A rapid detection method for steel plate sandblasting rust removal based on retroreflection coefficient

Abstract

The present invention relates to the technical field of sandblasting and rust removal detection of steel plates on steel bridge decks, and in particular to a rapid detection method for sandblasting and rust removal of steel plates based on retroreflection coefficient. The method comprises the following steps: using a high-resolution camera and a laser scanner to obtain the surface geometric information and the initial retroreflection coefficient of the steel plate; using a vehicle-mounted shot blasting machine to sandblast and remove rust, and recording the sandblasting time, pressure and medium consumption; using an retroreflection coefficient measuring instrument and multi-spectral imaging technology to measure the retroreflection coefficient after treatment; comparing and analyzing the initial and processed data, using a machine learning algorithm to evaluate the rust removal effect, and generating a quality report; according to the quality report, adaptively adjusting the sandblasting parameters; using augmented reality technology to visualize and compare the data, and generate a three-dimensional effect diagram for display. The present invention realizes efficient, accurate, and automated detection and optimization of the sandblasting and rust removal process of steel plates on steel bridge decks, improves the rust removal quality, ensures the surface treatment effect of steel plates on steel bridge decks, and improves the overall work efficiency.

Description

Technical Field

[0001] The invention relates to the technical field of sandblasting and rust removal detection of steel bridge deck steel plates, in particular to a rapid detection method for sandblasting and rust removal of steel plates based on retroreflection coefficient. Background Art

[0002] At present, a large number of large-span steel box girder bridges have been built and put into use in China. Most of the bridge deck pavement layers have experienced varying degrees of displacement and delamination before reaching the design service life. The main cause of the disease is the insufficient bonding between the steel plate and the waterproof bonding layer. The quality of steel plate sandblasting and rust removal has an extremely important influence on the bonding force. Therefore, it is necessary to do a good job in steel plate sandblasting and rust removal detection and strictly control the quality of steel plate sandblasting and rust removal construction. At present, rust removal detection has the following problems: the traditional steel plate sandblasting and rust removal detection method relies on manual inspection and simple measuring tools, with low detection efficiency and insufficient accuracy, making it difficult to provide detailed surface quality information, which easily leads to unstable rust removal quality; the evaluation of rust removal effect and surface quality mostly relies on manual experience, lacks systematic data processing and evaluation methods, and is difficult to achieve fast and accurate quality evaluation and feedback. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a rapid detection method for steel plate sandblasting and rust removal based on retroreflection coefficient.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A rapid detection method for steel plate sandblasting rust removal based on retroreflection coefficient comprises the following steps:

[0006] S1: Use high-resolution cameras and laser scanners to conduct preliminary assessment of the steel plate surface to obtain geometric information and initial retroreflection coefficient data of the steel bridge deck surface;

[0007] S2: Use a mobile shot blasting machine to blast the steel bridge deck to remove rust, and record the parameters of the blasting process in real time, including blasting time, blasting pressure and media consumption;

[0008] S3: Use a retroreflection coefficient measuring instrument combined with multispectral imaging technology to quickly measure the surface of the treated steel plate to obtain the retroreflection coefficient data after rust removal;

[0009] S4: Compare and analyze the initial retroreflection coefficient data with the retroreflection coefficient data after rust removal, use machine learning algorithms to perform data mining and processing, automatically evaluate the rust removal effect and surface quality, and generate a quality assessment report;

[0010] S5: Based on the quality assessment report, the blasting parameters, including blasting time, pressure and media type, are dynamically adjusted through an adaptive optimization algorithm;

[0011] S6: Using augmented reality technology, the initial retroreflection coefficient data and the data after rust removal are visualized for comparative analysis, and a three-dimensional rendering is generated through an interactive interface to be displayed to relevant personnel.

[0012] Furthermore, in step S1, the geometric information includes the roughness, flatness and corrosion distribution of the surface of the steel deck steel plate.

