Nuclear wastewater discharge metering and monitoring device

Abstract

The invention relates to the field of nuclear wastewater discharge, in particular to a nuclear wastewater discharge metering and monitoring device which comprises a detector module, a temperature sensing module, a data processing module, a machine learning prediction module and a self-adaptive calibration module. The detector monitors radioactivity and outputs gamma energy spectrum data, the temperature sensing module collects temperature data, the data processing module carries out data fusion processing to generate a characteristic matrix, the machine learning module predicts detector deviation based on a space-time diagram convolution network, and the self-adaptive calibration module dynamically adjusts detector voltage according to the deviation. All the modules cooperate to achieve precise monitoring and intelligent calibration. According to the device, through temperature compensation, data mining, intelligent prediction and self-adaptive calibration, the nuclear wastewater radioactivity monitoring precision and stability can be effectively improved, and the safety of nuclear wastewater discharge is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of nuclear wastewater discharge, and in particular to a nuclear wastewater discharge metering and monitoring device. Background Art

[0002] Radioactivity monitoring of nuclear wastewater discharges is a core component of the nuclear energy safety system. Its core mission is to accurately measure the activity concentration and spectral characteristics of radionuclides in the wastewater to ensure that discharge indicators meet international safety standards. The core technology for radioactivity monitoring relies on gamma spectroscopy. High-purity germanium detectors and other precision instruments are used to capture the energy distribution of gamma rays and, combined with Monte Carlo simulation algorithms, analyze the nuclide species and activity. However, the nuclear wastewater discharge environment is highly complex. On the one hand, the uneven distribution of radioactive materials in wastewater storage vessels and pipelines leads to dynamic changes in the radiation field. On the other hand, monitoring equipment is constantly exposed to high radiation, high humidity, and electromagnetic interference. Temperature fluctuations in electronic components can cause detector spectral drift, resulting in shifts in characteristic peak positions or distortions in peak areas. Furthermore, periodic maintenance operations during nuclear power plant operation can alter the shielding structure, further exacerbating the unpredictability of the radiation field distribution. These factors make it difficult for traditional fixed monitoring systems to achieve long-term, stable, and high-precision measurements. Multi-source data fusion and dynamic calibration technologies are urgently needed to improve monitoring reliability.

[0003] The existing nuclear wastewater radioactivity monitoring system has an inadequate temperature drift compensation mechanism in its technical implementation. The temperature sensitivity of the detector's electronic components has not been effectively suppressed. Changes in ambient temperature or the equipment's own heating can lead to unstable high-voltage power supply output, which in turn causes energy spectrum peak position drift. Traditional systems rely solely on periodic manual calibration or static temperature compensation coefficients, and are unable to achieve real-time dynamic calibration. Secondly, the level of intelligence in data processing is low. Existing solutions mostly use threshold filtering or simple linear regression analysis, which makes it difficult to handle the multi-physics coupling effects of temperature, electromagnetic noise, and radiation fields, resulting in incomplete elimination of abnormal data and insufficient feature extraction. The calibration strategy lacks closed-loop feedback. Traditional calibration relies on offline reference sources or manual intervention, with a long calibration cycle and the inability to dynamically adjust parameters based on real-time monitoring data. Cumulative errors are prone to occur in continuous emission scenarios. Summary of the Invention

[0004] The present invention provides a nuclear wastewater discharge metering and monitoring device to solve the defect in the prior art that the detector calibration lacks the ability to compensate for the energy spectrum peak position drift caused by the ambient temperature change or the heating of the device itself in real time.

[0005] A nuclear wastewater discharge metering and monitoring device provided by the present invention comprises:

[0006] Detector module, used to monitor the radioactivity of nuclear wastewater and output gamma spectrum array data;

[0007] The temperature sensing module is integrated into the periphery of the electronic components of the detector module and is used to collect temperature time series data in real time;

[0008] The data processing module is used to collect the temperature time series data output by the temperature sensing module and the gamma spectrum array data output by the detector module, and generate an alignment feature matrix containing temperature features and energy spectrum peak features through spatiotemporal alignment, feature extraction and normalization processing;

[0009] The machine learning prediction module builds a temperature-energy spectrum drift prediction model based on a spatiotemporal graph convolutional network, inputs an alignment feature matrix, and outputs a detector measurement deviation prediction value;

[0010] The adaptive calibration module is used to dynamically adjust the detector's operating voltage according to the deviation prediction value.

