Metal waste detection and identification method and system

Abstract

The invention relates to the field of metal waste detection, in particular to a metal waste detection and identification method and system. The method comprises the following steps: applying high-frequency ultrasonic waves to a waste area to be detected, and collecting high-frequency ultrasonic wave response signals; performing metal vibration mode parameter mining on the high-frequency ultrasonic response signal to obtain metal vibration mode characteristics in the region; performing ultrasonic excitation path identification on the high-frequency ultrasonic response signal, performing ultrasonic spatial propagation modeling, and constructing a spatial ultrasonic trajectory propagation model; carrying out sound beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration modal characteristics in the region, and carrying out propagation error compensation calculation to obtain a response beam propagation error compensation value; and performing ultrasonic triangulation positioning measurement on the space ultrasonic trajectory propagation model to obtain accurate three-dimensional coordinates of the metal wastes. According to the invention, efficient and accurate detection, identification and positioning of the metal wastes are realized.

Description

Technical Field

[0001] The present invention relates to the field of metal waste detection, and in particular to a metal waste detection and identification method and system. Background Art

[0002] With the continuous advancement of industrialization, the recycling and treatment of metal waste has become increasingly important. Metal waste is generated in the process of industrial production, building demolition, aging of mechanical equipment, etc. It is of various types and forms. In traditional waste recycling and treatment, how to efficiently and accurately identify and locate metal waste has become a technical problem that needs to be solved urgently. The treatment of metal waste is not only related to the reuse of resources, but also directly affects environmental protection and the sustainable use of resources. Therefore, the accurate detection, identification and classification of metal waste is an important task that cannot be ignored in the field of modern industry and environmental protection.

[0003] Traditional metal waste detection methods often rely on manual visual inspection, mechanical contact, and visual imaging. These methods have certain limitations. The first is low efficiency. Manual detection not only requires a lot of manpower, but is also prone to human errors. Secondly, traditional methods are difficult to accurately identify metal waste with irregular surface morphology and diverse material properties. In addition, conventional detection equipment often cannot guarantee good performance under harsh working environments, such as high temperature, dust, and moisture. In practical applications, the types and forms of metal waste are complex and varied. How to extract effective information from complex vibration signals and accurately identify the type and location of metal waste is still a technical problem. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method and system for detecting and identifying metal wastes to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a method for detecting and identifying metal waste, comprising the following steps: Step S1: applying high-frequency ultrasonic waves to the waste area to be inspected, collecting high-frequency ultrasonic wave response signals; mining the metal vibration modal parameters of the high-frequency ultrasonic wave response signals, thereby obtaining the metal vibration modal characteristics in the area; Step S2: performing ultrasonic excitation path identification on the high-frequency ultrasonic response signal, and performing ultrasonic spatial propagation modeling to construct a spatial ultrasonic trajectory propagation model; Step S3: performing acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the area, and performing propagation error compensation calculation to obtain a response beam propagation error compensation value; Step S4: Perform ultrasonic triangulation measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the accurate three-dimensional coordinates of the metal waste; Step S5: Collect multi-spectral images based on the accurate three-dimensional coordinates of the metal waste, and then perform quantitative deduction of metal components, so as to obtain the quantitative information of the metal waste components; Step S6: Perform metal waste feature classification and enhanced detection learning optimization based on the quantitative information of the metal waste components, so as to construct an intelligent detection model for metal waste.

