On-line data analysis method for fusion radiation calorimetry diagnostics and system thereof

Abstract

The application discloses an online data analysis method and system for fusion radiation calorimetric diagnosis, and belongs to the technical field of nuclear fusion experiments. The method firstly establishes a fusion radiation calorimetric diagnosis expert rule base based on multi-dimensional parameter data fusion, creates a paging man-machine interface and a unified canvas framework embedded in the interface homepage; then, a current discharge gun number is acquired online, multi-signal source time sequence data is acquired from a current pulse tree of a remote pulse database and stored into a local pulse library; then, data is processed and analyzed according to diagnostic expert rules, automatic identification and compensation of abnormal channels of the diagnosis are realized, and radiation intensity and total radiation power are calculated online; finally, various signal data are integrated and visualized under the unified canvas framework; the system realizes an efficient radiation calorimetric diagnosis online data analysis method and provides reliable data support for fusion experiments.

Description

Technical Field

[0001] This invention belongs to the field of nuclear fusion experimental technology, and specifically relates to an online data analysis method and system for fusion radiation calorimetry diagnosis. Background Technology

[0002] Radiation calorimetry diagnostic systems have been established or are about to be established at numerous nuclear fusion experimental facilities both domestically and internationally, including EAST, JET, DIII-D, ASDEX-U, JT-60U, Alcator C-Mod, BEST, and ITER. At the EAST fully superconducting tokamak, radiation calorimetry, as a key diagnostic tool for plasma discharge parameters, primarily measures the total radiated power and its distribution during plasma discharge experiments, employing metal film resistive radiation calorimeters to measure radiation across a wide energy range. In the complex electromagnetic environment of the EAST facility, the spatially distributed multi-detector arrays of radiation calorimetry can generate abnormal channel combinations due to variations in plasma discharge conditions corresponding to the current discharge gun number, thus affecting the quality of radiation calorimetry data and its application effectiveness. Furthermore, during the operation of large tokamak devices, the frequency of plasma discharges is high, and the number of daily discharges is large. Therefore, achieving efficient and reliable online data analysis is crucial for multi-channel radiation calorimetry diagnostic systems.

[0003] Currently, online data analysis systems for radiation calorimetry diagnostics meet the basic requirements for data analysis, but their main shortcomings are:

[0004] (1) It is impossible to automatically identify and compensate for abnormal channels in multi-channel radiation calorimetry diagnosis based on the current discharge gun number;

[0005] (2) The UAV interactive interface cannot automatically load, set and save the various parameters required for online calculation of radiation intensity and total radiation power; the various parameters required for calculating total radiation power include physical parameters and control parameters such as fusion experimental device and plasma discharge experiment related parameters, radiation calorimetry diagnostic detector characteristic parameters, detector related installation location information, etc.

[0006] (3) Experimental data of the original visual multi-channel radiation calorimetry diagnostic signal, the multi-channel radiation intensity signal obtained online, and the total radiation power signal obtained online were not integrated under a unified framework. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention proposes an online data analysis method and system for fusion radiation calorimetry diagnostics. By applying diagnostic expert rules from a fusion radiation calorimetry diagnostic expert rule base based on multidimensional parameter data fusion, it aims to achieve efficient online data analysis for fusion radiation calorimetry diagnostics. This lays the foundation for providing accurate absolute radiation power measurement data in nuclear fusion plasma discharge experiments and provides reliable data support for studying plasma energy loss mechanisms and developing related control technologies.

[0008] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0009] An online data analysis method for fusion radiation calorimetry diagnostics includes the following steps:

[0010] S1. Establish an expert rule base for fusion radiation calorimetry diagnosis based on multidimensional parameter data fusion;

[0011] S2. Create a pagination-based human-computer interface, and nest a unified canvas frame and its internal sub-images within the homepage of the pagination-based human-computer interface;

[0012] S3. Obtain the current discharge gun number of the fusion plasma discharge experiment online;

[0013] S4. Based on the current discharge gun number, acquire the time-series data of multiple signal sources in the fusion experiment remote pulse database; the multiple signal sources include the multi-channel raw voltage signal of the radiation calorimetry diagnostic detector array, the multi-channel input power signal of the plasma heating system, and the plasma discharge current signal; the multiple signals of the multiple signal sources have different sampling rates, data start times, data end times, and data durations;

[0014] S5. Store the acquired timing data from multiple signal sources into a local pulse database;

[0015] S6. According to the expert rules for fusion radiation calorimetry diagnosis, process the time-series data of multiple signal sources, identify and compensate for abnormal channels in radiation calorimetry diagnosis, calculate the radiation intensity of multiple channels using the compensated radiation calorimetry diagnosis data, and finally calculate the total radiation power of the current discharge gun number based on the radiation intensity data of multiple channels.

[0016] S7. Based on a unified canvas framework, integrate and visualize multiple signal data of the current discharge gun number;

[0017] S8. Return to S3 and continue waiting to obtain the next current discharge cannon number until all discharge cannon numbers are obtained.

[0018] An online data analysis system for fusion radiation calorimetry diagnostics includes:

[0019] The pulse database operation module reads, writes, and creates multi-signal source data in the current pulse tree named after the current discharge gun number;

[0020] The plasma discharge current data processing module calculates the time-domain statistical characteristic value and the actual discharge duration of the plasma discharge current corresponding to the current discharge gun number, classifies the plasma discharge current type, and determines the plasma discharge current type of the current discharge gun number.

[0021] The diagnostic raw data preprocessing module preprocesses the raw radiation calorimetry diagnostic data, segments the data according to the actual discharge duration, data duration and data purpose, calculates the time domain statistical characteristic values ​​of all channels, detects and statistically analyzes the waveform change types of the time series curves of all channels, and calculates the detector characteristic value.

[0022] The correlation analysis data module calculates the correlation coefficient matrix and p-value matrix between multiple channels;

[0023] The abnormal channel identification and compensation module identifies the abnormal radiation calorimetry diagnostic channel of the current discharge gun number multiple times according to the diagnostic expert rules; and compensates for the abnormal channel in multiple stages.

[0024] The online calculation module for radiation intensity and total radiation power is used to calculate the multi-channel radiation intensity based on the radiation calorimetry diagnostic data after abnormal channel compensation, the characteristic parameters of the multi-detector array and its spatial location parameters, and to calculate the total radiation power online by performing area integral and volume integral through weighted summation.

