that the actual sampling process of that measurement point overlaps with the exclusion time range of the corresponding transient event, and marks the original measurement data corresponding to that measurement point as disturbed data. Measurement points marked as disturbed data are not used as data points in subsequent steady-state profile processing.

[0185] When a sampling time interval overlaps with multiple time exclusion intervals, the executing entity only needs to determine that there is at least one overlap relationship, that is, mark the corresponding measurement point as the interfered data, without having to mark the same measurement point repeatedly.

[0186] In one implementation, when the sampling time interval and the time exclusion interval only touch at the interval boundary and do not have a common time range with a duration, the boundary contact is not judged as time overlap.

[0187] In another implementation, the executing entity sets a minimum overlap duration threshold and determines the duration for which the sampling time interval and the time exclusion interval actually overlap. When the duration reaches the minimum overlap duration threshold, the corresponding measurement point is marked as interfered data.

[0188] For measurement points that do not overlap with any time exclusion interval, the executing entity does not mark them as disturbed data, but uses them as valid data points for subsequent normalized flux coordinate calculations. Thus, the time exclusion results generated by transient event detection are further mapped to specific profile measurement data, realizing the transformation from the transient event time range to the set of valid measurement data.

[0189] Step S134: When the original measurement data does not record the sampling start and end times and integration duration, the original measurement data corresponding to the measurement points whose nominal times fall within any of the time exclusion intervals are marked as interfered data.

[0190] In this embodiment, when the executing entity reads the original measurement data and determines that a certain measurement point has not recorded the sampling start time and sampling end time, and has not recorded the integration duration, it is impossible to obtain the sampling time interval with duration for that measurement point according to step S131 or step S132.

[0191] In this scenario, the executing entity reads the nominal time corresponding to the measurement point and compares it sequentially with the start and end times of each time exclusion interval. When the nominal time falls within the time range defined by any time exclusion interval, the executing entity determines that the measurement point is affected by the corresponding transient event and marks the original measurement data corresponding to that measurement point as disturbed data.

[0192] When the nominal time is not within any time exclusion interval, the executing entity does not exclude the measurement point based on the transient event time exclusion rule.

[0193] In this embodiment, the executing entity also records the state where the measurement point lacks sampling duration information, in order to distinguish between measurement points that complete the overlap judgment through the actual sampling time interval and measurement points that complete the time matching through the nominal time.

[0194] In some implementations, the step of calculating the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point in the original measurement data that was not marked as disturbed data, based on the magnetic axis flux and boundary flux in the magnetic balance data, through interpolation, includes:

[0195] Step S141: Determine the two-dimensional flux grid, magnetic axis flux, and boundary flux in the magnetic balance data.

[0196] In this embodiment, the executing entity reads the magnetic balance data corresponding to the current valid data point and determines the two-dimensional magnetic flux grid, magnetic axis flux, and boundary flux from the magnetic balance data. The two-dimensional magnetic flux grid characterizes the poloidal magnetic flux distribution corresponding to different spatial locations of the plasma cross-section; the magnetic axis flux is the poloidal magnetic flux value corresponding to the magnetic axis position; and the boundary flux is the poloidal magnetic flux value corresponding to the last closed magnetic surface.

[0197] Specifically, the two-dimensional magnetic flux grid includes multiple grid nodes arranged along two-dimensional spatial coordinates and the corresponding pole magnetic flux value of each grid node. The execution entity reads the spatial coordinate range of the two-dimensional magnetic flux grid, the grid node positions, and the corresponding pole magnetic flux value of each grid node, and establishes the correspondence between spatial positions and pole magnetic flux values.

[0198] For each valid data point that is not marked as interfered data after processing, the execution entity simultaneously reads the spatial position corresponding to each valid data point. The spatial position is the geometric measurement position of the corresponding measurement point of the profile diagnostic device, which is used to determine the grid region where the valid data point is located in the two-dimensional magnetic flux grid.

