Method and system for real-time judgment of compliance of effluent quality of radioactive liquid waste treatment system

Abstract

The application provides a method and system for real-time judgment of compliance of effluent water quality of a radioactive waste liquid treatment system, and relates to the technical field of data processing. The method comprises the following steps: dividing an ultrasonic diffraction topological surface to obtain a plurality of parameter microzones; using three groups of characteristic parameters to perform spatial recursion according to the geometric center coordinates of each parameter microzone to obtain a local response index corresponding to each parameter microzone; fusing all the local response indexes to obtain a global distortion variable; inputting the three groups of characteristic parameters and the global distortion variable into a pre-trained multiple regression mapping model to convert the characteristic parameters into a nuclide concentration index value of the effluent water quality; and comparing the nuclide concentration index value with a preset compliance threshold value. If the nuclide concentration index value is lower than the compliance threshold value, it is determined that the effluent water quality is compliant, otherwise, it is determined that the effluent water quality is non-compliant. The application can realize stable and up-to-standard effluent water quality after radioactive waste liquid treatment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for real-time judgment of the compliance of effluent water quality in radioactive wastewater treatment systems. Background Technology

[0002] In the treatment of radioactive waste, compliance monitoring of effluent quality is a crucial step in ensuring environmental safety and the stable operation of the treatment system. This is especially true when treating high-level radioactive waste or complex radioactive waste, where achieving an ideal uniform mixing state of the fluid within the effluent pipe is often difficult. Due to differences in density and particle size among different radionuclides in the waste, and the influence of factors such as pipe diameter variations and elbow resistance, the flow velocity distribution within the pipe is often uneven, and some nuclides may exhibit localized distribution differences along the radial cross-section of the pipe. For example, radionuclides with relatively high density may experience slight enrichment at the bottom of the pipe under gravity; while the fluid velocity is relatively fast in the middle of the pipe, making it difficult for some nuclides to remain there, potentially forming localized low-concentration areas. When the operating conditions of the treatment system fluctuate slightly, transient high-concentration nuclide clusters may also occur. These localized abnormal distributions are usually difficult to eliminate quickly using conventional pipe mixing structures.

[0003] Currently, most existing effluent water quality testing methods rely on single-point sampling or single-sensor online monitoring. Single-point sampling requires a series of processes including sampling, testing, and analysis, making it difficult to achieve real-time assessment of water quality compliance. Furthermore, sampling points are often fixed, failing to comprehensively cover the entire pipe cross-section and thus possibly not fully reflecting the overall water quality across the pipe. While single-sensor online monitoring can achieve real-time monitoring to some extent, its detection range is limited, typically only acquiring water quality signals from a fixed point within the pipe. When uneven distribution of nuclides or localized high-concentration flow masses appear on the pipe cross-section, this single-point signal often fails to accurately characterize the average water quality across the entire pipe cross-section, potentially leading to misjudgments of effluent water quality compliance and posing potential environmental safety hazards. This may not fully meet the actual needs for real-time, accurate, and compliant assessment of effluent water quality during radioactive waste treatment. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for real-time judgment of the compliance of effluent water quality in radioactive waste liquid treatment system, which can achieve stable compliance of effluent water quality after radioactive waste liquid treatment.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a method for real-time assessment of the compliance of effluent quality from a radioactive wastewater treatment system, the method comprising:

[0007] Step 1: On the outer wall of the radioactive wastewater treatment effluent pipe, three detector array elements are staggered along the axial and circumferential directions, located at the top of the upstream of the pipe, the side of the middle of the pipe, and the bottom of the downstream of the pipe, respectively, so that the three detector array elements are not collinear; the raw ultrasonic echo signal of the effluent water quality at each detector array element is acquired, and the raw ultrasonic echo signal of each detector array element is preprocessed to obtain three sets of ultrasonic characteristic signals.

[0008] Step 2: Extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset.

[0009] Step 3: Construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; divide the ultrasonic diffraction topological surface to obtain multiple parameter micro-regions; based on the geometric centroid coordinates of each parameter micro-region, use the three sets of characteristic parameters to perform spatial recursion to obtain the local response index corresponding to each parameter micro-region; fuse all local response indices to obtain the global distortion variable.

[0010] Step 4: Input the three sets of feature parameters and global distorted variables into the pre-trained multivariate regression mapping model to convert the feature parameters into the nuclide concentration index values ​​of the effluent water quality.

[0011] Step 5: Compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

[0012] Secondly, a real-time system for assessing the compliance of effluent quality in radioactive wastewater treatment systems includes:

[0013] The acquisition module is used to arrange three detection array elements on the outer wall of the radioactive waste liquid treatment effluent pipe in an axial and circumferential staggered manner, located at the top of the upstream of the pipe, the side of the middle of the pipe, and the bottom of the downstream of the pipe, so that the three detection array elements are not collinear; to acquire the raw ultrasonic echo signal of the effluent water quality at each detection array element, and to preprocess the raw ultrasonic echo signal of each detection array element to obtain three sets of ultrasonic characteristic signals.

[0014] The processing module is used to extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset.

[0015] The calculation module is used to construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; to divide the ultrasonic diffraction topological surface into multiple parameter micro-regions; to obtain the local response index corresponding to each parameter micro-region by spatial recursion using the three sets of characteristic parameters based on the geometric centroid coordinates of each parameter micro-region; and to fuse all local response indices to obtain the global distortion variable.

[0016] The training module is used to input three sets of feature parameters and global distorted variables into a pre-trained multivariate regression mapping model, and convert the feature parameters into radionuclide concentration index values ​​of effluent water quality.

[0017] The judgment module is used to compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

[0018] Thirdly, a computing device includes:

[0019] One or more processors;

[0020] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0021] The above-described solution of the present invention has at least the following beneficial effects:

[0022] This invention acquires raw ultrasonic echo signals simultaneously through three detection array elements. After preprocessing, feature extraction, topological surface construction, and model mapping, ultrasonic characteristic parameters can be quickly converted into radionuclide concentration index values. The entire process eliminates the need for offline sampling and complex laboratory analysis, enabling real-time assessment of effluent water quality compliance. It shortens the monitoring cycle, promptly captures transient water quality anomalies, and meets the practical needs for real-time water quality monitoring during radioactive waste treatment. By constructing a plastic potential function and performing iterative minimization, the initial characteristic parameters are optimized, effectively correcting errors caused by signal interference, propagation attenuation, and other factors, thus improving the accuracy of the characteristic parameters. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the real-time judgment method for the compliance of effluent water quality in a radioactive waste liquid treatment system provided in an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of a real-time water quality compliance assessment system for a radioactive wastewater treatment system provided in an embodiment of the present invention. Detailed Implementation

[0025] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0026] like Figure 1 As shown, embodiments of the present invention propose a method for real-time judgment of the compliance of effluent water quality in a radioactive wastewater treatment system. The method includes the following steps:

[0027] Step 1: On the outer wall of the radioactive wastewater treatment effluent pipe, three detector array elements are staggered along the axial and circumferential directions, located at the top of the upstream of the pipe, the side of the middle of the pipe, and the bottom of the downstream of the pipe, respectively, so that the three detector array elements are not collinear; the raw ultrasonic echo signal of the effluent water quality at each detector array element is acquired, and the raw ultrasonic echo signal of each detector array element is preprocessed to obtain three sets of ultrasonic characteristic signals.

[0028] Step 2: Extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset.

[0029] Step 3: Construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; divide the ultrasonic diffraction topological surface to obtain multiple parameter micro-regions; based on the geometric centroid coordinates of each parameter micro-region, use the three sets of characteristic parameters to perform spatial recursion to obtain the local response index corresponding to each parameter micro-region; fuse all local response indices to obtain the global distortion variable.

[0030] Step 4: Input the three sets of feature parameters and global distorted variables into the pre-trained multivariate regression mapping model to convert the feature parameters into the nuclide concentration index values ​​of the effluent water quality.

[0031] Step 5: Compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

[0032] In this embodiment of the invention, the original ultrasonic echo signal is acquired synchronously by three detection array elements. After processing, feature extraction, topological surface construction, and model mapping, the ultrasonic feature parameters can be quickly converted into radionuclide concentration index values. The entire process does not require offline sampling and complex laboratory analysis, enabling real-time judgment of effluent water quality compliance. It shortens the monitoring cycle, can promptly capture transient water quality anomalies, and meets the actual needs of real-time water quality monitoring in the treatment of radioactive waste. By constructing a plastic potential function and performing a minimization iterative solution, the initial feature parameters are optimized, effectively correcting errors caused by signal interference, propagation attenuation, and other factors, and improving the accuracy of the feature parameters.