[0013] Furthermore, in step S3, the operation process of the multispectral imaging technology includes the following steps:

[0014] Multispectral imaging technology is used to obtain multispectral image data of the steel plate surface, which includes multiple spectral channels and covers different bands from visible light to near infrared;

[0015] Preprocess the collected multispectral image data, including noise removal, spectral correction and image registration;

[0016] Perform feature extraction on the preprocessed multispectral image, use image processing algorithms to extract surface texture, color differences and spectral reflectance characteristics, and use principal component analysis or independent component analysis techniques to reduce data dimensions;

[0017] Combined with the data from the retroreflection coefficient measuring instrument, the extracted characteristic information is analyzed to calculate the retroreflection coefficient of each spectral channel;

[0018] Through multispectral data fusion technology, the retroreflection coefficients of different bands are integrated to obtain a comprehensive retroreflection coefficient value.

[0019] Furthermore, the calculation formula of the comprehensive retroreflection coefficient value is as follows:

[0020]

[0021] Among them, R c Represents the comprehensive retroreflection coefficient value; R i represents the retroreflection coefficient of the i-th spectral channel; w i represents the weight factor of the i-th spectral channel; T i represents the sandblasting time corresponding to the i-th spectral channel; T max represents the maximum sandblasting time; α represents the attenuation coefficient of the effect of sandblasting time change on the retroreflection coefficient; P i represents the sandblasting pressure corresponding to the i-th spectral channel; P max represents the maximum sandblasting pressure; β represents the proportional coefficient of the effect of sandblasting pressure on the retroreflection coefficient; C i represents the medium consumption corresponding to the i-th spectral channel; γ represents the proportional coefficient of the influence of medium consumption on the retroreflection coefficient; n represents the number of spectral channels.

[0022] Furthermore, the step S4 specifically includes the following steps:

[0023] The acquired initial retroreflection coefficient data and the retroreflection coefficient data after rust removal are input into a multivariate adaptive regression spline model to capture the nonlinear relationship of the data;

[0024] Use principal component regression to extract features from input data and identify key features of rust removal effects;

[0025] The time series data were processed using time series decomposition algorithm to analyze the effect of parameter changes during sandblasting on the retroreflection coefficient.

[0026] Perform multi-model fusion of feature extraction and time series analysis results to generate comprehensive evaluation results;

[0027] Based on the comprehensive evaluation results, the model is optimized using kernel ridge regression, and a quality assessment report is generated, including rust removal effect, surface quality assessment and parameter adjustment suggestions.

[0028] Furthermore, the formula of the multivariate adaptive regression spline model is as follows:

[0029]

[0030] Where f(x) represents the predicted evaluation value of the rust removal effect of the steel plate; x represents the input variables, including the initial retroreflection coefficient data and the retroreflection coefficient data after rust removal, as well as the parameters in the sandblasting process; β0 represents the constant term of the model, which is the predicted value when all input variables are zero; β m is the weight of each basis function, indicating the contribution of the basis function to the final prediction value; h m (x) represents the basis function, which is used to capture the nonlinear relationship between the input variable x and the target value f(x); M represents the number of basis functions.

[0031] Furthermore, the formula of the kernel ridge regression is as follows:

[0032]

[0033] in, represents the comprehensive quality evaluation value predicted based on the input data; K is the kernel matrix, which represents the similarity matrix of the input data; λ represents the regularization parameter, which is used to control the complexity of the model; I is the unit matrix; y represents the actually measured inverse reflection coefficient data.

[0034] Furthermore, the adaptive optimization algorithm includes a particle swarm optimization algorithm and / or a genetic algorithm.

[0035] Furthermore, the augmented reality technology is implemented through an application program of a mobile device or tablet computer, combined with a GPS positioning function, and the three-dimensional rendering is displayed to the user through an interactive interface.