[0199] In this embodiment, the normalized magnetic flux coordinates use the magnetic axis flux and boundary flux as normalization references. The normalized magnetic flux coordinates corresponding to the magnetic axis position are 0, and the normalized magnetic flux coordinates corresponding to the final closed magnetic surface are 1, thereby converting the geometric spatial position of each valid data point into a unified magnetic flux coordinate representation.

[0200] Step S142: Based on the spatial position of each valid data point, perform bilinear interpolation on the two-dimensional magnetic flux grid to obtain the poloidal magnetic flux value corresponding to each valid data point.

[0201] In this embodiment, the execution entity processes each valid data point sequentially. For the current valid data point, a target grid region including the spatial location is determined in the two-dimensional magnetic flux grid based on the spatial location of the valid data point, and the adjacent grid nodes constituting the target grid region and the corresponding poloidal magnetic flux values ​​of the adjacent grid nodes are read.

[0202] Specifically, for a valid data point located between two adjacent first direction grid coordinates and two adjacent second direction grid coordinates, the execution entity reads the four grid nodes surrounding the valid data point and the corresponding polar magnetic flux values ​​of the four grid nodes.

[0203] The execution entity performs bilinear interpolation based on the relative positions of valid data points within the target grid region. First, based on the position of the valid data point in the first coordinate direction, linear interpolation is performed on the poloidal magnetic flux values ​​of the corresponding grid nodes on both sides of the target grid region, yielding two intermediate poloidal magnetic flux values. Then, based on the position of the valid data point in the second coordinate direction, linear interpolation is performed again on these two intermediate poloidal magnetic flux values. The result of this second linear interpolation is determined as the poloidal magnetic flux value corresponding to the current spatial position of the valid data point.

[0204] When the spatial location of a valid data point coincides with the location of a grid node in a two-dimensional magnetic flux grid, the pole magnetic flux value corresponding to that grid node can be directly read as the pole magnetic flux value corresponding to that valid data point.

[0205] Step S143: Based on the first difference between the polar magnetic flux value and the magnetic axis magnetic flux and the second difference between the boundary magnetic flux and the magnetic axis magnetic flux, the ratio of the first difference to the second difference is determined as the normalized magnetic flux coordinate corresponding to each valid data point.

[0206] In this embodiment, for each valid data point, the execution entity reads the pole magnetic flux value corresponding to the valid data point obtained in step S142, and reads the magnetic axis flux and boundary flux determined in step S141.

[0207] The executing entity calculates the difference between the pole magnetic flux value and the magnetic axis flux corresponding to the valid data point, and determines the obtained difference as the first difference. Simultaneously, it calculates the difference between the boundary magnetic flux and the magnetic axis flux, and determines the obtained difference as the second difference. Then, the ratio of the first difference to the second difference is determined as the normalized magnetic flux coordinate corresponding to the valid data point.

[0208] The normalized magnetic flux coordinates are expressed as:

[0209] ;

[0210] in, For spatial location Normalized magnetic flux coordinates at the location; To determine the spatial position of the two-dimensional magnetic flux grid ( The poloidal flux value obtained by bilinear interpolation at the location; For magnetic flux along the magnetic axis; For boundary magnetic flux.

[0211] Specifically, when the pole magnetic flux value corresponding to a valid data point equals the magnetic flux along the magnetic axis, the first difference is zero, and the corresponding normalized magnetic flux coordinate is 0. When the pole magnetic flux value corresponding to a spatial location equals the boundary magnetic flux, the first difference equals the second difference, and the corresponding normalized magnetic flux coordinate is 1. Thus, valid data points at different geometric spatial locations are uniformly mapped to a normalized magnetic flux coordinate system with the magnetic axis as the starting point and the last closed magnetic surface as the boundary.