[0011] Through the collaborative work of a detector module, temperature sensor module, data processing module, machine learning prediction module, and adaptive calibration module, this monitoring device achieves real-time and accurate monitoring of nuclear wastewater radioactivity. This effectively mitigates the impact of temperature changes on detector measurement accuracy, particularly in complex environmental conditions, ensuring the reliability of nuclear wastewater discharge monitoring. Machine learning technology enables intelligent prediction and calibration of detector measurement deviations, enhancing the monitoring device's intelligence and long-term stability. This helps to accurately and timely assess the radioactivity level in nuclear wastewater, providing a strong basis for nuclear wastewater discharge control.

[0012] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: a temperature sensing module comprising:

[0013] The temperature sensor unit is formed by arranging a plurality of resistance temperature sensors in an annular array in an embedded manner around the electronic components of the detector module;

[0014] A data acquisition unit, used to synchronously record the temperature data of each resistance temperature sensor according to a preset sampling frequency;

[0015] Interference compensation circuit is used to eliminate temperature measurement errors caused by environmental electromagnetic interference.

[0016] The temperature sensing module uses an embedded ring array layout of multiple resistance temperature sensors, combined with a data acquisition unit and interference compensation circuit, which can fully and accurately collect temperature data around the electronic components of the detector module, effectively reducing environmental electromagnetic interference errors, and providing a more accurate temperature basis for subsequent data processing and calibration, thereby improving the measurement accuracy of the detector and ensuring the accuracy of nuclear wastewater radioactivity monitoring results.

[0017] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: a method for a data processing module to perform data processing, comprising:

[0018] Collect the temperature time series data output by the temperature sensor module and the gamma spectrum array data output by the detector module;

[0019] Cubic spline interpolation was used to fill missing values in temperature time series data. The improved 3σ criterion was applied to γ spectrum data to eliminate abnormal counts and calculate characteristic peak areas and peak-to-background ratios.

[0020] Align the timestamps of temperature data and energy spectrum data based on the dynamic time warping algorithm; construct a multidimensional feature matrix containing temperature gradient, energy spectrum peak area and peak-to-background ratio;

[0021] The multidimensional feature matrix is hierarchically normalized, where the energy spectrum feature is normalized using Min-Max and the temperature feature is normalized using Z-score. The training set, validation set, and test set are divided according to the preset ratio.

[0022] The method of the data processing module can efficiently process temperature time series data and gamma energy spectrum array data. By filling missing values, eliminating abnormal counts, time alignment and feature extraction, it constructs a high-quality multidimensional feature matrix and performs reasonable normalization processing to make the data more in line with the input requirements of the machine learning model, thereby improving the training effect and prediction accuracy of the temperature-energy spectrum drift prediction model, ensuring the reliability of the detector measurement deviation prediction, and providing strong support for adaptive calibration.

[0023] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: a spatiotemporal graph convolutional network architecture comprising:

[0024] Input layer, used for the multi-dimensional feature matrix output by the data processing module;

[0025] The network architecture consists of several layers of spatiotemporal convolution blocks connected in series with LSTM timing units. Each convolution kernel uses a Chebyshev polynomial basis combined with dilated convolution to capture long-term temperature drift patterns.

[0026] The output layer uses a quantile regression structure to output the voltage deviation prediction value and confidence interval.

[0027] The spatiotemporal graph convolutional network architecture effectively captures the long-term temperature drift law and accurately models the complex relationship between temperature and energy spectrum by combining the Chebyshev polynomial basis with the convolution kernel of the void convolution and the LSTM timing unit. The output layer uses a quantile regression structure to further improve the reliability and comprehensiveness of the predicted value, which can more accurately predict the detector measurement deviation, provide more precise guidance for adaptive calibration, and enhance the adaptability and stability of nuclear wastewater monitoring equipment in complex environments.

[0028] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: a calculation formula of a loss function in a network architecture is:

[0029]

[0030] Among them, ρ 0.5 represents the robust quantile loss function, which is used to handle high noise and outliers of data in kernel environment through asymmetric weights; ω t =e -γt , represents the time-decreasing weight; Φ p represents the γ energy spectrum radiation flux predicted by the model; Φ M represents the flux result of Monte Carlo simulation; λ1 represents the weight of the physical constraint term; ΔV represents the working voltage adjustment of the detector; δ represents the preset safety threshold; λ2 represents the weight of the geometric reset constraint.

