Method and system for dynamic modeling of the radiation field in a radioisotope laboratory environment

Abstract

The application discloses a radioactive separation laboratory environment radiation field dynamic modeling method and system, and belongs to the field of chemical process modeling. Step 1, the concentration distribution of a non-radioactive process tracer physically associated with a target radionuclide in a process fluid is obtained; step 2, based on the concentration distribution and the macroscopic residence time distribution of the material, the pre-judgment source term of the radionuclide is obtained through time translation and decay scaling; step 3, the radiation field observation waveform collected by a heterogeneous radiation sensitive field sensor network and independent of the pre-judgment source term is obtained; step 4, the theoretical radiation field change waveform calculated based on the pre-judgment source term is obtained; the application can obtain the concentration distribution of the non-radioactive tracer physically associated with the target radionuclide, eliminate the inherent bias of the source term from the source, and accurately depict the time-varying characteristics of the source term caused by the continuous migration of the material without making steady-state or quasi-static simplification assumptions on fluid turbulence, two-phase interface dynamic change and other processes.

Description

Technical Field

[0001] This invention relates to the field of chemical process modeling, and more specifically, to a method and system for dynamic modeling of environmental radiation fields in radioactive separation laboratories. Background Technology

[0002] Radioactive separation laboratories routinely conduct continuous flow material operations such as continuous dissolution, continuous countercurrent extraction, and column chromatography separation. During these operations, radionuclides migrate continuously with the fluid and are dynamically distributed in the two-phase system, forming a highly dynamic radiation field that changes in both space and time. Accurate dynamic modeling of this radiation field is the technical foundation for optimizing laboratory radiation protection schemes, controlling occupational doses for operators, and providing real-time early warning of abnormal radiation conditions.

[0003] Currently, most methods for dynamic modeling of radiation fields in radioactive sites employ piecewise steady-state or quasi-static assumptions. These methods discretize the intensity, spatial location, and geometry of the radioactive source, simulating the dynamic process by stitching together static snapshots of multiple steady-state conditions. However, these methods cannot realistically depict the time-varying characteristics of source terms, diffusion effects, and local concentration fluctuations caused by continuous material transfer. This results in significant lag in the calculation of the spatiotemporal distribution of the radiation field, leading to large spatiotemporal errors between the dose rate distribution predictions and the actual situation on site. Consequently, these methods fail to meet the real-time radiation safety management requirements of continuous flow operations. Existing modeling methods generally adopt a forward modeling path of "first determining the macroscopic source term distribution, then solving the radiation field." Under this path, the accuracy of radiation field calculations is highly dependent on the accuracy of the macroscopic source term distribution of radionuclides. However, obtaining macroscopic source terms requires making steady-state or quasi-static simplifications regarding complex fluid turbulence, dynamic changes at two-phase interfaces, and interphase distribution of nuclides. These simplifications inevitably introduce source term errors, which are ultimately directly transmitted to the radiation field calculation results, making it impossible to fundamentally eliminate inherent biases.

[0004] Meanwhile, in typical continuous flow operation scenarios such as continuous countercurrent extraction, existing technologies pursue refined hard input of source term parameters such as the position and intensity of the instantaneous concentration peak of radionuclides in the mixing chamber. However, microscopic disturbances such as fluid turbulence pulsation and phase interface fluctuations are directly amplified into violent fluctuations in the source term parameters. This leads to a significant decrease in the stability and deterioration of the prediction accuracy of the local radiation field model built based on the source term, causing the model to fall into an over-response to noise signals. It is unable to effectively capture the decisive changes in the radiation field brought about by the overall migration of materials, and it is difficult to adapt to the full-process dynamic modeling requirements of continuous flow operation in radioactive separation laboratories. Summary of the Invention

[0005] To address the problems existing in the prior art, the purpose of this invention is to provide a method and system for dynamic modeling of the environmental radiation field in a radioactive separation laboratory. This method can obtain the concentration distribution using non-radioactive tracers that are physically associated with the target radionuclide, without requiring steady-state or quasi-static simplifications for processes such as fluid turbulence and dynamic changes at two-phase interfaces. It eliminates the inherent bias of the source term from the source and accurately characterizes the time-varying characteristics of the source term caused by the continuous migration of materials.

[0006] To solve the above problems, the present invention adopts the following technical solution:

[0007] The first aspect is a method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory, which specifically includes the following steps:

[0008] Step 1: Obtain the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclide in the process fluid;

[0009] Step 2: Based on the concentration distribution and the macroscopic residence time distribution of the material, the predicted source terms of the radionuclide are obtained by time translation and decay scaling.

[0010] Step 3: Obtain the radiation field observation waveforms, independent of the predicted source terms, collected by the heterogeneous radiation sensitive field sensor network;

[0011] Step 4: Perform wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtain the spatial offset correction amount for the predicted source term by inverting the phase difference between the two.

[0012] Step 5: Based on the corrected source term after spatial offset correction, calculate and publish the radiation field using the radiation calculation engine;

[0013] Step 6: Compare the published radiation field with the observed waveforms of the radiation field in subsequent time slices. Based on the deviation of the radiation field change waveform relative to the theoretical radiation field change waveform in Step 4, dynamically calibrate the macroscopic dwell time distribution and the response function library of the radiation calculation engine.

[0014] Further, step 1 includes:

[0015] Step 11: Inject a non-radioactive process tracer with a predetermined timing pulse code into the process fluid, so that the tracer carries a physical marker corresponding to the injection time;

[0016] Step 12: Collect the mixed concentration response signal triggered by the tracer carrying the physical label at the predetermined detection node to obtain the original response waveform;

[0017] Step 13: According to the predetermined timing pulse code, the original response waveform is deconvolved and analyzed to obtain the tracer concentration distribution.

[0018] Further, step 2 includes:

[0019] Step 21: Perform flow regime decomposition on the tracer concentration distribution obtained in step 13 to obtain the multipath transport time spectrum from the injection point to each detection node;

[0020] Step 22: Based on the multipath transport time spectrum, dynamically decay-weight the activity of the initial radioactive material entering from the inlet to obtain the path activity components.

[0021] Step 23: The path activity components are remapped and synthesized according to the physical spatial topology of the process equipment to obtain the predicted source terms of the radionuclides.

[0022] Further, step 3 includes:

[0023] Step 31: Obtain the raw response signals collected by the heterogeneous sensor network in the key areas of the laboratory. The raw response signals constitute a hybrid spatiotemporal response matrix containing radiation source direction information and energy information.

[0024] Step 32: Spatial decoupling analysis is performed on the hybrid spatiotemporal response matrix. Based on the difference in response timing, the direct signal and the scattered signal are distinguished. The non-characteristic energy components in the scattered signal are removed by combining energy discrimination features to obtain the observed waveform of the radiation field.

[0025] Further, step 4 includes:

[0026] Step 41: Perform radiation calculation on the predicted source term obtained in Step 2 to generate the theoretical radiation field change waveform. At the same time, perform spatiotemporal feature decomposition on the radiation field observation waveform obtained in Step 3 to extract the wavefront feature unit set of the theoretical radiation field change waveform and the radiation field observation waveform.

[0027] Step 42: Perform spatiotemporal correlation matching between the theoretical wavefront feature set and the observed wavefront feature set to obtain feature set pairs originating from the same physical disturbance event;

[0028] Step 43: Calculate the spatiotemporal phase difference of each pair of feature units, and inversely derive the hydrodynamic disturbance parameter field based on the distribution characteristics of the spatiotemporal phase difference.

[0029] Furthermore, step 4 also includes:

[0030] Step 44: Based on the fluid dynamics disturbance parameter field, the spatial offset correction field is calculated using the pre-established source term flow field coupling transfer function.

[0031] Step 45: Superimpose the spatial offset correction field onto the predicted source term to generate the correction source term.

[0032] Further, step 5 includes:

[0033] Step 51: Perform spatial feature decomposition on the generated modified source term, and identify and segment the key constraint region for radiation field calculation in the source term based on the distribution pattern of the radionuclide concentration gradient.

[0034] Step 52: Based on the feature labels of each key constraint region, call the corresponding response operators from the pre-generated multi-resolution response operator library, perform parallel radiation calculations on each region, and synthesize the calculation results according to spatial topological relationships to generate the final radiation field distribution and publish it.