[0212] Before performing normalization, the executing entity also determines whether the second difference between the boundary magnetic flux and the magnetic axis flux can be used for normalization calculation. When the absolute value of the second difference is less than a preset threshold, it indicates that the boundary magnetic flux and the magnetic axis flux are too close, and the current magnetic balance data cannot form a valid normalization reference. Therefore, the normalized magnetic flux coordinates of the corresponding valid data points are not calculated based on the magnetic balance data.

[0213] For valid data points that can undergo normalization calculations, the executing entity associates and saves the calculated normalized magnetic flux coordinates with the measured value, spatial location, and temporal information of the valid data point, thereby forming a data record containing the measured value and normalized magnetic flux coordinates. After each valid data point completes the above magnetic coordinate mapping, profile data represented by unified normalized magnetic flux coordinates is obtained, which is used for subsequent data filtering and profile model fitting based on a preset magnetic flux coordinate range.

[0214] In some implementations, the step of selecting valid data points that meet a preset magnetic flux coordinate range, performing profile model fitting on the valid data points, and outputting plasma profile parameters includes:

[0215] Step S151: Retain valid data points whose normalized magnetic flux coordinates are within the preset magnetic flux coordinate range.

[0216] In this embodiment, the execution entity reads each valid data point and its corresponding normalized magnetic flux coordinates obtained in step S143, and compares the normalized magnetic flux coordinates of each valid data point with the preset magnetic flux coordinate range to select data points for subsequent profile model fitting.

[0217] In this embodiment, the normalized magnetic flux coordinates use the magnetic flux coordinate 0 corresponding to the magnetic axis position and the magnetic flux coordinate 1 corresponding to the last closed magnetic surface as the normalization reference. In one embodiment, the preset magnetic flux coordinate range includes the magnetic flux coordinate range from 0 to 1, and the executing entity retains valid data points whose normalized magnetic flux coordinates fall within the preset magnetic flux coordinate range.

[0218] Considering the deviations caused by magnetic flux interpolation and plasma boundary reconstruction, preset tolerances can be set on both sides of the preset magnetic flux coordinate range. Data points whose normalized magnetic flux coordinates exceed 0 to 1 but are still within the preset tolerance range can be marked as boundary data. Data points whose normalized magnetic flux coordinates exceed the preset tolerance range are removed from the dataset used for subsequent profile model fitting.

[0219] In this embodiment, after the screening is completed, the executing entity constructs a profile to be fitted based on the normalized magnetic flux coordinates and measured values ​​corresponding to the retained valid data points. Specifically, when the original measurement data is electron temperature data, the measured value is electron temperature; when the original measurement data is electron density data, the measured value is electron density; and when the original measurement data is ion temperature data, the measured value is ion temperature.

[0220] In this embodiment, the preset four-parameter model adopts the four-parameter mtanh model. The four-parameter model is used to characterize the change process of the profile from the top side of the platform to the bottom side of the platform. It includes the top value of the platform, the bottom value of the platform, the center position of the platform, and the width of the platform.

[0221] The four-parameter mtanh model is expressed as:

[0222] ;

[0223] Where ψ is the normalized flux coordinate corresponding to the valid data point; y(ψ) is the profile parameter value at the corresponding normalized flux coordinate position; A is the top value of the abutment; B is the bottom value of the abutment; x0 is the center position of the abutment; and w is the width of the abutment.

[0224] Step S152: Estimate the initial value of the sump top value in the preset four-parameter model based on the statistics of the retained valid data points in the core magnetic flux region, and estimate the initial value of the sump base value in the four-parameter model based on the statistics of the retained valid data points in the boundary magnetic flux region.

[0225] In this embodiment, the executing entity divides the valid data points into data points located in the core magnetic flux region and data points located in the boundary magnetic flux region according to the normalized magnetic flux coordinates corresponding to the valid data points retained in step S151.

[0226] For the core magnetic flux region, the execution entity acquires the profile measurement values ​​of each valid data point in the region, and determines the level of the top side profile of the pedestal based on the statistics of the profile measurement values. The obtained statistical results are used as the initial value of the top value A of the pedestal in the four-parameter model.