[0033] In a preferred embodiment of the present invention, step 1 involves arranging three detector elements on the outer wall of the radioactive wastewater treatment effluent pipe, staggered axially and circumferentially, located at the top upstream of the pipe, the side midstream of the pipe, and the bottom downstream of the pipe, respectively, ensuring that the three detector elements are not collinear; acquiring the raw ultrasonic echo signal of the effluent water quality at each detector element, and preprocessing the raw ultrasonic echo signal of each detector element to obtain three sets of ultrasonic characteristic signals, which may include:

[0034] Step 101: Based on the ultrasonic pulses and their reflected echoes synchronously emitted by the first detection array element at the top of the upstream of the effluent pipe, the second detection array element on the side of the middle of the pipe, and the third detection array element at the bottom of the downstream of the pipe, the three original echo timing data of the three detection array elements are obtained. Specifically, this includes: first, completing the fixed deployment of the three detection array elements. All three detection array elements are firmly installed on the outer wall of the radioactive waste liquid treatment effluent pipe. The first detection array element is precisely deployed at a preset position at the top of the upstream of the effluent pipe, the second detection array element is precisely deployed at a preset position on the side of the middle of the effluent pipe, and the third detection array element is precisely deployed at a preset position at the bottom of the downstream of the effluent pipe. The three detection array elements are strictly staggered along the axial and circumferential directions of the pipe, and it is ensured that the three are not collinear.

[0035] After deployment, the control module issues a synchronous control command to control the three detection array elements to simultaneously emit ultrasonic pulses of a fixed frequency. The ultrasonic pulses penetrate the pipe wall and enter the radioactive waste liquid inside the pipe. Some ultrasonic pulses collide and reflect with radionuclide particles and suspended impurities in the waste liquid, while others are reflected by the inner wall of the pipe, forming reflected echoes. The three detection array elements receive the reflected echoes corresponding to their respective emitted ultrasonic pulses. At the same time, the data acquisition module synchronously records the time sequence of the reflected echoes received by each detection array element. That is, starting from the emission of the ultrasonic pulse, the amplitude of the reflected echo is recorded at fixed time intervals for a preset duration, and finally, the first raw echo time sequence data corresponding to the first detection array element, the second raw echo time sequence data corresponding to the second detection array element, and the third raw echo time sequence data corresponding to the third detection array element are obtained.

[0036] Step 102: Filter the three original echo timing data streams to obtain three sets of filtered echo signals. Perform time gain compensation on the three sets of filtered echo signals to correct the attenuation differences caused by different sound wave propagation distances, resulting in three sets of compensated echo signals. Specifically, this includes: filtering the three original echo timing data streams; the core purpose of filtering is to remove irrelevant interference signals contained in the original echo timing data streams; for each original echo timing data stream, a low-pass filter adapted to ultrasonic signals is used, with a fixed filter cutoff frequency set that matches the transmission frequency of the ultrasonic pulse; signals in the original echo timing data streams whose frequencies are within the effective frequency range of the ultrasonic signal are selected one by one, accurately eliminating interference signals whose frequencies exceed the effective range. After filtering, three sets of filtered echo signals are obtained. The filtered echo signals can more clearly present the effective signal characteristics related to water quality.

[0037] After filtering, time gain compensation was performed on the three sets of filtered echo signals. Because the three detector elements were positioned at different locations in the pipeline, the propagation distances of the ultrasonic pulses from the detector elements into the waste liquid and back to the detector elements varied significantly. The third detector element (downstream bottom) had the longest propagation distance, while the first detector element (upstream top) had the shortest. The longer the propagation distance, the greater the attenuation of the ultrasonic signal in the waste liquid. This attenuation difference resulted in inconsistent echo signal amplitudes received by the three detector elements, making direct feature extraction and comparative analysis impossible. During compensation, the ultrasonic propagation distance corresponding to each filtered echo signal was measured and recorded. This involved measuring the vertical distance from each detector element to the reflection interface on the inner wall of the pipeline, and then multiplying this vertical distance by 2 to obtain the total propagation distance of the ultrasonic wave from the detector element to the reflection interface and back to the detector element. Simultaneously, the standard attenuation coefficient of the ultrasonic wave in pure water was recorded. As a reference benchmark, the standard attenuation coefficient is corrected based on the temperature parameters of the waste liquid in the pipeline to obtain an actual attenuation coefficient suitable for the current waste liquid conditions. Then, based on the propagation distance, the actual attenuation coefficient, and the current amplitude of the filtered echo signal, the amplitude compensation amount for each signal at each time point is calculated. The specific calculation process is as follows: Amplitude compensation amount = Current amplitude of the filtered echo signal × (Actual attenuation coefficient × Propagation distance). The longer the propagation distance, the larger the product of the actual attenuation coefficient and the propagation distance, and the larger the corresponding amplitude compensation amount. After the calculation is completed, the amplitude of each filtered echo signal is corrected point by point according to the calculated compensation amount for each time point. The corrected amplitude = Current amplitude of the filtered echo signal + Amplitude compensation amount. This accurately corrects the attenuation differences caused by different sound wave propagation distances, ensuring that the amplitudes of the three sets of compensated echo signals are within the same reasonable range, providing a stable and consistent signal basis for subsequent feature waveform extraction, and finally obtaining three sets of compensated echo signals.

[0038] Step 103: Based on the three sets of compensated echo signals, feature waveforms are extracted, and the main peak segment containing water quality information is extracted to obtain three sets of effective echo segments. Feature enhancement processing is performed on the three sets of effective echo segments to obtain three sets of ultrasonic feature signals. Specifically, based on the three sets of compensated echo signals, feature waveforms are extracted for each set of compensated echo signals. The compensated echo signals contain multiple peaks and troughs. Only the main peak segment contains the richest and most accurate radioactive waste liquid water quality information. The other secondary peaks and troughs are mostly interference signals or invalid reflection signals, which cannot accurately reflect the actual water quality status of the waste liquid. In practice, each compensated echo signal is analyzed point by point to identify each peak in the signal and record the amplitude, width and occurrence time of each peak. By comparing the amplitude and width of each peak, the main peak with the largest amplitude and the width that best matches the interaction characteristics between ultrasound and waste liquid is selected. Then, the complete segment containing the main peak is extracted. The extraction range starts from the preset time point before the main peak appears and ends at the preset time point after the main peak ends. This segment is the effective segment containing the core water quality information, thus obtaining three sets of effective echo segments. Each set of effective echo segments corresponds to the core water quality information of a detection array element.

[0039] After extracting the effective echo segments, feature enhancement processing was performed on the three groups of effective echo segments to further highlight water quality-related features and weaken residual weak interference signals. An amplitude amplification algorithm was used: first, the average amplitude of each group of effective echo segments was calculated; then, an amplification reference factor was set, and the amplification factor = amplification reference factor × (amplitude of the main wave peak of the effective echo segment ÷ average amplitude of the effective echo segment). The larger the ratio of the main wave peak amplitude to the average amplitude, the smaller the amplification factor, to avoid excessive distortion due to over-amplification of the main wave peak amplitude. Then, the effective echo segments were... The amplitude at each time point in the segment is multiplied by the corresponding amplification factor to obtain the amplified amplitude. At the same time, an interference suppression algorithm is used to first calculate the amplitude standard deviation of the effective echo segment. Signals with amplitudes less than (average amplitude - 0.5 × amplitude standard deviation) are identified as weak interference signals. The amplitude of such interference signals is suppressed point by point. The suppressed amplitude = current amplitude of the interference signal × interference suppression coefficient. The interference suppression coefficient is set to 0.2~0.3. Through the above specific calculations of amplitude amplification and interference suppression, after feature enhancement processing, three sets of ultrasonic characteristic signals are finally obtained.

[0040] This embodiment, through the reasonable arrangement of three non-collinear detection array elements, realizes the acquisition of water quality signals in different areas of the entire cross section of the pipeline, avoiding the shortcomings of single-point monitoring that cannot capture uneven local nuclide distribution and instantaneous high-concentration nuclide clusters.

