A method and system for laser cleaning optimization of product containers in a uranium enrichment plant

Abstract

This invention provides an optimized method and system for laser cleaning of product containers in uranium enrichment plants. It proposes a method for optimizing laser cleaning parameters. In the actual laser cleaning process, real-time data is input into the model to obtain the prediction results, i.e., the optimal laser cleaning parameter configuration, so as to achieve the best cleaning effect. A double wedge mirror spiral scanning method is used to achieve full coverage scanning of the inner wall of the container. The cleaning speed can reach up to 9000 mm / s, which improves the speed and efficiency of cleaning.

Description

Technical Field

[0001] This invention belongs to the field of container laser cleaning, specifically relating to an optimized method and system for laser cleaning of product containers in uranium enrichment plants. Background Technology

[0002] Uranium enrichment plant product containers require cleaning when residual levels exceed limits, pressure testing time (5 years), container enrichment levels exceed 5% and change enrichment, or containers are defective. Before cleaning, a vacuum recovery process is performed. After deep evacuation, the weight of residual material inside the container is generally less than 2 kg, mainly consisting of metallic fluorides, non-volatile uranium, fluorine compounds, and rust. Currently, there are two methods for cleaning uranium hexafluoride containers: chemical cleaning and high-pressure water cleaning. Domestically, chemical cleaning is mainly used. This involves adding no more than 30L of water for hydrolysis, followed by multiple alkaline washes with a hot solution of 5% sodium carbonate and approximately 30% hydrogen peroxide, then multiple water washes, followed by rust removal with a 2% oxalic acid solution, and finally drying, baking, and vacuum leak testing. Containers must also undergo a hydrostatic test upon reaching the end of their lifespan. This method consumes many types and large quantities of reagents, and the amount of radioactive wastewater is approximately four times the container volume. High-pressure water cleaning of a 740L uranium hexafluoride container has been successfully tested, using high-pressure water jets to spray the inner wall of the container instead of alkaline washing, water washing, and rust removal. Water jets, with high impact kinetic energy, continuously act on the inner surface of the container, causing residues and rust to fall off, thus achieving the cleaning purpose. This method does not consume reagents, and the amount of radioactive wastewater is approximately 0.6 times the container volume. Both the chemical cleaning method and the high-pressure cleaning method generate large amounts of radioactive wastewater. This wastewater requires wastewater treatment processes, and the treatment process produces a large amount of radioactive lime residue, which presents certain difficulties in storage and treatment, and does not meet the goal of minimizing radioactive waste.

[0003] Laser cleaning is a dry cleaning method that uses a high-energy laser beam to irradiate the surface of an object. Through shock waves and thermoelastic effects, the contaminants absorb energy and vaporize, separating from the surface and achieving cleaning. This method does not produce radioactive wastewater and poses no safety concerns, making it a green and safe cleaning technology.

[0004] Therefore, laser cleaning of uranium containers is a green, environmentally friendly, and safe cleaning method, which is of great significance for solving the problems existing in the wet process of cleaning containers in uranium enrichment plants. How to optimize the laser cleaning of uranium enrichment plant product containers to make the laser cleaning of containers more effective and controllable is an urgent problem to be solved. Summary of the Invention

[0005] To address the shortcomings and improvement needs of existing product container cleaning processes, this invention provides an optimized laser cleaning method and system for uranium enrichment plant product containers. This solves the problem of generating large amounts of radioactive wastewater from current wet cleaning methods, thereby achieving a cleaning effect where the inner surface of the container is clean and dry after washing, free of oil, rust, slag, and visible deposits.

[0006] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0007] An optimized method for laser cleaning of product containers in a uranium enrichment plant includes the following steps:

[0008] S1. Install scanning mechanisms at the valve and plug of the container to be cleaned, and debug and calibrate the scanning mechanisms. The scanning mechanism includes a focusing lens group, a galvanometer, a scanning wedge, and an optical window. The beam input of the focusing lens group is connected to an external collimation output device. The beam output of the focusing lens group is calibrated and controlled for the first time through two sets of galvanometers. The beam output of the galvanometer is calibrated and controlled for the second time through two sets of scanning wedges. The beam output of the scanning wedge is output through the optical window.