[0227] For the boundary magnetic flux region, the execution entity obtains the profile measurement values ​​of each valid data point in the region, and determines the level of the base side profile based on the statistics of the profile measurement values. The obtained statistical results are used as the initial value of the base value B in the four-parameter model.

[0228] The initial values ​​of the sump top value and the sump base value are determined by using data from the core flux region and the boundary flux region, respectively, so that the initial values ​​of A and B correspond to the core side amplitude level and the boundary side amplitude level of the current profile to be fitted, respectively.

[0229] Step S153: Based on the position of the maximum gradient of the profile formed by the retained valid data points and the position corresponding to the preset amplitude ratio of the profile amplitude, the initial value of the platform center position in the four-parameter model is estimated by fusing.

[0230] In this embodiment, the executing entity can arrange the valid data points retained in step S151 according to the normalized magnetic flux coordinates, determine the gradient distribution of the profile along the normalized magnetic flux coordinate direction based on the changes in the profile measurement values ​​corresponding to adjacent magnetic flux positions, and determine the position with the most significant profile change from the gradient distribution as the maximum gradient position.

[0231] The executing entity can also determine the effective amplitude range of the profile based on the amplitude levels at the top and bottom of the platform, and determine the profile amplitude corresponding to a preset amplitude ratio within the effective amplitude range. In this embodiment, the preset amplitude ratio includes 80% and 50%, and the executing entity determines the normalized flux position corresponding to when the profile reaches 80% amplitude level and the normalized flux position corresponding to when the profile reaches 50% amplitude level, respectively.

[0232] The executing entity uses the maximum gradient position, the 80% amplitude position, and the 50% amplitude position together as candidate estimated positions for the base center position, and fuses the candidate estimated positions to obtain an initial estimated result for the base center position. The initial estimated result is then used as the initial value for the base center position in the four-parameter model.

[0233] In this embodiment, by simultaneously introducing the location of the maximum gradient and the locations corresponding to different amplitude levels, it is possible to avoid relying solely on a single profile feature to determine the location of the platform center, so that the initial value of the platform center location can comprehensively reflect the location characteristics of profile gradient changes and profile amplitude transitions.

[0234] Step S154: Estimate the initial value of the platform width in the four-parameter model based on the gradient half-width of the profile.

[0235] In this embodiment, the execution entity reads the profile gradient distribution obtained in step S153 and determines the maximum value of the absolute gradient and the normalized magnetic flux position corresponding to the maximum value.

[0236] The executing entity uses half of the maximum absolute value of the gradient as the gradient half-height level, and uses the position of the maximum gradient as a reference to search for the position where the gradient reaches the gradient half-height level in both the direction of the smaller normalized magnetic flux coordinate and the direction of the larger normalized magnetic flux coordinate.

[0237] In this embodiment, the magnetic flux coordinate interval between the two normalized magnetic flux positions corresponding to the gradient half-height level on both sides of the maximum gradient position is determined as the gradient half-height width of the profile, and the initial value of the pedestal width in the four-parameter model is estimated based on the gradient half-height width.

[0238] Step S155: Generate the fitting boundaries for each parameter based on the range of magnetic flux coordinates and the range of profile amplitude of the retained valid data points.

[0239] In this embodiment, the executing entity can obtain the normalized magnetic flux coordinates corresponding to all valid data points retained in step S151, and determine the distribution range of the magnetic flux coordinates. Simultaneously, it obtains the profile measurement values ​​corresponding to each valid data point, and determines the range of profile amplitude formed by the measurement values.

[0240] In this embodiment, the executing entity generates fitting boundaries corresponding to the top value of the sump, the bottom value of the sump, the center position of the sump, and the width of the sump, respectively, based on the magnetic flux coordinate range and the profile amplitude range.