[0025] The visualization module is used to create a unified canvas framework and its subplots that integrate and display time-series curves of multiple signals. Subplots are used to draw time-series curves of multiple signals and play them automatically in the order of the subplots.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] 1. To meet the needs of researching plasma energy loss mechanisms and developing related control technologies on the EAST (Experimental Advanced Superconducting Tokamak) device, this invention proposes a novel data-driven technical solution and systematic design approach, successfully developing an online data analysis method and system for fusion radiation calorimetry diagnosis. This method first establishes a fusion radiation calorimetry diagnosis expert rule base based on multi-dimensional parameter data fusion, a paginated human-machine interface, and a unified canvas framework with a nested homepage. Then, based on the current discharge gun number, it acquires the multi-signal source time-series data of the current pulse tree in the remote pulse database of the fusion experiment and transfers it to the local pulse database. Finally, it performs data processing and analysis according to the fusion radiation calorimetry diagnosis expert rules, achieving… The radiation calorimetry diagnostic system automatically identifies and compensates for abnormal channels, then calculates the radiation intensity and total radiation power of multiple channels online. Finally, it integrates and visualizes experimental data of various signals under a unified canvas framework. This online data analysis system for radiation calorimetry diagnostics mainly includes 12 modules: pulse database operation module, plasma discharge current data processing module, diagnostic raw data preprocessing module, correlation analysis data module, abnormal channel identification module, abnormal channel compensation module, online calculation module of radiation intensity and total radiation power, heating system data processing module, paginated human-machine interface module based on Qt framework, integrated visualization module based on unified canvas framework, parameter operation module based on JSON file, and discharge gun number information processing module.

[0028] 2. This invention designs and implements an expert rule base for fusion radiation calorimetry diagnosis based on multi-dimensional parameter data fusion. The diagnostic expert rules in this expert rule base are mainly used for plasma discharge current type classification, data segmentation, preprocessing of raw radiation calorimetry diagnostic data, abnormal channel identification, abnormal channel compensation, online calculation of multi-channel radiation intensity, online calculation of total radiation power, and elimination of temperature drift effects that may exist under long-pulse discharge. The multi-dimensional parameters used in the data fusion mainly include: online acquired multi-signal source data, time-series statistical characteristic values ​​of various signals, waveform change types and trends of time-series curves, radiation calorimetry diagnostic detector characteristics, and correlation data between various signals. According to the diagnostic expert rules in this diagnostic expert rule base, experimental data is processed and analyzed online, thereby realizing automatic identification and compensation of abnormal radiation calorimetry diagnostic channels of the current discharge gun number, and realizing online calculation of multi-channel radiation intensity and total radiation power.

[0029] 3. This invention designs and implements automatic identification of abnormal channels in radiation calorimetry diagnosis based on the discharge gun number. Abnormal channel identification is performed in multiple stages, including: pre-identification of abnormal channels based on the time-domain statistical characteristics of multi-channel radiation calorimetry diagnostic data, the waveform change type of time-series curves, and the characteristics of the metal film resistive radiation calorimeter; secondary identification of abnormal channels based on plasma discharge current type, discharge duration, and correlation analysis data between various signals; and tertiary identification of abnormal channels based on plasma discharge current type, discharge duration, and trend data of multi-channel radiation intensity time-series curves. Simultaneously, automatic compensation of abnormal channels in radiation calorimetry diagnosis is achieved by replacing the automatically identified abnormal channel data with the nearest neighbor normal channel data, or by replacing it with data obtained by averaging adjacent normal channel data.

[0030] 4. This invention designs and implements a paginated human-machine interface based on the Qt framework. This paginated human-machine interface, combined with a parameter operation module based on JSON files, enables the online data analysis system for fusion radiation calorimetry diagnosis to visualize physical and control parameters, automatically load parameter information, and update and record it online or offline, thereby improving the system's generalization ability. At the same time, this paginated human-machine interface, combined with an integrated visualization module based on a unified canvas framework, enables the integrated visualization of multi-signal experimental data.

[0031] 5. This invention designs and implements integrated visualization of multiple signal data based on a unified canvas framework. By creating a unified canvas framework and its sub-graphs, the sub-graphs plot multiple signal time series data curves and automatically play multiple signal time series curves in the order of the sub-graphs. Attached Figure Description

[0032] Figure 1 This is a flowchart of an online data analysis method for fusion radiation calorimetry diagnostics according to the present invention;

[0033] Figure 2 This is a schematic diagram of the components and data flow between modules of an online analysis system for fusion radiation calorimetry diagnostics.

[0034] Figure 3 A flowchart for the raw data preprocessing module of radiation calorimetry diagnostics;

[0035] Figure 4 The diagram shows the module composition and workflow of the abnormal channel identification module.

[0036] Figure 5-1 For discharge gun number 159002, the time-series curves of radiation calorimetry diagnostic channels 22, 24 and 38 after Z-score normalization and plasma discharge current are compared.

[0037] Figure 5-2For discharge gun number 158501, the time-series curves of radiation calorimetry diagnostic channels 22, 24 and 38 after Z-score normalization and plasma discharge current are compared.

[0038] Figure 6-1 Correlation coefficient diagram between all normal channels for radiation calorimetry diagnosis after pre-identification of plasma discharge current and abnormal channels for discharge gun No. 159002;

[0039] Figure 6-2 Correlation coefficients between all normal channels and between all normal channels and plasma discharge current after pre-identification of the abnormal channel of discharge gun No. 159002 for radiation calorimetry diagnosis.

[0040] Figure 7 For discharge gun number 159002, the radiation intensity time series curves obtained by online calculation of radiation calorimetry diagnostic channels 24, 36, 38 and 39 are shown.

[0041] Figure 8-1 A subplot of the timing curve of the input power signal of the plasma discharge heating system at discharge gun number 159002;

[0042] Figure 8-2 This is a subplot of the time-series curve of the total radiation power signal for radiation calorimetry diagnostics obtained online at discharge gun number 159002. Detailed Implementation

[0043] 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 accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other. To achieve the above objectives, this invention adopts the following technical solution.

[0044] This invention provides an online data analysis method for fusion radiation calorimetry diagnostics, comprising the following steps:

[0045] S1. Establish an expert rule base for fusion radiation calorimetry diagnosis based on multidimensional parameter data fusion;

[0046] S2. Create a pagination-based human-computer interface, and create a unified canvas frame and its internal sub-graphs nested within the homepage of the pagination-based human-computer interface;

[0047] S3. Obtain the current discharge gun number of the fusion plasma discharge experiment online;

[0048] S4. Based on the current discharge gun number, acquire the time-series data of multiple signal sources in the fusion experiment remote pulse database; the multiple signal sources include the multi-channel raw voltage signal of the radiation calorimetry diagnostic detector array, the multi-channel input power signal of the plasma heating system, and the plasma discharge current signal; the multiple signals of the multiple signal sources have different sampling rates, data start times, data end times, and data durations;

[0049] S5. The acquired timing data of the multiple signal sources is stored in a local pulse database;

[0050] S6. According to the fusion radiation calorimetry diagnostic expert rules, process the time-series data of the multiple signal sources, identify and compensate for abnormal radiation calorimetry diagnostic channels, calculate the multi-channel radiation intensity using the compensated radiation calorimetry diagnostic data, and finally calculate the total radiation power of the current discharge gun number based on the multi-channel radiation intensity data.