[0036] The beneficial effects of the present invention are:

[0037] The present invention uses a high-resolution camera and a laser scanner to preliminarily evaluate the geometric information and retroreflection coefficient data of the steel plate surface, thereby realizing a rapid and accurate evaluation of the state of the steel plate, and providing a reliable data basis for subsequent sandblasting and rust removal. During the sandblasting and rust removal process, the vehicle-mounted mobile shot blasting machine can record parameters such as sandblasting time, pressure, and medium consumption in real time, realizing comprehensive monitoring and fine control of the sandblasting process, and helping to ensure the stability of the rust removal quality. By using a retroreflection coefficient meter combined with multi-spectral imaging technology, the surface of the treated steel plate is quickly measured to obtain detailed retroreflection coefficient data. Combined with machine learning algorithms for data analysis, automated evaluation of rust removal effects and surface quality is realized, greatly improving detection efficiency and accuracy. Based on the quality assessment report, the sandblasting parameters such as sandblasting time, pressure, and medium type are dynamically adjusted through an adaptive optimization algorithm, realizing intelligent regulation of the sandblasting process, ensuring the best rust removal effect and surface quality. Using augmented reality technology, the initial retroreflection coefficient data and the data after rust removal are visualized for comparative analysis, and a three-dimensional rendering is generated through an interactive interface, so that relevant personnel can intuitively understand the rust removal effect and surface quality changes, improving the scientificity and convenience of management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The present invention is a schematic flow chart of a method for rapid detection of steel plate sandblasting and rust removal based on retroreflection coefficient.

[0039] Figure 2 It is a flowchart of step S3 provided in one embodiment of the present invention.

[0040] Figure 3 It is a flowchart diagram of step S4 provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0041] See also Figure 1-3 As shown, the present invention relates to a rapid detection method for steel plate sandblasting and rust removal based on retroreflection coefficient.

[0042] Example

[0043] A rapid detection method for steel plate sandblasting rust removal based on retroreflection coefficient comprises the following steps:

[0044] S1: Use a high-resolution camera and a laser scanner to perform a preliminary assessment of the steel plate surface to obtain geometric information and initial retroreflection coefficient data of the steel bridge deck steel plate surface; the geometric information includes the roughness, flatness and rust distribution of the steel bridge deck steel plate surface.

[0045] Specifically, select suitable high-resolution cameras (e.g., industrial cameras with more than 50 million pixels) and laser scanners (e.g., 3D laser scanners with micron-level accuracy). Ensure that the equipment is in normal working condition and perform necessary calibration and inspection to ensure measurement accuracy. Select suitable installation points at different locations on the steel bridge deck to ensure that all areas of the steel plate surface can be covered. For large steel plate surfaces, multiple cameras and scanners can be used to work together to ensure the comprehensiveness and accuracy of the data. Adjust the shooting angle and distance of the high-resolution camera so that it can capture high-definition images of the steel plate surface and avoid interference from reflected light. Determine the scanning range and distance of the laser scanner to ensure that the scanning beam can evenly cover the steel plate surface.

[0046] Start a high-resolution camera to take all-round photos of the steel plate surface of the steel bridge deck to obtain high-definition images from multiple angles. During the image acquisition process, ensure that there are no obstructions or moving interference on the steel plate surface to ensure image quality. Start a laser scanner to scan the steel plate surface with high precision and generate a detailed three-dimensional geometric model. During the scanning process, gradually move the scanner to cover the entire steel plate surface and record the three-dimensional coordinate data of each point. Use image processing software to analyze the collected high-definition images and extract the roughness and rust distribution of the steel plate surface. Use three-dimensional modeling software to process the laser scanning data, generate a three-dimensional geometric model of the steel plate surface, and calculate parameters such as surface flatness and roughness.

[0047] Measure different areas of the steel plate surface and record the initial retroreflection coefficient data of each area. Make multiple measurements to ensure the reliability of the data, and take the average value as the final initial retroreflection coefficient data. Match and store the geometric information with the initial retroreflection coefficient data to establish a detailed initial data archive. Use a database system to save all measurement data for subsequent comparative analysis and quality assessment.