[0241] The fitting boundaries for the sump top and sump base values ​​can be determined based on the amplitude distribution range of the current profile. The fitting boundary for the sump center position can be determined based on the normalized flux coordinate range covered by the retained valid data points. The fitting boundary for the sump width can be determined based on the normalized flux coordinate range and the radial variation range of the current profile.

[0242] In this embodiment, the fitting boundary is dynamically generated according to the actual magnetic flux coordinate range and amplitude range of the current profile to be fitted, and is used to constrain the search range of each parameter in the subsequent model fitting process, so that the model parameters remain within the range corresponding to the current profile data.

[0243] Step S156: Perturb the initial values ​​of each parameter to obtain multiple sets of initial parameters, call the four-parameter model, and fit the retained effective data points based on each set of initial parameters and the fitting boundary. Select the best fitting result from each fitting result according to the goodness of fit.

[0244] In this embodiment, the executing entity can read the initial values ​​of the pedestal top, pedestal base, pedestal center position, and pedestal width obtained in steps S152 to S154, and perform perturbation based on the initial values ​​of the parameters to form multiple sets of different initial parameters.

[0245] In this embodiment, each set of initial parameters may include the initial value of the apex top, the initial value of the apex base, the initial value of the apex center position, and the initial value of the apex width. By perturbing the initial values ​​of each parameter, different sets of initial parameters enter the model fitting process from different parameter starting points.

[0246] For each set of initial parameters, the execution entity can call the four-parameter model and use the normalized magnetic flux coordinates of the valid data points retained in step S151 as the model independent variables, the corresponding profile measurement values ​​as the fitting target, and simultaneously use the parameters generated in step S155 to fit the variation range of the boundary constraints of the top value of the abutment, the base value of the abutment, the center position of the abutment, and the width of the abutment, and iteratively fit the four model parameters.

[0247] After each set of initial parameters is fitted, the execution entity obtains the fitted values ​​of the pedestal top value, pedestal base value, pedestal center position, and pedestal width corresponding to that set of initial parameters, and generates the corresponding fitted profile based on the parameter fitted values.

[0248] In this embodiment, the executing entity can determine the goodness of fit corresponding to each fitting based on the difference between the fitted profile and each valid data point. The fitting quality evaluation index may include one or more of the following: coefficient of determination, adjusted coefficient of determination, mean square error, root mean square error, and mean absolute error, and the fitting quality evaluation results corresponding to each fitting are recorded.

[0249] After the execution entity completes the model fitting for multiple sets of initial parameters, it compares the goodness of fit of each fitting result and selects the fitting result with the best fitting quality from multiple fitting results as the best fitting result for the current profile.

[0250] By perturbing the initial parameters multiple times and performing model fitting separately, the influence of a single initial parameter on the model fitting result can be reduced, allowing the final output model parameters to be determined from multiple candidate fitting results.

[0251] Step S157: Output the top value, bottom value, center position, and width of the abutment from the optimal fitting result as the plasma profile parameters.

[0252] In this embodiment, the execution entity reads the optimal fitting result determined in step S156 and extracts the pedestal top value, pedestal base value, pedestal center position, and pedestal width corresponding to the optimal fitting result.

[0253] Specifically, the top value of the sump is used to characterize the profile parameter level inside the sump transition region, the bottom value of the sump is used to characterize the profile parameter level outside the sump transition region, the center position of the sump is used to characterize the center position of the main transition region of the sump profile in the normalized magnetic flux coordinates, and the width of the sump is used to characterize the characteristic width of the sump transition region in the direction of the normalized magnetic flux coordinates.

[0254] In this embodiment, the executing entity can output the top value of the sump, the bottom value of the sump, the center position of the sump, and the width of the sump as plasma profile parameters of the current profile.

[0255] Furthermore, the executing entity can determine the pedestal height based on the difference between the pedestal top value and the pedestal base value, and output the pedestal height in association with the pedestal center position, pedestal width, and fitting quality evaluation results.