[0009] S2. Turn on the pulsed laser source. The pulsed laser is transmitted through a flexible armored fiber optic cable to output the laser beam. The position information of the working point on the inner wall of the container to be cleaned is obtained through illumination imaging monitoring.

[0010] S3. The scanning mechanism is recalibrated through centralized control so that the working point position of the inner wall of the container to be cleaned is sequentially scanned at the preset point position;

[0011] In step S1, the two sets of scanning mechanisms scan and clean the container to be cleaned. The plug and the valve are respectively located on both sides of the container to be cleaned. The container to be cleaned is fully scanned at the plug and the valve by the corresponding spiral scanning method of the two sets of scanning mechanisms.

[0012] The scanning path for the full-coverage scan is as follows:

[0013] For the plug: the scanning mechanism scans the upper inner wall of the container, the inner wall on the side where the valve is installed, and the lower inner wall of the container sequentially; for the valve: the scanning mechanism scans the upper inner wall of the container, the inner wall on the side where the plug is installed, and the lower inner wall of the container sequentially; wherein, step S3 also includes a process step of centralized control to optimize the parameter configuration of laser cleaning:

[0014] S301. Collect real-time data, including input parameters and output results during the laser cleaning process. The input parameters include at least: laser energy (energy intensity of the laser beam); cleaning speed (speed at which the laser beam moves across the container surface); cleaning time (duration of the cleaning process); container internal conditions (environmental parameters such as temperature and humidity); container initial state (including degree of contamination and coating type); and laser beam position and angle (irradiation position and angle inside the container). The output results include at least: optimal laser energy (model-predicted optimal laser energy under current conditions); and optimal cleaning speed (model-predicted optimal cleaning speed under current conditions).

[0015] S302. Data preprocessing;

[0016] S303. Select a machine learning model based on the nature of the problem, data characteristics, etc.

[0017] S304. Based on the problem and the selected model, perform feature engineering to extract, transform, or select appropriate features for model training.

[0018] S305. Train the model using the training set; optimize model performance by adjusting hyperparameters;

[0019] S306. Deploy the trained model into the actual laser cleaning system. During the actual laser cleaning process, input real-time data into the model to obtain the prediction results, i.e., the optimal laser cleaning parameter configuration.

[0020] Specifically, step S303 includes:

[0021] S3031. Input Feature (X): Laser Energy (Unit: Joules)

[0022] X=[100, 150, 200, 250, 300];

[0023] Output (Y): Cleaning effect (unit: percentage)

[0024] Y=[80, 85, 90, 88, 92];

[0025] S3032. A linear regression model will be used to fit the data, specifically:

[0026] Y = β0 + β1·X + ε;

[0027] Wherein, Y represents the predicted value of the cleaning effect, X represents the laser energy, β0 represents the intercept, β1 represents the coefficient of the laser energy, and ε represents the error term;

[0028] S3033. Find the optimal β0 and β1 so that the model's predictions are closest to the actual observations;

[0029] S3034. The least squares method is used as the loss function for the linear regression model, with the goal of minimizing the squared error between the predicted and actual values. Specifically:

[0030]

[0031] Where m represents the number of samples, and h β (X (i) Y represents the model's prediction for the i-th sample. (i) This represents the actual value of the i-th sample;

[0032] S3035. Minimize the loss function using the gradient descent algorithm to find the optimal model parameters. Specifically:

[0033]

[0034]

[0035] Wherein, α represents the learning rate, used to control the step size of parameter updates; and These represent the partial derivatives of the loss function with respect to the parameters;

[0036] S3036. Parameter Update: Through multiple iterations, the parameters are updated until the loss function reaches its minimum, yielding the optimal β0 and β1. Specifically:

[0037]

[0038]

[0039] Furthermore, step S2 also includes extracting and discharging the dust and fumes generated during the laser action inside the container to be cleaned, maintaining a low negative pressure state inside the container.

[0040] Furthermore, in step S2, the pulsed laser source enters the scanning mechanism after being output through a flexible armored optical fiber cable, and the laser beam is adjusted by beam shaping to perform a full-coverage scan of the container to be cleaned, directly acting on the dirt on the surface of the part to be cleaned to cause the dirt to burn, vaporize and decompose.