[0041] In a preferred embodiment of the present invention, step 2 involves extracting water-quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters. These feature parameters include ultrasonic velocity, attenuation coefficient, and peak frequency shift, and may include:

[0042] Step 201: Perform transit time analysis, amplitude attenuation analysis, and spectrum analysis on the first ultrasonic characteristic signal to obtain the first initial sound velocity, the first initial attenuation coefficient, and the first initial peak frequency offset. Similarly, perform the same analysis on the second and third ultrasonic characteristic signals to obtain the second initial sound velocity, the second initial attenuation coefficient, the second initial peak frequency offset, the third initial sound velocity, the third initial attenuation coefficient, and the third initial peak frequency offset. Specifically, this includes: performing initial extraction of feature parameters for the three sets of ultrasonic characteristic signals, where the first set of ultrasonic characteristic signals corresponds to the first detector element, the second set of ultrasonic characteristic signals corresponds to the second detector element, and the third set of ultrasonic characteristic signals corresponds to the third detector element. The extraction methods for the three sets of signals are completely identical. Perform transit time analysis, amplitude attenuation analysis, and spectrum analysis on the first ultrasonic characteristic signal to obtain the first initial sound velocity, the first initial attenuation coefficient, and the first initial peak frequency offset. The transit time analysis is used to obtain the first initial sound velocity. Specifically, the signal analysis unit first records the ultrasonic pulse emitted from the first detector element. The total time from the arrival of the reflected echo to the arrival of the echo is called the transit time. The distance from the first detection element to the reflecting interface on the inner wall of the pipe is then measured beforehand. The propagation distance of the ultrasonic wave in the waste liquid is equal to twice the distance from the detection element to the reflecting interface. Dividing the calculated propagation distance by the transit time yields the first initial sound velocity. The specific calculation process is: First initial sound velocity = (2 × distance from the first detection element to the reflecting interface) ÷ transit time. Amplitude attenuation analysis is used to obtain the first initial attenuation coefficient. In practice, the emission data from the first detection element is first recorded. The initial amplitude of the ultrasonic pulse is recorded, and the amplitude of the received reflected echo is also recorded. The amplitude attenuation ratio is obtained by dividing the reflected echo amplitude by the initial amplitude. Then, combined with the propagation distance of the ultrasonic wave, the first initial attenuation coefficient is calculated based on the linear relationship between amplitude attenuation and propagation distance. In the formula, -ln is the negative of the natural logarithm, where ln is the natural logarithm with a constant e as the base. This quantifies the logarithmic relationship of the amplitude attenuation ratio, transforming the nonlinear amplitude attenuation into a linear relationship for accurate calculation of the attenuation coefficient. The specific calculation process is as follows: First initial attenuation coefficient = The calculation first calculates the ratio of the reflected echo amplitude to the initial amplitude, then takes the natural logarithm of this ratio, and takes a negative number. Finally, this negative number is divided by the propagation distance to obtain the first initial attenuation coefficient. Spectral analysis is used to obtain the first initial peak frequency offset. Specifically, a Fast Fourier Transform (FFT) is performed on the first ultrasonic characteristic signal. The specific calculation process is as follows: first, the first ultrasonic characteristic signal is sampled, and a fixed sampling frequency and number of sampling points are determined. The continuous ultrasonic characteristic signal is then discretized to obtain discretized signal data. Next, the sampled discrete signal is zero-padded to the nearest power of 2 (the number of sampling points). (b is an integer) to improve the accuracy of spectrum analysis and reduce signal spectrum distortion; then, Fourier transform is performed on the zero-padded discrete signal to convert the discrete signal from the time domain to the frequency domain, decomposing it into each frequency component and its corresponding amplitude; then, all frequency components and their corresponding amplitudes are integrated to form the frequency distribution curve of the signal, and the initial peak frequency of the signal is determined from the frequency distribution curve. The initial peak frequency is then compared with the transmission frequency of the ultrasonic pulse, and the difference between the two is calculated. This difference is the first initial peak frequency offset. The specific calculation process is as follows: first initial peak frequency offset = initial peak frequency - ultrasonic transmission frequency; using the same method as above, transit time analysis, amplitude attenuation analysis, and spectrum analysis are performed on the second ultrasonic characteristic signal to obtain the second initial sound velocity, the second initial attenuation coefficient, and the second initial peak frequency offset in turn; the same three analyses are performed on the third ultrasonic characteristic signal to obtain the third initial sound velocity, the third initial attenuation coefficient, and the third initial peak frequency offset in turn.

[0043] Step 202: Based on the first, second, and third initial sound velocities, and combined with the spatial coordinates of the three detection array elements, a plastic potential function for the sound velocity field is constructed. The plastic potential function is then iteratively minimized to obtain the optimized first, second, and third sound velocities. Based on the optimized first, second, and third sound velocities, and combined with the first, second, and third initial attenuation coefficients, a plastic potential function for the attenuation coefficients is constructed. The plastic potential function is then iteratively minimized to obtain the optimized first, second, and third attenuation coefficients. Attenuation coefficients; then, based on the optimized first, second, and third attenuation coefficients, combined with the first, second, and third initial peak frequency offsets, a plastic potential function for the frequency offset is constructed. The plastic potential function is then iteratively minimized to obtain the optimized first, second, and third peak frequency offsets. Specifically, based on the first, second, and third initial sound velocities, and combined with the spatial coordinates of the three detector elements, a plastic potential function for the sound velocity field is constructed, the expression of which is: ,in, Let be the plastic potential function of the sound velocity field. Let these be the spatial coordinates of any point inside the pipe. Each corresponds to one of the three detection array elements. The weighting coefficient (with a value ranging from 0.9 to 1.1 to ensure balanced signal weights for the three detector elements) is used. Let be the initial sound velocity of the i-th detection element. Let i be the spatial coordinates of the i-th detection element. Let be the sound velocity output value of the sound velocity field function at the position of the i-th detector element; during the construction process, the spatial coordinates of the three detector elements are... As input variables, the three initial sound speeds As an output variable reference, and based on the fundamental principles of the plastic potential function, the non-uniformity of the sound velocity field is incorporated into the function considerations, ensuring that the constructed plastic potential function accurately reflects the spatial distribution characteristics of the sound velocity within the pipe. After construction, the plastic potential function of the sound velocity field is iteratively minimized. During the solution process, initial iteration parameters and a deviation threshold (set to 0.01 m / s) are first defined, and then the weighting coefficients in the function are continuously adjusted. Calculate the output value of the function after each adjustment and the three initial sound velocities. If the deviation is greater than the preset threshold, the parameters are adjusted until the deviation reaches the preset reasonable range. The three sound speed values ​​obtained at this time are the optimized first sound speed, second sound speed and third sound speed.

[0044] The optimized attenuation coefficient parameter is constructed by using the optimized first, second, and third sound velocities, combined with the first, second, and third initial attenuation coefficients, to create a plastic potential function for the attenuation coefficient. Its expression is as follows: ,in, Let be the plastic potential function of the attenuation coefficient. Let be the initial attenuation coefficient of the j-th detector element. This represents the output value of the attenuation coefficient field function at the j-th detector element position. This represents the sound velocity of the j-th detector element after optimization; the meanings of the remaining letters are consistent with the plastic potential function of the sound velocity field. , , correspond During construction, the optimized three sound speeds will be used. As a correlation variable, since there is an inherent relationship between sound velocity and attenuation coefficient, changes in the sound velocity of the waste liquid in the pipeline will affect the degree of ultrasonic attenuation. This correlation is incorporated into the plastic potential function, and three initial attenuation coefficients are used. As an output reference, this ensures the function accurately reflects the spatial distribution of the attenuation coefficient. The plastic potential function of the attenuation coefficient is iteratively minimized using the same iterative method as the plastic potential function of the sound velocity field, with the same deviation threshold set. Parameters are continuously adjusted to reduce the deviation between the function output value and the initial attenuation coefficient until a preset accuracy is achieved, resulting in the optimized first, second, and third attenuation coefficients. The optimized peak frequency offset parameter is constructed based on the optimized first, second, and third attenuation coefficients, combined with the first, second, and third initial peak frequency offsets, to create a plastic potential function for the frequency offset, expressed as follows: ,in, Let be the frequency shift plastic potential function. Let be the initial peak frequency offset of the j-th detector element. This represents the output value of the frequency offset field function at the j-th detector element. This refers to the attenuation coefficient of the j-th detector element after optimization; the meanings of the remaining letters are consistent with the two plastic potential functions mentioned above. During the construction process, the three optimized attenuation coefficients... As a correlation variable, since changes in the attenuation coefficient affect the frequency characteristics of ultrasound, and thus the peak frequency shift, this correlation is incorporated into the function, along with three initial peak frequency shifts. As an output reference, the plastic potential function of the frequency offset is minimized iteratively, and the parameters are adjusted to reduce the deviation until the preset accuracy is achieved, so as to obtain the optimized first peak frequency offset, second peak frequency offset and third peak frequency offset.