[0035] Further, step 6 includes:

[0036] Step 61: Compare the final radiation field distribution published in Step 52 with the radiation field observation waveforms obtained in Step 32 for each spatial point to obtain the composite time-domain deviation waveform.

[0037] Step 62: Perform frequency domain feature decomposition on the composite time-domain deviation waveform, and separate the low-frequency deviation component and the high-frequency deviation component based on the distribution differences in spatial frequency and time frequency.

[0038] Step 63: Invert the low-frequency deviation component into the multipath transport time spectrum extraction process in step 21, obtain the correction amount of the macroscopic residence time distribution parameter through deconvolution calculation, and update the multipath transport time spectrum.

[0039] Furthermore, step 6 also includes:

[0040] Step 64: Inject the high-frequency deviation component into the multi-resolution response operator library of step 52 in reverse. By perturbation injection and response comparison, locate and correct the local response function of the corresponding spatial region in the multi-resolution response operator library.

[0041] Step 65: Solidify and replace the corresponding parameters in steps 21 and 52 with the updated multipath transport time spectrum and the updated multi-resolution response operator library, respectively.

[0042] Secondly, the present invention also provides a dynamic modeling system for the environmental radiation field of a radioactive separation laboratory, comprising: a concentration acquisition module for acquiring the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclides in the process fluid;

[0043] The source term prediction module is used to obtain the predicted source terms of radionuclides based on the concentration distribution and the macroscopic residence time distribution of the material, through time translation and decay scaling.

[0044] The waveform acquisition module is used to acquire radiation field observation waveforms that are independent of the predicted source terms, collected by the heterogeneous radiation sensitive field sensor network.

[0045] The offset correction module is used to perform wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtain the spatial offset correction amount for the predicted source term by inverting the phase difference between the two.

[0046] The radiation publishing module is used to calculate and publish the radiation field based on the corrected source term after spatial offset correction using the radiation calculation engine.

[0047] The dynamic calibration module is used to compare the published radiation field with the observed waveforms of the radiation field in subsequent time slices. Based on the deviation of the comparison waveform relative to the theoretical radiation field change waveform, it dynamically calibrates the macroscopic dwell time distribution and the response function library of the radiation calculation engine.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] (1) This scheme uses non-radioactive tracers that are physically associated with the target radionuclides to obtain the concentration distribution. There is no need to make steady-state or quasi-static simplifications for processes such as fluid turbulence and dynamic changes at the two-phase interface. This eliminates the inherent bias of the source term from the source and accurately characterizes the time-varying characteristics of the source term caused by the continuous migration of materials.

[0050] (2) This scheme obtains the source term spatial offset correction by observing the measured radiation field waveform independent of the predicted source term, and then obtains it through wavefront matching and phase difference inversion. This avoids the source term parameter jitter caused by fluid micro-disturbance, solves the problem of the model over-responding to noise signals, and improves the model stability and prediction accuracy.

[0051] (3) This scheme identifies key constraint regions for radiation field calculation by source term spatial feature decomposition, and performs parallel radiation calculation by matching multi-resolution response operators. While ensuring the calculation accuracy of key regions, it significantly reduces the overall calculation volume and meets the needs of real-time radiation safety management of the entire continuous flow operation process.

[0052] (4) This scheme separates the deviation components from different sources by comparing the predicted radiation field with the subsequent measured waveform point by point, and calibrates the macroscopic residence time distribution and radiation response operator library respectively, forming a closed-loop iterative optimization mechanism that can adapt to the dynamic changes of process and environment in the long term and maintain the long-term running accuracy of the model. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0054] Figure 1 This is a flowchart of the dynamic modeling method for the environmental radiation field in a radioactive separation laboratory according to the present invention;

[0055] Figure 2 This is a data flow diagram between various modules in the dynamic modeling system for the environmental radiation field of the radioactive separation laboratory of this invention. Detailed Implementation

[0056] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0057] Example 1:

[0058] Please see Figure 1 A method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory, comprising the following steps:

[0059] Step 1: Obtain the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclide in the process fluid. The specific operation is as follows:

[0060] The physical co-occurrence relationship refers to the fact that the non-radioactive process tracer and the target radionuclide exhibit completely consistent hydrodynamic behavior during the continuous flow of the process fluid. In all operational stages involved in the process, such as material dissolution, continuous countercurrent extraction, and column chromatography separation, the distribution characteristics of the two in different phases are completely matched, and their diffusion and migration behaviors in turbulent environments are completely synchronized. This allows the spatiotemporal concentration distribution of the tracer in the process fluid to correspond without deviation to the distribution state of the target radionuclide in the same time and space. By measuring the concentration of the non-radioactive tracer, the interference of the environmental radiation field on the measurement process when directly measuring the radionuclide is avoided. At the same time, there is no need to make steady-state or quasi-static simplifications about the flow state of the process fluid and the interphase distribution of the nuclides. This ensures that the input information for subsequent source term calculations comes from the measured results of the process rather than theoretical simplifications.

[0061] Step 1 also includes the following steps:

[0062] Step 11: Inject a non-radioactive process tracer with a predetermined timing pulse code into the process fluid, so that the tracer carries a physical marker corresponding to the injection time. The specific operation is as follows:

[0063] The predetermined timing pulse code is a pre-set pulse injection sequence with a unique time correspondence. This sequence consists of multiple pulse units with fixed time intervals and fixed injection doses. The injection start time, duration, and injection dose of each pulse unit are pre-set and do not repeat, so that each injected tracer pulse can correspond to a unique injection time node, forming a time-bound physical marker. The selected non-radioactive process tracer must meet the physical association characteristics with the target radionuclide in the process system. In all the operation units involved in the process, the distribution behavior of the tracer and the target radionuclide between different phases is completely consistent, and the diffusion and migration behavior in turbulent flow is completely synchronized, ensuring that the transport process of the tracer in the fluid is completely synchronized with the transport process of the target radionuclide.

[0064] The tracer is injected at the inlet of the process fluid, which is also the starting point for the radioactive material to enter the process system. This ensures that the tracer and the target radionuclide enter the process system under exactly the same initial conditions and undergo completely identical process operations, avoiding transport path deviations caused by differences in injection location. The timing pulse coding design needs to match the macroscopic residence time range of the process system. The total coding duration covers the maximum residence time of the process system, and the time width of a single pulse is less than one-tenth of the minimum residence time of the process system. This ensures that tracers of different pulses do not completely overlap during transport, and that the physical marker corresponding to each pulse can be accurately identified by subsequent detection nodes.

[0065] Step 12: At the predetermined detection node, acquire the mixed concentration response signal triggered by the tracer carrying the physical label to obtain the original response waveform. The specific operation is as follows:

[0066] The predetermined detection nodes must cover all critical operational units of the process system, including the material dissolution unit, the mixing and clarification chambers of the countercurrent extraction unit, the inlet and outlet positions of the column chromatography separation unit, and the critical turning points of the process fluid delivery pipeline. Each detection node is equipped with a concentration detection element that is in direct contact with the process fluid. This detection element has a specific response to the selected non-radioactive process tracer and is not affected by other non-target components in the process fluid or by the environmental radiation field. The sampling frequency of the concentration detection element must be higher than ten times the reciprocal of the pulse time width in the predetermined timing pulse code to ensure that the response signal corresponding to each injection pulse can be completely acquired without sampling distortion.

[0067] During the acquisition process, each detection node is synchronously sampled according to a unified time axis. The sampling data of all detection nodes are bound to the corresponding sampling timestamps, forming time-series response data corresponding to the injection time markers. The mixed concentration response signal is an electrical signal output by the detection element when the tracer flows through the detection node with the process fluid, which is linearly related to the instantaneous concentration of the tracer. The continuous change sequence of this signal over time constitutes the original response waveform. The original response waveform completely records the time, concentration change amplitude and duration of the tracer carrying different injection time physical markers flowing through the corresponding detection node. It reflects the comprehensive result of all fluid dynamic behaviors such as fluid mixing, diffusion and interphase distribution that the tracer undergoes during the transport process from the injection point to the detection node.