[0051] S7. Based on the unified canvas framework, integrate and visualize multiple signal data of the current discharge gun number;

[0052] S8. Return to S3 and continue waiting to obtain the next current discharge cannon number until all discharge cannon numbers are obtained.

[0053] Furthermore, the expert rules for fusion radiation calorimetry diagnosis based on multi-dimensional parameter data fusion are used for: plasma discharge current type classification, data segmentation, preprocessing of raw radiation calorimetry diagnostic data, abnormal channel identification, abnormal channel compensation, online calculation of multi-channel radiation intensity, online calculation of total radiation power, and elimination of temperature drift effects that may exist under long-pulse discharge. Abnormal channel identification is performed multiple times, including pre-identification of abnormal channels, secondary identification of abnormal channels based on plasma discharge current type, discharge duration, and correlation analysis, and tertiary identification of abnormal channels based on plasma discharge current type, discharge duration, and multi-channel radiation intensity trend data. The multi-dimensional parameters include online acquired time-series data from multiple signal sources, time-series statistical characteristic values ​​of various signals, waveform change types and trend data of time-series curves, characteristics of radiation calorimetry diagnostic detectors, and correlation analysis data between various signals.

[0054] Furthermore, the data segments are divided into three types according to different uses: data segments for calculating correlation analysis data, data segments for calculating time-series statistical characteristic values, and data segments for calculating radiation intensity and total radiation power in radiation calorimetry. The data segments for calculating correlation analysis data include data segments for correlation analysis between plasma discharge current and multi-channel radiation calorimetry signals, and data segments for correlation analysis between multiple channels in radiation calorimetry. The data segments for calculating time-series statistical characteristic values ​​include data segments for plasma discharge current characteristic values ​​and data segments for calculating multi-channel radiation calorimetry characteristic values.

[0055] Furthermore, the preprocessing of the raw data for radiation calorimetry includes: online removal of baseline drift and zero-phase low-pass filtering of the raw data for multi-channel radiation calorimetry; data segmentation according to application; calculation of time-domain statistical characteristic values ​​of the radiation calorimetry data; calculation of characteristic values ​​such as the characteristics of the detector used for radiation calorimetry; and detection and statistical analysis of waveform change types of the time-series curves of multi-channel radiation calorimetry using an overlapping sliding window. The time-domain statistical characteristic values ​​include maximum value, minimum value, mean, median, variance, maximum first-order difference, minimum first-order difference, maximum second-order difference, minimum second-order difference, skewness, and kurtosis. The characteristic values ​​of the detector used for radiation calorimetry include thermal time constant, baseline drift rate, proportion of high and low frequency noise, ambient temperature fluctuation amplitude, and signal-to-noise ratio.

[0056] Furthermore, the abnormal channel pre-identification specifically includes identifying abnormal channels that do not reflect any physical information and have obvious abnormal characteristics based on the statistical analysis of the time-domain statistical characteristic values ​​and time-series curve waveform change types of multi-channel radiation calorimetry diagnostic data; and identifying abnormal channels whose data is unreliable based on the characteristic values ​​of the detectors used in multi-channel radiation calorimetry diagnostics.

[0057] Furthermore, the specific steps for secondary identification of abnormal channels based on plasma discharge current type, discharge duration, and correlation analysis include:

[0058] A1. Calculate the actual discharge duration of the current discharge gun number based on the plasma discharge current Ip data;

[0059] A2. Based on the actual discharge duration and diagnostic expert rules, the raw data of plasma discharge current and radiation calorimetry are segmented and processed, and the processed plasma discharge current and radiation calorimetry data are aligned.

[0060] A3. Calculate the time-domain statistical characteristic value of the plasma discharge current Ip;

[0061] A4. Based on the time-domain statistical characteristic value of the plasma discharge current Ip and the diagnostic expert rules, classify the Ip type and determine the Ip type of the current discharge shot number; the Ip type is divided into six categories, including Class I test shot, Class II non-breakdown current waveform, Class III breakdown but quickly ruptured current waveform, Class IV breakdown but incomplete current waveform, Class V breakdown and complete current waveform, and Class VI long pulse discharge;

[0062] A5. Eliminate the abnormal channels obtained from the abnormal channel pre-identification to obtain the normal channel data set after the abnormal channel pre-identification for the current discharge gun number;

[0063] A6. Correlation analysis data acquisition, including correlation analysis data such as correlation coefficient matrix and p-value matrix between multiple normal channels of radiation calorimetry, calculated based on the Spearman rank correlation method and the pre-identified normal channel data set and plasma discharge current data, as well as correlation analysis data such as correlation coefficient matrix and p-value matrix between each normal channel of radiation calorimetry and plasma discharge current;

[0064] A7. By combining the plasma discharge current type and discharge duration, anomaly detection is performed on the correlation analysis data to obtain the abnormal channel after secondary identification of the current discharge gun number in the radiation calorimetry diagnosis.

[0065] Furthermore, the abnormal channel compensation refers to the method of compensating for the data of the automatically identified abnormal channels in radiation calorimetry diagnosis by replacing them with the data of the nearest neighbor channel or by taking the average value of the data of the adjacent normal channels, according to the abnormal channel compensation strategy in the diagnostic expert rule base.

[0066] Furthermore, the integration and visualization of multiple signal data of the current discharge gun number based on the unified canvas framework specifically includes: creating a unified canvas framework and its sub-graphs and molecular graphs to automatically draw multiple signal time-series curves, and automatically playing them in the order of the sub-graphs; the multiple signals drawn by the molecular graph include the multi-channel raw data of the radiation calorimetry diagnostic detector array and its multi-channel data after removing baseline drift, the online calculated multi-channel radiation intensity, the online calculated total radiation power, the input power of different plasma heating systems, and the plasma discharge current; the input power signals of different plasma heating systems have different sampling rates, data start times, data end times, and data durations.

[0067] The present invention also provides an online data analysis system for fusion radiation calorimetry diagnostics, the system comprising:

[0068] The pulse database operation module connects to the remote pulse database via the fusion experiment's local area network and reads and writes multi-signal source data in the current pulse tree named after the current discharge gun number. It also creates a local current pulse tree named after the current discharge gun number in the local pulse database and reads and writes the signals in it.

[0069] The plasma discharge current data processing module calculates the time-domain statistical characteristic value and the actual discharge duration of the plasma discharge current corresponding to the current discharge gun number, classifies the plasma discharge current type, and determines the plasma discharge current type of the current discharge gun number. The time-domain statistical characteristic value of the discharge current includes the maximum value, minimum value, mean, median, variance, and data duration.