[0045] Step 203: Based on the optimized first sound velocity, first attenuation coefficient, and first peak frequency offset, a first set of characteristic parameters is formed. Similarly, the optimized second sound velocity, second attenuation coefficient, and second peak frequency offset are combined to form a second set of characteristic parameters, and the optimized third sound velocity, third attenuation coefficient, and third peak frequency offset are combined to form a third set of characteristic parameters, resulting in three sets of characteristic parameters. Specifically, the optimized first sound velocity, first attenuation coefficient, and first peak frequency offset are integrated and combined to form the first set of characteristic parameters. This set of characteristic parameters fully reflects the radioactive waste liquid at the top upstream position of the pipeline where the first detection array element is located. The same combination method is used to integrate the optimized second sound velocity, second attenuation coefficient, and second peak frequency offset to form a second set of characteristic parameters. This set of characteristic parameters fully reflects the water quality characteristics of the radioactive waste liquid at the midstream side of the pipeline where the second detection element is located. The optimized third sound velocity, third attenuation coefficient, and third peak frequency offset are integrated to form a third set of characteristic parameters. This set of characteristic parameters fully reflects the water quality characteristics of the radioactive waste liquid at the downstream bottom of the pipeline where the third detection element is located. Through this combination, the three core water quality characteristics of each detection element are integrated together to form three sets of characteristic parameters.

[0046] This embodiment eliminates the influence of various error factors by extracting and optimizing the initial feature parameters, thereby improving the accuracy and reliability of the feature parameters and enabling the extracted feature parameters to truly reflect the actual water quality of the waste liquid in the pipeline.

[0047] In a preferred embodiment of the present invention, step 3 involves constructing an ultrasonic diffraction topological surface based on the spatial coordinates of the three detector array elements and three sets of characteristic parameters; subdividing the ultrasonic diffraction topological surface to obtain multiple parameter micro-regions; using the three sets of characteristic parameters to perform spatial recursion based on the geometric centroid coordinates of each parameter micro-region to obtain the local response index corresponding to each parameter micro-region; and fusing all local response indices to obtain the global distortion variable, which may include:

[0048] Step 301: Construct the first spatial feature point based on the spatial coordinates of the first detection array element and its corresponding first set of feature parameters; construct the second spatial feature point based on the spatial coordinates of the second detection array element and its corresponding second set of feature parameters; construct the third spatial feature point based on the spatial coordinates of the third detection array element and its corresponding third set of feature parameters. Specifically, this includes: obtaining the spatial coordinates of the first detection array element, which are the three-dimensional coordinates of the upstream top detection position of the pipeline that have been pre-measured and recorded; simultaneously obtaining the optimized first sound velocity, first attenuation coefficient, and first peak frequency offset; integrating and combining the three-dimensional spatial coordinates of the first detection array element with these three optimized feature parameters by sequentially associating the three components of the spatial coordinates with the three feature parameters to form the first spatial feature point. The first feature point contains both the detection location information and the core water quality characteristics of that location. Then, the spatial coordinates of the second detection element (i.e., the three-dimensional coordinates of the detection location on the midstream side of the pipeline), along with the optimized second sound velocity, second attenuation coefficient, and second peak frequency offset, are obtained. Using the same combination method as the first spatial feature point, the three-dimensional spatial coordinates of the second detection element are integrated with these three optimized feature parameters to form the second spatial feature point. Similarly, the spatial coordinates of the third detection element (i.e., the three-dimensional coordinates of the detection location at the downstream bottom of the pipeline), along with the optimized third sound velocity, third attenuation coefficient, and third peak frequency offset, are obtained. Again, using the same combination method, the three-dimensional spatial coordinates of the third detection element are integrated with these three optimized feature parameters to form the third spatial feature point.

[0049] Step 302: Based on the spatial coordinates of the three spatial feature points, determine the primitive computational domain enclosed by the three spatial feature points; extend the feature parameters at the three spatial feature points into the primitive computational domain to obtain the initial damage field; based on the initial damage field, divide the primitive computational domain into multiple computational cells to obtain the geometric centroid coordinates and initial damage values ​​of each computational cell. Specifically, this includes: based on the three-dimensional spatial coordinates of the three spatial feature points, determining the spatial region enclosed by the three through spatial geometric calculations; this region is the primitive computational domain, and the extent of the primitive computational domain is completely... It covers the pipe cross-sectional area corresponding to the three detection array elements, which can comprehensively include the main distribution range of waste liquid in the pipe. The optimized feature parameters at the three detection array elements are extended into the primitive calculation domain. The extension method is to use linear interpolation based on the difference of feature parameters of the three spatial feature points. The specific interpolation process is as follows: first, the primitive calculation domain is divided into several uniform interpolation nodes according to a preset step size. Each interpolation node has a unique three-dimensional spatial coordinate. Then, interpolation calculation is performed for each feature parameter (sound velocity, attenuation coefficient, peak frequency offset).

[0050] Taking sound speed interpolation as an example, the spatial distance from each interpolation node to three spatial feature points is first calculated. Then, the interpolation weight of the sound speed value of each spatial feature point is determined based on the spatial distance. The closer the point is to the interpolation node, the larger the weight; the farther the point is, the smaller the weight. The weight is calculated as the reciprocal of the spatial distance from the feature point to the interpolation node. The three weights are then normalized to ensure that the sum of the three weights is one. The interpolation weight of each spatial feature point ranges from 0.1 to 0.8, and the three weights are not equal to the sum of the interpolation weights. The weights of spatial feature points are set to 1. Points closer to the interpolation node have weights close to 0.8, points farther from the node have weights close to 0.1, and points at medium distances have weights between 0.3 and 0.5. The sound velocity value of each spatial feature point is multiplied by its corresponding interpolation weight, and the three weighted sound velocity values ​​are summed to obtain the sound velocity value of that interpolation node. Using the same interpolation method and weight range, the attenuation coefficient and peak frequency offset of each interpolation node are calculated sequentially, ensuring that each element in the computational domain... Each location has corresponding characteristic parameter values, thus obtaining the initial damage field within the entire primitive computational domain. The initial damage field is used to characterize the degree of water quality damage at different locations within the primitive computational domain. Its magnitude is positively correlated with the deviation of the characteristic parameters; that is, the greater the difference between the characteristic parameters and the average value of the characteristic parameters at the three detection array elements, the larger the initial damage field value. Based on the distribution of the initial damage field, the primitive computational domain is divided into multiple uniformly sized computational cells. During the division, it is ensured that the size of each computational cell is moderate, which can guarantee computational accuracy while avoiding excessive computational load. After the division, the geometric centroid coordinates of each computational cell are calculated one by one. The calculation method is to take the average value of the coordinates of all vertices of each computational cell, that is, to add the x-coordinates of each vertex and divide by the number of vertices to obtain the x-coordinate of the centroid. Similarly, the y-coordinates and z-coordinates of the centroid are calculated. At the same time, based on the value of the initial damage field within the computational cell, a corresponding initial cell damage value is assigned to each computational cell. The initial cell damage value is calculated by taking the average value of the initial damage field values ​​of all interpolation nodes within the computational cell.