[0068] Step 13: Based on the predetermined timing pulse code, perform deconvolution analysis on the original response waveform to obtain the tracer concentration distribution. The specific operation is as follows:

[0069] The injection signal corresponding to the predetermined timing pulse code serves as the system input function, the original response waveform serves as the system output function, and the fluid dynamics behavior of the process system from the injection point to the corresponding detection node constitutes the system transfer function. The deconvolution analysis process involves solving for the system transfer function based on the known system input and output functions, and then using the system transfer function to reconstruct the instantaneous concentration distribution of the tracer throughout the entire space of the process system. The transport, mixing, and diffusion processes of the tracer in the process system satisfy linear time-invariant characteristics under fixed process operating parameters. The response signals of tracer pulses injected at different times exhibit linear superposition, and the transport characteristics of the system do not change with time. Therefore, the relationship between the input, output, and system transfer function can be described by a convolution relationship, and the corresponding mathematical expression is:

[0070] ;

[0071] The formula is abstracted from the fundamental physical principles of linear time-invariant systems. The system's output response is equal to the convolution of the input signal and the system's unit impulse response. This relationship accurately describes the transport and superposition process of the tracer in the process system. The pulse signals injected by the tracer at different times in the process system will linearly superimpose with the transport process, ultimately forming a continuously acquireable response waveform at the detection node. This formula can be used to describe the intrinsic correlation between tracer injection and detection response. The tracer concentration value corresponding to the raw response signal collected by the detection node at time t, in kilograms per cubic meter; The injection dose rate of the tracer at time t is expressed in kilograms per second. is the unit impulse response function of the process system from the injection point to the corresponding detection node, which is also the residence time distribution density function, in units of seconds; t is the time variable, in units of seconds.

[0072] The deconvolution analysis process involves using a regularized numerical solution method to obtain the unit impulse response function from the known input and output functions. The regularization method used in the solution process is used to suppress the interference of high-frequency noise introduced during sampling on the solution results, ensuring that the obtained unit impulse response function can accurately reflect the true transport characteristics of the tracer in the process system. After obtaining the unit impulse response function corresponding to each detection node, the unit impulse response function of each detection node is spatially mapped and interpolated based on the physical spatial topology of the process system to obtain the tracer transport time distribution in the entire space of the process system. Then, combined with the total dose of injected tracer, the instantaneous concentration distribution of tracer in the entire space and time dimensions of the process system is calculated. The entire analysis process does not require any steady-state or quasi-static simplifications regarding the flow state and phase distribution behavior of the process fluid. It is based entirely on the measured input and output signals to obtain the true concentration distribution, ensuring that the obtained tracer concentration distribution completely matches the true flow state of the process fluid.

[0073] In a preferred embodiment of the present invention, step 2 is further included: based on the concentration distribution and the macroscopic residence time distribution of the material, the predicted source term of the radionuclide is obtained by time shifting and scaling according to decay. The specific operation is as follows:

[0074] Non-radioactive process tracers exhibit completely identical hydrodynamic behavior to the target radionuclides. The migration path, residence time, and spatial distribution characteristics of the tracer within the process system are perfectly synchronized with those of the target radionuclides. Therefore, the concentration distribution of the tracer directly corresponds to the spatial migration pattern of the target radionuclides within the process system. Time-shifting operations, based on the macroscopic residence time distribution of the material, map the concentration distribution of the tracer at different spatial locations to the time point at which the radionuclides arrive at the corresponding spatial location after entering the system from the inlet, thus realizing the correlation between the tracer concentration distribution and the radionuclide migration. Precise correspondence to time shift; the decay scaling operation is based on the inherent decay constant of the target radionuclide. The activity of the nuclide corresponding to the concentration distribution after time shift is corrected according to its residence time in the system, eliminating the activity decay caused by spontaneous decay of the radionuclide during migration. The concentration distribution measured by the tracer replaces the nuclide source term distribution obtained by steady-state or quasi-static assumptions in the existing technology. It does not require simplified assumptions about fluid turbulence, dynamic changes at the two-phase interface, and interphase distribution of nuclides, thus avoiding source term deviation caused by simplified assumptions from the source.

[0075] Step 2 also includes the following steps:

[0076] Step 21: Perform flow regime decomposition on the tracer concentration distribution obtained in Step 13 to obtain the multipath transport time spectrum from the injection point to each detection node. The specific operation is as follows:

[0077] During continuous flow within the process system, the tracer from the injection point to the same detection node will travel through multiple different transport paths due to factors such as equipment structure, two-phase flow, and turbulent mixing. The residence time, transport velocity, and mixing degree of the fluid corresponding to different paths are all different, and these differences will be reflected in the concentration distribution waveform of the tracer. The flow decomposition process is based on the unit impulse response function obtained by deconvolution analysis to perform multi-scale decomposition of the time-domain waveform of the tracer concentration distribution. The decomposition process adopts a modal decomposition method based on time-domain features to decompose the continuous concentration response waveform into multiple independent single-mode response components, each of which corresponds to an independent fluid transport path.

[0078] The temporal characteristics of each single-mode response component, including the peak occurrence time, full width at half maximum (FWHM), and peak shape distribution, correspond to the average residence time, residence time distribution width, and fluid mixing degree within the respective transport path. All decomposed single-mode response components are sorted according to the order of their peak occurrence times. The residence time interval and path transport proportion corresponding to each component are statistically analyzed, forming a multi-path transport time spectrum from the injection point to the corresponding detection node. This multi-path transport time spectrum completely covers all possible transport paths of the tracer from the injection point to the detection node, accurately records the residence time distribution and fluid transport proportion corresponding to each path, and fully restores the true flow characteristics of the fluid within the process system without requiring any simplified assumptions about the fluid flow state.

[0079] Step 22: Based on the multipath transport time spectrum, dynamically decay-weight the activity of the initial radioactive material entering from the inlet to obtain the path activity components. The specific operation is as follows:

[0080] Initial radioactivity refers to the activity of the target radionuclide in a unit volume of fluid at the moment the radioactive material enters the process system inlet. This value can be directly measured by the online radioactivity detection device at the inlet, and corresponds synchronously with the tracer injection process. In the multi-path transport time spectrum, each transport path corresponds to a specific residence time interval. This residence time interval is the time it takes for the target radionuclide to be transported from the system inlet to the detection node along the corresponding path, which is the decay time of the radionuclide within that path. The dynamic decay weighting process is to calculate the decay coefficient of the target radionuclide within the residence time of each independent transport path based on the residence time of that path, and then multiply the initial radioactivity activity by the decay coefficient and the fluid transport ratio of the corresponding path to obtain the path activity component corresponding to that path. The decay of radionuclides follows inherent nuclear physics laws, with activity changing exponentially over time. This relationship is determined by the spontaneous decay characteristics of radionuclides. The decay constant of a nuclide is only related to the type of nuclide and is not affected by external conditions such as the flow regime, temperature, and pressure of the process fluid. Therefore, the activity decay of a nuclide over a given residence time can be accurately calculated using a fixed exponential decay formula. For a radionuclide migrating along a fixed transport path, its activity at the end of the path is equal to the initial activity multiplied by the decay coefficient corresponding to the residence time, and then multiplied by the proportion of fluid transport along that path. The specific calculation is as follows:

[0081] ;

[0082] The activity of radionuclides decays exponentially over time. This law is unaffected by the external environment and is a fundamental law in nuclear physics. For radionuclides migrating along a specific transport path, their activity decay is determined only by the nuclide's intrinsic decay constant and its residence time within the path. At the same time, the total activity of the nuclide within the path is directly related to the proportion of fluid transport in that path. The higher the proportion of transport, the higher the proportion of the nuclide activity within the path. By combining these two physical relationships, we can obtain the formula for calculating the path activity component corresponding to a single transport path. This formula can accurately describe the remaining activity of a radionuclide after it has been transported along a specific path.

[0083] in the formula Let be the path activity component corresponding to the i-th transport path, in becquerels per cubic meter; The initial activity of radioactive material at the moment of entry into the process system, expressed in becquerels per cubic meter; λ represents the proportion of fluid transport corresponding to the i-th transport path, which is a dimensionless quantity. Its value is equal to the ratio of the integral area of ​​the single-mode response component corresponding to the path to the integral area of ​​the total response waveform; λ is the intrinsic decay constant of the target radionuclide, in units of per second, and its value is determined only by the type of target radionuclide. is the average residence time corresponding to the i-th transport path, in seconds, and its value is equal to the peak occurrence time of the single-mode response component corresponding to that path; e is the base of the natural logarithm; for each transport path included in the multi-path transport time spectrum, the path activity component is calculated according to the above calculation logic. The path activity components corresponding to all paths completely cover the activity distribution corresponding to all transport paths during the process of radionuclides being transported from the system inlet to the corresponding detection node, accurately reflecting the activity decay caused by radionuclides during migration, as well as the activity distribution caused by different transport paths.