[0031] The loss function definition fully considers the characteristics of high noise and outliers in nuclear environmental data. It uses robust quantile loss function and asymmetric weight processing, combined with time-decreasing weights and physical constraints, so that the model can better adapt to the characteristics of nuclear environmental data during training, effectively suppress overfitting, and improve the accuracy and stability of the model in predicting the γ energy spectrum radiation flux.

[0032] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: a self-adaptive calibration module adjusting the detector working voltage in detail comprising:

[0033] When the measurement deviation output by the machine learning prediction module exceeds a preset threshold, the adaptive calibration module calculates the initial voltage adjustment value through the temperature-voltage transfer function;

[0034] To address the energy spectrum drift phenomenon, the adjustment voltage is applied step by step through the high-voltage power supply controller;

[0035] The collected and calibrated energy spectrum data is fed back to the machine learning prediction module to calculate the residual voltage deviation prediction value;

[0036] If the voltage deviation prediction value exceeds the preset threshold, iterative calibration is started until the measured deviation reaches the accuracy.

[0037] The detailed steps of the adaptive calibration module calculate the initial voltage adjustment value through the temperature-voltage transfer function, and adopt measures such as step-by-step application of adjustment voltage and closed-loop feedback. It can efficiently and accurately calibrate the detector operating voltage, so that the detector can always maintain the best working state under complex environmental conditions, effectively reduce measurement deviation, improve the accuracy and reliability of nuclear wastewater radioactivity monitoring, and ensure the credibility of the monitoring results.

[0038] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: an initial voltage adjustment value V of a detector module; new The calculation method is as follows:

[0039] Vnew =V base +ΔV

[0040] Among them, V base Indicates nominal voltage; E ref represents the reference energy of the γ radiation source; α represents the attenuation factor; V max represents the maximum safe voltage, ΔE represents the voltage deviation prediction value; β represents the temperature sensitivity coefficient; ΔT represents the difference between the current temperature and the reference temperature; T ref Indicates the reference temperature value.

[0041] This method for calculating the initial voltage adjustment value comprehensively considers multiple factors such as the nominal voltage, reference energy of the gamma radiation source, attenuation factor, maximum safe voltage, temperature sensitivity coefficient, and the difference between the current and reference temperatures. It accurately calculates the initial voltage adjustment value based on the physical model, providing a scientific and reasonable adjustment basis for the adaptive calibration module.

[0042] The present invention provides a nuclear wastewater discharge metering and monitoring device, comprising: an adaptive calibration module comprising:

[0043] A voltage adjustment value calculation unit, configured to calculate an initial voltage adjustment value of the detector module based on the voltage deviation prediction value output by the machine learning prediction module;

[0044] The high-voltage power supply controller is used to convert the initial voltage adjustment value digital instruction into an analog voltage reference value, and drive the high voltage to gradually increase or decrease through the PWM signal;

[0045] The closed-loop feedback unit is used to monitor the detector output energy spectrum data in real time and feed it back to the machine learning prediction module to evaluate the residual deviation after each voltage adjustment.

[0046] The voltage adjustment value calculation unit accurately calculates the initial voltage adjustment value, the high-voltage power supply controller realizes efficient conversion of digital instructions to analog voltage, and monitors the energy spectrum data feedback evaluation in real time through the closed-loop feedback unit to form a closed-loop control. It can quickly and accurately complete the dynamic adjustment of the detector working voltage, effectively improve the measurement accuracy of the detector, and enhance the intelligence level and adaptability of the nuclear waste water monitoring device.

[0047] The present invention also provides a nuclear wastewater discharge metering and monitoring device, comprising: a step of applying an adjustment voltage in steps by a high-voltage power supply controller, comprising:

[0048] The high-voltage power supply controller receives the initial voltage adjustment value output by the voltage adjustment value calculation unit and decomposes the total adjustment amount into multiple small steps according to a preset single-step adjustment upper limit;

[0049] The digital instructions are converted into analog voltage reference values through the DAC module, and the adjustment voltage is applied step by step by the high-voltage amplifier;

[0050] After each voltage adjustment, wait for the preset time to ensure that the energy spectrum data is stable.