[0041] A laser cleaning optimization system for product containers in a uranium enrichment plant, the system being implemented based on any of the aforementioned laser cleaning optimization methods for product containers in a uranium enrichment plant, comprising a laser cleaning subsystem and a container to be cleaned, wherein the laser cleaning subsystem cleans the interior of the container to be cleaned by laser irradiation with a laser beam, wherein:

[0042] The laser cleaning subsystem is located outside the container to be cleaned. The laser beam of the laser cleaning subsystem has two outputs, which are respectively located at the valve and the plug of the container to be cleaned. The valve and the plug are respectively located on both sides of the container to be cleaned.

[0043] The laser cleaning subsystem includes two sets of scanning mechanisms. Each scanning mechanism includes a focusing lens group, a galvanometer, a scanning wedge, and an optical window. The beam input of the focusing lens group is connected to an external collimation output device. The beam output of the focusing lens group is first calibrated and controlled by the two sets of galvanometers. The beam output of the galvanometer is second calibrated and controlled by the two sets of scanning wedges. The beam output of the scanning wedge is output and scanned through the optical window.

[0044] The laser cleaning subsystem also includes a distance measurement module, which measures the distance information of the working point position. The scanning mechanism adjusts and controls the focal point position based on the distance information of the working point position collected by the distance measurement module.

[0045] Furthermore, the laser cleaning subsystem also includes an illumination imaging module, which is used to perform online image monitoring of the working point position on the inner wall of the container to be cleaned during the laser irradiation cleaning process.

[0046] Furthermore, the laser cleaning subsystem also includes a centralized control module, which is used to perform automated control of the system and optimize the parameter configuration of laser cleaning through centralized control software. The automated control includes centralized control of the focusing, scanning, shaping, imaging and measurement systems.

[0047] Furthermore, the laser cleaning subsystem also includes a pulsed laser source module, which is used to output pulsed laser light and transmits the output laser beam through a flexible armored optical fiber cable.

[0048] Furthermore, the laser cleaning subsystem also includes an extraction module, which is used to extract and discharge the dust and fumes generated inside the container during the laser action, maintaining a low negative pressure state inside the container.

[0049] Compared with the prior art, the present invention has at least the following beneficial effects:

[0050] (1) The present invention proposes a laser cleaning optimization method and system for uranium enrichment plant product containers. It proposes a method for optimizing the parameter configuration of laser cleaning. In the actual laser cleaning process, real-time data is input into the model to obtain the prediction result, that is, the optimal laser cleaning parameter configuration, so that the cleaning effect is optimal. The double wedge mirror spiral scanning method is used to achieve full coverage scanning of the inner wall of the container. The cleaning speed can reach up to 9000 mm / s, which improves the cleaning speed and efficiency.

[0051] (2) The present invention proposes an optimized method and system for laser cleaning of product containers in uranium enrichment plants. It adopts centralized control software for the automated control of the entire system, including centralized control of focusing, scanning, shaping and imaging and measurement systems, so as to realize the automation of the cleaning process.

[0052] (3) The laser cleaning optimization method and system for uranium enrichment plant product containers proposed in this invention is a non-contact cleaning method compared with traditional cleaning methods. The beam shaping can adjust the output laser beam and can continuously adjust the size of the laser beam as needed. In addition, the precision measuring component is used to measure the distance information of the working point position online. It adopts a high-precision laser ranging method and has strong applicability in materials such as pipelines, closed systems and pumps.

[0053] (4) The present invention proposes an optimized method and system for laser cleaning of product containers in uranium enrichment plants. Based on pulsed laser as the laser source of the cleaning system, it outputs high-power, high-energy, and high-repetition-rate pulsed laser to provide pulsed laser for the laser cleaning system. It adopts a flexible armored optical fiber cable output method. The high-energy laser beam irradiates the surface of the object. The residue inside the container absorbs energy and is rapidly vaporized and separated from the surface of the object, thereby improving the cleaning efficiency and saving costs. Attached Figure Description

[0054] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings. In the exemplary embodiments, the same reference numerals generally denote the same components.

[0055] Figure 1 This is a system structure diagram of an optimized laser cleaning system for product containers in a uranium enrichment plant, as proposed in an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram of the structure of the container to be cleaned in a laser cleaning optimization system for uranium enrichment plant product containers according to an embodiment of the present invention;

[0057] Figure 3 This is a scanning schematic diagram of an optimized laser cleaning system for product containers in a uranium enrichment plant, as proposed in an embodiment of the present invention.