[0084] Step 23: Remap and synthesize the path activity components according to the physical spatial topology of the process equipment to obtain the predicted source terms of the radionuclide. The specific operations are as follows:

[0085] The physical spatial topology of process equipment refers to the set of physical spatial parameters such as the three-dimensional spatial coordinates, connection relationships, internal cavity structure, and effective flow volume of all equipment, pipelines, and operating units within the process system. This set fully describes the spatial structural characteristics of the process system, clarifies the positions of different detection nodes in three-dimensional space, and the spatial connection relationships between detection nodes. The remapping process involves mapping the activity components of each path corresponding to each detection node to the spatial grid corresponding to the three-dimensional spatial topology of the process equipment, according to the spatial direction of the corresponding transport path. Each path activity component corresponds to all grid cells distributed along the transport path in the spatial grid. The activity value corresponding to each grid cell is determined by the path activity component, the relative position of the grid cell within the path, and the instantaneous concentration ratio of the tracer at that position.

[0086] The synthesis process involves linearly superimposing activity values ​​mapped from different transport paths and detection nodes within the same spatial grid cell to obtain the instantaneous activity concentration of the target radionuclide within that grid cell. This mapping and superposition calculation is performed on all spatial grid cells corresponding to the three-dimensional spatial topology of the process equipment, yielding the instantaneous activity concentration distribution of the target radionuclide across the entire three-dimensional space of the process system. This distribution represents the predicted source term of the radionuclide. The spatial resolution of the predicted source term directly corresponds to the accuracy of the spatial grid division, and the temporal resolution directly corresponds to the sampling frequency of the tracer concentration detection in step 1. This comprehensively covers the changes in radionuclide activity distribution across the entire spatial and temporal dimensions of the process system, eliminating the need for discretization and steady-state processing of the radionuclide spatial distribution. It accurately characterizes the time-varying characteristics of the source term resulting from continuous material transfer.

[0087] In a preferred embodiment of the present invention, step 3 is further included: acquiring the radiation field observation waveform, independent of the predicted source term, collected by the heterogeneous radiation sensitive field sensor network. The specific operation is as follows:

[0088] The heterogeneous radiation-sensitive field sensor network is deployed throughout the entire space of the radioactive separation laboratory. It can directly collect real radiation response signals in the laboratory environment. The acquisition process does not rely on any input parameters and calculation processes related to the construction of the predicted source term, such as tracer concentration distribution, material residence time distribution, and radionuclide activity calculation. The obtained radiation field observation waveform maintains complete data independence from the predicted source term. The radiation field observation waveform is a continuous time series of radiation dose rate changes with time and spatial location in the laboratory environment, which fully records the real-time changes of the environmental radiation field caused by the migration of radionuclides in the process system.

[0089] Step 3 also includes the following steps:

[0090] Step 31: Obtain the raw response signals collected by the heterogeneous sensor network in the key areas of the laboratory. The raw response signals constitute a hybrid spatiotemporal response matrix containing radiation source direction information and energy information. The specific operations are as follows:

[0091] The heterogeneous radiation sensitive field sensor network consists of multiple radiation detector units with different radiation detection characteristics. Different detector units have different matching in terms of energy response range, particle type discrimination ability, and time resolution to incident radiation, which can cover all radiation types and energy ranges released by the decay of target radionuclides. The key areas of the laboratory include the periphery of process equipment operation units, key nodes of material conveying pipelines, areas of routine personnel activities, weak points in the shielding structure, and areas for temporary storage of radioactive materials. The detector units are fixedly deployed in each key area according to a preset spatial topology. Each detector unit has a unique three-dimensional spatial coordinate and detection sensitive volume orientation, forming a distributed detection array with full spatial coverage.

[0092] All detector units are connected to a unified system synchronization clock and continuously and synchronously acquire data at the same sampling frequency. The raw response signal output by each detector unit is a continuous time-series electrical signal corresponding to the energy deposited by the incident radiation within the sensitive volume. Each signal sampling point is bound to a unique sampling timestamp, detector unit spatial coordinates, and detection type identifier. The raw response signals acquired synchronously by all detector units are arranged in a structured manner according to the detector spatial coordinate order and sampling time order, forming a two-dimensional hybrid spatiotemporal response matrix. The row dimension of the matrix corresponds to detector units at different spatial locations, and the column dimension corresponds to the synchronous time sampling nodes. Each element in the matrix is ​​the raw response amplitude of the corresponding detector unit at the corresponding sampling time. The signal amplitude distribution within the hybrid spatiotemporal response matrix contains information about the source direction of the incident radiation. The time and amplitude differences of radiation particles generated by the same radiation event reaching detector units at different spatial locations reflect the spatial location and propagation direction of the radiation source. The response differences of different types of detector units within the matrix contain energy information of the incident radiation. There is a fixed correspondence between the energy deposited by radiation particles of different energies in heterogeneous detector units and the response efficiency.

[0093] Step 32: Spatial decoupling analysis is performed on the hybrid spatiotemporal response matrix. Based on the difference in response time sequence, the direct signal and the scattered signal are distinguished. Then, non-characteristic energy components in the scattered signal are removed by combining energy discrimination features to obtain the observed waveform of the radiation field. The specific operation is as follows:

[0094] The physical basis of spatial decoupling analysis is that the propagation speed of radiation particles in air is constant. For radiation particles generated by the same radiation event, those that travel directly to the detector unit have the shortest propagation path and arrive earliest. Radiation particles scattered by media such as laboratory walls, process equipment, and shielding structures have longer propagation paths and arrive at the same detector unit later than the direct signal from the same event. These two types of signals exhibit distinguishable temporal differences in the time-domain response of the same detector unit. Direct radiation particles emitted by the same radiation source propagate in a straight line in uniform air at the speed of light. The time difference between their arrival at two detectors in different spatial locations is determined solely by the spatial position difference between the two detectors relative to the radiation source. This relationship is derived from the fundamental physical laws of uniform linear motion and accurately describes the correspondence between the arrival time of the direct radiation signal and the spatial position of the detector. In contrast, the propagation path of scattered signals is non-linear, and their arrival time difference does not conform to this fixed relationship. Therefore, the temporal and spatial correspondence established by this relationship can distinguish between direct and scattered signals. The formula for calculating this correspondence is:

[0095] ;

[0096] In the formula, Δt is the time difference between the arrival of the direct signal of the same radiation event at the first detector and the second detector, in seconds; is the three-dimensional spatial position vector of the radiation source point, in meters; This represents the three-dimensional spatial position vector of the first detector, in meters. denoted as , where is the three-dimensional spatial position vector of the second detector, in meters; c is the speed of light in air, in meters per second; the spatial decoupling analysis process involves fitting the wavefront arrival direction to the response signals of all detector units within the same time window in the hybrid spatiotemporal response matrix, based on the aforementioned temporal and spatial correspondence, selecting signal components that conform to the linear propagation temporal relationship and labeling them as direct signals, and labeling signal components that do not conform to the linear propagation temporal relationship and whose arrival time is later than the direct signals of the same event as scattered signals; the radiation particles released during the decay of the target radionuclide possess fixed characteristic energy values, which are determined only by the decay type and energy level structure of the nuclide and are not affected by external environment and process conditions.

[0097] After scattering, radiation particles lose energy, deviating from the inherent characteristic energy of the radionuclide and forming non-characteristic energy components. Simultaneously, background radiation, detector electronic noise, and interference from irrelevant radionuclides also contribute to non-characteristic energy components. By combining the energy response characteristics of different types of detector units in a heterogeneous sensor network, energy analysis is performed on the labeled scattered signal. By comparing the response amplitude and energy response efficiency curves of the same signal in different heterogeneous detector units, non-characteristic energy components whose signal energy does not match the characteristic energy value of the target radionuclide are identified and completely removed from the mixed spatiotemporal response matrix, thus completing the process. After filtering direct signals and removing non-feature energy components, the remaining effective response signals in the hybrid spatiotemporal response matrix are interpolated and reconstructed according to the three-dimensional spatial coordinates of the detector unit and the synchronous time axis to generate a time series sequence of radiation dose rate continuously changing with time throughout the entire laboratory space. This sequence is the radiation field observation waveform. The temporal resolution of the radiation field observation waveform is consistent with the sampling frequency of the detector unit, and the spatial resolution is matched with the deployment density of the sensor network. It completely restores the real dynamic change process of the radiation field in the laboratory environment, and all data comes from environmental measurements and does not contain any calculation assumptions related to the predicted source terms.