[0006] By applying high-frequency ultrasonic waves to the waste area, the present invention can effectively stimulate the vibration modes on and inside the metal surface. Different vibration modal characteristics of the metal exhibit different signal characteristics in the acoustic wave reflection, which can help identify the physical and structural properties of the waste. By mining the metal vibration modal parameters, characteristic information of the metal can be obtained, thereby improving the accuracy of identifying the types and properties of waste metals. The propagation path of ultrasonic waves is an important factor affecting the signal accuracy. Path recognition and modeling help clarify the propagation path of acoustic waves from the source point to the detection point, reducing signal attenuation and offset errors during propagation. The spatial ultrasonic trajectory propagation model can accurately describe the propagation characteristics of acoustic waves in the waste area, thereby enhancing the accuracy of subsequent positioning and component analysis. Through acoustic beam positioning detection, the accurate position of the waste in space can be effectively determined. Especially in complex environments, precise acoustic beam positioning can help avoid positioning deviations caused by propagation path errors. By compensating for the propagation errors, the signal accuracy can be greatly improved, ensuring the reliability of the measurement results, and thus providing more accurate data support for subsequent positioning and component analysis. Through ultrasonic triangulation technology, three-dimensional spatial precise positioning of metal waste can be achieved, providing a basis for subsequent classification and component analysis. By further optimizing the error compensation, the accuracy of the positioning results can be significantly improved, avoiding errors caused by environmental factors or other interferences, and thus ensuring the high precision of the entire detection process. Through multi-spectral image acquisition, spectral data of metal waste can be obtained. Combining with the component deduction technology, the specific components of metal waste can be deduced. This not only helps with the quantitative analysis of metals but also helps determine the type, purity, and state of metals. The combination of spectral data and ultrasonic data provides comprehensive information support for the identification of metal waste, further improving the accuracy of waste component analysis. By combining the precise three-dimensional coordinates of metal waste with the component data, more intelligent and efficient waste characteristic classification can be achieved. The system can automatically identify different types of waste based on the learned characteristics. Based on big data and deep learning technologies, the waste classification model can be continuously optimized to make the detection of different types of waste more precise and efficient, further improving the overall detection performance. The intelligent detection model can gradually improve the waste identification and detection capabilities according to environmental changes and its own detection experience, realizing autonomous optimization and adaptation.

[0007] Preferably, step S1 includes the following steps: Step S11: Apply high-frequency ultrasonic waves to the waste area to be detected through a microarray ultrasonic sensor, and collect high-frequency ultrasonic response signals; Step S12: Perform noise filtering and noise reduction on the high-frequency ultrasonic response signals, and extract filtered and noise-reduced ultrasonic signals; Step S13: Perform multi-scale frequency decomposition on the filtered and noise-reduced ultrasonic signal to obtain acoustic frequency domain diagrams of different frequencies; Step S14: Based on the acoustic frequency domain diagrams of different frequencies, perform regional vibration distribution evolution to generate vibration distribution characteristics of the area to be detected; Step S15: Mine the metal vibration mode parameters of the vibration distribution characteristics of the area to be detected to obtain the metal vibration mode characteristics within the area.

[0008] The microarray ultrasonic sensor of the present invention can perform all-round and high-resolution ultrasonic scanning on the waste area in multiple directions and angles, thereby improving the ability to identify the shape and position of complex metal wastes. Through the synergistic effect of the microarray sensors, uniform high-frequency ultrasonic waves can be applied within the area, which provides comprehensive signal data support for subsequent precise vibration analysis. Since high-frequency ultrasonic waves have a short wavelength and can detect the microstructures inside metal materials, it helps to detect fine cracks, defects or irregular shapes of the wastes. In the actual environment, ultrasonic signals are often affected by noise interference, especially environmental noise and electromagnetic interference. Through noise filtering and noise reduction technology, the purity of the signal can be significantly improved, and the influence of irrelevant noise on signal analysis can be reduced. By removing low-frequency noise and other interference components, the contrast of the effective signal is increased, thereby providing clearer data input for subsequent analysis and enhancing the reliability of detection. By decomposing the signal into different frequency bands, detailed information at each frequency level can be obtained, which is of great significance for carefully identifying the complex vibration behavior and material properties of metal wastes. Multi-scale decomposition helps to improve the time and space resolution in the analysis process, making the detection of metal wastes more refined and comprehensive, especially in the detection of wastes with complex or unconventional shapes. Through the frequency domain diagram, the vibration distribution of the waste area can be more clearly displayed, helping to judge the structural characteristics, surface conditions and possible crack or defect positions of the metal waste. Through the evolution process of the regional vibration distribution, the vibration changes of the waste at different time points or under different working conditions can be monitored, providing more dynamic and real-time detection data, which helps to detect problems or potential risks at an early stage. The generation of vibration distribution characteristics helps to reveal the spatial characteristics of the waste area, thereby improving the accuracy of spatial positioning in the subsequent analysis process and ensuring more accurate detection results of metal wastes. By mining the metal vibration mode characteristics, physical properties such as the material, density, and elastic modulus of the waste can be revealed, helping to more accurately identify the metal type and its state of the waste. The vibration mode characteristics can provide an important basis for constructing the physical model of the waste, helping to understand the vibration behavior of the waste under the action of ultrasonic waves, and providing support for subsequent component analysis and positioning. The metal vibration mode characteristics provide reliable data support for subsequent metal classification and identification, helping to perform intelligent waste classification based on vibration modes and being able to distinguish different types of metal materials.