[0048] S2: Use a mobile shot blasting machine to blast the steel bridge deck to remove rust, and record the parameters of the blasting process in real time, including blasting time, blasting pressure and media consumption;

[0049] It should be noted that the vehicle-mounted mobile shot blasting machine selected by the system is BLASTRAC 2-4800DHMKV, which includes a shot blasting wheel, an engine and a vacuum cleaner. It can clean a 1.22-meter-wide path at one time and load 544.31 kg of steel shot, which is suitable for the pretreatment of all horizontal surfaces. According to the preliminary evaluation results of the steel plate, set the initial parameters such as sandblasting time, sandblasting pressure and medium consumption. Select suitable sandblasting media (such as steel shot or sand) to ensure that its particle size and hardness meet the requirements.

[0050] Start the shot blasting machine for rust removal and monitor the blasting process in real time through the control system. Use built-in sensors to record parameters such as blasting time, pressure and media consumption, and upload the data to the central control system in real time. Set the blasting time for each area to ensure that the blasting time is long enough to remove rust but not damage the steel plate surface. Use a timer to accurately record the blasting time and automatically control the end of blasting. Dynamically adjust the blasting pressure according to the degree of rust in different areas to ensure the best blasting effect. Monitor the blasting pressure in real time through the pressure sensor, and feed the data back to the control system to automatically adjust the blasting pressure. Record the amount of media consumed during each blasting process to ensure sufficient media supply and avoid interruptions during the blasting process. Monitor the media consumption in real time through the weight sensor, and upload the data to the central control system for storage and analysis. During the blasting process, use cameras and sensors to monitor the state of the steel plate surface in real time to ensure that the rust is effectively removed. When it is detected that the surface has achieved the expected effect, the blasting is automatically stopped to avoid over-processing. All parameter data recorded during the blasting process are stored in the central database to form a complete blasting process data archive. The data includes blasting time, pressure, media consumption and surface condition monitoring results.

[0051] S3: Use a retroreflection coefficient measuring instrument combined with multispectral imaging technology to quickly measure the surface of the treated steel plate to obtain the retroreflection coefficient data after rust removal;

[0052] Wherein, in step S3, the operation process of the multispectral imaging technology includes the following steps:

[0053] Multispectral imaging technology is used to obtain multispectral image data of the steel plate surface, which includes multiple spectral channels and covers different bands from visible light to near infrared;

[0054] Specifically, a high-resolution multispectral imager is used to cover different bands from visible light to near infrared. The specific bands include multiple spectral channels such as visible light (400-700nm) and near infrared (700-1000nm). The multispectral imager is fixed on a stable platform and aimed at the treated steel plate surface. Make sure that the optical axis of the imager is perpendicular to the steel plate surface to reduce image distortion. Start the imager and collect multispectral image data of the steel plate surface. Each spectral channel generates an image, which together constitute a multispectral image dataset.

[0055] Preprocess the collected multispectral image data, including noise removal, spectral correction and image registration;

[0056] Perform feature extraction on the preprocessed multispectral image, use image processing algorithms to extract surface texture, color differences and spectral reflectance characteristics, and use principal component analysis or independent component analysis techniques to reduce data dimensions;

[0057] It should be noted that surface texture extraction: image processing algorithms (such as gradient directional histogram, Gabor filter, etc.) are used to extract texture features of the steel plate surface.

[0058] Color difference extraction: Calculate the color difference on each spectral channel and extract color features using color space conversion (such as HSV, Lab, etc.).

[0059] Spectral reflectance characteristic extraction: Calculate the reflectance of each spectral channel and use the spectral reflectance curve to analyze the spectral reflectance characteristics of the steel plate surface.

[0060] Dimensionality reduction: Use principal component analysis (PCA) or independent component analysis (ICA) technology to perform dimensionality reduction on the extracted feature data to reduce data dimensions and improve computing efficiency.

[0061] Combined with the data from the retroreflection coefficient measuring instrument, the extracted characteristic information is analyzed to calculate the retroreflection coefficient of each spectral channel;

[0062] It should be noted that the retroreflection coefficient measuring instrument is a BJZJ-R03 retroreflection measuring instrument with an incident angle of -4°, an observation angle of 0.2°, a measuring area of ​​φ32mm, a measuring range of 0 to 2000CD·LX-1·M-2, and a light source color temperature of 2856±50K (standard A light source), which is suitable for measuring the photometric properties of retroreflective materials.