[0070] The diagnostic raw data preprocessing module removes baseline drift and zero-phase low-pass filtering from the raw radiation calorimetry diagnostic data online, normalizes it, performs various data segmentations based on the actual discharge duration of the current discharge gun number, the data duration of the raw radiation calorimetry diagnostic data, and the data purpose, calculates the time-domain statistical characteristic values ​​of all channels of radiation calorimetry diagnostics, detects and statistically analyzes the waveform change types of all channel time-series curves, and calculates the characteristic values ​​of the radiation calorimetry diagnostic detector.

[0071] The correlation analysis data module calculates the correlation coefficient matrix and p-value matrix between multiple channels of radiation calorimetry diagnosis. After aligning the plasma discharge current and multi-channel radiation calorimetry diagnosis signals, it calculates the correlation coefficient matrix and p-value matrix between the plasma discharge current and multiple channels of radiation calorimetry diagnosis, as well as other correlation analysis data.

[0072] The abnormal channel identification module identifies abnormal channels for radiation calorimetry diagnosis of the current discharge gun number multiple times according to diagnostic expert rules.

[0073] The abnormal channel compensation module compensates the abnormal channels identified by the abnormal channel identification module in stages according to the abnormal channel compensation strategy.

[0074] The online calculation module for radiation intensity and total radiation power calculates the chord integral radiation intensity of the plasma along the measurement string, i.e., the multi-channel radiation intensity, based on the radiation calorimetry diagnostic data after abnormal channel compensation, the characteristic parameters of the radiation calorimetry diagnostic multi-detector array, and its spatial position parameters. Finally, based on the multi-channel radiation intensity, the total radiation power is calculated online by performing area integral and volume integral using a weighted summation method.

[0075] The heating system data processing module is used to align the multiple input power signals of the heating system using a unified time, frequency reduction and edge copying and padding method, calculate the total input power signal, and integrate all input power signals to obtain a final input power data set of uncertain dimension.

[0076] An integrated visualization module based on a unified canvas framework is used to create a unified canvas framework and its subplots that integrate and display time-series curves of multiple signals. Subplots are used to draw time-series curves of multiple signals and play them automatically in the order of the subplots.

[0077] The pagination human-computer interface module is used to create a human-computer interaction interface for pagination parameters and data visualization, and to nest the unified canvas framework and its sub-graphs on the home page of the pagination human-computer interface.

[0078] The parameter operation module is used to update and record various physical and control parameters involved in the online data analysis system for fusion radiation calorimetry diagnosis, either online or offline.

[0079] The discharge gun number information processing module is used to monitor the current discharge gun number information transmitted on the fusion experiment's local area network, parse the received information packets, and save the discharge gun number information.

[0080] Furthermore, the abnormal channel identification module includes a pre-identification module, a secondary identification module, and a tertiary identification module. The pre-identification module, based on all channel time-domain statistical characteristic values, time-series curve waveform change types, and detector characteristic values ​​detected and statistically analyzed by the diagnostic raw data preprocessing module, applies the diagnostic expert rules to pre-identify abnormal channels, thereby obtaining the pre-identified abnormal channel combination BadChan_GP1. The secondary identification module, based on the correlation analysis data obtained from the correlation analysis data module, the plasma discharge current type and discharge duration of the current discharge gun number, employs robust probability analysis and the IRQ quartile method, and applies the diagnostic expert rules... The abnormal channel is identified a second time, resulting in the abnormal channel combination BadChan_GP2. The third identification module first uses a linear regression method based on signal amplitude to obtain multi-channel radiation intensity trend data, and then combines the plasma discharge current type and discharge duration to apply the diagnostic expert rules to identify the abnormal channel a third time, resulting in the abnormal channel combination BadChan_GP3. The abnormal channel combinations BadChan_GP1, BadChan_GP2, and BadChan_GP3 are deduplicated and sorted to obtain the final abnormal channel combination BadChan_GP corresponding to the current discharge gun number.

[0081] The following are specific examples:

[0082] like Figure 1 As shown, this invention provides an online data analysis method for fusion radiation calorimetry diagnostics, comprising:

[0083] The first step is to construct an expert rule base for fusion radiation calorimetry diagnosis based on multi-dimensional parameter data fusion. This expert rule base is primarily used for: plasma discharge current type classification, data segmentation, preprocessing of raw radiation calorimetry diagnostic data, anomaly channel identification, anomaly channel compensation, online calculation of multi-channel radiation intensity, online calculation of total radiation power, and elimination of temperature drift effects under long-pulse discharge. Anomaly channel identification is performed multiple times, including pre-identification, secondary identification based on plasma discharge current type and discharge duration and correlation analysis, and tertiary identification based on plasma discharge current type, discharge duration, and multi-channel radiation intensity trend data. The multi-dimensional parameters mainly include online acquired multi-signal source data, time-series statistical characteristic values ​​of various signals, waveform change types and trends of time-series curves, characteristic values ​​of radiation calorimetry diagnostic detectors, and correlation analysis data between various signals.

[0084] The second step involves obtaining the current discharge gun number information and then reading the timing data of multiple signal sources from the current pulse tree named after the current discharge gun number in the fusion experiment remote pulse database. The multiple signal sources include the multi-channel raw voltage signal of the radiation calorimetry diagnostic detector array, the multi-channel input power signal of the plasma heating system, and the plasma discharge current signal.

[0085] The third step is to create a current pulse tree named after the current discharge gun number in the local pulse database, and store the multi-signal source timing data read in the second step into the local current pulse tree;

[0086] The fourth step involves processing and analyzing the multi-signal source time-series data stored locally according to the fusion radiation calorimetry diagnostic expert rules, automatically identifying and compensating for abnormal radiation calorimetry diagnostic channels, calculating the multi-channel radiation intensity using the compensated radiation calorimetry diagnostic data, and finally calculating the total radiation power based on the multi-channel radiation intensity data.

[0087] The fifth step integrates and visualizes various signal data of the current discharge gun number based on a unified canvas framework; finally, return to the second step to obtain the next current discharge gun number.

[0088] In this embodiment, as Figure 2 As shown, this invention provides an online data analysis system for fusion radiation calorimetry diagnosis, including a pulse database operation module, a plasma discharge current data processing module, a diagnostic raw data preprocessing module, a correlation analysis data module, an abnormal channel identification module, an abnormal channel compensation module, an online calculation module for radiation intensity and total radiation power, a heating system data processing module, a paginated human-machine interface module based on the Qt framework, an integrated visualization module based on a unified canvas framework, a parameter operation module based on JSON files, and a discharge gun number information processing module.