[0051] Step 303: Establish the implicit integral control equation for damage evolution based on the initial cell damage value. Discretize the control equation in the time domain using backward Euler discretization to obtain the incremental form of the damage evolution equation for each computational cell. Specifically, this includes: establishing the implicit integral control equation for damage evolution corresponding to each computational cell based on the initial cell damage value of each computational cell. The specific expression is as follows: Where D is the damage variable for calculating the cell, which characterizes the degree of water quality damage to the radioactive waste liquid in the area corresponding to the cell; To calculate the initial value of cell damage; To calculate the damage variable of a cell at time t; This is the damage evolution rate function, used to describe how quickly the damage variable changes over time. Its value is related to the damage variable D and the stress of the waste liquid in the pipeline. And related to time t; The time step is defined as follows: Based on the initial value of cell damage, the equation considers the influence of factors such as waste liquid flow and radionuclide diffusion on damage changes, and is used to describe the evolution of cell damage as a function of the calculation process. The established implicit integral control equation for damage evolution is discretized in the time domain using the backward Euler method. The specific discretization process is as follows: first, the continuous time domain is divided into a time step of a preset value. Divided into multiple discrete time nodes, the time nodes are denoted as follows: ,in This is the initial calculation time. Each time step The value range is from 0.01s to 0.1s, specifically set according to the calculation accuracy and the flow rate of the waste liquid in the pipeline; the core discretization principle of the backward Euler method is clarified, namely, assuming a constant damage evolution rate within each time step, and using the damage evolution rate at the end of the current time step as the constant rate for the entire time step, rather than the initial rate, thereby improving the accuracy of the discretization calculation and adapting to the need to capture subtle changes in the water quality of the waste liquid in the pipeline; the implicit integral equation is discretized and derived, and the integration interval is... Corresponding to the time step, that is, the previous time step Damage variables As the lower bound of integration, the current time step Damage variables As the upper limit of integration, the left side of the integral equation is approximately: The derivation logic of this approximation is the increment of the damage variable. This is equal to the damage evolution rate multiplied by the time step, simplified by combining the physical meaning of the integral equation; substituting the above approximation into the original implicit integral control equation, we obtain... After rearranging and transforming the equation, we finally obtain the incremental form of the damage evolution equation corresponding to each computational cell, the specific expression of which is as follows: ;in, The damage variable at the current time step; The damage variable is from the previous time step; The damage evolution rate at the current time step is related to the damage variable, stress, and time at the current time step. This incremental equation can clearly define the incremental change of the cell damage variable within each time step, that is, the damage variable at the current time step is equal to the damage variable at the previous time step plus the damage increment at the current time step.

[0052] Step 304: Assemble all incremental damage evolution equations into a global nonlinear equation system. After iterative convergence, obtain the updated damage variables. Extract isosurfaces based on the updated damage variables, extracting spatial continuous surfaces where the damage variables equal a preset threshold to obtain the ultrasonic diffraction topological surface. Specifically, this includes: assembling all incremental damage evolution equations of all computational cells as a whole. The assembly method is to integrate the incremental damage equations of each computational cell sequentially to form a global nonlinear equation system containing the damage variables of all computational cells. The specific expression is as follows: ,in, It is the residual vector of the global nonlinear equation system, and each component corresponds to the residual of the incremental form of the damage evolution equation of the computation cell. A vector consisting of all computed cell damage variables, containing the damage variable for each computed cell at the current time step. The unknowns in this system of equations are the damage variables of all computational cells, and the number of equations is the same as the number of computational cells. The global nonlinear equation system is iteratively calculated. The iterative process involves assuming an initial set of damage variable values, substituting them into the equation system to calculate the deviation value of each equation. If the deviation value is greater than a preset convergence threshold (ranging from 0.001 to 0.01, specifically set according to the required computational accuracy), the damage variable values ​​are adjusted based on the deviation value. This process is repeated, continuously adjusting the damage variables and calculating the deviation, until the deviation values ​​of all equations are less than the preset convergence threshold. At this point, the iteration converges, yielding the updated damage variables of all computational cells. Isosurface extraction is performed based on the updated damage variables. The extraction process involves selecting all spatial points where the damage variables are equal to a preset threshold (ranging from 0.1 to 0.3), a critical value used to distinguish the degree of water quality damage. These spatial points are then connected sequentially to form a continuous spatial surface, which is the ultrasonic diffraction topological surface.

[0053] Step 305: Perform finite element meshing on the ultrasonic diffraction topological surface, dividing it into multiple parametric micro-regions to obtain the geometric centroid coordinates, boundary, and area of ​​each micro-region. Specifically, this involves: performing finite element meshing on the ultrasonic diffraction topological surface, dividing it into multiple uniformly sized parametric micro-regions, each a planar structure. During the meshing process, ensure seamless connection between adjacent micro-regions, with no overlap or omissions, covering the entire ultrasonic diffraction topological surface; after meshing, determine the geometric centroid coordinates of each parametric micro-region individually, using the same calculation method as in step 302 for calculating the cell centroid. The coordinate method is consistent: the average coordinates of all vertices in each parametric micro-region are taken to obtain the x, y, and z coordinates of the centroid. Simultaneously, the boundary range of each parametric micro-region is defined, i.e., the coordinates of all vertices in each micro-region are determined, thus defining the spatial range of the micro-region. Furthermore, the area of ​​each parametric micro-region is calculated by determining the planar area based on the coordinates of all vertices. Additionally, a shape symmetry detection algorithm is performed on each parametric micro-region. Specifically, a local coordinate system is established with the geometric centroid of the parametric micro-region as the center of symmetry, and the coordinates of all vertices in the micro-region are converted to coordinates in this local coordinate system, denoted as . , , Calculate the coordinates of the point symmetric to each vertex about the centroid. The formula for calculating the coordinates of the symmetric point is: ,in Let p = 1, 2, 3 be the geometric centroid coordinates of the parametric microregion. Calculate the Euclidean distance between each symmetric point and the other two vertices of the microregion to determine whether the symmetric point lies on the boundary or inside the microregion. Simultaneously, calculate the lengths of each side of the microregion, denoted as […]. Calculate the length deviation of the symmetrical side. The formula for calculating the deviation is: , and The edges are symmetrical. A symmetry determination threshold is set, with the threshold value ranging from 0.05 to 0.1. If the length deviation of all symmetrical edges is less than the threshold and the symmetrical points all fall on the boundary or inside of the micro-region, then the parameter micro-region is determined to be a symmetrical micro-region; otherwise, it is an asymmetrical micro-region.

[0054] Step 306: Based on the geometric centroid coordinates of each parameter micro-region, and combined with the three sets of feature parameters at the first, second, and third detection array elements, perform spatial recursive calculations to obtain local feature estimates. Specifically, this includes: for the geometric centroid coordinates of each parameter micro-region, calculating the spatial distance from the centroid to the first, second, and third detection array elements. The calculation method is based on the centroid coordinates and the coordinates of each detection array element, using the spatial distance formula, i.e., the spatial distance is equal to the square root of the sum of the squares of the differences between the corresponding coordinates of the centroid and the detection array element; based on the calculated three spatial distances, determining the spatial distance weight of each detection array element's feature parameter, with a larger weight for closer distances and a smaller weight for farther distances. The weight calculation method is that the weight of each detection array element is equal to the reciprocal of the spatial distance from the detection array element to the centroid; then, normalizing the three weights by dividing each weight by the sum of the three weights to ensure that the three weights are equal. The sum is one; combined with the shape symmetry detection results obtained in step 305, the weights are corrected. If it is a symmetrical micro-region, the weight correction coefficient is 1.0; if it is an asymmetrical micro-region, the weight correction coefficient is 0.9 to 0.95. The correction method is to multiply the normalized weight of each detector element by the correction coefficient and then perform normalization again to ensure that the sum of the three weights is still one after correction. The three sets of feature parameters of the three detector elements are weighted according to the corrected spatial distance weights. That is, each feature parameter of the first detector element is multiplied by the corresponding weight, each feature parameter of the second detector element is multiplied by the corresponding weight, and each feature parameter of the third detector element is multiplied by the corresponding weight. Then the weighted values ​​of the corresponding feature parameters of the three detector elements are added together to obtain the local feature estimate of each feature parameter in the parameter micro-region. That is, the local feature estimate of each parameter micro-region contains three components, which correspond to the local estimates of sound velocity, attenuation coefficient, and peak frequency offset, respectively.

[0055] Step 307: Calculate the local response index based on the local feature estimates and the micro-area area; fuse the local response index and the micro-area area, and sum the local response indices according to the corresponding micro-area area weights to obtain the global distortion variable. Specifically, this includes: calculating the local response index of each micro-area based on its local feature estimates and micro-area area, combined with the shape symmetry detection results from step 305. The calculation method is as follows: Let the three components of the local feature estimates be... (Local estimate of sound speed) (Local estimate of attenuation coefficient) (Local estimate of peak frequency offset): First, take the absolute value of each of the three components, then add the three absolute values ​​to obtain the total local characteristic deviation. The calculation formula is as follows: = Multiply the total deviation of local features by the area A of the micro-region of the parameter, and simultaneously multiply by the symmetry correction factor. Among them, the symmetric micro-region correction coefficient =1.0, asymmetric micro-area correction coefficient This highlights the impact of asymmetric micro-regions (areas where water quality heterogeneity is more likely to occur). The formula for calculating the local response index Q is Q = The larger the local response index, the greater the deviation between the water quality of that micro-region and the water quality at the detection point, and the more obvious the water quality inhomogeneity. The local response indices of all parameter micro-regions are accumulated and fused according to the weight of their corresponding micro-region areas. Assuming there are M parameter micro-regions in total, the local response index of the s-th micro-region is... The area of ​​the micro-region is The specific method of cumulative fusion is to first calculate the weighted local response index of each micro-region. Then, sum all the weighted local response exponents of all micro-regions. The formula for calculating the global distorted variable U is as follows: .