[0063] Through multispectral data fusion technology, the retroreflection coefficients of different bands are integrated to obtain a comprehensive retroreflection coefficient value.

[0064] Furthermore, the calculation formula of the comprehensive retroreflection coefficient value is as follows:

[0065]

[0066] Among them, R c Represents the comprehensive retroreflection coefficient value; R i represents the retroreflection coefficient of the i-th spectral channel; w i represents the weight factor of the i-th spectral channel; T i represents the sandblasting time corresponding to the i-th spectral channel; T max represents the maximum sandblasting time; α represents the attenuation coefficient of the effect of sandblasting time change on the retroreflection coefficient; P i represents the sandblasting pressure corresponding to the i-th spectral channel; P max represents the maximum sandblasting pressure; β represents the proportional coefficient of the effect of sandblasting pressure on the retroreflection coefficient; C i represents the medium consumption corresponding to the i-th spectral channel; γ represents the proportional coefficient of the influence of medium consumption on the retroreflection coefficient; n represents the number of spectral channels.

[0067] S4: Compare and analyze the initial retroreflection coefficient data with the retroreflection coefficient data after rust removal, use machine learning algorithms to perform data mining and processing, automatically evaluate the rust removal effect and surface quality, and generate a quality assessment report;

[0068] Wherein, the step S4 specifically includes the following steps:

[0069] The acquired initial retroreflection coefficient data and the retroreflection coefficient data after rust removal are input into a multivariate adaptive regression spline model to capture the nonlinear relationship of the data;

[0070] Use principal component regression to extract features from input data and identify key features of rust removal effects;

[0071] The time series data were processed using time series decomposition algorithm to analyze the effect of parameter changes during sandblasting on the retroreflection coefficient.

[0072] Perform multi-model fusion of feature extraction and time series analysis results to generate comprehensive evaluation results;

[0073] Specifically, the parameters (time, pressure, and medium consumption) in the sandblasting process are processed using a time series decomposition algorithm (such as STL decomposition) to analyze their impact on the retroreflection coefficient. The decomposed time series data provides the trend, seasonality, and residual of the sandblasting process parameter changes. The features extracted by PCR and the results of time series decomposition are fused into multiple models to generate a comprehensive evaluation result. The fusion model can comprehensively consider the information from different data sources and improve the accuracy of the evaluation.

[0074] Based on the comprehensive evaluation results, the model is optimized using kernel ridge regression, and a quality assessment report is generated, including rust removal effect, surface quality assessment and parameter adjustment suggestions.

[0075] It should be noted that the rust removal effect: The report includes a comparison of the initial retroreflection coefficient and the retroreflection coefficient after rust removal, showing the rust removal effect.

[0076] Surface quality assessment: Based on the comprehensive evaluation results of multi-model fusion and KRR, a surface quality assessment is generated, including surface smoothness, rust residue, etc.

[0077] Parameter adjustment suggestions: Based on the model analysis results, optimization suggestions for sandblasting parameters (time, pressure, media type) are put forward to help improve the quality of future rust removal.

[0078] Furthermore, the formula of the multivariate adaptive regression spline model is as follows:

[0079]

[0080] Wherein, f(x) represents the predicted evaluation value of the rust removal effect of the steel plate; x represents the input variables, including the initial retroreflection coefficient data and the retroreflection coefficient data after rust removal, as well as the parameters in the sandblasting process, including: initial retroreflection coefficient data, retroreflection coefficient data after rust removal, sandblasting time, sandblasting pressure and medium consumption; β0 represents the constant term of the model, which is the predicted value when all input variables are zero; β m is the weight of each basis function, indicating the contribution of the basis function to the final prediction value; h m (x) represents the basis function, which is used to capture the nonlinear relationship between the input variable x and the target value f(x); M represents the number of basis functions;