[0089] During the plasma discharge experiment of the nuclear fusion tokamak device, the discharge gun number information processing module of the online data analysis system for fusion radiation calorimetry receives and parses the current discharge gun number information through the fusion experiment's local area network, thereby triggering online data analysis based on the current discharge gun number and continuing to monitor the information of the next discharge gun number. The pulse database operation module connects to the current pulse tree in the remote pulse database of the fusion experiment through the fusion experiment's local area network, obtains the time-series data of various signal sources required for online data analysis, and transfers it to the current pulse tree created in the local pulse database. The paginated human-machine interface module based on the Qt framework loads, sets, and saves the parameter information required for radiation calorimetry online data analysis through the parameter operation module based on JSON files. The plasma discharge current data processing module calculates based on the plasma discharge current data. The actual discharge duration of the current discharge gun number is obtained by segmenting the plasma discharge current data into a data set IpData, with the discharge zero time T_Ip0 and the plasma discharge current data end time T_IpDataEnd as the start and end times. Then, the time-series statistical characteristic value of the plasma discharge current is calculated, and the plasma discharge current type of the current discharge gun number is determined according to the time-series statistical characteristic value set and the diagnostic expert rules. The time-series statistical characteristic value of the plasma discharge current includes the maximum value, minimum value, mean, median, variance, and data duration. The plasma discharge current type is divided into six categories, including Class I test gun, Class II non-breakdown current waveform, Class III breakdown but quickly ruptured current waveform, Class IV breakdown but incomplete current waveform, Class V breakdown and complete current waveform, and Class VI long pulse discharge.

[0090] like Figure 3 As shown, the diagnostic raw data preprocessing module first performs baseline drift removal and zero-phase low-pass filtering on the raw radiation calorimetry diagnostic data. Then, it segments the data according to different applications: the first segment (corresponding to the time from zero discharge time to zero discharge current time), the second segment (corresponding to the time from zero discharge current time to the end of radiation calorimetry diagnostic data), and the third segment (corresponding to the time from -1 second discharge time to the end of radiation calorimetry diagnostic data). For the first and second segment data, the time-domain statistical characteristic values ​​of all channels in the radiation calorimetry diagnostic data are calculated. For the first segment data... The overlapping sliding window is used to detect and statistically analyze the waveform change types of all channel time-series curves. After Z-SCORE normalization of the front-end and back-end data segments, the characteristic values ​​of the radiation calorimetry diagnostic detector are calculated. The time-domain statistical characteristics of radiation calorimetry include maximum value, minimum value, mean, median, variance, maximum first-order difference, minimum first-order difference, maximum second-order difference, minimum second-order difference, skewness, and kurtosis. The characteristic values ​​of the radiation calorimetry diagnostic detector include thermal time constant, baseline drift rate, high and low frequency noise ratio, ambient temperature fluctuation amplitude, and signal-to-noise ratio.

[0091] The correlation analysis data module, based on the Spearman rank correlation method, calculates correlation coefficient matrices and p-value matrices among multiple channels in radiation calorimetry diagnostics, and aligns them accordingly. Figure 3 After obtaining the plasma discharge current IpData and multi-channel radiation calorimetry diagnostic data from the front-end data shown, the correlation coefficient matrix and p-value matrix between the plasma discharge current and the multi-channel radiation calorimetry diagnostic data are calculated. The p-value matrix calculated by the online data analysis system for fusion radiation calorimetry diagnostics has extremely low values, indicating that the obtained correlation coefficients are statistically significant. Therefore, the p-value matrix is ​​generally not discussed further.

[0092] like Figure 4 As shown, the abnormal channel identification module includes three sub-modules. The abnormal channel pre-identification module, based on the time-domain statistical characteristic values ​​of the multi-channel radiation calorimetry diagnostics, the waveform change type of the time-series curve, and the characteristic values ​​of the radiation calorimetry diagnostic detector obtained from the diagnostic raw data preprocessing module, applies diagnostic expert rules to pre-identify abnormal radiation calorimetry diagnostic channels, thus obtaining the pre-identified abnormal channel combination BadChan_GP1. The abnormal channel secondary identification module, based on the correlation coefficient matrix and p-value matrix obtained from the correlation analysis data module, as well as the plasma discharge current type and actual discharge duration, uses the IRQ quartile method and robust probability analysis, applying diagnostic expert rules... The abnormal channel is identified a second time, resulting in the abnormal channel combination BadChan_GP2. The abnormal channel identification module identifies the abnormal channel a third time based on the plasma discharge current type, discharge duration, and multi-channel radiation intensity trend data obtained by using a linear regression method based on signal amplitude. Combined with diagnostic expert rules, the abnormal channel is identified a third time, resulting in the abnormal channel combination BadChan_GP3. The abnormal channel combinations BadChan_GP1, BadChan_GP2, and BadChan_GP3 identified in the third identification are deduplicated and sorted to obtain the final abnormal channel combination BadChan_GP for the current discharge gun number. The abnormal channel compensation first uses the abnormal channel combinations BadChan_GP1 and BadChan_GP2 obtained from the abnormal channel pre-identification and secondary abnormal channel identification, as well as diagnostic expert rules, to perform compensation. Then, the online calculation module for radiation intensity and total radiation power initially calculates the multi-channel radiation intensity, and uses a linear regression method based on signal amplitude to obtain its trend data. Finally, the abnormal channel compensation is performed again based on the final abnormal channel combination BadChan_GP and diagnostic expert rules, and then the online calculation module for radiation intensity and total radiation power completes the online calculation of multi-channel radiation intensity and online calculation of total radiation power.

[0093] The heating system data processing module aligns the heating system input power signal data obtained from the current pulse tree of the fusion experiment's remote pulse database using a unified time, frequency reduction, and edge replication complement method. After data integration, it obtains a heating system input power signal data set of varying dimensions. The integrated visualization module, based on a unified canvas framework, creates a unified canvas framework and its subgraphs, plotting various signal time-series curves and automatically playing them in subgraph order. The paginated human-machine interface module, based on the Qt framework, interacts with other modules and radiation calorimetry experimental personnel to visualize the physical and control parameters of the fusion radiation calorimetry online data analysis system, automatically load parameter information, and update and record parameter information online or offline. The paginated human-machine interface module, based on the Qt framework, nests the unified canvas framework created by the integrated visualization module within the first page of the human-machine interface, thus achieving integrated visualization of the experimental data.