[0058] In the diagram, 1-the container to be cleaned, 2-the valve, 3-the plug, 4-the focusing lens group, 5-the galvanometer, 6-the scanning wedge mirror, 7-the distance measurement module, 8-the pulsed laser source module, and 9-the extraction module. Detailed Implementation

[0059] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are merely preferred embodiments of the present invention, and not all embodiments, nor are they intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications or equivalent variations based on the disclosed technical content. However, any simple modifications, equivalent variations, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention's technical solution shall still fall within the protection scope of the present invention's technical solution.

[0060] like Figure 1 , Figure 2 This embodiment proposes an optimized laser cleaning system for uranium enrichment plant product containers, including a laser cleaning subsystem and a container 1 to be cleaned. The laser cleaning subsystem cleans the interior of the container 1 by laser irradiation with a laser beam, wherein:

[0061] The laser cleaning subsystem is located outside the container 1 to be cleaned. The laser beam of the laser cleaning subsystem has two outputs, which are respectively located at the valve 2 and the plug 3 of the container 1 to be cleaned. The valve 2 and the plug 3 are respectively located on both sides of the container 1 to be cleaned.

[0062] The laser cleaning subsystem includes two scanning mechanisms, each comprising a focusing lens group 4, a galvanometer 5, a scanning wedge mirror 6, and an optical window. The beam input of the focusing lens group 4 is connected to an external collimation output device. The beam output of the focusing lens group 4 undergoes a first calibration control via the two galvanometer mirrors 5. The beam output of the galvanometer mirrors 5 undergoes a second calibration control via the two scanning wedge mirrors 6. The beam output of the scanning wedge mirrors 6 is then scanned through the optical window. Wherein, as... Figure 3 The specific scanning process is as follows: when the scanning head is in the valve position (A): the scanning area is CBD; when the scanning head is in the plug position (B): the scanning area is CAD.

[0063] The laser cleaning subsystem also includes a distance measurement module 7, which measures the distance information of the working point position. The scanning mechanism adjusts and controls the focal point position based on the distance information of the working point position collected by the distance measurement module.

[0064] Furthermore, as a preferred embodiment of this example, the container 1 to be cleaned is selected as a dedicated UF6 liquefied material receiving and transport container, used for transfer between uranium enrichment plants and nuclear fuel element plants. The container has a valve and a plug installed at each end cap. The valve is a 1-inch right-angle valve, connected to the container via an NPT1 thread. The valve is used for receiving and discharging material from the container. The 30B container consists of a container body, valves, plugs, and valve protective covers. The container is 1930mm long, 760mm in diameter, 12mm thick, weighs approximately 630kg, and has a maximum loading capacity of 2200kg.

[0065] Furthermore, the laser cleaning subsystem also includes an illumination imaging module, which is used to perform online image monitoring of the working point position on the inner wall of the container to be cleaned 1 during the laser irradiation cleaning process.

[0066] Furthermore, the laser cleaning subsystem also includes a centralized control module, which is used to automatically control the system through centralized control software. The automated control includes centralized control of the focusing, scanning, shaping, imaging, and measurement systems.

[0067] Furthermore, the laser cleaning subsystem also includes a pulsed laser source module 8, which is used to output high-power, high-energy, high-repetition-rate pulsed laser light and transmits the output laser beam through a flexible armored fiber optic cable. Specifically, in this embodiment, the laser used as the pulsed laser source is a continuous pulsed laser. The pulsed laser can be output via a flexible armored fiber optic cable, with an output power adjustable between 10W and 500W, a spectral width less than 30nm, and an output pulse width adjustable from 0-500ns.

[0068] Furthermore, a preferred embodiment of this invention is proposed, employing an optical transmission component for transmitting the laser beam output from the light source. This component includes a reflector, a dichroic mirror, a polarizer, a glass slide, a high-precision optical frame, and their structural components. This allows for laser beam connection between the various components, with a focusing range of 1-3m and an injection spot size of <20mm.