[0051] The high-voltage power supply controller applies the adjustment voltage in steps, decomposing the total adjustment amount into small steps, applying the adjustment voltage step by step, and waiting for the energy spectrum data to stabilize after each step, avoiding the over-adjustment and instability problems that may be caused by large step adjustments, making the detector working voltage adjustment process smoother and more accurate, and effectively improving the accuracy and reliability of the detector calibration.

[0052] The present invention also provides a nuclear wastewater discharge metering and monitoring device, including: the anti-interference design of the device includes: setting a metal shielding cover on the periphery of the detector to suppress electromagnetic interference within the device; deploying the PTP precise time protocol to ensure the synchronization accuracy of multiple detectors.

[0053] This technical solution proposes an innovative nuclear wastewater discharge metering and monitoring device. Through the collaborative operation of multiple modules, it achieves efficient and accurate monitoring of nuclear wastewater radioactivity and intelligent detector calibration. The device's core components include a detector module, a temperature sensor module, a data processing module, a machine learning prediction module, and an adaptive calibration module. The detector module monitors the radioactivity of nuclear wastewater and outputs gamma spectrum array data. The temperature sensor module, integrated within the detector module's electronic components, collects temperature time series data in real time. Its design utilizes multiple resistive temperature sensors arranged in an embedded ring array and equipped with an interference compensation circuit to eliminate temperature measurement errors caused by environmental electromagnetic interference, ensuring the accuracy of temperature data. The data processing module processes the collected temperature and gamma spectrum data. Through spatiotemporal alignment, feature extraction, and normalization, it generates an aligned feature matrix containing temperature and gamma spectrum peak features, providing a high-quality data foundation for subsequent prediction models. The machine learning prediction module constructs a temperature-spectrum drift prediction model based on a spatiotemporal graph convolutional network. It takes the aligned feature matrix as input and outputs predicted detector measurement deviations. This model employs optimization strategies such as a robust quantile loss function to improve prediction accuracy and reliability. The adaptive calibration module dynamically adjusts the detector's operating voltage based on the predicted deviation value. It includes a voltage adjustment value calculation unit, a high-voltage power supply controller, and a closed-loop feedback unit. Through a closed-loop feedback mechanism, it monitors the detector's output energy spectrum data in real time and feeds it back to the machine learning prediction module to evaluate residual deviations, achieving precise calibration of the detector's operating voltage and ensuring the accuracy of the monitoring results. Furthermore, the device features anti-interference features, such as a metal shield and the PTP precision time protocol, to suppress electromagnetic interference and ensure multi-detector synchronization accuracy, further improving the device's stability and reliability in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 It is a structural schematic diagram of a nuclear wastewater discharge metering and monitoring device provided by an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram of the installation of a nuclear wastewater discharge metering and monitoring device provided by an embodiment of the present invention.

[0057] Figure 3 This is a flow chart for obtaining a detector deviation prediction value in a nuclear wastewater discharge metering and monitoring device provided by an embodiment of the present invention.

[0058] Figure 4 This is a flow chart for adjusting the working voltage of a detector in a nuclear wastewater discharge metering and monitoring device provided by an embodiment of the present invention.

[0059] Figure 5 This is a circuit structure diagram of an adaptive calibration module in a nuclear wastewater discharge metering and monitoring device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0061] The following combination Figure 1-Figure 5 The invention describes a nuclear waste water discharge metering and monitoring device.

[0062] like Figure 1-Figure 5 As shown, an embodiment of the present invention provides a nuclear wastewater discharge metering and monitoring device, comprising:

[0063] The detector module, in this embodiment, comprises a 3×3 annular array of high-purity germanium detectors. Each detector is equipped with a preamplifier and a high-voltage power supply. These detectors monitor the radioactivity of the nuclear wastewater and output gamma spectrum array data. During operation, the detectors continuously detect gamma rays from the environment surrounding the nuclear wastewater outlet or within the discharge pipeline, collecting gamma photons and converting them into electrical signals for gamma spectrum array data, which are then transmitted to the data processing module.

[0064] The temperature sensing module is integrated into the electronic components of the detector module and is used to collect temperature time series data in real time. The temperature sensing module includes:

[0065] The temperature sensor unit in this embodiment uses multiple PT1000 platinum resistance temperature sensors arranged in a circular array in an embedded manner around the electronic components of the detector module; when the detector electronic components are working, the PT1000 platinum resistance temperature sensor senses these temperature changes in real time, converts the temperature signal into a resistance change value, and connects it to the data acquisition unit through a dedicated four-wire terminal.