[0256] For different time slices or different types of profile data, the execution entity executes steps S151 to S157 respectively, and establishes a correlation between the obtained plasma profile parameters and the corresponding shot number, diagnostic type and time slice, thereby forming the profile parameter results corresponding to each effective time slice.

[0257] In one specific implementation, the executing entity also performs statistical analysis on the time exclusion results. The executing entity merges overlapping time exclusion intervals to obtain a set of non-overlapping exclusion intervals. The ratio of the total duration of the exclusion interval set to the duration of the original H-mode time window is determined as the removal ratio. The ratio of the remaining time after deducting the total duration from the original H-mode time window duration to the original H-mode time window duration is determined as the net H-mode time ratio. The ratio of the number of measurement points located within the original H-mode time window and whose sampling time interval does not overlap with any time exclusion interval to the total number of measurement points within the original H-mode time window is determined as the effective time point ratio.

[0258] The duration of the original H-mode time window is expressed as follows:

[0259] ;

[0260] in, The original H-mode time window duration. The start time of the H-mode. This is the end time of the H-mode.

[0261] The total duration of the set of non-overlapping exclusion intervals is expressed as:

[0262] ;

[0263] in, The total duration of the set of non-overlapping exclusion intervals. To exclude the k-th time interval after merging, K represents the duration of the time exclusion interval, and K represents the number of time exclusion intervals after merging. The portion of the time exclusion interval that exceeds the start and end boundaries of the H module is not included in the statistics.

[0264] The removal ratio is expressed as:

[0265] ;

[0266] in, To remove the proportion, The total duration of the set of non-overlapping exclusion intervals. This represents the duration of the original H-mode time window.

[0267] The ratio of net H-mode time to net H-mode time is expressed as follows:

[0268] ;

[0269] ;

[0270] in, The net H-model time is the remaining time after deducting the time exclusion interval from the original H-model time window. The net H-mode time ratio. When only transient events are excluded as a time clipping criterion, the relationship between the net H-mode time ratio and the removal ratio is expressed as:

[0271] ;

[0272] The proportion of effective time points is expressed as follows:

[0273] ;

[0274] Accordingly, the proportion of excluded measurement points is expressed as:

[0275] ;

[0276] in, This represents the total number of measurement points with valid measurements within the original H-mode time window. Measurement points lacking profile diagnostic data, with invalid measurements, or whose sampling time is outside the H-mode time window are excluded. ; The number of measurement points whose sampling time intervals do not overlap with any time exclusion interval from the total number of measurement points; The proportion of effective time points; The proportion of measurement points that were excluded.

[0277] In one specific implementation, for multiple candidate transient events clustered into the same event cluster by time clustering, the executing entity can weight their occurrence times according to the peak amplitude of each candidate transient event, and determine the weighted median of the amplitude at each occurrence time as the transient event time corresponding to that event cluster. The peak width of the transient event is taken as the half-width at half maximum (FWHM) of the corresponding boundary radiation signal or a preset width estimate.

[0278] In one specific implementation, the executing entity sorts the fused transient events according to their peak times. When the time interval between two adjacent transient events is less than or equal to 5ms, they are considered as repeated events generated by the same transient disturbance and merged. The merged transient event time is preferentially selected based on the event time with the largest peak amplitude within the corresponding time window. When multiple peak amplitudes are close, the event time with the largest rising slope or the weighted average of the candidate event times is used. For multiple consecutive transient events with adjacent time intervals not exceeding 5ms, they are merged as the same repeated event group.

[0279] In one specific implementation, the model used in step S15 to perform profile model fitting can also be a nine-parameter segmented pedestal model, a Bayesian fitting model, an EPED constraint model, or a machine learning surrogate model, to replace the four-parameter mtanh model, and continue to output the corresponding pedestal parameters based on the effective data points obtained by filtering the H-mode time window and time exclusion interval, as well as the normalized magnetic flux coordinates.