[0009] Preferably, the specific steps of step S14 are as follows: Calculate the acoustic vibration frequency of the acoustic frequency domain diagrams of different frequencies to extract the vibration frequency of each frequency domain diagram; Calculate the vibration peak value of each frequency for the acoustic frequency domain diagrams of different frequencies to obtain the vibration peak value of each frequency domain diagram; Perform peak width analysis on the vibration peaks of each frequency-domain graph to obtain the vibration peak widths of each frequency-domain graph; Perform vibration frequency offset identification based on the vibration peak widths of each frequency-domain graph to generate the local frequency offset features of each frequency-domain graph; Perform temporal vibration change mining on the local frequency offset features of each frequency-domain graph and the vibration frequencies of each frequency-domain graph, thereby generating the temporal vibration change features of each frequency-domain graph; Perform regional vibration distribution evolution based on the temporal vibration change features of each frequency-domain graph to generate the vibration distribution features of the region to be detected.

[0010] The calculation of the vibration frequency of the present invention can reveal the specific characteristics of the vibration within the metal waste area. Each metal and structure will exhibit different vibration frequencies. Therefore, by accurately calculating the acoustic wave vibration frequencies of different frequencies, it can help identify the physical properties and states of the materials. Different metal wastes usually have different vibration frequencies under the same ultrasonic excitation. Through this calculation step, different types of metal wastes can be better distinguished. By calculating the vibration peaks in each frequency domain diagram, the resonance points and specific vibration characteristics of the metal waste at each frequency can be revealed. These peaks reflect the elasticity, stiffness, and other mechanical properties of the metal. The vibration peak is an important indicator of the change in the physical state of the metal. Therefore, accurately calculating the peak value of each frequency can effectively improve the accuracy of waste identification and help detect potential structural defects (such as cracks, voids, etc.). The change in the peak width reflects the non-uniformity or defects of the material. For example, a wider peak may indicate the presence of cracks or other irregularities in the material, while a narrower peak may indicate that the material is more uniform. By analyzing the peak width, the structure and physical properties of the metal waste can be further analyzed. The analysis of the peak width helps to understand the vibration characteristics of the material at the microscopic level, especially in the detection of complex or damaged metal wastes, and can provide detailed feedback information. The shift of the vibration frequency is usually caused by internal defects (such as cracks, corrosion, deformation) or external factors (such as stress, temperature change) within the material. By identifying the frequency shift, these defects or abnormal areas can be accurately located. The local frequency shift characteristics can improve the sensitivity to small changes, especially in the vibration analysis of complex wastes, and can effectively distinguish relatively weak structural changes and damages. The vibration behavior of metal wastes may change at different time periods or under different environmental conditions. By mining the time-series vibration changes, the state changes of the metal wastes can be tracked in real time, helping the detection personnel to discover problems in a timely manner. The analysis of the time-series vibration changes not only helps in the identification of the current state but also can provide a trend prediction for future vibration changes, identifying potential faults or damages in advance. Through the time-series vibration change characteristics, an evolution map of the vibration distribution in the waste area can be generated, helping the detection personnel to clearly see the changes in the vibration characteristics of the waste at different times or in different areas, thereby better understanding the structure and performance of the waste. This process can reveal the local changes in the metal waste, such as which areas have abnormal vibrations, which helps to locate the problems, especially in an environment with multiple different metal wastes or multiple defects. By combining the time-series and spatial vibration characteristic analyses, a more comprehensive vibration distribution map can be obtained, which helps to globally understand the vibration behavior of the waste and its potential damage areas.

[0011] Preferably, the specific steps of step S2 are as follows: Step S21: Identify multi-sensor nodes based on the microarray ultrasonic sensor; Step S22: Calculate the acoustic wave reflection response of the multi-sensor nodes based on the high-frequency ultrasonic response signal, so as to obtain the acoustic wave propagation time of each node; Step S23: Identify the ultrasonic excitation paths of the multi-sensor nodes, so as to generate the ultrasonic excitation trajectories of each node; Step S24: Perform ultrasonic spatial propagation modeling on the ultrasonic excitation trajectories based on each node according to the acoustic wave propagation time of each node, and construct a spatial ultrasonic trajectory propagation model.