[0094] In specific implementation, taking the current discharge gun number 159002 as an example, the plasma discharge current data processing module calculates the actual discharge duration to be 12 seconds, and determines that the plasma discharge current type is Class V breakdown and the current waveform is complete. Figure 5-1 As shown, the plasma discharge current has a complete initial level (approximately 0 kA), a slow rise, a flat top, a slow fall to the recovery level (approximately 0 kA) waveform; Figure 5-1 The timing curves of radiation calorimetry diagnostic channels 22, 24, and 38 after Z-SCORE normalization show waveform changes that are roughly similar to those of plasma discharge current timing curves. Based on the calculation results of the diagnostic raw data preprocessing module, the abnormal channel pre-identification module first identifies the abnormal channel combination [5, 18, 20, 37] based on the multi-channel time-domain statistical characteristic values ​​of radiation calorimetry. This is because the data from radiation calorimetry channels Chan5, Chan18, Chan20, and Chan37 do not reflect any physical information, representing a detector channel fault. Next, based on the statistical results of the waveform change types of all channel time-series curves in radiation calorimetry, the waveform of channel [18, 20] remains horizontal throughout the entire discharge process, indicating that the detector signal in that channel is extremely weak or faulty. Finally, based on the characteristic values ​​such as the detector characteristic values ​​in radiation calorimetry, the abnormal channel combination [5, 17, 18, 20, 37] is obtained, which makes the data unreliable due to anomalies in the detector's thermal time constant or signal-to-noise ratio. The abnormal channel pre-identification module, according to diagnostic expert rules, integrates the multi-channel time-domain statistical characteristic values, the waveform change types of all channel time-series curves, and the detector characteristic values ​​to obtain the pre-identified abnormal channel combination BadChan_GP1=[5, 17, 18, 20, 37].

[0095] The correlation analysis data module employs the Spearman rank correlation method to calculate correlation coefficient matrices and p-value matrices among all normal channels in the radiation calorimetry diagnostics after anomaly channel pre-identification, including the correlation coefficient matrix and p-value matrix. The p-value matrix data indicates that the correlation coefficient matrices are statistically significant. Figure 6-1 The correlation coefficients between the plasma discharge current of gun number 159002 and the radiation calorimetry diagnosis after abnormal channel pre-identification are all positive, and all except channel 23 are moderately positive. Figure 6-2 The radiation calorimetric diagnosis after pre-identifying the abnormal channel for gun number 159002 shows that all correlation coefficients between normal channels and between all normal channels and plasma discharge current are positive, with over 90% of normal channels showing a moderate positive correlation. The abnormal channel secondary identification module, based on the correlation coefficient matrix obtained from the correlation analysis data module, as well as the plasma discharge current type and actual discharge duration, uses the IRQ quartile method and robust probability analysis, applying diagnostic expert rules to identify abnormal channels a second time. The resulting abnormal channel combination BadChan_GP2 is empty. Deduplication and sorting of the abnormal channel combinations BadChan_GP1 and BadChan_GP2 yields the abnormal channel combination BadChan_GP12. The abnormal channel compensation module performs initial compensation on the abnormal channels in BadChan_GP12 according to the abnormal channel compensation strategy in the diagnostic expert rule base. It replaces the data of abnormal channels 17 and 18 with the data of the nearest normal channels, namely channels 16 and 19, respectively. At the same time, it replaces the data of abnormal channels 5, 20 and 37 with the average of the data of the adjacent normal channels, namely 4 and 6, 19 and 21 and 36 and 38, respectively, thus obtaining the radiation calorimetry diagnostic signal data after initial compensation.

[0096] The online calculation module for radiation intensity and total radiation power uses the radiation calorimetry diagnostic signal data obtained after the initial compensation by the abnormal channel compensation module, and applies diagnostic expert rules to calculate the preliminary multi-channel radiation intensity for radiation calorimetry diagnosis. The abnormal channel tri-identification module uses the preliminary multi-channel radiation intensity data and a linear regression method based on the radiation intensity signal amplitude to calculate the multi-channel radiation intensity trend data. Then, combining the plasma discharge current type and discharge duration with diagnostic expert rules, it identifies the abnormal channel combination BadChan_GP3=[38, 39]. Figure 7The figure shows the radiation intensity time series curves of radiation calorimetry diagnostic channels 24, 36, 38 and 39 calculated online for gun number 159002. It can be seen that the radiation intensity time series curves of channels 24 and 36 are obviously close to zero and stable, while the radiation intensity time series curves of channels 38 and 39 are not close to zero and have a clear upward trend.

[0097] The abnormal channel combinations BadChan_GP1, BadChan_GP2, and BadChan_GP3 are deduplicated and sorted to obtain the final abnormal channel combination BadChan_GP=[5,17,18,20,37,38,39] for gun number 159002. The abnormal channel compensation module performs abnormal channel compensation again based on the abnormal channel compensation strategy in the radiation calorimetry diagnostic expert rule base and BadChan_GP. In addition to compensating for [5, 17, 18, 20, 37], the data for abnormal channels 38 and 39 are replaced with the data from the nearest normal channels, namely channels 37 and 40, respectively, thus obtaining the final radiation calorimetry diagnostic signal data. The online calculation module for radiation intensity and total radiation power then calculates the final multi-channel radiation intensity online based on the final radiation calorimetry diagnostic signal data, and then calculates the total radiation power online based on the final multi-channel radiation intensity.

[0098] In specific implementation, the fusion experiment remote pulse database contains four types of heating systems, with a total of 12 output power signals. At discharge gun number 159002, the pulse database operation module only obtains PECRH3I and PECRH4I output power signals from the remote pulse database. Since there is no issue of inconsistent sampling rate and time at this time, the heating system data processing module does not need to perform data processing such as time unification, frequency reduction, and edge replication padding. It directly calculates the total output power PHeatIn of the heating system and then integrates the time-series data of PECRH3I, PECRH4I, and PHeatIn to obtain the data set PHeatInGP. The paginated human-machine interface module based on the Qt framework nests the canvas framework and its subgraphs created by the integrated visualization module based on the unified canvas framework within the homepage of the human-machine interface. In the paginated human-machine interface homepage of the online data analysis system for fusion radiation calorimetry diagnosis, the integrated visualization module based on the unified canvas framework creates a unified canvas framework and multiple subgraphs within it, drawing a diagram like... Figure 8-1 The timing curve of the input power signal of the plasma discharge heating system is shown, and it is plotted in another subplot as follows. Figure 8-2 The time-series curve of the total radiation power signal obtained from online calculation for radiation calorimetry is shown.