[0081] The formula for the kernel ridge regression is as follows:

[0082]

[0083] in, represents the comprehensive quality evaluation value predicted based on the input data; K is the kernel matrix, which represents the similarity matrix of the input data; λ represents the regularization parameter, which is used to control the complexity of the model; I is the unit matrix; y represents the actual measured inverse reflection coefficient data;

[0084] It should be noted that K is the kernel matrix, which represents the similarity matrix of the input data. Specifically, each element K in the kernel matrix ij It can be calculated by the kernel function:

[0085] Among them, x i and x j respectively Data points include the following: initial retroreflectivity data, retroreflectivity data after rust removal, blasting time, blasting pressure, media consumption, geometric information such as roughness, flatness and rust distribution.

[0086] S5: Based on the quality assessment report, dynamically adjust the sandblasting parameters, including sandblasting time, pressure and medium type, through an adaptive optimization algorithm; the adaptive optimization algorithm includes a particle swarm optimization algorithm and / or a genetic algorithm.

[0087] In one embodiment, key data are extracted from the quality assessment report, including blasting time (T), blasting pressure (P), media type (M), rust removal effect (E) and surface quality (Q). These data will be used as input and applied to the particle swarm optimization algorithm, including the following steps:

[0088] Each particle represents a possible blasting parameter combination, including blasting time (T), blasting pressure (P) and media type (M). The particle group is initialized, including the position (i.e., blasting parameter combination) and velocity of the particle.

[0089] The fitness function is defined as a comprehensive evaluation function of rust removal effect (E) and surface quality (Q):

[0090] F = w1·E+w2·Q;

[0091] Among them, w1 and w2 are weight coefficients in the fitness function, reflecting the importance of rust removal effect and surface quality.

[0092] Update particle positions and velocities:

[0093] Particle velocity update formula:

[0094] v i (t+1)=w·v i (t)+c1·r1·(p best，i -x i (t))+c2·r2·(g best -x i (t))

[0095] Among them, v i (t+1) represents the velocity of particle i at time t+1, i.e., the updating speed of the sandblasting parameters; v i (t) represents the velocity of particle i at time t; x i (t) represents the position of particle i at time t; p best，i represents the best historical position of particle i; g bestrepresents the global best position, that is, the best sandblasting parameter combination among all particles; w represents the inertia weight; c1 and c2 represent the learning factors, which represent the particle's ability to learn its own historical best position and the global best position respectively; r1 and r2 represent random numbers;

[0096] Particle position update formula:

[0097] x i (t+1)=x i (t)+v i (t+1)

[0098] Among them, x i (t+1) represents the position of particle i at time t+1;

[0099] Through multiple iterations, the particles gradually tend to the optimal sandblasting parameter combination. The termination condition is that the preset maximum number of iterations is reached or the fitness value is no longer significantly improved.

[0100] S6: Using augmented reality technology, the initial retroreflection coefficient data and the data after rust removal are visualized for comparative analysis, and a three-dimensional rendering is generated through an interactive interface and displayed to relevant personnel; the augmented reality technology is implemented through an application on a mobile device or tablet computer, combined with a GPS positioning function, and the three-dimensional rendering is displayed to the user through an interactive interface.

[0101] Specifically, it is equipped with high-performance mobile devices or tablets that support AR functions and have specific AR applications installed. The application needs to be able to access the initial and derusted retroreflectivity data stored in the database and have 3D modeling and rendering capabilities.

[0102] Use graphics processing technology to generate a 3D model of the initial and post-processing states of the steel plate surface. During the model generation process, the retroreflection coefficient data will be mapped to the 3D model surface and displayed as color changes or other visualization effects. The initial 3D model shows the state of the steel plate before rust removal, including surface roughness, rust distribution, and retroreflection coefficient distribution. The 3D model after rust removal shows the state of the steel plate after treatment, highlighting the change in retroreflection coefficient, smooth surface and rust-free areas.