[0280] According to an embodiment of the present invention, an electronic device is provided. The electronic device in this embodiment may include one or more of the following components: a processor, a network interface, memory, non-volatile memory, and one or more application programs, wherein the one or more application programs may be stored in the non-volatile memory and configured to be executed by one or more processors, and the one or more programs are configured to perform the methods as described in the foregoing method embodiments.

[0281] According to embodiments of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the method described in any of the above embodiments.

[0282] According to embodiments of the present invention, a computer program product comprising instructions is also provided, which, when executed by a computer, cause the computer to perform a method in any of the above embodiments.

[0283] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for processing plasma profile data, characterized in that, include: Acquire raw measurement data, boundary radiation signals, and corresponding magnetic balance data from the profile diagnostic equipment; wherein, the raw measurement data includes time information and spatial location information of each measurement point; Transient events are detected in the boundary radiation signal, and time exclusion intervals are constructed based on the detected transient events; Based on the sampling duration of the original measurement data, determine the sampling time interval corresponding to each measurement point, and determine whether there is an overlap between each sampling time interval and the time exclusion interval. If there is an overlap, mark the corresponding original measurement data as interfered data and remove it. For valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data. Valid data points that meet the preset magnetic flux coordinate range are selected, and profile model fitting is performed on the valid data points to output plasma profile parameters.

2. The method according to claim 1, characterized in that, The transient event detection of the boundary radiation signal includes: The boundary radiation signal is smoothed, and the first derivative sequence of the smoothed signal is calculated. Local fluctuations that satisfy the sequence characteristics are identified as the first candidate transient events. The sequence characteristics are characterized by a rapid rise, reaching a peak, and then falling back. Local maxima in the smoothed signal are identified, and local maxima with peak heights greater than a preset minimum peak height are identified as second candidate transient events; wherein the amplitude of the local maxima is greater than the signal value in a preset neighborhood. The local background level and local fluctuation of the boundary radiation signal are calculated based on the sliding time window. A dynamic threshold that changes with time is constructed based on the local background level and local fluctuation. Signal segments whose amplitude exceeds the dynamic threshold are identified as the third candidate transient events. The first, second, and third candidate transient events are fused, and the candidate transient events retained after fusion are used as the detected transient events.

3. The method according to claim 2, characterized in that, The fusion of the first candidate transient event, the second candidate transient event, and the third candidate transient event includes: The first candidate transient event, the second candidate transient event, and the third candidate transient event are merged into a candidate event set; Perform temporal clustering on the candidate event set, grouping candidate transient events whose occurrence time interval is less than or equal to a preset time tolerance into the same event cluster; For each event cluster, the number of detection methods from which the candidate transient events within the event cluster originate is counted, which is used as the support number for that event cluster. Event clusters with a number of supports greater than or equal to a preset support threshold are retained, while event clusters with a number of supports less than the preset support threshold undergo morphological verification. Event clusters that pass the morphological verification and the directly retained event clusters are used together as detected transient events.

4. The method according to claim 1, characterized in that, The construction of time exclusion intervals based on detected transient events includes: For each detected transient event, the rising slope of the transient event is determined based on the signal changes before the peak time, and the falling slope of the transient event is determined based on the signal changes after the peak time. Transient events that only have a rising edge but no falling edge, whose rising or falling slope does not reach the corresponding preset slope threshold, or whose peak duration is less than the preset minimum duration are removed as noise spikes. For transient events retained after removing noise spikes, determine the peak time, peak amplitude, and local background value before the peak time of the transient event; Based on the local background value, the target amplitude level is determined according to the effective amplitude of the peak amplitude relative to the local background value and the first preset ratio. The signal sequence is searched forward and backward from the peak time, and the time when the signal amplitude first crosses the target amplitude level is taken as the left boundary time and the right boundary time, respectively. The reference peak width is determined based on the left boundary time and the right boundary time. Centered on the peak moment, the length of the second preset proportion of the reference peak width is extracted as the time exclusion interval corresponding to the transient event.