[0099] In specific implementation, taking the current discharge gun number 158501 as an example, the plasma discharge current data processing module calculates the actual discharge duration to be 7.9 seconds and determines that the plasma discharge current type is Class IV breakdown, but the current waveform is incomplete; for example... Figure 5-2As shown, the plasma discharge current of discharge gun number 158501 exhibits a waveform with an initial level (approximately 0 kA), a slow rise, a flat top, a rapid drop due to plasma rupture, and finally a recovery to the level (approximately 0 kA). Based on the statistical results of the time-domain characteristic values ​​and time-series curve waveform change types calculated by the diagnostic raw data preprocessing module, as well as the characteristic values ​​of the radiation calorimetry detector, the abnormal channel pre-identification module identifies the abnormal channel combination BadChan_GP1=[5,17,18,20,37]. The correlation analysis data module calculates the correlation coefficient matrix and p-value matrix between the plasma discharge current and all normal channels of radiation calorimetry after abnormal channel pre-identification, and calculates the correlation coefficient matrix and p-value matrix, etc., of all normal channels of radiation calorimetry after abnormal channel pre-identification. The p-value matrix data indicates that the correlation coefficient matrix is ​​statistically significant. Based on the correlation coefficient matrix among all normal channels after the pre-identification of abnormal channels, the total number of normal channels with weak positive correlation, no correlation, or negative correlation with other normal channels is counted, resulting in a matrix of weakly correlated channels. The secondary identification module for abnormal channels, based on the correlation coefficient matrix, the matrix of weakly correlated channels, the plasma discharge current type, and the actual discharge duration obtained from the correlation analysis data module, employs robust probability analysis and the IRQ quartile method, and applies diagnostic expert rules to identify abnormal channels a second time, resulting in the secondary identified abnormal channel combination BadChan_GP2=[9,14,29]. The abnormal channel combinations BadChan_GP1 and BadChan_GP2 are deduplicated and sorted to obtain the abnormal channel combination BadChan_GP12=[5, 9,14, 17, 18, 20, 29, 37]. The abnormal channel compensation module performs initial compensation on the abnormal channels in BadChan_GP12, resulting in the initially compensated radiation calorimetry diagnostic signal data. The online calculation module for radiation intensity and total radiation power calculates the preliminary multi-channel radiation intensity based on the radiation calorimetry diagnostic signal data obtained after initial compensation, applying diagnostic expert rules. The abnormal channel triple identification module calculates the multi-channel radiation intensity trend data based on the preliminary multi-channel radiation intensity data using a linear regression method based on the radiation intensity signal amplitude. Then, combining the plasma discharge current type and discharge duration with diagnostic expert rules, it determines that the radiation intensity time-series curves of channels 4, 16, and 17 have not returned to zero and have a significant downward trend, thus identifying BadChan_GP3=[4,16,17].The abnormal channel combinations BadChan_GP1, BadChan_GP2, and BadChan_GP3 obtained from the three identifications were deduplicated and sorted to obtain the final abnormal channel combination BadChan_GP=[4,5,9,14,16,17,18,20,29,37] for discharge gun number 158501.

[0100] In summary, this invention employs a data-driven, systematic technical solution. By creating an expert rule base for fusion radiation calorimetry diagnosis based on multi-dimensional parameter data fusion, and applying the diagnostic expert rules within it, experimental data processing and analysis are performed based on the current discharge gun number. This enables automatic identification and compensation of abnormal channels in radiation calorimetry diagnosis, as well as online calculation of multi-channel radiation intensity and total radiation power. Simultaneously, by creating a paginated human-machine interface based on the Qt framework and a unified canvas framework and its subgraphs nested within the interface's homepage, combined with parameter operation methods based on JSON files, the control of this online data analysis system for radiation calorimetry diagnosis is achieved, along with the automatic loading, online setting, and recording of physical parameters. Furthermore, experimental data from multiple signals are integrated and visualized within the unified framework. This online data analysis method and system for fusion radiation calorimetry diagnosis boasts advantages such as high automation, a user-friendly paginated human-machine interface, and fully integrated and visualized parameters. It not only provides reliable online absolute radiation power measurement data but also operates efficiently and possesses a certain degree of generalization capability, providing strong data support for the experimental operation of large tokamak devices.

Claims

1. An online data analysis method for fusion radiation calorimetry diagnostics, characterized in that, Includes the following steps: S1. Establish an expert rule base for fusion radiation calorimetry diagnosis based on multidimensional parameter data fusion; S2. Create a pagination-based human-computer interface, and nest a unified canvas frame and its internal sub-images within the homepage of the pagination-based human-computer interface; S3. Obtain the current discharge gun number of the fusion plasma discharge experiment online; S4. Based on the current discharge gun number, acquire the time-series data of multiple signal sources in the fusion experiment remote pulse database; the multiple signal sources include the multi-channel raw voltage signal of the radiation calorimetry diagnostic detector array, the multi-channel input power signal of the plasma heating system, and the plasma discharge current signal; Multiple signals from multiple signal sources have different sampling rates, data start times, data end times, and data durations; S5. Store the acquired timing data from multiple signal sources into a local pulse database; S6. According to the expert rules for fusion radiation calorimetry diagnosis, process the time-series data of multiple signal sources, identify and compensate for abnormal channels in radiation calorimetry diagnosis, calculate the radiation intensity of multiple channels using the compensated radiation calorimetry diagnosis data, and finally calculate the total radiation power of the current discharge gun number based on the radiation intensity data of multiple channels. S7. Based on a unified canvas framework, integrate and visualize multiple signal data of the current discharge gun number; S8. Return to S3 and continue waiting to obtain the next current discharge cannon number until all discharge cannon numbers are obtained.

2. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 1, characterized in that, The expert rules for fusion radiation calorimetry diagnosis based on multi-dimensional parameter data fusion are used for: plasma discharge current type classification, data segmentation, preprocessing of raw data for radiation calorimetry diagnosis, abnormal channel identification, abnormal channel compensation, online calculation of multi-channel radiation intensity, online calculation of total radiation power, and elimination of temperature drift effects that may exist under long-pulse discharge. The abnormal channel identification process is conducted in multiple stages, including pre-identification of abnormal channels, secondary identification of abnormal channels based on plasma discharge current type and discharge duration and correlation analysis, and tertiary identification of abnormal channels based on plasma discharge current type, discharge duration and multi-channel radiation intensity trend data. The multi-dimensional parameters include online acquired time-series data of multiple signal sources, time-series statistical characteristic values ​​of multiple signals, waveform change types and trend data of time-series curves, characteristics of radiation calorimetry diagnostic detectors, and correlation analysis data between multiple signals.

3. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 2, characterized in that, The data segments are divided into three types according to different uses: data segments for calculating correlation analysis data, data segments for calculating time-series statistical characteristic values, and data segments for calculating radiation intensity and total radiation power for radiation calorimetry diagnosis. The data segments for calculating correlation analysis data include data segments for correlation analysis between plasma discharge current and multi-channel radiation calorimetry diagnostic signals, and data segments for correlation analysis between multiple channels in radiation calorimetry diagnosis. The data segments for calculating time-series statistical characteristic values ​​include data segments for plasma discharge current characteristic values ​​and data segments for calculating multi-channel radiation calorimetry diagnostic characteristic values.

4. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 2, characterized in that, The preprocessing of the raw data for radiation calorimetry includes: online removal of baseline drift and zero-phase low-pass filtering of the raw data for multi-channel radiation calorimetry; data segmentation according to application; calculation of time-domain statistical characteristics of the radiation calorimetry data; calculation of the characteristic values ​​of the detector used for radiation calorimetry; and detection and statistical analysis of waveform change types of the time-series curves for multi-channel radiation calorimetry using an overlapping sliding window. The time-domain statistical characteristics include maximum value, minimum value, mean, median, variance, maximum first-order difference, minimum first-order difference, maximum second-order difference, minimum second-order difference, skewness, and kurtosis. The characteristic values ​​of the detector used for radiation calorimetry include thermal time constant, baseline drift rate, proportion of high and low frequency noise, ambient temperature fluctuation amplitude, and signal-to-noise ratio.

5. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 2, characterized in that, The pre-identification of abnormal channels specifically includes identifying abnormal channels that do not reflect any physical information and have obvious abnormal characteristics based on the statistical analysis of the time-domain statistical characteristic values ​​and time-series curve waveform change types of multi-channel radiation calorimetry diagnostic data; and identifying abnormal channels whose data are unreliable based on the characteristic values ​​of the detectors used in multi-channel radiation calorimetry diagnostics.

6. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 2, characterized in that, The specific steps for secondary identification of abnormal channels based on plasma discharge current type, discharge duration, and correlation analysis include: A1. Calculate the actual discharge duration of the current discharge gun number based on the plasma discharge current Ip data; A2. Based on the actual discharge duration and diagnostic expert rules, the raw data of plasma discharge current and radiation calorimetry are segmented and processed, and the processed plasma discharge current and radiation calorimetry data are aligned. A3. Calculate the time-domain statistical characteristic value of the plasma discharge current Ip; A4. Based on the time-domain statistical characteristic value of the plasma discharge current Ip and the diagnostic expert rules, classify the Ip type and determine the Ip type of the current discharge shot number; the Ip type is divided into six categories, including Class I test shot, Class II non-breakdown current waveform, Class III breakdown but quickly ruptured current waveform, Class IV breakdown but incomplete current waveform, Class V breakdown and complete current waveform, and Class VI long pulse discharge; A5. Eliminate the abnormal channels obtained from the abnormal channel pre-identification to obtain the normal channel data set after the abnormal channel pre-identification for the current discharge gun number; A6. Correlation analysis data acquisition: Based on the pre-identified normal channel data set and plasma discharge current data, the correlation coefficient matrix and p-value matrix of multiple normal channels for radiation calorimetry diagnosis are calculated as correlation analysis data, and the correlation coefficient matrix and p-value matrix of each normal channel for radiation calorimetry diagnosis and plasma discharge current are calculated as correlation analysis data. A7. By combining the plasma discharge current type and discharge duration, anomaly detection is performed on the correlation analysis data to obtain the abnormal channel after secondary identification of the current discharge gun number in the radiation calorimetry diagnosis.

7. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 2, characterized in that, The abnormal channel compensation refers to the method of compensating for the data of the automatically identified abnormal channels in radiation calorimetry diagnosis by replacing them with the data of the nearest neighbor channel or by taking the average value of the data of the adjacent normal channels, according to the abnormal channel compensation strategy in the diagnostic expert rule base.

8. The online data analysis method for fusion radiation calorimetry diagnostics according to claim 1, characterized in that, The integrated visualization of multiple signal data of the current discharge gun number based on the unified canvas framework specifically includes: creating a unified canvas framework and its sub-graphs and molecular graphs to automatically draw multiple signal time-series curves, and automatically playing them in the order of the sub-graphs; the multiple signals drawn by the molecular graphs include the multi-channel raw data of the radiation calorimetry detector array and its multi-channel data after removing baseline drift, the online calculated multi-channel radiation intensity, the online calculated total radiation power, the input power of different plasma heating systems, and the plasma discharge current; the input power signals of different plasma heating systems have different sampling rates, data start times, data end times, and data durations.

9. An online data analysis system for fusion radiation calorimetry diagnostics, characterized in that, include: The pulse database operation module reads, writes, and creates multi-signal source data in the current pulse tree named after the current discharge gun number; The plasma discharge current data processing module calculates the time-domain statistical characteristic value and the actual discharge duration of the plasma discharge current corresponding to the current discharge gun number, classifies the plasma discharge current type, and determines the plasma discharge current type of the current discharge gun number. The diagnostic raw data preprocessing module preprocesses the raw radiation calorimetry diagnostic data, segments the data according to the actual discharge duration, data duration and data purpose, calculates the time domain statistical characteristic values ​​of all channels, detects and statistically analyzes the waveform change types of the time series curves of all channels, and calculates the detector characteristic value. The correlation analysis data module calculates the correlation coefficient matrix and p-value matrix between multiple channels; The abnormal channel identification and compensation module identifies the abnormal radiation calorimetry diagnostic channel of the current discharge gun number multiple times according to the diagnostic expert rules. Compensation for abnormal channels in stages; The online calculation module for radiation intensity and total radiation power is used to calculate the multi-channel radiation intensity based on the radiation calorimetry diagnostic data after abnormal channel compensation, the characteristic parameters of the multi-detector array and its spatial location parameters, and to calculate the total radiation power online by performing area integral and volume integral through weighted summation. The visualization module is used to create a unified canvas framework and its subplots that integrate and display time-series curves of multiple signals. Subplots are used to draw time-series curves of multiple signals and play them automatically in the order of the subplots.

10. The online data analysis system for fusion radiation calorimetry diagnostics according to claim 9, characterized in that, The abnormal channel identification module includes a pre-identification module, a secondary identification module, and a tertiary identification module. The pre-identification module, based on all channel time-domain statistical characteristic values, time-series curve waveform change types, and detector characteristics detected and statistically analyzed by the diagnostic raw data preprocessing module, applies the diagnostic expert rules to pre-identify abnormal channels, thereby obtaining the pre-identified abnormal channel combination BadChan_GP1. The secondary identification module, based on the correlation analysis data obtained from the correlation analysis data module, the plasma discharge current type and discharge duration of the current discharge gun number, employs robust probability analysis and the IRQ quartile method, and applies the diagnostic expert rules to further identify abnormal channels. The normal channel is used to obtain the abnormal channel combination BadChan_GP2 identified in the second identification. The third identification module first uses a linear regression method based on signal amplitude to obtain multi-channel radiation intensity trend data, and then combines the plasma discharge current type and discharge duration to apply the diagnostic expert rules to identify abnormal channels three times, thereby obtaining the abnormal channel combination BadChan_GP3 identified in the third identification. The abnormal channel combinations BadChan_GP1, BadChan_GP2 and BadChan_GP3 are deduplicated and sorted to obtain the final abnormal channel combination BadChan_GP corresponding to the current discharge gun number.

Images