[0012] The present invention can deploy multiple sensor nodes within the waste area through a microarray ultrasonic sensor, thereby covering a larger area and enabling omnidirectional detection, enhancing the detection range and sensitivity. By identifying multiple sensor nodes, cooperation between multiple sensors can be achieved, which makes signal acquisition and analysis more flexible and efficient, reducing the signal blind area caused by the angle limitation of a single sensor. By calculating the acoustic wave reflection response of each node, the acoustic wave propagation time of each sensor node can be accurately obtained. The acoustic wave propagation time is directly related to the shape, position, and physical properties of the metal waste, which is crucial for positioning and composition analysis. The accurate acoustic wave propagation time provides a basis for subsequent spatial positioning and trajectory analysis, helping to eliminate errors caused by different signal propagation paths, thereby improving the positioning accuracy. By calculating the propagation times of multiple nodes, multiple propagation paths can be identified, providing multi-dimensional data support for subsequent modeling and enhancing the overall modeling accuracy. Identifying the ultrasonic excitation path helps to understand the propagation path of ultrasonic waves from the excitation source to the sensor nodes, and can reveal the propagation characteristics of different paths. Especially in complex environments (such as waste accumulation or irregularly shaped metals), it can help locate abnormal areas. By identifying the excitation path, errors caused by factors such as path loss, reflection, or refraction can be excluded, thereby improving the accuracy of signal decoding and data analysis. After identifying the excitation paths of multiple sensor nodes, it can provide basic information for subsequent multi-path modeling, especially in complex spatial structures, helping to obtain a more accurate propagation model. Using the acoustic wave propagation time and excitation path information, an accurate ultrasonic trajectory propagation model can be constructed in space. This model can comprehensively describe the propagation characteristics of ultrasonic waves in complex environments, providing strong support for accurately locating metal waste. The spatial ultrasonic trajectory propagation model helps to eliminate the influence of factors such as signal attenuation and scattering during propagation, making the final detection results more accurate. Through model optimization, the accuracy of waste identification and composition analysis can be improved. The spatial propagation model fuses the signal information of multiple nodes, comprehensively considering multi-dimensional data, and can accurately identify information such as the spatial position, shape, and size of metal waste. In complex detection environments (such as waste accumulation and irregular metal surface morphology), this spatial propagation model can cope with the interference of various factors, improving the reliability and effectiveness of modeling and avoiding misjudgment caused by a single path or signal deviation.

[0013] Preferably, the specific steps of step S3 are as follows: Step S31: Perform acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the region to extract the acoustic response beam of the metal vibration mode; Step S32: Perform acoustic wave attenuation analysis on the acoustic response beam of the metal vibration mode to generate the acoustic response beam attenuation characteristics; Step S33: Perform diffusion multipath reflection evolution on the acoustic response beam of the metal vibration mode to obtain the multipath reflection evolution data of the acoustic response beam; Step S34: Conduct propagation interference error analysis based on the acoustic response beam attenuation characteristics and the multipath reflection evolution data of the acoustic response beam to obtain the propagation interference error parameters of the response beam; Step S35: Perform propagation error compensation calculation on the propagation interference error parameters of the response beam to obtain the propagation error compensation value of the response beam.