[0066] The data acquisition unit is used to synchronously record temperature data from each resistance temperature sensor at a preset sampling frequency. This embodiment uses a PCI-1716UX data acquisition card to ensure synchronous and high-precision data acquisition from multiple temperature sensors. It simultaneously records temperature time series data from each PT1000 platinum resistance temperature sensor at a sampling frequency of 10Hz.

[0067] The interference compensation circuit is used to eliminate temperature measurement errors caused by environmental electromagnetic interference. In this embodiment, the interference compensation circuit is composed of a low-pass filter circuit and an electromagnetic shielding layer.

[0068] Through the rational selection of temperature sensors and precise layout and connection methods, combined with effective interference compensation circuits, high-precision, real-time and reliable measurement of the temperature of the detector module's electronic components is achieved, providing accurate temperature timing data parameters for subsequent temperature-energy spectrum drift prediction and detector adaptive calibration.

[0069] Data processing module. In this embodiment, the data processing module includes an MPSoC chip, a 24-bit ADC and Gigabit Ethernet. The 24-bit ADC is used for multi-channel data acquisition, and data is transmitted with the machine learning prediction module via Gigabit Ethernet. It is used to collect the temperature time series data output by the temperature sensing module and the gamma energy spectrum array data output by the detector module. Through spatiotemporal alignment, feature extraction and normalization processing, an alignment feature matrix containing temperature features and energy spectrum peak features is generated and sent to the machine learning prediction module.

[0070] The data processing module performs data processing in the following ways:

[0071] Collect the temperature time series data output by the temperature sensor module and the gamma spectrum array data output by the detector module;

[0072] Cubic spline interpolation was used to fill missing values in temperature time series data. The improved 3σ criterion was applied to γ spectrum data to eliminate abnormal counts and calculate characteristic peak areas and peak-to-background ratios.

[0073] Align the timestamps of temperature data and energy spectrum data based on the dynamic time warping algorithm; construct a multidimensional feature matrix containing temperature gradient, energy spectrum peak area and peak-to-background ratio;

[0074] The multidimensional feature matrix is hierarchically normalized, with energy spectrum features using Min-Max normalization and temperature features using Z-score normalization. The training, validation, and test sets are then divided according to a preset ratio. In this embodiment, the training, validation, and test sets are divided in a ratio of 7:2:1 to obtain the normalized training, validation, and test sets.

[0075] Through a scientific data processing process, data quality and usability were effectively improved. The data cleaning step removed outliers and filled missing values, making the data more complete and accurate. The spatiotemporal alignment and feature extraction steps ensured data consistency and relevance, facilitating subsequent analysis and modeling. The normalization and dataset partitioning steps provided a sound data foundation for training and evaluating machine learning models, improving the model's performance and generalization capabilities, thereby enhancing the overall intelligence and accuracy of the facility's nuclear wastewater discharge monitoring.

[0076] The machine learning prediction module builds a temperature-energy spectrum drift prediction model based on the spatiotemporal graph convolutional network, inputs the alignment feature matrix, and outputs the detector measurement deviation prediction value ΔE.

[0077] The architecture of the spatiotemporal graph convolutional network includes:

[0078] Input layer, used for the multi-dimensional feature matrix output by the data processing module;

[0079] The network architecture consists of several layers of spatiotemporal convolution blocks connected in series with LSTM timing units. Each convolution kernel uses a Chebyshev polynomial basis combined with dilated convolution to capture long-term temperature drift patterns.

[0080] The output layer uses a quantile regression structure to output the voltage deviation prediction value and confidence interval.

[0081] The calculation formula of the loss function in the network architecture is:

[0082]

[0083] Among them, ρ 0.5 represents the robust quantile loss function, which is used to handle high noise and outliers of data in kernel environment through asymmetric weights; ω t =e -γt , represents the time-decreasing weight; Φ p represents the γ energy spectrum radiation flux predicted by the model; Φ Mrepresents the flux result of Monte Carlo simulation; λ1 represents the weight of the physical constraint term; ΔV represents the working voltage adjustment of the detector; δ represents the preset safety threshold; λ2 represents the weight of the geometric reset constraint.