[0069] Furthermore, the laser cleaning subsystem also includes an extraction module 9, which is used to extract and discharge the dust and fumes generated inside the container 1 during the laser action, maintaining a low negative pressure state inside the container.

[0070] Furthermore, a laser cleaning principle of this embodiment is proposed: a high-energy laser beam irradiates the workpiece surface, causing surface dirt, rust, or coatings to evaporate or peel off instantly, thereby quickly and effectively removing surface deposits or coatings to achieve a cleaning effect. The cleaning force (thermal stress) must be greater than the adhesive force (van der Waals force) for the particles to detach from the substrate. Van der Waals force: the adhesion of micron-sized particles to the substrate surface is mainly due to van der Waals force, while particles larger than 50μm are mainly due to electrostatic force. For small particles, the van der Waals force exceeds gravity, and the formula for calculating the adhesion force F per unit area on the solid surface is as follows:

[0071]

[0072] Wherein, r represents the particle radius, h represents the van der Waals constant of the material, δ represents the adhesion area radius, and z represents the interatomic distance between the particle and the object surface, which is approximately 0.4 nm.

[0073] Pulsed laser irradiation causes heat accumulation, resulting in uneven temperature distribution and a rapid temperature rise across the particles, generating thermal stress within and near the particles. The cleaning force per unit area is:

[0074] f = γEΔT(d, t);

[0075] Wherein, f represents the cleaning force per unit area; represents the surface tension of the γ liquid; E represents the porosity of the material; ΔT(d, t) represents the temperature difference; d represents the depth; and t represents the time.

[0076] Furthermore, based on different laser parameters, it can be divided into two methods: continuous laser cleaning and short-pulse laser cleaning.

[0077] Continuous laser cleaning (including long pulses and microsecond pulses): The thermal effect caused by the laser on the material is significant, and its main cleaning mechanism is laser ablation. High-power continuous lasers have a large heat input, increasing the degree of damage to the substrate. Therefore, it is suitable for cleaning large steel structures, pipelines, etc., which are bulky, have fast heat dissipation, and do not require high tolerance for damage to the substrate.

[0078] Short-pulse laser (nanoseconds to hundreds of nanoseconds) cleaning: The mechanical effects of laser on materials are quite pronounced. Its main decontamination mechanism is the shock wave effect and thermoelastic effect of plasma. Pulsed lasers can control heat input and prevent the substrate temperature from becoming too high or causing micro-melting. It is suitable for applications that require strict control of substrate temperature rise and require the substrate to remain undamaged. A high-repetition-rate short-pulse laser beam acts on the surface of the workpiece and is absorbed by the rust layer, paint layer, and contamination layer, forming a rapidly expanding plasma (a highly ionized unstable gas). At the same time, it generates a shock wave, which breaks the contaminants into fragments and removes them.

[0079] The key indicator determining the effectiveness of laser rust removal is the "peak power density." Below the damage threshold, a higher peak power density results in better rust removal. Sufficient energy density is required to ensure that surface contaminants absorb enough energy to reach their vaporization temperature. Peak power density is a crucial indicator of laser cleaning effectiveness; referring to the process parameters for laser rust and paint removal on carbon steel surfaces, its peak power density ranges from 10⁷ to 10⁹ W / cm². 2 (100ns); Based on experience, if it is only surface oil stains and the surface oxide layer does not need to be removed, the cleaning threshold can be 1-2 levels lower; The peak power density range required for this embodiment is 10⁵-10⁷ W / cm². 2 (100ns).

[0080] A method for optimizing laser cleaning of product containers in a uranium enrichment plant, the method being implemented based on any of the laser cleaning optimization systems for product containers in a uranium enrichment plant, comprising the following steps:

[0081] S1. Install scanning mechanisms at the valve 2 and plug 3 of the container to be cleaned 1 respectively, and debug and calibrate the scanning mechanisms. The scanning mechanism includes a focusing lens group 4, a galvanometer 5, a scanning wedge 6 and an optical window. The beam input of the focusing lens group 4 is connected to an external collimation output device. The beam output of the focusing lens group 4 is calibrated and controlled for the first time through two galvanometers 5. The beam output of the galvanometers 5 is calibrated and controlled for the second time through two scanning wedges 6. The beam output of the scanning wedges 6 is output through the optical window.