[0014] Through acoustic beam localization based on the vibration modal characteristics of metals, the vibration characteristics of metals within the waste area can be accurately identified and located. The metal vibration mode reflects the natural frequency and vibration characteristics of metal waste under the action of ultrasonic waves. Through this information, it is possible to better extract the acoustic response beam and accurately determine the position of metal waste. Through acoustic beam localization detection, it helps to extract effective acoustic response signals from complex environments, filter out irrelevant noises, and improve the signal-to-noise ratio of signal analysis. The metal vibration modal characteristics can distinguish different types of metal materials and forms. Through acoustic beam localization, targeted waste detection can be achieved, avoiding the generation of redundant data and improving the detection efficiency. Acoustic wave attenuation reflects the degree to which the signal is affected by the medium, material, and obstacles during propagation. By analyzing the attenuation characteristics of the acoustic response beam, the physical characteristics (such as density, surface roughness, corrosion degree, etc.) of the area where the metal waste is located can be revealed, helping to accurately judge the structure and material state of the waste. Attenuation analysis can effectively identify which areas of the signal are no longer valid due to large attenuation, thus helping to optimize the signal reception and subsequent analysis, and improving the accuracy of overall signal analysis. The attenuation characteristics not only reflect the loss of signal propagation but may also be closely related to the microstructure and defects within the material. Therefore, attenuation analysis provides strong support for subsequent analysis of the composition and structure of metal materials. The ultrasonic signals in the metal waste area often undergo multiple reflections and refractions, resulting in the formation of multiple reflection paths. Through multi-path reflection evolution, these reflection paths can be captured, thus providing a more comprehensive understanding of the signal propagation process in space. The multi-path reflection evolution data can reveal the complexity of acoustic wave propagation within the waste area, thus helping to identify and distinguish signals of different paths, making the positioning and identification of waste more accurate. The multi-path reflection evolution data provides support for the comprehensive perception of space, especially in the case of complex waste accumulation or occlusion, helping to improve the robustness and accuracy of overall detection. During the acoustic wave propagation process, due to factors such as non-uniform medium, obstacles, and reflections, propagation interference errors often occur. By integrating the attenuation characteristics and multi-path reflection evolution data, the error sources during the propagation process can be accurately identified and these errors can be quantified, providing a basis for subsequent error compensation. The propagation interference error analysis can help to accurately identify the specific types of errors (such as time delay, path loss, etc.), thus improving the data accuracy in subsequent measurement processes and avoiding the influence of errors on the final result. After analyzing the propagation interference errors, the acoustic wave propagation model can be corrected, making subsequent positioning, identification, and composition analysis more accurate and reliable. By compensating and calculating the propagation interference error parameters of the response beam, the errors caused by factors such as multi-path reflection, attenuation, and interference can be effectively corrected, thus improving the accuracy of acoustic wave signals. The signal obtained after error compensation is more accurate, reducing positioning or identification errors caused by errors. After compensation calculation, the propagation error of the response beam is effectively corrected, making the spatio-temporal characteristics of the signal more real and reliable.This is very important for the precise identification and positioning of metal waste, especially in complex environments, which can effectively improve the credibility of the results. The propagation error compensation ensures more accurate data in subsequent processes such as detection, positioning, and classification, laying the foundation for the intelligent analysis of the entire detection system and improving the quality of the final analysis results.

[0015] Preferably, the specific steps of step S4 are as follows: Step S41: Quantify the multi-sensor reflection distance difference based on the acoustic wave response beam of the metal vibration mode, so as to obtain the distance from the response beam to each sensor node. Step S42: Perform ultrasonic triangulation measurement on the spatial ultrasonic trajectory propagation model according to the distance from the response beam to each sensor node, so as to obtain the three-dimensional positioning coordinates of the metal waste. Step S43: Optimize the positioning error accuracy of the three-dimensional positioning coordinates of the metal waste based on the response beam propagation error compensation value, so as to obtain the precise three-dimensional coordinates of the metal waste.

[0016] The present invention can accurately calculate the distance from each sensor node to the response beam by quantifying the multi-sensor reflection distance of the acoustic wave response beam. The distance data provides an accurate spatial reference for subsequent positioning, making the positioning of metal waste more reliable. The combination of the reflection distance data of multiple sensor nodes can eliminate the errors of a single sensor and improve the overall detection accuracy. This multi-point information fusion ensures efficient measurement in complex environments. Whether in waste accumulation areas or complex structures, the analysis of the reflection distance difference of multiple sensors can cope with the influence of different propagation paths and improve the adaptability and robustness of the detection system. Through ultrasonic triangulation measurement, the position of the metal waste in the three-dimensional space can be accurately calculated. Using the reflection distance data between multiple sensor nodes for triangulation measurement, the obtained positioning result is more accurate than a single method. This method does not rely solely on single-plane or linear data, but through three-dimensional space positioning calculations, it can more comprehensively reflect the spatial distribution of the waste and improve the accuracy of identification and positioning. The triangulation method usually has strong real-time performance and can quickly respond to dynamic changes, providing support for the real-time monitoring and management of metal waste and being applicable to environments that require rapid response. By compensating for the propagation error, the errors caused by factors such as multi-path reflection and signal attenuation can be effectively corrected, thereby improving the accuracy of metal waste positioning. This is a key step in the calculation of the precise three-dimensional coordinates of metal waste. Error compensation ensures that even in the case of signal interference or complex propagation paths, a high positioning accuracy can still be maintained, improving the performance of the system in complex environments.