[0103] Users can interact with the 3D model by touching the screen or using gestures. For example, clicking on an area can display detailed retroreflectivity data and surface quality information for that area. The application provides multiple viewing options, including viewing the initial state, the treated state, or a comparison of the two. Users can switch views freely to gain a deeper understanding of the rust removal effect. The AR application provides data analysis tools, and users can view the trend chart of the retroreflectivity in real time. The chart shows the comparison of the initial and treated retroreflectivity values ​​of each measurement point to help users evaluate the rust removal effect. Users can choose different color schemes and display modes to more clearly observe and analyze the changes on the steel plate surface. The AR application has a report generation function, and users can generate detailed PDF or image reports based on the results of the visual comparison analysis. The report includes a 3D rendering, a retroreflectivity change chart, and a summary of key data. The generated report can be shared with relevant personnel by email or other means for subsequent quality inspection and decision support.

[0104] In summary, the present invention uses a high-resolution camera and a laser scanner to perform a preliminary assessment of the steel plate surface to obtain detailed geometric information and initial retroreflection coefficient data. Multi-angle high-definition images and high-precision three-dimensional geometric models ensure the comprehensiveness and accuracy of the data and improve the efficiency and accuracy of surface state assessment. A vehicle-mounted mobile shot blasting machine is used for sandblasting and rust removal, and parameters such as sandblasting time, sandblasting pressure, and medium consumption are recorded in real time. Through real-time monitoring of the control system, the controllability and consistency of the sandblasting process are ensured, human errors are reduced, and the rust removal effect is improved.

[0105] The present invention combines a retroreflection coefficient measuring instrument and multispectral imaging technology to quickly obtain the retroreflection coefficient data of the treated steel plate surface. Through multispectral data fusion technology, the retroreflection coefficients of each spectral channel are integrated to obtain a more accurate comprehensive retroreflection coefficient value, thereby improving the measurement accuracy and efficiency. The initial and processed retroreflection coefficient data are compared and analyzed using a machine learning algorithm to automatically evaluate the rust removal effect and surface quality. Through comprehensive analysis of a multivariate adaptive regression spline model and a kernel ridge regression model, a quality assessment report is generated to provide detailed rust removal effect and surface quality information. Based on the quality assessment report, an adaptive optimization algorithm is used to dynamically adjust the sandblasting parameters. The particle swarm optimization algorithm gradually optimizes the sandblasting time, pressure, and medium type through iterative updates to ensure that each sandblasting and rust removal treatment achieves the best effect, thereby improving the efficiency and quality of the sandblasting process.

[0106] The present invention uses augmented reality technology to visualize and compare the initial and processed retroreflection coefficient data, and generates a three-dimensional rendering to show to relevant personnel. Users can intuitively understand the changes on the surface of the steel plate through the interactive interface, which improves the readability and analysis effect of the data and enhances decision support. The AR application has a report generation function, and users can generate a detailed report based on the results of the visual comparison analysis, which is convenient for sharing and subsequent quality inspection. The report contains a summary of key data and a three-dimensional rendering, providing comprehensive quality assessment information.