[0056] This embodiment, through the construction of ultrasonic diffraction topological surfaces, micro-region subdivision, and spatial recursive calculation, can accurately reflect the spatial distribution characteristics of radioactive waste liquid in the pipeline, capture minute changes and uneven distribution inside the waste liquid, and solve the problem that single-point monitoring cannot cover the entire cross-section of the pipeline.

[0057] In a preferred embodiment of the present invention, step 4, which involves inputting the three sets of feature parameters and the global distorted variable into a pre-trained multivariate regression mapping model to convert the feature parameters into radionuclide concentration index values ​​of the effluent water quality, may include:

[0058] Step 401: Based on the first set of feature parameters, the second set of feature parameters, the third set of feature parameters, and the global distorted variable, a feature parameter fusion set is obtained. Specifically, this includes: obtaining the optimized first set of feature parameters, the second set of feature parameters, the third set of feature parameters, and the global distorted variable calculated in step 307; integrating these parameters by arranging the three components of the first set of feature parameters, the three components of the second set of feature parameters, the three components of the third set of feature parameters, and the global distorted variable in a fixed order to form a complete feature vector, which is the feature parameter fusion set.

[0059] Step 402a: Based on the pre-trained multivariate regression mapping model, obtain the regression coefficient vector and intercept term; multiply each feature component in the feature parameter fusion set with the corresponding regression coefficient in the regression coefficient vector to obtain the weighted eigenvalue of each feature component. Specifically, this includes: calling the pre-trained multivariate regression mapping model, which is trained based on a large number of radioactive waste liquid detection samples and can achieve accurate mapping from ultrasonic feature parameters to nuclide concentration index values; obtaining the regression coefficient vector and intercept term from the model, where each regression coefficient in the regression coefficient vector corresponds to a feature component in the feature parameter fusion set, used to characterize the degree of influence of the feature component on the nuclide concentration; multiplying each feature component in the feature parameter fusion set with the corresponding regression coefficient in the regression coefficient vector, i.e., the first feature component multiplied by the first regression coefficient, the second feature component multiplied by the second regression coefficient, and so on, until the weighted eigenvalue of each feature component is obtained.

[0060] Step 402b involves summing all weighted eigenvalues ​​to obtain a weighted eigenvalue sum; then adding the weighted eigenvalue sum to the intercept term to obtain the initial nuclide concentration index value. Specifically, this includes summing all weighted eigenvalues ​​obtained in step 402a, i.e., summing all weighted eigenvalues ​​of each eigencomponent to obtain a weighted eigenvalue sum, which integrates the contributions of all eigencomponents to the nuclide concentration; finally, adding the weighted eigenvalue sum to the intercept term obtained from the multiple regression mapping model, i.e., adding the intercept term to the weighted eigenvalue sum, and the result is the initial nuclide concentration index value, which is the preliminary concentration value obtained through model mapping.

[0061] Step 403: Perform dimensional reduction processing based on the initial nuclide concentration index values ​​to convert them into final nuclide concentration index values ​​under actual dimensions, thereby obtaining the nuclide concentration index values ​​for the effluent water quality. Specifically, this includes: obtaining the dimensionally standardized parameters used during model training. These parameters need to be saved in advance along with the model parameters, and include two parts: firstly, the maximum nuclide concentration in the training samples. and minimum value Secondly, the maximum and minimum values ​​corresponding to each characteristic parameter (nine components of three sets of characteristic parameters and global distortion variables) are determined. Among them, the standardized parameters of the nuclide concentration are the core basis for dimensional reduction. The standardized parameters of the characteristic parameters are used to check the consistency of the reduction logic and avoid parameter confusion. Then, based on the dimensional standardized parameters, the initial nuclide concentration index values ​​obtained in step 402b are... To perform dimensional reduction calculations, we must first clarify the core logic of dimensional reduction, which is to reverse the standardization process during model training. The standardization formula for nuclide concentration during model training is as follows: (in This represents the standardized nuclide concentration. (This refers to the actual nuclide concentration). Therefore, the dimensional reduction formula is the inverse operation of the standardized formula. Specifically, the calculation method is to first calculate the difference between the maximum and minimum nuclide concentrations. = Then use the initial nuclide concentration index value Multiply by the difference The concentration deviation after reduction is obtained, and finally the minimum value of the nuclide concentration is added. ,Right now ,in This refers to the nuclide concentration index value in actual dimensions; the accuracy of the parameter values ​​needs to be verified during the calculation process to ensure accuracy. and The sample parameters used during model training are completely consistent with those used to avoid distortion of the restoration results due to parameter deviation. Through this dimensional restoration process, the initial nuclide concentration index value (standardized value in the range of 0 to 1) is converted into a nuclide concentration value that conforms to the actual detection standard (the unit is consistent with the actual detection), and finally the nuclide concentration index value corresponding to the effluent water quality is obtained.

[0062] The process of constructing, training, and implementing a multivariate regression mapping model:

[0063] Model construction involves determining the model's input and output. The input is a fusion set of feature parameters, consisting of nine components from three sets of feature parameters plus a global distorter, totaling ten feature components. The output is the nuclide concentration index value. Then, the basic structure of a multiple linear regression model is constructed. The model expression is that the nuclide concentration index value equals each input feature component multiplied by its corresponding regression coefficient, plus an intercept term, ensuring the model can linearly fit the relationship between the input features and the nuclide concentration. Simultaneously, considering the characteristics of radioactive waste liquid, the model is optimized by adding feature interaction terms—product terms between different feature components—to capture the intrinsic correlation between features, improve the model's fitting accuracy, and ensure the model can adapt to complex operating conditions such as fluctuating nuclide concentrations and uneven water quality within the pipeline.

[0064] Model training involves collecting a large number of radioactive waste liquid detection samples. The samples need to cover different nuclide concentrations, different water quality conditions, and different pipeline operating conditions. Each sample contains the characteristic parameters of three detection array elements, global distortion variables, and the corresponding actual nuclide concentration measurements to ensure the diversity and representativeness of the samples. The collected samples were then preprocessed. The feature parameters and measured nuclide concentrations in the samples were standardized by subtracting the minimum value from each parameter and then dividing by the difference between the maximum and minimum values, converting all parameters to the range of 0 to 1 to avoid the influence of parameters with different dimensions on model training. The preprocessed samples were divided into training and testing sets. The training set was used to train the model parameters, and the testing set was used to verify the model's accuracy, with a ratio of 80% for the training set and 20% for the testing set. The training set was input into the constructed multivariate regression mapping model, and the model was trained through iterative optimization. The training process involved continuously adjusting the regression coefficients and intercept term of the model to minimize the deviation between the predicted nuclide concentration values ​​output by the model and the actual measured nuclide concentration values ​​in the samples. Specifically, the square of the difference between the predicted and actual values ​​of all training samples was calculated. The model parameters are obtained by iteratively adjusting the regression coefficients and intercept term to minimize the sum of squares. The trained model is then validated using a test set. The deviation between the predicted and actual values ​​of the test set samples is calculated. If the deviation is within a preset accuracy range, the model training is considered successful. If the deviation exceeds the preset range, the model structure is readjusted, the sample size is increased, and training is repeated until the model accuracy meets the requirements. Model implementation involves saving the model parameters (regression coefficient vector and intercept term) after successful training. In actual detection, the feature parameter fusion set obtained in step 401 is input into the saved model. The initial nuclide concentration index value is obtained through calculations in steps 402a and 402b. After dimensional reduction processing in step 403, the actual nuclide concentration index value is obtained, achieving a rapid and accurate conversion from feature parameters to nuclide concentration.

[0065] This embodiment achieves accurate conversion of ultrasonic characteristic parameters to radionuclide concentration index values ​​by constructing, training and implementing a multivariate regression mapping model, thereby improving the accuracy and stability of water quality index calculation.