[0098] In a preferred embodiment of the present invention, step 4 is further included: performing wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtaining the spatial offset correction amount for the predicted source term by inverting the phase difference between the two. The specific operation is as follows:

[0099] The predicted source term is constructed based on the measured concentration distribution of the tracer and the decay law of the nuclide, avoiding the inherent bias caused by the steady-state or quasi-static assumptions in the existing technology. However, the instantaneous dynamic changes of the fluid in the process system will cause a spatiotemporal offset between the actual spatial distribution of radionuclides and the predicted source term. This offset will be directly transmitted to the radiation field calculation results. The change in the spatial distribution of radionuclides will directly correspond to the wavefront change of the environmental radiation field. The phase difference between the theoretical radiation field waveform and the observed radiation field waveform is essentially the difference in the spatiotemporal response of the radiation field caused by the spatial offset of the nuclide source term. By matching the same source wavefront and inverting the phase difference, the spatial offset correction of the source term can be directly extracted from the measured radiation field data. There is no need to perform fine hard input processing on the source term parameters, avoiding the problem of model stability degradation caused by micro-disturbances of the fluid. At the same time, the source term is corrected by measured data independent of the predicted source term, further reducing the deviation between the source term and the actual nuclide distribution.

[0100] Step 4 also includes the following steps:

[0101] Step 41: Perform radiation calculations on the predicted source terms obtained in Step 2 to generate the theoretical radiation field change waveform. Simultaneously, perform spatiotemporal feature decomposition on the radiation field observation waveform obtained in Step 3 to extract the wavefront feature unit set of the theoretical radiation field change waveform and the radiation field observation waveform. The specific operations are as follows:

[0102] Based on the predicted source term and the full three-dimensional spatial distribution of nuclide activity concentration, a continuous time series of radiation dose rate changes over time at each spatial location in the laboratory is calculated according to the spatial coverage and synchronous time axis that are completely consistent with the observed radiation field waveform. This series is the theoretical radiation field change waveform. The temporal and spatial resolutions of the theoretical radiation field change waveform are completely matched with the corresponding parameters of the observed radiation field waveform, ensuring that the two sets of waveforms are comparable in spatiotemporal dimensions. The spatiotemporal feature decomposition process is performed synchronously for the theoretical radiation field change waveform and the observed radiation field waveform, using completely consistent decomposition rules and judgment thresholds. The decomposition process identifies the spatiotemporal regions where the dose rate undergoes continuous step changes by calculating the gradient changes of the radiation dose rate in the time and spatial domains. Each continuous gradient change region constitutes an independent wavefront feature unit. All wavefront feature units are structurally integrated according to spatiotemporal coordinates to form the theoretical wavefront feature unit set corresponding to the theoretical radiation field change waveform and the observed wavefront feature unit set corresponding to the observed radiation field waveform. Each wavefront feature unit carries corresponding spatiotemporal attribute parameters, including wavefront start time, peak time, spatial distribution range, and dose rate change amplitude.

[0103] Step 42: Perform spatiotemporal correlation matching between the theoretical wavefront feature set and the observed wavefront feature set to obtain feature set pairs originating from the same physical disturbance event. The specific operation is as follows:

[0104] A physical disturbance event refers to a change in the local source term activity caused by radionuclides reaching a specific process unit during their migration within the process system. This change synchronously generates corresponding wavefront responses in both the theoretical and measured radiation fields. The two sets of responses share a common physical origin and differ only in phase in the spatiotemporal dimension. The matching process first screens all wavefront feature units in the two feature unit sets, selecting those with dose rate change amplitude differences within a preset range, spatial distribution overlap that meets requirements, and adjacent occurrence time windows. This forms a candidate set of feature unit pairs. For each pair of feature units in the candidate set, the spatiotemporal correlation is calculated. The correlation calculation is based on the wavefront's temporal rising edge slope, peak half-width, and spatial distribution topology. Feature unit pairs with a correlation exceeding a preset threshold are determined to be valid matching pairs originating from the same physical disturbance event. The entire matching process is based on a unified three-dimensional spatial coordinate system and a system-synchronized time axis in the laboratory, ensuring that the spatiotemporal coordinates of all feature units have a unified comparison benchmark. At the same time, false feature units caused by environmental background fluctuations and detection noise are eliminated, ensuring that the input data for subsequent phase difference calculations are all valid signals from the same source.

[0105] Step 43: Calculate the spatiotemporal phase difference for each pair of feature elements, and inversely derive the hydrodynamic disturbance parameter field based on the distribution characteristics of the spatiotemporal phase difference. The specific operation is as follows:

[0106] For each pair of validly matched feature units, the normalized time-domain waveforms corresponding to the theoretical and observed wavefronts are extracted. The time phase difference between the two waveforms is calculated through cross-correlation. The time phase difference corresponds to the time offset of the two waveforms reaching their peak values. When two time-domain waveforms from the same source have only a time-dimensional shift, their cross-correlation function will reach its maximum value at the position where the shift equals the true time phase difference. This principle is the basic principle of time delay estimation in signal processing, and it is not affected by signal amplitude attenuation, thus accurately extracting the time offset of two waveforms from the same source. The theoretical and observed wavefronts caused by the same physical disturbance event have completely identical waveform shapes, with only a time shift. The time phase difference between the two can be directly calculated through the peak position of the cross-correlation function. The specific calculation is as follows:

[0107] ;

[0108] in the formula τ is the time phase difference between the theoretical wavefront waveform and the observed wavefront waveform, in seconds; τ is the time shift, in seconds. For time integration, the unit is seconds; The normalized time-domain waveform corresponding to the theoretical wavefront characteristic unit is a dimensionless quantity. To observe the normalized time-domain waveform corresponding to the wavefront characteristic unit, it is a dimensionless quantity.

[0109] The spatial distribution centers of two sets of wavefronts are extracted, and their spatial position offset vectors are calculated. This vector is the spatial phase difference. The spatiotemporal phase differences of all feature unit pairs are distributed throughout the entire space of the process system according to their corresponding spatial positions, forming a set of spatiotemporal phase differences with spatial distribution characteristics. The fluid dynamics disturbance parameter field refers to the deviation distribution between the actual fluid transport parameters and the nominal parameters used when constructing the predicted source terms within the entire space of the process system. This deviation is the fundamental cause of the spatiotemporal offset of the nuclide source terms. The inversion process is based on the spatial distribution characteristics of the spatiotemporal phase difference. Combining the physical spatial topology of the process system and the basic laws of fluid transport, a quantitative correlation between the spatiotemporal phase difference and the fluid transport parameter deviation is established. The fluid transport parameter deviation distribution within the entire space is obtained by least squares fitting. This distribution is the fluid dynamics disturbance parameter field.

[0110] Step 44: Based on the fluid dynamics disturbance parameter field, the spatial offset correction field is calculated using the pre-established source term flow field coupling transfer function. The specific operation is as follows:

[0111] The source term flow field coupling transfer function is a pre-established quantitative function used to describe the correspondence between fluid dynamics parameter deviations and the spatial distribution shift of radionuclides. This function is constructed based on the physical spatial structure of the process system, fluid transport characteristics, and the physical association between the tracer and the target radionuclide. During the process system commissioning phase, tracer transport experiments under multiple operating conditions were conducted to measure the tracer concentration distribution shift corresponding to changes in different fluid parameters. The parameters of the function were determined through data fitting to ensure that the function can accurately reflect the quantitative correlation between fluid disturbance and source term spatial shift. In the calculation process, the fluid dynamics disturbance parameter field is input into the source term flow field coupling transfer function. For each grid cell in the full three-dimensional spatial grid of the process system, the corresponding source term spatial shift vector is calculated. The spatial shift vectors of all grid cells are integrated to form a spatial shift correction field. The spatial resolution of the spatial shift correction field is completely consistent with the spatial grid resolution of the predicted source term. The shift vector of each grid cell contains shift components in three directions in three-dimensional space, which can completely describe the positional shift of the source term in the entire spatial range.