5. The method according to claim 1, characterized in that, The step of determining the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, and determining whether there is any overlap between each sampling time interval and the time exclusion interval, includes: When the original measurement data record has a sampling start time and a sampling end time, the time interval between the sampling start time and the sampling end time is determined as the sampling time interval of the corresponding measurement point; When the original measurement data records the nominal time and integration duration but not the sampling start and end times, the time interval centered on the nominal time and with the integration duration as the time length will be determined as the sampling time interval for the corresponding measurement point. When the sampling time interval of any measurement point overlaps with any of the time exclusion intervals, the original measurement data corresponding to that measurement point is marked as interfered data. When the original measurement data does not record the sampling start and end times and the integration duration, the original measurement data corresponding to the measurement points whose nominal times fall within any of the time exclusion intervals are marked as interfered data.

6. The method according to claim 1, characterized in that, For valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial positions of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data, including: Determine the two-dimensional flux grid, magnetic axis flux, and boundary flux in the magnetic balance data; Based on the spatial location of each valid data point, bilinear interpolation is performed on the two-dimensional magnetic flux grid to obtain the poloidal magnetic flux value corresponding to each valid data point; Based on the first difference between the pole magnetic flux and the magnetic axis flux, and the second difference between the boundary magnetic flux and the magnetic axis flux, the ratio of the first difference to the second difference is determined as the normalized magnetic flux coordinate corresponding to each valid data point.

7. The method according to claim 6, characterized in that, The process involves selecting valid data points that meet the preset magnetic flux coordinate range, performing profile model fitting on the valid data points, and outputting plasma profile parameters, including: Retain valid data points whose normalized flux coordinates are within the preset flux coordinate range; The initial value of the sump top value in the preset four-parameter model is estimated based on the statistics of the retained effective data points in the core magnetic flux region, and the initial value of the sump base value in the four-parameter model is estimated based on the statistics of the retained effective data points in the boundary magnetic flux region. Based on the position of the maximum gradient of the profile formed by the retained valid data points and the position corresponding to the preset amplitude ratio of the profile amplitude, the initial value of the platform center position in the four-parameter model is estimated by fusing the data. The initial value of the platform width in the four-parameter model is estimated based on the gradient half-width of the profile. Based on the range of magnetic flux coordinates and the range of profile amplitude of the retained valid data points, the fitting boundaries of each parameter are generated; The initial values ​​of each parameter are perturbed to obtain multiple sets of initial parameters. The four-parameter model is called, and the retained effective data points are fitted based on each set of initial parameters and the fitting boundary. The best fitting result is selected from each fitting result according to the goodness of fit. The apex top value, apex base value, apex center position, and apex width from the optimal fitting result are output as the plasma profile parameters.

8. A plasma profile data processing device, characterized in that, include: Acquisition module: Acquires raw measurement data, boundary radiation signals, and corresponding magnetic balance data from the profile diagnostic equipment; wherein, the raw measurement data includes time information and spatial location information of each measurement point; Detection module: performs transient event detection on boundary radiation signals and constructs time exclusion intervals based on the detected transient events; Elimination module: Determine the sampling time interval corresponding to each measurement point based on the sampling duration of the original measurement data, determine whether there is an overlap between each sampling time interval and the time exclusion interval, and if there is an overlap, mark the corresponding original measurement data as interfered data and eliminate it; Calculation module: For valid data points in the original measurement data that were not marked as disturbed data, the normalized magnetic flux coordinates corresponding to the spatial location of each valid data point are calculated by interpolation based on the magnetic axis flux and boundary flux in the magnetic balance data. Output module: Filter out valid data points that meet the preset magnetic flux coordinate range, perform profile model fitting on the valid data points, and output plasma profile parameters.

9. An electronic device, characterized in that, include: A memory, and one or more processors communicatively connected to the memory; The memory stores instructions that can be executed by the one or more processors to cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images