[0084] By constructing a spatiotemporal graph convolutional network, we can fully exploit the spatiotemporal characteristics and complex relationships in the data, effectively capture long-term dependencies, and improve the accuracy of predicting detector measurement deviations. This loss function definition fully considers the high noise and outlier characteristics of nuclear environmental data. By utilizing a robust quantile loss function and asymmetric weighting, combined with time-decreasing weights and physical constraints, the model can better adapt to the characteristics of nuclear environmental data during training, effectively suppress overfitting, and improve the accuracy and stability of the model's gamma spectral radiation flux predictions.

[0085] The training process of the machine learning prediction module is as follows:

[0086] Input data construction: The normalized multidimensional feature matrix containing temperature gradient, energy spectrum peak area and peak-to-background ratio is fused into graph structure data. The node feature is the temperature value, and the edge weight is dynamically calculated based on the detector array consistency coefficient and environmental parameters.

[0087] Network architecture configuration: 4 layers of spatiotemporal convolution blocks use Chebyshev polynomial basis and dilated convolution in sequence, with the number of output channels in each layer being 64, 128, 256, and 512 respectively; LSTM units with 128 hidden nodes capture temporal dependencies; the output layer predicts the deviation value through a quantile regression structure.

[0088] Loss function optimization: The AdamW optimizer is used to jointly optimize the robust quantile loss, Monte Carlo physical constraints, and geometric reset penalty terms. During training, multi-unit data is aggregated through a federated learning framework, and the weights of the physical constraints are retained to ensure that the prediction results conform to the laws of radiation transmission.

[0089] The training data was divided into 7:2:1, and 5-fold cross validation was used to prevent overfitting.

[0090] The adaptive calibration module is used to dynamically adjust the detector's operating voltage based on the deviation prediction value. The adaptive calibration module includes:

[0091] A voltage adjustment value calculation unit, configured to calculate an initial voltage adjustment value of the detector module based on the voltage deviation prediction value output by the machine learning prediction module;

[0092] The high-voltage power supply controller is used to convert the initial voltage adjustment value digital instruction into an analog voltage reference value, and drive the high voltage to gradually increase or decrease through the PWM signal;

[0093] The closed-loop feedback unit is used to monitor the detector output energy spectrum data in real time and feed it back to the machine learning prediction module to evaluate the residual deviation after each voltage adjustment.

[0094] The detailed steps for the adaptive calibration module to adjust the detector position and operating voltage include:

[0095] When the measurement deviation output by the machine learning prediction module exceeds 3%, the adaptive calibration module calculates the initial voltage adjustment value through the temperature-voltage transfer function; the initial voltage adjustment value V new The calculation method is as follows:

[0096] V new =V base +ΔV

[0097] Among them, V base Indicates nominal voltage; E ref represents the reference energy of the γ radiation source; α represents the attenuation factor; V max represents the maximum safe voltage, ΔE represents the voltage deviation prediction value; β represents the temperature sensitivity coefficient; ΔT represents the difference between the current temperature and the reference temperature; T ref Indicates the reference temperature value. In this embodiment, the attenuation factor α=1×e -3 / ℃, temperature sensitivity coefficient β=0.005V / ℃, reference energy of γ radiation source is 1332.5keV corresponding to the γ ray energy of Co-60, and reference temperature value is set to 25℃, the ambient temperature of the detector during calibration.

[0098] To address the energy spectrum drift phenomenon, a high-voltage power supply controller is used to apply an adjustment voltage in steps. The steps of applying the adjustment voltage in steps by the high-voltage power supply controller include:

[0099] The high-voltage power supply controller receives the initial voltage adjustment value output by the voltage adjustment value calculation unit and decomposes the total adjustment amount into multiple small steps according to a preset single-step adjustment upper limit;

[0100] The digital instructions are converted into analog voltage reference values through the DAC module, and the adjustment voltage is applied step by step by the high-voltage amplifier;

[0101] After each voltage adjustment, wait for the preset time to ensure that the energy spectrum data is stable.

[0102] The collected and calibrated energy spectrum data is fed back to the machine learning prediction module to calculate the residual voltage deviation prediction value;

[0103] If the voltage deviation prediction value exceeds 1%, iterative calibration is started until the measured deviation reaches the accuracy.