[0112] Step 45: Superimpose the spatial offset correction field onto the predicted source term to generate the correction source term. The specific operation is as follows:

[0113] The superposition process is performed on each grid cell in the three-dimensional spatial grid corresponding to the predicted source term. Based on the spatial offset correction vector corresponding to the grid cell, the nuclide activity concentration value in the grid cell is mapped to the offset spatial grid position, completing the spatial translation correction of the source term distribution. When the activity concentration values ​​of multiple grid cells are mapped to the same target grid cell, the final activity concentration value of the target grid cell is calculated by linear superposition. For blank grid cells after mapping, the activity concentration values ​​of adjacent grid cells are used for linear interpolation to fill them, ensuring that the corrected source term distribution has continuity throughout the entire space. The spatial grid structure, temporal resolution, and activity concentration measurement unit of the corrected source term are completely consistent with the predicted source term. Its spatial distribution is corrected by inversion of measured radiation field data, eliminating the spatial position deviation of the source term caused by hydrodynamic disturbances. It can accurately reflect the true spatial distribution of radionuclides in the process system and avoid source term parameter jitter caused by microscopic fluid disturbances.

[0114] In a preferred embodiment of the present invention, step 5 is further included: based on the corrected source term after spatial offset correction, the radiation field is calculated and published using a radiation calculation engine. The specific operation is as follows:

[0115] The corrected source terms fully reflect the true spatiotemporal distribution of radionuclides within the process system, eliminating spatial offset deviations caused by fluid dynamic disturbances and providing a reliable input basis for radiation field calculations. The radiation calculation engine, built upon the fundamental physical laws of radiation transport, can transform the three-dimensional spatial distribution of source term activity concentrations into a radiation dose rate distribution across the entire laboratory space. Through regional feature decomposition and differentiated calculations, it improves computational efficiency while ensuring the accuracy of radiation field calculations, meeting the time response requirements for real-time radiation safety management throughout the continuous flow operation of the radioactive separation laboratory. The final radiation field distribution generated after calculation is simultaneously published to the laboratory radiation safety management system, providing data support for occupational dose control of operators and early warning of abnormal radiation conditions. It also serves as a comparison benchmark for subsequent dynamic calibration, used for iterative optimization of the macroscopic residence time distribution and the radiation calculation engine's response function library.

[0116] Step 5 also includes the following steps:

[0117] Step 51: Perform spatial feature decomposition on the generated modified source term. Based on the distribution pattern of the radionuclide concentration gradient, identify and segment the key constraint region for radiation field calculation in the source term. The specific operation is as follows:

[0118] The corrected source terms are stored in the form of a three-dimensional spatial grid. Each grid cell corresponds to a unique three-dimensional spatial coordinate and a radionuclide activity concentration value. The first step of spatial feature decomposition is to calculate the radionuclide concentration gradient of each grid cell in the entire space. The concentration gradient value is the ratio of the activity concentration difference between adjacent grid cells to the spatial step size of the grid cell. After calculating the gradient components in the three orthogonal directions in three-dimensional space, the modulus is taken to obtain the concentration gradient scalar of the grid cell. The distribution pattern of the concentration gradient fully reflects the degree of change of radionuclide activity concentration in the spatial range. The region with drastic change in activity concentration has a higher concentration gradient modulus and a greater weight in influencing the distribution of the surrounding spatial radiation field. It is the key region that determines the accuracy of the radiation field calculation.

[0119] The identification process traverses all grid cells in the entire space. Based on the distribution characteristics of the concentration gradient magnitude, continuously distributed grid cells are clustered to form continuous spatial regions with similar concentration gradient characteristics. The segmentation process, based on the clustering results and combined with the physical spatial topology of the process equipment, delineates the spatial boundaries of each region to ensure that the region boundaries match the physical boundaries of the process equipment's cavities, pipelines, and operating units, thus avoiding disruption of the physical spatial continuity of the source term distribution during segmentation. After segmentation, each independent spatial region is assigned a corresponding feature label, which includes attribute information such as the region's concentration gradient range, mean activity concentration, and spatial volume percentage.

[0120] Step 52: Based on the feature labels of each key constraint region, call the corresponding response operators from the pre-generated multi-resolution response operator library to perform parallel radiation calculations on each region, and synthesize the calculation results according to spatial topological relationships to generate and publish the final radiation field distribution. The specific operations are as follows:

[0121] The multi-resolution response operator library is a pre-built set of radiation response coefficients. The construction process is based on the decay radiation type and characteristic energy parameters of the target radionuclide, as well as the spatial medium distribution, shielding structure parameters, and equipment material properties of the radioisotope separation laboratory, and is completed through numerical calculations of radiation transport. The operator library contains multiple sets of response operators with different resolutions. Each set of response operators corresponds to specific source term feature labels, calculation accuracy levels, and calculation efficiency levels. The core of the operator is a pre-calculated response coefficient matrix from source term activity concentration to spatial dose rate, which can be directly used for rapid calculation of the radiation field in the corresponding region. The basic physical principle of radiation field calculation is the linear superposition principle of radiation transport. The radiation dose rate at any point in space is linearly superimposed by the radiation contributions of all radionuclide voxels distributed in three-dimensional space. This relationship can be expressed as:

[0122] ;

[0123] The transport and energy deposition processes of radiated particles in a medium satisfy the linear superposition property. The total dose rate contribution of multiple independent radioactive sources to a point in space is equal to the sum of the dose rate contributions of each individual radioactive source to that point. For a given nuclide type, a fixed space medium, and a shielding structure, the dose rate contribution of a source element per unit activity to a fixed point in space is a constant value. This constant value is the response coefficient, which can be pre-calculated and solidified into a response operator. This eliminates the need to repeatedly solve the complete radiation transport equation for each radiation field calculation, significantly improving computational efficiency. In the formula... Spatial location The radiation dose rate at the location, measured in gray per second; This is the three-dimensional spatial position vector of the dose calculation point, in meters; is the three-dimensional spatial position vector of the source element, in meters; For source item element The activity concentration of radionuclides at a given location, expressed in becquerels per cubic meter; For source item element For calculation points The radiation response coefficient, in units of gray cubic meters per becquerel second, is predetermined by the radionuclide radiation characteristics, spatial medium distribution, and shielding structure parameters, and is stored in the multi-resolution response operator library; V is the total spatial volume region of the source term distribution, in cubic meters; dV' is the volume element of the source term volume element, in cubic meters.

[0124] During operator invocation, the most suitable response operator is matched from the operator library based on the feature labels of each key constraint region. Regions with high concentration gradients and high activity concentrations are matched with high-resolution and high-precision response operators, while regions with low concentration gradients and low activity concentrations are matched with low-resolution and high-efficiency response operators. This ensures the accuracy of calculations in key regions while reducing the overall computational load. The parallel computing process allocates the radiation field calculation task of each key constraint region to an independent computing thread. The calculation processes of each region are executed synchronously without interference. The boundary coupling conditions between regions are pre-set during the construction of response operators, eliminating the need for iterative interaction during the calculation process.

[0125] After the calculations are completed, the radiation dose rate calculation results of each region are stitched together in a unified three-dimensional spatial coordinate system according to the physical spatial topology of the process equipment. The boundaries of adjacent regions are processed with linear smooth transition to ensure the continuity of the final radiation field distribution in the entire space. After the synthesis is completed, a three-dimensional radiation dose rate distribution of the entire laboratory space is generated, which is the final radiation field distribution. The spatial resolution of this distribution is consistent with the spatial grid resolution of the correction source term, and the temporal resolution is matched with the system's synchronous sampling frequency. The final radiation field distribution is released synchronously after completion. On the one hand, it is output to the laboratory radiation safety management system to provide data support for the optimization of radiation protection schemes, occupational dose control of operators, and real-time early warning of abnormal radiation conditions. On the other hand, it is output to the dynamic calibration stage as benchmark data for comparison with the observed waveform of the measured radiation field.