[0066] In a preferred embodiment of the present invention, step 5, comparing the radionuclide concentration index value with a preset compliance threshold, and determining that the effluent water quality is compliant if the radionuclide concentration index value is lower than the compliance threshold, and otherwise non-compliant, may include:

[0067] Step 501: Based on the nuclide concentration index value of the effluent water quality, obtain the current nuclide concentration index value as the value to be judged; perform intensity reduction processing on the value to be judged, and successively reduce the value to be judged according to the preset reduction coefficient to obtain the reduced nuclide concentration index value. Specifically, this includes: extracting the current nuclide concentration index value from the final nuclide concentration index value obtained in step 403. This value is used as the value to be judged. This value directly reflects the nuclide content of the wastewater effluent from the pipeline at the current moment, and its unit is consistent with the actual detection standard. Then, the value to be judged is gradually reduced according to a preset reduction factor. The reduction factor is a pre-set fixed value, ranging from 0.85 to 0.95, with 0.9 being preferred. The reduction factor needs to be set according to the pipeline operating conditions, nuclide type, and detection accuracy to ensure that the reduction process closely matches the actual nuclide concentration decay law. The reduction method is as follows: for the first reduction, the value to be judged is used... Multiply by the reduction factor The nuclide concentration index value after the first reduction was obtained. For the second reduction, the nuclide concentration index value after the first reduction is used. Multiply by the reduction factor The nuclide concentration index value after the second reduction was obtained. This process is repeated multiple times, with the number of reductions ranging from 3 to 8, and 5 under normal operating conditions, until the nuclide concentration index value after the last reduction is significantly lower than the preset compliance threshold, and the difference between the reduced value and the compliance threshold is not less than 10% of the compliance threshold, ensuring that the extreme reduction of nuclide concentration can be fully simulated; after each reduction, the corresponding reduction number and the nuclide concentration index value after that reduction are recorded.

[0068] Step 502: Compare the reduced nuclide concentration index value with the preset compliance threshold to obtain the comparison result under the reduced state. Specifically, this includes: obtaining the preset nuclide concentration compliance threshold R. This threshold is determined according to the preset detection standard and design requirements and is the core standard for judging whether the effluent water quality is compliant. The specific value range is 0.1 Bq / L to 1.0 Bq / L. Different nuclide types correspond to different thresholds. For example, the compliance threshold for Cs-137 ranges from 0.3 Bq / L to 0.5 Bq / L, and the compliance threshold for Co-60 ranges from 0.2 Bq / L to 0.4 Bq / L. The threshold needs to be entered into the detection system in advance and calibrated regularly; then, the reduced nuclide concentration index value obtained in step 501 ( , ... , Each reduction (number of times) is compared with a preset compliance threshold R. The comparison process uses precise numerical comparison to ensure no calculation error. The comparison result for each time is recorded, and the comparison result is divided into two types: one is that the reduced nuclide concentration index value is lower than the compliance threshold, i.e. < R (n=1,2,...,r), and secondly, the reduced nuclide concentration index value reaches or exceeds the compliance threshold, i.e. Simultaneously, accurately record the number of reductions that occur when the value after reduction first falls below the compliance threshold. , The value ranges from 1 to 8.

[0069] Step 503: Perform a limit analysis based on the comparison results under the reduced state. If the reduced nuclide concentration index value is lower than the preset compliance threshold, the current effluent water quality is determined to be compliant; if the reduced nuclide concentration index value reaches or exceeds the preset compliance threshold, the current effluent water quality is determined to be non-compliant. This specifically includes: based on the comparison results obtained in step 502, the reduced value sequence, and the number of reductions required for the first time to meet the standard. Limit analysis is conducted to determine the maximum reduction capacity and safety margin of nuclide concentration. Combining the reduction factor, the number of reductions, and the compliance threshold, the analysis quantifies the tolerance range of current nuclide concentration fluctuations. If, after a preset number of reductions, at least one reduction results in a nuclide concentration value lower than the preset compliance threshold R, and the difference between the final reduction value and the compliance threshold is not less than 10% of the compliance threshold, it indicates that the current effluent wastewater has a sufficient safety margin in nuclide concentration. Even if the nuclide concentration fluctuates to a certain extent (within the range of fluctuation not exceeding the attenuation ratio corresponding to the reduction factor), it can still meet compliance requirements, and the current effluent water quality is deemed compliant. If, after the preset maximum number of reductions (8 times), all reduction results still reach or exceed the preset compliance threshold R, it indicates that the current effluent wastewater has an excessively high nuclide concentration, exceeding the control range. Even with maximum reduction simulation, it cannot meet compliance requirements, posing an environmental safety hazard. Therefore, the current effluent water quality is deemed non-compliant, and an abnormal alarm is triggered, prompting staff to handle the situation promptly. At the same time, based on the number of reductions that first occur when the reduced value falls below the compliance threshold. To determine the stability of water quality: when When the water quality is stable and the nuclide concentration decreases rapidly, compliance is more assured; when When the water quality is stable, it indicates that the water quality is in good condition and meets the requirements for normal operation; when This indicates poor water quality stability.

[0070] This embodiment, through compliance judgment methods of intensity reduction and limit analysis, effectively avoids misjudgments caused by instantaneous concentration fluctuations, improves the reliability of compliance judgment, can promptly detect water quality anomalies, avoid environmental safety risks, and ensure the stable operation of the radioactive waste liquid treatment system.

[0071] like Figure 2 As shown, embodiments of the present invention also provide a real-time judgment system for the compliance of effluent water quality in a radioactive wastewater treatment system, including:

[0072] The acquisition module is used to arrange three detection array elements on the outer wall of the radioactive waste liquid treatment effluent pipe in an axial and circumferential staggered manner, located at the top of the upstream of the pipe, the side of the middle of the pipe, and the bottom of the downstream of the pipe, so that the three detection array elements are not collinear; to acquire the raw ultrasonic echo signal of the effluent water quality at each detection array element, and to preprocess the raw ultrasonic echo signal of each detection array element to obtain three sets of ultrasonic characteristic signals.

[0073] The processing module is used to extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset.

[0074] The calculation module is used to construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; to divide the ultrasonic diffraction topological surface into multiple parameter micro-regions; to obtain the local response index corresponding to each parameter micro-region by spatial recursion using the three sets of characteristic parameters based on the geometric centroid coordinates of each parameter micro-region; and to fuse all local response indices to obtain the global distortion variable.

[0075] The training module is used to input three sets of feature parameters and global distorted variables into a pre-trained multivariate regression mapping model, and convert the feature parameters into radionuclide concentration index values ​​of effluent water quality.

[0076] The judgment module is used to compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

[0077] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0078] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0079] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for real-time judgment of the compliance of effluent water quality in a radioactive waste liquid treatment system, characterized in that, The method includes: Step 1: Three detection array elements are staggered axially and circumferentially arranged on the outer wall of the radioactive wastewater treatment effluent pipe, located at the top upstream, the side midstream, and the bottom downstream of the pipe, respectively, ensuring that the three detection array elements are not collinear; the raw ultrasonic echo signals of the effluent water quality at each detection array element are acquired, and the raw ultrasonic echo signals of each detection array element are preprocessed to obtain three sets of ultrasonic characteristic signals, including: ultrasonic pulses synchronously emitted by the first detection array element at the top upstream of the effluent pipe, the second detection array element at the side midstream of the pipe, and the third detection array element at the bottom downstream of the pipe, respectively. The reflected echoes yielded three raw echo time-series data points corresponding to the three detection array elements. These three raw echo time-series data points were then filtered to obtain three sets of filtered echo signals. Time gain compensation was then applied to each of the three sets of filtered echo signals to correct for attenuation differences caused by varying sound wave propagation distances, resulting in three sets of compensated echo signals. Feature waveform extraction was performed on each of the three sets of compensated echo signals, extracting the main wave peak segment containing water quality information to obtain three sets of effective echo segments. Feature enhancement processing was then performed on each of the three effective echo segments to obtain three sets of ultrasonic feature signals. Step 2: Extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset. Step 3: Construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; divide the ultrasonic diffraction topological surface to obtain multiple parameter micro-regions; based on the geometric centroid coordinates of each parameter micro-region, use the three sets of characteristic parameters to perform spatial recursion to obtain the local response index corresponding to each parameter micro-region; fuse all local response indices to obtain the global distortion variable. Step 4: Input the three sets of feature parameters and global distorted variables into the pre-trained multivariate regression mapping model to convert the feature parameters into the nuclide concentration index values ​​of the effluent water quality. Step 5: Compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