[0104] The adaptive calibration module calculates the initial voltage adjustment value using the temperature-voltage transfer function. This calculation method comprehensively considers multiple factors, including the nominal voltage, gamma-ray source reference energy, attenuation factor, maximum safe voltage, temperature sensitivity coefficient, and the temperature difference between the current and reference temperatures. Based on a physical model, this precise calculation of the initial voltage adjustment value provides a scientific and rational basis for the adaptive calibration module. Furthermore, by employing step-by-step application of the adjustment voltage and closed-loop feedback, the module efficiently and accurately calibrates the detector operating voltage, ensuring that the detector maintains optimal operating conditions under complex environmental conditions. This effectively reduces measurement deviation, improves the accuracy and reliability of nuclear wastewater radioactivity monitoring, and ensures the credibility of monitoring results. The high-voltage power supply controller applies the adjustment voltage stepwise, breaking the total adjustment into small steps. The adjustment voltage is applied stepwise, and the energy spectrum data is allowed to stabilize after each step. This avoids the overshoot and instability that can occur with large step-by-step adjustments, resulting in a smoother and more precise detector operating voltage adjustment process and effectively improving the accuracy and reliability of detector calibration.

[0105] In addition, the device's anti-interference design includes: setting up a metal shielding cover around the detector to suppress electromagnetic interference below 200kV / m; deploying the PTP precise time protocol to ensure the synchronization accuracy of multiple detectors.

[0106] The present invention provides a nuclear wastewater discharge metering and monitoring device. The beneficial effects of this technical solution are primarily reflected in the following aspects: First, through the close integration of the temperature sensing module and the detector module, the impact of temperature changes on detector measurements can be monitored in real time. Furthermore, through the refined processing of the data processing module, the measurement error caused by temperature changes is effectively reduced, thereby improving the accuracy and stability of nuclear wastewater radioactivity monitoring. Second, the introduction of a machine learning prediction module provides the device with intelligent prediction and calibration capabilities, dynamically adjusting the detector's operating state based on real-time data to adapt to the complex and changing nuclear wastewater discharge environment, greatly improving the adaptability and reliability of monitoring. Third, the closed-loop feedback design of the adaptive calibration module ensures that the detector always maintains optimal operating conditions during long-term operation, reducing maintenance costs and workload while improving the credibility of monitoring data. Finally, the anti-interference design of the entire device enhances its stability and safety in practical applications, enabling it to operate stably in complex industrial environments. It provides accurate and reliable data support for the metering and monitoring of nuclear wastewater discharge, helps to protect environmental safety and public health, and has important practical application value for the supervision of nuclear wastewater treatment and discharge.

[0107] The device embodiments described above are merely illustrative. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., located in one location or distributed across multiple network units. Some or all of these modules may be selected based on actual needs to achieve the objectives of the present embodiments. Persons of ordinary skill in the art will understand and implement these embodiments without inventive effort.

[0108] Through the description of the above embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by means of software plus a necessary general-purpose hardware platform, or of course, by means of hardware. Based on this understanding, the above technical solution is essentially embodied in the form of a software product, or in other words, the portion that contributes to the prior art. This computer software product is stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, a server, or a network device) to execute the methods of each embodiment or certain portions of the embodiments.

[0109] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will appreciate that modifications may be made to the technical solutions described in the aforementioned embodiments, or that some of the technical features therein may be replaced with equivalents. However, such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A nuclear wastewater discharge metering and monitoring device, characterized in that: include: Detector module, used to monitor the radioactivity of nuclear wastewater and output gamma spectrum array data; The temperature sensing module is integrated into the periphery of the electronic components of the detector module and is used to collect temperature time series data in real time; a data processing module, configured to collect the temperature time series data output by the temperature sensing module and the gamma spectrum array data output by the detector module, and generate an alignment feature matrix containing temperature features and energy spectrum peak features through spatiotemporal alignment, feature extraction, and normalization processing; a machine learning prediction module, which constructs a temperature-energy spectrum drift prediction model based on a spatiotemporal graph convolutional network, inputs the alignment feature matrix, and outputs a detector measurement deviation prediction value; An adaptive calibration module is used to dynamically adjust the operating voltage of the detector according to the deviation prediction value.

2. A nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The temperature sensing module includes: The temperature sensor unit is formed by arranging a plurality of resistance temperature sensors in an annular array in an embedded manner around the electronic components of the detector module; A data acquisition unit, used to synchronously record the temperature data of each resistance temperature sensor according to a preset sampling frequency; Interference compensation circuit is used to eliminate temperature measurement errors caused by environmental electromagnetic interference.