[0082] S2. Turn on the pulsed laser source. The pulsed laser is transmitted through the flexible armored fiber optic cable to output the laser beam. The position information of the working point on the inner wall of the container to be cleaned is obtained through illumination imaging monitoring.

[0083] S3. The scanning mechanism is recalibrated through centralized control so that the working point position of the inner wall of the container to be cleaned is sequentially scanned at the preset point position;

[0084] In step S1, two sets of scanning mechanisms scan and clean the container 1 to be cleaned. The plug 3 and valve 2 are respectively located on both sides of the container 1 to be cleaned. The two sets of scanning mechanisms scan and clean in a corresponding manner to achieve full coverage scanning of the container 1 to be cleaned. Specifically, the scanning angle range is 0-50° and the scanning accuracy is 0.1.

[0085] Furthermore, step S2 also includes extracting and discharging the dust and fumes generated during the laser action inside the container 1 to be cleaned, maintaining a low negative pressure state inside the container.

[0086] Furthermore, in step S2, the pulsed laser source enters the scanning mechanism after being output through a flexible armored fiber optic cable. The laser beam is then shaped to perform a full-coverage scan of the container 1 to be cleaned, directly acting on the contaminants on the surface of the item to be cleaned, causing the contaminants to burn, vaporize, and decompose. For example, the beam shaping allows for continuous adjustment of the laser beam size as needed, with an output transmission wavelength of 500-1200 nm and a scanning speed range of 1000-20000 mm / s.

[0087] Furthermore, as a preferred embodiment of this invention, laser cleaning of the inner wall of a uranium hexafluoride container is proposed. Specifically, the device includes a pulsed laser module, an optical transmission module, a focusing assembly, a laser beam, an illumination and imaging module, a precision measurement assembly, an extraction system, and a centralized control system. The pulsed laser module is the laser source of the cleaning system, used to output high-power, high-energy, high-repetition-rate pulsed laser light, employing a flexible armored fiber optic cable output method. The focusing assembly is used for beam control of the system, including a focusing lens group, a precision guide rail, a stepper motor, and its control board. It focuses the output laser light from the pulsed laser module to achieve the laser cleaning threshold and adjusts the focal position of the laser beam, enabling large-scale dynamic adjustment and control of the focal position. The illumination and imaging module consists of a 532nm green illumination beam, a narrow-band filter, and a high-speed CCD industrial camera. The precision measurement assembly consists of a measurement light source and a photoelectric sensor. A high-precision laser ranging method is used to provide positional information reference for the scanning system and the focusing assembly, achieving efficient laser cleaning of irregularly shaped inner walls of containers. The extraction system, implemented using a metering pump or vacuum pump, primarily removes and discharges dust and fumes generated during laser processing within the container, maintaining a low negative pressure environment. The optical transmission module, comprising a reflector, dichroic mirror, polarizer, glass slide, high-precision optical frame, and its structural components, enables laser beam connection between various components. It is mainly used for transmitting the laser beam output from the light source. The centralized control system includes a main control card and a control panel. Operators perform laser cleaning operations via a touchscreen control panel. It is primarily used for automated control of the entire system, including centralized control of focusing, scanning, shaping, imaging, and measurement systems, automating the cleaning process.

[0088] Furthermore, as a preferred embodiment of this invention, in step S3, a method for centralized control of parameter configuration optimization for laser cleaning is proposed, which includes the following steps:

[0089] S301. Collect real-time data, including input parameters and output results during the laser cleaning process. The input parameters include at least: laser energy (energy intensity of the laser beam); cleaning speed (speed at which the laser beam moves across the container surface); cleaning time (duration of the cleaning process); container internal conditions (environmental parameters such as temperature and humidity); container initial state (including degree of contamination and coating type); and laser beam position and angle (irradiation position and angle inside the container). The output results include at least: optimal laser energy (model-predicted optimal laser energy under current conditions); and optimal cleaning speed (model-predicted optimal cleaning speed under current conditions).

[0090] S302. Data preprocessing;

[0091] S303. Select a machine learning model based on the nature of the problem, data characteristics, etc.

[0092] S304. Based on the problem and the selected model, perform feature engineering to extract, transform, or select appropriate features for the model.