[0126] In a preferred embodiment of the present invention, step 6 is further included: comparing the published radiation field with the observed radiation field waveforms of subsequent time slices; dynamically calibrating the macroscopic dwell time distribution and the response function library of the radiation calculation engine based on the deviation generated by the comparison relative to the theoretical radiation field change waveform in step 4; the specific operations are as follows:

[0127] The radiation field distribution published in the preceding steps is compared with the measured radiation field waveforms collected in subsequent time slices. The deviation data between the two is extracted. The difference between this deviation data and the theoretical radiation field change waveform in step 4 can distinguish between temporary deviations caused by single source term spatial offsets and systematic deviations caused by model basic parameters. Based on the different sources of deviation, the macroscopic residence time distribution on which the predicted source term is based and the response function library used by the radiation calculation engine are dynamically calibrated to eliminate the inherent model deviations caused by factors such as slow changes in the flow state of the process system, local adjustments in the laboratory radiation transport environment, and long-term drift of equipment parameters. This enables the modeling method to continuously adapt to the dynamic changes of continuous flow operation in the radioactive separation laboratory, maintain the accuracy and response speed of radiation field calculation in the long term, and meet the long-term operational requirements of real-time radiation safety management throughout the entire process.

[0128] Step 6 also includes the following steps:

[0129] Step 61: Compare the final radiation field distribution published in Step 52 with the radiation field observation waveforms obtained in Step 32 for subsequent time slices, point by point, to obtain the composite time-domain bias waveform. The specific operation is as follows:

[0130] Subsequent time slices are adjacent sampling time slices that maintain strict temporal synchronization with the initial time slices processed in steps 4 and 5. The radiation field observation waveform and the final radiation field distribution within this time slice have completely consistent temporal resolution, spatial coverage, and three-dimensional spatial coordinate system, ensuring complete consistency of the spatiotemporal reference for the comparison process. The point-by-point spatial comparison process traverses all corresponding spatial grid cells within the unified coordinate system. For each grid cell, the predicted dose rate value corresponding to the final radiation field distribution at the same sampling time is extracted, along with the measured dose rate value corresponding to the radiation field observation waveform. The difference between the two is calculated, and the continuous change sequence of this difference with the sampling time constitutes the temporal deviation waveform of the corresponding spatial grid cell. The temporal deviation waveforms of all spatial grid cells are structurally integrated according to the three-dimensional spatial topology to form a composite temporal deviation waveform covering the entire laboratory space. The composite temporal deviation waveform completely contains all deviation information between the predicted and measured values ​​of the radiation field, covering the systematic deviation of the source term transport process, the local deviation of radiation transport calculation, and the random deviation caused by environmental background fluctuations.

[0131] Step 62: Perform frequency domain feature decomposition on the composite time-domain deviation waveform. Based on the differences in spatial and temporal frequency distributions, separate the low-frequency deviation component and the high-frequency deviation component. The specific operation is as follows:

[0132] Deviation signals from different physical origins exhibit distinguishable distribution characteristics in both spatial and temporal frequency dimensions. Deviations originating from macroscopic residence time distribution mismatches are caused by long-period factors such as slow changes in the flow regime of the process system and continuous drift of material properties. In the time domain, they manifest as slow, continuous changes across the entire spatial range, with both spatial and temporal frequencies in the lower range. Deviations originating from radiation calculation response function mismatches are caused by local factors such as changes in the local shielding structure of the laboratory, adjustments in the scattering characteristics of the medium, and changes in the spatial position of equipment. In the spatial domain, they manifest as rapid changes in local areas, and in the temporal domain, they manifest as short-period fluctuations, with either spatial or temporal frequencies in the higher range.

[0133] The frequency domain eigenvalue decomposition process first performs a two-dimensional Fourier transform on the composite time-domain deviation waveform, converting the deviation signal in the time-spatial dimension to a two-dimensional frequency domain space composed of spatial frequency and time frequency. Based on a pre-set frequency threshold, the two-dimensional frequency domain space is divided into low-frequency and high-frequency intervals. The frequency threshold is set according to the time constant of the flow state change in the process system and the characteristic scale of the spatial change of the laboratory radiation field, ensuring the effective separation of the two types of deviation signals. Then, an inverse two-dimensional Fourier transform is performed on the frequency domain signals in the low-frequency and high-frequency intervals, converting the signals back to the time-spatial dimension, obtaining the low-frequency deviation component and high-frequency deviation component with full spatial coverage. The low-frequency deviation component corresponds to the systematic deviation of the source term transport process, and the high-frequency deviation component corresponds to the local deviation of the radiation transport calculation.

[0134] Step 63: The low-frequency deviation component is injected in reverse into the multipath transport time spectrum extraction process in step 21. The correction amount for the macroscopic residence time distribution parameters is obtained through deconvolution calculation, and the multipath transport time spectrum is updated. The specific operation is as follows:

[0135] The physical origin of the low-frequency deviation component is that the macroscopic residence time distribution used when constructing the predictive source term is systematically mismatched with the actual material transport residence time distribution in the current process system. This mismatch leads to an overall temporal shift in the predicted source term of radionuclides, which in turn causes a long-period continuous deviation in the radiation field prediction results across the entire space. The reverse injection process first converts the low-frequency deviation component into the corresponding temporal deviation of the activity concentration of the radionuclide source term based on the principle of linear superposition of radiation transport. This conversion process is achieved through the inverse operation of the radiation response operator, ensuring that the temporal deviation and the low-frequency deviation component have a strict linear correspondence.

[0136] The temporal deviation of the source term activity concentration obtained by conversion is injected as a correction input into the multipath transport time spectrum extraction process in step 21. Together with the tracer concentration distribution and predetermined temporal pulse code obtained in step 13, it serves as the input parameters for deconvolution calculation. The physical basis of deconvolution calculation is the transport response relationship of a linear time-invariant system. The tracer injection signal, the system's unit impulse response function, and the original response waveform of the detection node satisfy a fixed convolution operation relationship. The temporal deviation of the source term activity concentration directly corresponds to the systematic shift of the system's unit impulse response function, and the macroscopic residence time distribution parameter is the integral representation of the unit impulse response function. Through regularized deconvolution solution, the correction amount of the unit impulse response function can be calculated from the input parameters. This correction amount is the correction amount of the macroscopic residence time distribution parameter. Based on the obtained correction amount, the original multipath transport time spectrum in step 21 is updated, and the average residence time and fluid transport ratio corresponding to each transport path are adjusted to obtain the corrected multipath transport time spectrum, eliminating the systematic temporal deviation in the source term construction process.

[0137] Step 64: Inject the high-frequency bias component back into the multi-resolution response operator library of step 52. By comparing the perturbation injection with the response, locate and correct the local response function of the corresponding spatial region in the multi-resolution response operator library. The specific operation is as follows:

[0138] The physical origin of the high-frequency deviation component lies in the local response function stored in the multi-resolution response operator library, which exhibits a local mismatch with the actual radiation transport environment in the laboratory. This mismatch leads to localized, short-period deviations in the calculated radiation field results for the corresponding spatial region. The reverse injection process first spatially locates the high-frequency deviation component by traversing all grid cells in the entire space and selecting continuous spatial regions where the amplitude of the high-frequency deviation component exceeds a preset threshold. This region corresponds to the spatial region where the response function mismatch occurs. The preset threshold is determined based on the detector's measurement uncertainty and the nominal accuracy of the radiation field calculation to avoid false localization caused by random noise. For the localized spatial region, a perturbation injection and response comparison method is used to correct the response function. The perturbation injection process involves... The local response function corresponding to this region is injected with multiple sets of preset small parameter perturbations. Each perturbation corresponds to a unique parameter change direction and amplitude. Based on the perturbed response function, the radiation field of the corresponding region is calculated, resulting in multiple perturbed radiation field prediction sequences. These multiple perturbed radiation field prediction sequences are compared one by one with the observed radiation field waveforms of the corresponding region to obtain the deviation change corresponding to each perturbation. Based on the quantitative correspondence between the deviation change and the perturbation parameters, the optimal correction amount of the local response function is obtained through least squares fitting. Using the obtained optimal correction amount, the local response function of the corresponding spatial region in the multi-resolution response operator library is corrected, the response coefficient matrix corresponding to the response function is updated, and local bias in the radiation transport calculation process is eliminated.