[0107] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A rapid detection method for steel plate sandblasting and rust removal based on retroreflection coefficient, characterized in that: The following steps are involved: S1: Use high-resolution cameras and laser scanners to conduct preliminary assessment of the steel plate surface to obtain geometric information and initial retroreflection coefficient data of the steel bridge deck surface; S2: Use a mobile shot blasting machine to blast the steel bridge deck to remove rust, and record the parameters of the blasting process in real time, including blasting time, blasting pressure and media consumption; S3: Use a retroreflection coefficient measuring instrument combined with multispectral imaging technology to quickly measure the surface of the treated steel plate to obtain the retroreflection coefficient data after rust removal; S4: Compare and analyze the initial retroreflection coefficient data with the retroreflection coefficient data after rust removal, use machine learning algorithms to perform data mining and processing, automatically evaluate the rust removal effect and surface quality, and generate a quality assessment report; S5: Based on the quality assessment report, the blasting parameters, including blasting time, pressure and media type, are dynamically adjusted through an adaptive optimization algorithm; S6: Using augmented reality technology, the initial retroreflection coefficient data and the data after rust removal are visually compared and analyzed, and a three-dimensional rendering is generated through an interactive interface to be displayed to relevant personnel; In step S3, the operation process of the multispectral imaging technology includes the following steps: Multispectral imaging technology is used to obtain multispectral image data of the steel plate surface, which includes multiple spectral channels and covers different bands from visible light to near infrared; Preprocess the collected multispectral image data, including noise removal, spectral correction and image registration; Perform feature extraction on the preprocessed multispectral image, use image processing algorithms to extract surface texture, color differences and spectral reflectance characteristics, and use principal component analysis or independent component analysis techniques to reduce data dimensions; Combined with the data from the retroreflection coefficient measuring instrument, the extracted characteristic information is analyzed to calculate the retroreflection coefficient of each spectral channel; Through multispectral data fusion technology, the retroreflection coefficients of different bands are integrated to obtain a comprehensive retroreflection coefficient value; The calculation formula of the comprehensive retroreflection coefficient value is as follows: ； in, Indicates the comprehensive retroreflection coefficient value; represents the retroreflection coefficient of the i-th spectral channel; represents the weight factor of the i-th spectral channel; represents the sandblasting time corresponding to the i-th spectral channel; Indicates the maximum blasting time; The attenuation coefficient representing the effect of the change in blasting time on the retroreflection coefficient; represents the sandblasting pressure corresponding to the i-th spectral channel; Indicates the maximum blasting pressure; The proportionality factor representing the effect of blasting pressure on the retroreflection coefficient; represents the medium consumption corresponding to the i-th spectral channel; It represents the proportional coefficient of the effect of medium consumption on the retroreflection coefficient; n represents the number of spectral channels; Step S4 specifically includes the following steps: The acquired initial retroreflection coefficient data and the retroreflection coefficient data after rust removal are input into a multivariate adaptive regression spline model to capture the nonlinear relationship of the data; Use principal component regression to extract features from input data and identify key features of rust removal effects; The time series data were processed using time series decomposition algorithm to analyze the effect of parameter changes during sandblasting on the retroreflection coefficient. Perform multi-model fusion of feature extraction and time series analysis results to generate comprehensive evaluation results; Based on the comprehensive evaluation results, the model is optimized using kernel ridge regression, and a quality assessment report is generated, including rust removal effect, surface quality assessment, and parameter adjustment suggestions; The formula of the multivariate adaptive regression spline model is as follows: ； in, represents the predicted evaluation value of the rust removal effect of the steel plate; x represents the input variables, including the initial retroreflection coefficient data and the retroreflection coefficient data after rust removal, as well as the parameters in the sandblasting process; Represents the constant term of the model, the predicted value when all input variables are zero; is the weight of each basis function, indicating the contribution of the basis function to the final prediction value; Represents the basis function, which is used to capture the input variable x and the target value The nonlinear relationship between them; M represents the number of basis functions; The formula for the kernel ridge regression is as follows: ； in, represents the comprehensive quality evaluation value predicted based on the input data; K is the kernel matrix, which represents the similarity matrix of the input data; represents the regularization parameter, which is used to control the complexity of the model; I is the unit matrix; y represents the actually measured inverse reflection coefficient data.

2. A rapid detection method for steel plate sandblasting rust removal based on retroreflection coefficient according to claim 1, characterized in that: In step S1, the geometric information includes the roughness, flatness and corrosion distribution of the surface of the steel deck steel plate.

3. A rapid detection method for steel plate sandblasting and rust removal based on retroreflection coefficient according to claim 1, characterized in that: The adaptive optimization algorithm includes a particle swarm optimization algorithm and / or a genetic algorithm.

4. The method for rapid detection of steel plate sandblasting and rust removal based on retroreflection coefficient according to claim 1, characterized in that: The augmented reality technology is implemented through an application program on a mobile device or tablet computer, combined with a GPS positioning function, and a three-dimensional rendering is displayed to the user through an interactive interface.

Images