[0017] Preferably, the specific steps of step S5 are as follows: Step S51: Collect multispectral images based on the precise three-dimensional coordinates of metal waste to obtain high-resolution multispectral images of the metal surface; Step S52: Perform image detail sharpening on the high-resolution multispectral images of the metal surface to obtain sharpened and enhanced multispectral images; Step S53: Decompose the pixel points of the sharpened and enhanced multispectral images and extract all the pixel points in the spectral images; Step S54: Conduct multispectral channel feature analysis based on all the pixel points in the spectral images to obtain multispectral channel features; Step S55: Perform quantitative deduction of metal components on the multispectral channel features to obtain quantitative information on the components of metal waste.

[0018] The present invention can ensure the consistency of the acquisition position and viewing height of the image by collecting multi-spectral images based on the precise three-dimensional coordinates of metal waste, thus guaranteeing the spatial accuracy of the image data. This provides more accurate background information for subsequent analysis and helps reduce positional errors. By using high-resolution multi-spectral imaging technology, detailed information on the metal surface in different spectral bands can be obtained, especially in regions such as the visible light, infrared, and near-infrared bands. These information can reveal the microscopic surface features, structures, and compositions of metal waste. The multi-spectral images can capture the different reflection characteristics of the surface and internal structures of metal waste, providing multi-dimensional spectral information to assist in the analysis and identification of the composition of metal waste, especially surface contamination, corrosion, and composition distribution of the metal. Through sharpening processing, the details on the surface of metal waste in the image can be enhanced, making features such as tiny textures, cracks, corrosion signs, or surface scratches more prominent. These details are crucial for the quality assessment, damage detection, and composition analysis of metal waste. Sharpening enhancement can not only increase the contrast of the image but also improve the clarity, eliminating the effects caused by blurring or noise during the acquisition process and making the features of each pixel point more distinct. By performing pixel decomposition on the sharpened and enhanced multi-spectral images, the spectral information of each pixel point can be accurately extracted. Each pixel point represents the reflection characteristics of the metal surface in different bands. Through these data, the subtle differences in metal waste can be analyzed. Pixel decomposition provides the ability to deeply analyze each detailed area in the image. Different spectral channels (such as visible light, near-infrared, etc.) can reflect different material compositions. Therefore, the data of each pixel point will help reveal the chemical composition and physical properties of metal waste. The extracted pixel points will become the basic data for subsequent multi-spectral channel feature analysis and metal composition deduction, providing reliable data support for accurately identifying the types and compositions of metal waste. By analyzing the reflection characteristics of each pixel point in different spectral channels, features in different bands can be extracted. These features support multi-dimensional data fusion and help reveal the composition information of metal waste, such as metal types, oxidation degrees, etc. Each spectral channel represents different physical or chemical properties. Conducting multi-channel feature analysis can further distinguish metal waste from other non-metal substances, corrosives, or pollutants, thereby improving the accuracy of composition analysis. By analyzing the features of different spectral channels, multi-dimensional support can be provided for the classification and detection of metal waste, helping the system identify the specific types or classifications of metal waste (such as metals like iron, copper, aluminum, etc.). Based on the multi-spectral channel features, the specific composition ratios of metal waste (such as the contents of iron, copper, aluminum, etc.) can be deduced. This quantitative analysis helps determine the quality, purity, and physical and chemical properties of metal waste. By deducing the composition of metal waste, trace elements in metal waste can be accurately identified, providing refined classification information for metal waste, which is particularly significant in aspects such as recycling, resource reuse, and environmental treatment.Through the quantitative deduction of metal components, accurate data support can be provided for the subsequent treatment of waste, helping to select the best recycling, reuse or disposal solutions, improving resource utilization rate and reducing waste.

[0019] Preferably, the specific steps of step S6 are as follows: Step S61: Classify the characteristics of metal waste based on the quantitative information of metal waste components, so as to obtain the types of metal waste; Step S62: Dynamically visualize and mark the spatial ultrasonic trajectory propagation model based on the type of metal waste and the precise three-dimensional coordinates of the metal waste to obtain a dynamic visualization model; Step S63: Optimize the dynamic visualization model through enhanced detection learning to construct an intelligent detection model for metal waste.