[0139] Step 65: The updated multipath transport time spectrum and the updated multiresolution response operator library are solidified and used to replace the corresponding parameters in steps 21 and 52, respectively. The specific operations are as follows:

[0140] The solidification process involves writing the updated multipath transport time spectrum parameters and multi-resolution response operator library parameters into the system's non-volatile storage medium, forming a fixed parameter file that can be called for a long time. This replaces the original corresponding parameter file. The parameter replacement process follows a strict system synchronization sequence. After the full-process modeling calculation of the current time slice is completed, the full parameter replacement is completed before the modeling calculation of the next time slice starts, avoiding calculation timing misalignment and data chaos caused by the parameter replacement process. After the parameter replacement is completed, when the multipath transport time spectrum extraction process in step 21 is executed, the updated macroscopic residence time distribution parameters will be directly called. When the radiation field parallel calculation process in step 52 is executed, the updated multi-resolution response operator library will be directly called. The entire calibration and parameter update process is executed continuously and dynamically. The radiation field observation data of each subsequent time slice will be used for iterative optimization of the model parameters, so that the entire modeling method can continuously adapt to the dynamic changes of the process system and laboratory environment, maintain the accuracy and stability of the dynamic modeling of the radiation field in the long term, and meet the real-time radiation safety control requirements of the entire continuous flow operation process in the radioactive separation laboratory.

[0141] Example 2:

[0142] Please see Figure 2 Based on Example 1, this example provides a dynamic modeling system for the environmental radiation field of a radioactive separation laboratory, including:

[0143] The concentration acquisition module is used to acquire the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclide in the process fluid;

[0144] The source term prediction module is used to obtain the predicted source terms of radionuclides based on the concentration distribution and the macroscopic residence time distribution of the material, through time translation and decay scaling.

[0145] The waveform acquisition module is used to acquire radiation field observation waveforms that are independent of the predicted source terms, collected by the heterogeneous radiation sensitive field sensor network.

[0146] The offset correction module is used to perform wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtain the spatial offset correction amount for the predicted source term by inverting the phase difference between the two.

[0147] The radiation publishing module is used to calculate and publish the radiation field based on the corrected source term after spatial offset correction using the radiation calculation engine.

[0148] The dynamic calibration module is used to compare the published radiation field with the observed waveforms of the radiation field in subsequent time slices. Based on the deviation of the comparison waveform relative to the theoretical radiation field change waveform, it dynamically calibrates the macroscopic dwell time distribution and the response function library of the radiation calculation engine.

[0149] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory, characterized in that, include: Step 1, obtaining the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclide in the process fluid, also includes: Step 11: Inject a non-radioactive process tracer with a predetermined timing pulse code into the process fluid, so that the tracer carries a physical marker corresponding to the injection time; Step 12: Collect the mixed concentration response signal triggered by the tracer carrying the physical label at the predetermined detection node to obtain the original response waveform; Step 13: According to the predetermined timing pulse code, the original response waveform is deconvolved and analyzed to obtain the tracer concentration distribution; Step 2, based on the concentration distribution and the macroscopic residence time distribution of the material, obtains the predicted source term of the radionuclide through time shifting and decay scaling, and also includes: Step 21: Perform flow regime decomposition on the tracer concentration distribution obtained in step 13 to obtain the multipath transport time spectrum from the injection point to each detection node; Step 22: Based on the multipath transport time spectrum, dynamically decay-weight the activity of the initial radioactive material entering from the inlet to obtain the path activity components. Step 23: Remap and synthesize the path activity components according to the physical spatial topology of the process equipment to obtain the predicted source terms of the radionuclide. Step 3: Obtain the radiation field observation waveforms, independent of the predicted source terms, collected by the heterogeneous radiation sensitive field sensor network; Step 4: Perform wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtain the spatial offset correction amount for the predicted source term by inverting the phase difference between the two. Step 5: Based on the corrected source term after spatial offset correction, calculate and publish the radiation field using the radiation calculation engine; Step 6: Compare the published radiation field with the observed waveforms of the radiation field in subsequent time slices. Based on the deviation of the radiation field change waveform relative to the theoretical radiation field change waveform in Step 4, dynamically calibrate the macroscopic dwell time distribution and the response function library of the radiation calculation engine.

2. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 1, characterized in that, Step 3 includes: Step 31: Obtain the raw response signals collected by the heterogeneous sensor network in the key areas of the laboratory. The raw response signals constitute a hybrid spatiotemporal response matrix containing radiation source direction information and energy information. Step 32: Spatial decoupling analysis is performed on the hybrid spatiotemporal response matrix. Based on the difference in response timing, the direct signal and the scattered signal are distinguished. The non-characteristic energy components in the scattered signal are removed by combining energy discrimination features to obtain the observed waveform of the radiation field.

3. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 2, characterized in that, Step 4 includes: Step 41: Perform radiation calculation on the predicted source term obtained in Step 2 to generate the theoretical radiation field change waveform. At the same time, perform spatiotemporal feature decomposition on the radiation field observation waveform obtained in Step 3 to extract the wavefront feature unit set of the theoretical radiation field change waveform and the radiation field observation waveform. Step 42: Perform spatiotemporal correlation matching between the theoretical wavefront feature set and the observed wavefront feature set to obtain feature set pairs originating from the same physical disturbance event; Step 43: Calculate the spatiotemporal phase difference of each pair of feature units, and inversely derive the hydrodynamic disturbance parameter field based on the distribution characteristics of the spatiotemporal phase difference.

4. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 3, characterized in that, Step 4 also includes: Step 44: Based on the fluid dynamics disturbance parameter field, the spatial offset correction field is calculated using the pre-established source term flow field coupling transfer function. Step 45: Superimpose the spatial offset correction field onto the predicted source term to generate the correction source term.

5. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 4, characterized in that, Step 5 includes: Step 51: Perform spatial feature decomposition on the generated modified source term, and identify and segment the key constraint region for radiation field calculation in the source term based on the distribution pattern of the radionuclide concentration gradient. Step 52: Based on the feature labels of each key constraint region, call the corresponding response operators from the pre-generated multi-resolution response operator library, perform parallel radiation calculations on each region, and synthesize the calculation results according to spatial topological relationships to generate the final radiation field distribution and publish it.

6. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 5, characterized in that, Step 6 includes: Step 61: Compare the final radiation field distribution published in Step 52 with the radiation field observation waveforms obtained in Step 32 for each spatial point to obtain the composite time-domain deviation waveform. Step 62: Perform frequency domain feature decomposition on the composite time-domain deviation waveform, and separate the low-frequency deviation component and the high-frequency deviation component based on the distribution differences in spatial frequency and time frequency. Step 63: Invert the low-frequency deviation component into the multipath transport time spectrum extraction process in step 21, obtain the correction amount of the macroscopic residence time distribution parameter through deconvolution calculation, and update the multipath transport time spectrum.

7. The method for dynamic modeling of the environmental radiation field in a radioactive separation laboratory according to claim 6, characterized in that, Step 6 also includes: Step 64: Inject the high-frequency deviation component into the multi-resolution response operator library of step 52 in reverse. By perturbation injection and response comparison, locate and correct the local response function of the corresponding spatial region in the multi-resolution response operator library. Step 65: Solidify and replace the corresponding parameters in steps 21 and 52 with the updated multipath transport time spectrum and the updated multi-resolution response operator library, respectively.

8. A dynamic modeling system for the environmental radiation field of a radioactive separation laboratory, applied to the dynamic modeling method for the environmental radiation field of a radioactive separation laboratory as described in any one of claims 1-7, characterized in that, include: The concentration acquisition module is used to acquire the concentration distribution of non-radioactive process tracers that are physically associated with the target radionuclide in the process fluid; The source term prediction module is used to obtain the predicted source terms of radionuclides based on the concentration distribution and the macroscopic residence time distribution of the material, through time translation and decay scaling. The waveform acquisition module is used to acquire radiation field observation waveforms that are independent of the predicted source terms, collected by the heterogeneous radiation sensitive field sensor network. The offset correction module is used to perform wavefront matching between the theoretical radiation field change waveform calculated based on the predicted source term and the observed radiation field waveform, and obtain the spatial offset correction amount for the predicted source term by inverting the phase difference between the two. The radiation publishing module is used to calculate and publish the radiation field based on the corrected source term after spatial offset correction using the radiation calculation engine. The dynamic calibration module is used to compare the published radiation field with the observed waveforms of the radiation field in subsequent time slices. Based on the deviation of the comparison waveform relative to the theoretical radiation field change waveform, it dynamically calibrates the macroscopic dwell time distribution and the response function library of the radiation calculation engine.

Images