3. A nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The method for the data processing module to perform data processing includes: Collecting the temperature time series data output by the temperature sensing module and the gamma energy spectrum array data output by the detector module; The temperature time series data is interpolated using cubic spline to fill missing values; the gamma energy spectrum data is subjected to an improved 3σ criterion to eliminate abnormal counts, and characteristic peak areas and peak-to-background ratios are calculated; Align the timestamps of temperature data and energy spectrum data based on the dynamic time warping algorithm; construct a multidimensional feature matrix containing temperature gradient, energy spectrum peak area and peak-to-background ratio; The multidimensional feature matrix is subjected to hierarchical normalization, wherein the energy spectrum feature is normalized by Min-Max and the temperature feature is normalized by Z-score; and the training set, validation set and test set are divided according to a preset ratio.

4. The nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The architecture of the spatiotemporal graph convolutional network includes: Input layer, for the multi-dimensional feature matrix output by the data processing module; The network architecture consists of several layers of spatiotemporal convolution blocks connected in series with LSTM timing units. Each convolution kernel uses a Chebyshev polynomial basis combined with dilated convolution to capture long-term temperature drift patterns. The output layer uses a quantile regression structure to output the voltage deviation prediction value and confidence interval.

5. A nuclear wastewater discharge metering and monitoring device according to claim 4, characterized in that: The calculation formula of the loss function in the network architecture is: Among them, ρ 0.5 represents the robust quantile loss function, which is used to handle high noise and outliers of data in kernel environment through asymmetric weights; ω t =e -γt , represents the time-decreasing weight; Φ p represents the γ energy spectrum radiation flux predicted by the model; Φ M represents the flux result of Monte Carlo simulation; λ1 represents the weight of the physical constraint term; ΔV represents the working voltage adjustment of the detector; δ represents the preset safety threshold; λ2 represents the weight of the geometric reset constraint.

6. The nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The detailed steps of the adaptive calibration module adjusting the detector operating voltage include: When the measurement deviation output by the machine learning prediction module exceeds a preset threshold, the adaptive calibration module calculates an initial voltage adjustment value through a temperature-voltage transfer function; To address the energy spectrum drift phenomenon, the adjustment voltage is applied step by step through the high-voltage power supply controller; The collected and calibrated energy spectrum data is fed back to the machine learning prediction module to calculate the residual voltage deviation prediction value; If the voltage deviation prediction value exceeds the preset threshold, iterative calibration is started until the measured deviation reaches the accuracy.

7. A nuclear wastewater discharge metering and monitoring device according to claim 6, characterized in that: The initial voltage adjustment value V of the detector module new The calculation method is as follows: V new =V base +ΔV Among them, V base Indicates nominal voltage; E ref represents the reference energy of the γ radiation source; α represents the attenuation factor; V max represents the maximum safe voltage, ΔE represents the voltage deviation prediction value; β represents the temperature sensitivity coefficient; ΔT represents the difference between the current temperature and the reference temperature; T ref Indicates the reference temperature value.

8. The nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The adaptive calibration module includes: a voltage adjustment value calculation unit, configured to calculate an initial voltage adjustment value of the detector module according to the voltage deviation prediction value output by the machine learning prediction module; A high-voltage power supply controller, configured to convert the initial voltage adjustment value digital instruction into an analog voltage reference value, and drive the high voltage to gradually increase or decrease through a PWM signal; The closed-loop feedback unit is used to monitor the detector output energy spectrum data in real time and feed it back to the machine learning prediction module to evaluate the residual deviation after each voltage adjustment.

9. The nuclear wastewater discharge metering and monitoring device according to claim 8, characterized in that: The step of applying the adjustment voltage in steps by the high-voltage power supply controller includes: The high-voltage power supply controller receives the initial voltage adjustment value output by the voltage adjustment value calculation unit, and decomposes the total adjustment amount into multiple small steps according to a preset single-step adjustment upper limit; The digital instructions are converted into analog voltage reference values through the DAC module, and the adjustment voltage is applied step by step by the high-voltage amplifier; After each voltage adjustment, wait for the preset time to ensure that the energy spectrum data is stable.

10. The nuclear wastewater discharge metering and monitoring device according to claim 1, characterized in that: The anti-interference design of the device includes: setting a metal shielding cover on the periphery of the detector to suppress electromagnetic interference within the device; deploying the PTP precise time protocol to ensure the synchronization accuracy of multiple detectors.

Images