[0093] For training purposes;

[0094] S305. Train the model using the training set. Optimize the model by adjusting hyperparameters (such as learning rate, tree depth, etc.).

[0095] performance;

[0096] S306. Deploy the trained model into the actual laser cleaning system. During the actual laser cleaning process, input real-time data into the model to obtain the prediction results, i.e., the optimal laser cleaning parameter configuration.

[0097] Specifically, step S303 includes:

[0098] S3031. Input Feature (X): Laser Energy (Unit: Joules)

[0099] X=[100, 150, 200, 250, 300];

[0100] Output (Y): Cleaning effect (unit: percentage)

[0101] Y=[80, 85, 90, 88, 92];

[0102] S3032. A linear regression model will be used to fit the data, specifically:

[0103] Y = β0 + β1·X + ε;

[0104] Wherein, Y represents the predicted value of the cleaning effect; X represents the laser energy; β0 represents the intercept; β1 represents the coefficient of the laser energy; and ε represents the error term.

[0105] S3033. Find the optimal β0 and β1 so that the model's predictions are closest to the actual observations;

[0106] S3034. Use the least squares method as the loss function for the linear regression model, with the goal of minimizing...

[0107] The squared error between the predicted and actual values, specifically:

[0108]

[0109] Wherein, h β (X (i) ) represents the number of samples, m represents the model's prediction value for the i-th sample, and Y (i) This represents the actual value of the i-th sample;

[0110] S3035. Minimize the loss function using the gradient descent algorithm to find the optimal model parameters (slope and intercept). Specifically:

[0111]

[0112]

[0113] Wherein, α represents the learning rate, used to control the step size of parameter updates; and These represent the partial derivatives of the loss function with respect to the parameters;

[0114] S3036. Parameter Update: Through multiple iterations, the parameters are updated until the loss function reaches its minimum, yielding the optimal β0 and β1. Specifically:

[0115]

[0116]

[0117] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An optimized method for laser cleaning of product containers in a uranium enrichment plant, characterized in that, Includes the following steps: S1. Install scanning mechanisms at the valve and plug of the container to be cleaned, and debug and calibrate the scanning mechanisms. The scanning mechanism includes a focusing lens group, a galvanometer, a scanning wedge, and an optical window. The beam input of the focusing lens group is connected to an external collimation output device. The beam output of the focusing lens group is calibrated and controlled for the first time through two sets of galvanometers. The beam output of the galvanometer is calibrated and controlled for the second time through two sets of scanning wedges. The beam output of the scanning wedge is output through the optical window. S2. Turn on the pulsed laser source. The pulsed laser is transmitted through a flexible armored fiber optic cable to output the laser beam. The position information of the working point on the inner wall of the container to be cleaned is obtained through illumination imaging monitoring. S3. The scanning mechanism is recalibrated by centralized control so that the working point position of the inner wall of the container to be cleaned is sequentially scanned at the preset point position; In step S1, the two sets of scanning mechanisms scan and clean the container to be cleaned. The plug and the valve are respectively located on both sides of the container to be cleaned. The container to be cleaned is fully scanned at the plug and the valve by the corresponding spiral scanning method of the two sets of scanning mechanisms. Step S3 further includes a process step of centralized control for parameter configuration optimization of laser cleaning: S301. Collect real-time data, including input parameters and output results during the laser cleaning process, wherein the input parameters include at least laser energy, cleaning speed, cleaning time, container internal conditions, container initial state, laser beam position and angle, and the output results include at least optimal laser energy and optimal cleaning speed; S302. Data preprocessing; S303. Select a machine learning model; select appropriate features through feature engineering for model training; train the model using the training set and optimize model performance by adjusting hyperparameters; deploy the trained model to the actual laser cleaning system, input real-time data into the model during the actual laser cleaning process, and obtain the prediction results, i.e., the optimal laser cleaning parameter configuration. Step S303 specifically includes the following sub-steps: S3031. Input characteristics: Laser energy X = [100, 150, 200, 250, 300]; Output: Cleaning effect; Y=[80， 85， 90， 88， 92]； S3032. Use a linear regression model to fit the data, specifically: ； Among them, the The predicted value of the cleaning effect is expressed as a percentage. The laser energy is expressed in Joules. Represents the intercept, the The coefficient representing laser energy, the Indicates the error term; S3033. Find the best and This ensures that the model's predictions are as close as possible to the actual observed values. S3034. The least squares method is used as the loss function in the linear regression model, with the goal of minimizing the squared error between the predicted and actual values. Specifically: ； Among them, the Indicates the number of samples, the This represents the model's predicted value for the i-th sample. This represents the actual value of the i-th sample; S3035. Minimize the loss function using the gradient descent algorithm to find the optimal model parameters. Specifically: ； ； Among them, the The learning rate is used to control the step size for parameter updates; and These represent the partial derivatives of the loss function with respect to the parameters; S3036. Parameter Update: Through multiple iterations, the parameters are updated until the loss function reaches its minimum, thus obtaining the optimal result. and Specifically: ； 。 2. The laser cleaning optimization method for uranium enrichment plant product containers as described in claim 1, characterized in that, Step S2 also includes extracting and discharging the dust and fumes generated during the laser action inside the container to be cleaned, maintaining a low negative pressure state inside the container.