2. The method for real-time judgment of compliance of effluent water quality in a radioactive wastewater treatment system according to claim 1, characterized in that, Step 2 includes: The first ultrasonic characteristic signal is subjected to transit time analysis, amplitude attenuation analysis and spectrum analysis to obtain the first initial sound velocity, the first initial attenuation coefficient and the first initial peak frequency offset; similarly, the second ultrasonic characteristic signal and the third ultrasonic characteristic signal are subjected to the same analysis to obtain the second initial sound velocity, the second initial attenuation coefficient, the second initial peak frequency offset, the third initial sound velocity, the third initial attenuation coefficient and the third initial peak frequency offset. Based on the first, second, and third initial sound velocities, and combined with the spatial coordinates of the three detection array elements, a plastic potential function for the sound velocity field is constructed. The plastic potential function is then iteratively minimized to obtain the optimized first, second, and third sound velocities. Based on the optimized first, second, and third sound velocities, and combined with the first, second, and third initial attenuation coefficients, a plastic potential function for the attenuation coefficients is constructed. This plastic potential function is then iteratively minimized to obtain the optimized first, second, and third attenuation coefficients. Finally, based on the optimized first, second, and third attenuation coefficients, and combined with the first, second, and third initial peak frequency offsets, a plastic potential function for the frequency offset is constructed. This plastic potential function is then iteratively minimized to obtain the optimized first, second, and third peak frequency offsets. The optimized first sound velocity, first attenuation coefficient, and first peak frequency offset are combined to form the first set of feature parameters. Similarly, the optimized second sound velocity, second attenuation coefficient, and second peak frequency offset are combined to form the second set of feature parameters. The optimized third sound velocity, third attenuation coefficient, and third peak frequency offset are combined to form the third set of feature parameters, resulting in three sets of feature parameters.

3. The method for real-time judgment of the compliance of effluent water quality in a radioactive wastewater treatment system according to claim 2, characterized in that, Based on the spatial coordinates of the three detector elements and three sets of characteristic parameters, an ultrasonic diffraction topological surface is constructed, including: Based on the spatial coordinates of the first detection array element and its corresponding first set of feature parameters, a first spatial feature point is constructed; based on the spatial coordinates of the second detection array element and its corresponding second set of feature parameters, a second spatial feature point is constructed; based on the spatial coordinates of the third detection array element and its corresponding third set of feature parameters, a third spatial feature point is constructed. Based on the spatial coordinates of three spatial feature points, the primitive computational domain enclosed by the three spatial feature points is determined; the feature parameters at the three spatial feature points are extended into the primitive computational domain to obtain the initial damage field; based on the initial damage field, the primitive computational domain is divided into multiple computational cells to obtain the geometric centroid coordinates of each computational cell and the initial damage value of the cell. Based on the initial value of cell damage, an implicit integral control equation for damage evolution is established. The control equation is then discretized by backward Euler in the time domain to obtain the incremental form of the damage evolution equation for each computational cell. All incremental damage evolution equations are assembled into a global nonlinear equation system. After iterative convergence, the updated damage variables are obtained. Isosurfaces are extracted based on the updated damage variables. A spatial continuous surface with damage variables equal to a preset threshold is extracted to obtain the ultrasonic diffraction topological surface.

4. The method for real-time judgment of compliance of effluent water quality in a radioactive wastewater treatment system according to claim 3, characterized in that, The ultrasonic diffraction topological surface is divided to obtain multiple parametric micro-regions; based on the geometric centroid coordinates of each parametric micro-region, spatial recursion is performed using three sets of characteristic parameters to obtain the local response index corresponding to each parametric micro-region. By fusing all local response indices, a global distorted variable is obtained, including: Finite element meshing is performed based on the ultrasonic diffraction topological surface, dividing the ultrasonic diffraction topological surface into multiple parametric micro-regions, and obtaining the geometric centroid coordinates, micro-region boundary, and micro-region area of ​​each parametric micro-region. Based on the geometric centroid coordinates of each parameter micro-region, combined with the three sets of characteristic parameters at the first, second, and third detector array elements, spatial recursive calculation is performed to obtain the estimated local feature values. The local response index is calculated based on the estimated local features and the area of ​​the micro-region. The local response index and the area of ​​the micro-region are then fused together, and the local response index is accumulated according to the corresponding micro-region area weight to obtain the global distortion variable.

5. The method for real-time judgment of the compliance of effluent water quality in a radioactive wastewater treatment system according to claim 4, characterized in that, Three sets of feature parameters and global distorted variables are input into a pre-trained multivariate regression mapping model to convert the feature parameters into radionuclide concentration index values ​​of the effluent water quality, including: The feature parameters are fused based on the first set of feature parameters, the second set of feature parameters, the third set of feature parameters, and the global distorted variables to obtain the feature parameter fusion set; The feature parameter fusion set is input into a pre-trained multivariate regression mapping model, and the initial nuclide concentration index value is obtained by linear combination calculation through the multivariate regression mapping model. The initial nuclide concentration index value is converted into the final nuclide concentration index value under the actual dimensions by performing dimensional reduction processing to obtain the nuclide concentration index value of the effluent water quality.

6. The method for real-time judgment of compliance of effluent water quality in a radioactive wastewater treatment system according to claim 5, characterized in that, The fused set of feature parameters is input into a pre-trained multivariate regression mapping model. Linear combination calculations are then performed using the multivariate regression mapping model to obtain initial nuclide concentration index values, including: Based on the pre-trained multivariate regression mapping model, the regression coefficient vector and intercept term are obtained; each feature component in the feature parameter fusion set is multiplied by the corresponding regression coefficient in the regression coefficient vector to obtain the weighted eigenvalue of each feature component. The weighted eigenvalues ​​are summed to obtain the weighted eigenvalue sum; the weighted eigenvalue sum is then added to the intercept term to obtain the initial nuclide concentration index value.

7. The method for real-time judgment of compliance of effluent water quality in a radioactive wastewater treatment system according to claim 6, characterized in that, The radionuclide concentration index value is compared with a preset compliance threshold. If the radionuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant, including: Based on the nuclide concentration index value of the effluent water quality, the current nuclide concentration index value is obtained as the value to be judged; the intensity reduction processing is performed on the value to be judged, and the value to be judged is reduced one by one according to the preset reduction coefficient to obtain the reduced nuclide concentration index value. The comparison results under the reduced nuclide concentration index value are obtained by comparing it with the preset compliance threshold. Based on the comparison results under the reduced state, a limit analysis is performed. If the reduced nuclide concentration index value is lower than the preset compliance threshold, the current effluent water quality is determined to be compliant; if the reduced nuclide concentration index value reaches or exceeds the preset compliance threshold, the current effluent water quality is determined to be non-compliant.

8. A real-time system for judging the compliance of effluent water quality in a radioactive wastewater treatment system, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to arrange three detection array elements on the outer wall of the radioactive waste liquid treatment effluent pipe in an axial and circumferential staggered manner, located at the top of the upstream of the pipe, the side of the middle of the pipe, and the bottom of the downstream of the pipe, so that the three detection array elements are not collinear; to acquire the raw ultrasonic echo signal of the effluent water quality at each detection array element, and to preprocess the raw ultrasonic echo signal of each detection array element to obtain three sets of ultrasonic characteristic signals. The processing module is used to extract water quality-related feature parameters from each group of ultrasonic feature signals to obtain three groups of feature parameters, including ultrasonic velocity, attenuation coefficient and peak frequency offset. The calculation module is used to construct an ultrasonic diffraction topological surface based on the spatial coordinates of the three detection array elements and three sets of characteristic parameters; to divide the ultrasonic diffraction topological surface into multiple parameter micro-regions; to obtain the local response index corresponding to each parameter micro-region by spatial recursion using the three sets of characteristic parameters based on the geometric centroid coordinates of each parameter micro-region; and to fuse all local response indices to obtain the global distortion variable. The training module is used to input three sets of feature parameters and global distorted variables into a pre-trained multivariate regression mapping model, and convert the feature parameters into radionuclide concentration index values ​​of effluent water quality. The judgment module is used to compare the nuclide concentration index value with the preset compliance threshold. If the nuclide concentration index value is lower than the compliance threshold, the effluent water quality is judged to be compliant; otherwise, it is judged to be non-compliant.

9. A computing device, characterized in that, include: Multiple processors; A storage device for storing a plurality of programs, which, when executed by the plurality of processors, cause the plurality of processors to implement the method as described in any one of claims 1 to 7.

Images