3. The laser cleaning optimization method for uranium enrichment plant product containers as described in claim 1, characterized in that, The scanning path for the full-coverage scan is as follows: For the plug: the scanning mechanism scans the upper inner wall of the container, the inner wall on the side where the valve is installed, and the lower inner wall of the container in sequence; For the valve: the scanning mechanism scans the upper inner wall of the container, the inner wall on the side with the plug installed, and the lower inner wall of the container in sequence.

4. The laser cleaning optimization method for uranium enrichment plant product containers as described in claim 1, characterized in that, In step S2, the pulsed laser source enters the scanning mechanism after being output through a flexible armored optical fiber cable, and the laser beam is adjusted by beam shaping to perform a full-coverage scan of the container to be cleaned, directly acting on the dirt on the surface of the container to be cleaned to cause the dirt to burn, vaporize and decompose.

5. A laser cleaning optimization system for uranium enrichment plant product containers, the system being implemented based on the laser cleaning optimization method for uranium enrichment plant product containers according to any one of claims 1-4, comprising a laser cleaning subsystem and a container to be cleaned, wherein the laser cleaning subsystem cleans the interior of the container to be cleaned by laser irradiation with a laser beam, characterized in that: The laser cleaning subsystem is located outside the container to be cleaned. The laser beam of the laser cleaning subsystem has two outputs, which are respectively located at the valve and the plug of the container to be cleaned. The valve and the plug are respectively located on both sides of the container to be cleaned. The laser cleaning subsystem includes two sets of scanning mechanisms. Each scanning mechanism includes a focusing lens group, a galvanometer, a scanning wedge, and an optical window. The beam input of the focusing lens group is connected to an external collimation output device. The beam output of the focusing lens group is first calibrated and controlled by the two sets of galvanometers. The beam output of the galvanometer is second calibrated and controlled by the two sets of scanning wedges. The beam output of the scanning wedge is output and scanned through the optical window. The laser cleaning subsystem also includes a distance measurement module, which measures the distance information of the working point position. The scanning mechanism adjusts and controls the focal point position based on the distance information of the working point position collected by the distance measurement module.

6. The laser cleaning optimization system for uranium enrichment plant product containers as described in claim 5, characterized in that, The laser cleaning subsystem also includes an illumination imaging module, which is used to perform online image monitoring of the working point position on the inner wall of the container to be cleaned during the laser irradiation cleaning process.

7. The laser cleaning optimization system for uranium enrichment plant product containers as described in claim 5, characterized in that, The laser cleaning subsystem also includes a centralized control module, which is used to automatically control the system and optimize the parameters of laser cleaning through centralized control software. The automated control includes centralized control of the focusing, scanning, shaping, imaging and measurement systems.

8. The laser cleaning optimization system for uranium enrichment plant product containers as described in claim 5, characterized in that, The laser cleaning subsystem also includes a pulsed laser source module, which is used to output pulsed laser light and transmits the output laser beam through a flexible armored optical fiber cable.

9. The laser cleaning optimization system for uranium enrichment plant product containers as described in claim 5, characterized in that, The laser cleaning subsystem also includes an extraction module, which is used to extract and discharge the dust and fumes generated during the laser action inside the container to be cleaned, maintaining a low negative pressure state inside the container.

Images