[0020] Through the quantitative information of metal components, the present invention can accurately classify metal waste, such as iron, copper, aluminum, alloys, etc. This helps to identify different types of metals in the process of waste management and recycling, optimize the treatment plan. Clear classification helps to achieve efficient sorting and recycling of metal waste. Different types of metals can adopt the most suitable treatment technologies, such as smelting, alloying or physical separation, so as to improve resource utilization rate and reduce environmental pollution. In the links of waste treatment, logistics transportation, etc., accurate identification and classification of metal types can achieve more refined management and planning, helping to reduce unnecessary resource waste and costs. By combining the type and three-dimensional coordinate information of metal waste with the spatial ultrasonic trajectory model, a dynamic visualization model can be generated, which makes the spatial distribution and propagation path of metal waste intuitive and visible, facilitating understanding and monitoring. The dynamic visualization model can update the position and state of metal waste according to real-time changes, providing instant feedback for subsequent treatment, and is particularly suitable for scenarios that require real-time monitoring, such as industrial production lines, waste accumulation sites, etc. Through the visual trajectory model, the ultrasonic detection path can be optimized to ensure comprehensive signal coverage and high detection accuracy, which provides more effective technical support for the detection, evaluation and treatment of waste. Through continuous optimization of the detection model, reinforcement learning can automatically identify the characteristics and changes of metal waste, and gradually improve the detection accuracy and efficiency. This adaptive ability enables the system to cope with various complex and dynamically changing detection environments. Through enhanced detection learning, the model can gradually adapt to different types of waste, environmental changes and detection errors, so as to improve the detection accuracy and robustness of metal waste. The intelligent detection model can not only identify the waste type, but also automatically adjust the detection path, optimize resource allocation, reduce human intervention, and improve the automation level and working efficiency of the overall system.

[0021] In this specification, a detection and identification system for metal waste is provided, which is used to execute the detection and identification method for metal waste as described above, including: A metal vibration mode module, which is used to apply high-frequency ultrasonic waves to the area of waste to be detected, collect high-frequency ultrasonic response signals, and mine the metal vibration mode parameters of the high-frequency ultrasonic response signals, so as to obtain the metal vibration mode characteristics in the area. An ultrasonic trajectory module, which is used to identify the ultrasonic excitation path of the high-frequency ultrasonic response signal, perform ultrasonic spatial propagation modeling, and construct a spatial ultrasonic trajectory propagation model. A propagation error compensation module, which is used to perform acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the area, and perform propagation error compensation calculation, so as to obtain the response beam propagation error compensation value. A triangulation measurement module, which is used to perform ultrasonic triangulation measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the accurate three-dimensional coordinates of the metal waste. A quantitative deduction module, which is used to collect multi-spectral images based on the accurate three-dimensional coordinates of the metal waste, and then perform quantitative deduction of metal components, so as to obtain the quantitative information of the metal waste components. An intelligent detection module, which is used to classify the characteristics of the metal waste and optimize the enhanced detection learning based on the quantitative information of the metal waste components, so as to construct an intelligent detection model for the metal waste.

[0022] By extracting the metal vibration modes, the present invention can accurately reflect the physical properties of metals, such as the surface state, microstructure, and deformation characteristics of metals, which is of great significance for the early identification and qualitative analysis of metal waste. The metal vibration mode characteristics provide the "fingerprints" of different metal types or surface damages, enabling rapid identification of the regional characteristics of waste, providing key data for subsequent ultrasonic wave propagation modeling and error compensation. By mining the metal vibration modes, the response of ultrasonic sensors can be optimized, the signal quality can be enhanced, and a foundation for efficient positioning and detection can be laid. The identification of the ultrasonic wave propagation path can help construct an accurate propagation model, thereby avoiding the influence of errors during the waste detection process and ensuring that the signal can accurately cover the area to be detected. By establishing a propagation model, the propagation path and reflection of ultrasonic signals can be predicted in advance, the detection path can be optimized, and signal attenuation or errors can be reduced. Through the spatial ultrasonic wave trajectory model, the propagation trajectory of ultrasonic waves in space can be accurately described, providing precise path data support for subsequent triangulation positioning measurement. Through propagation error compensation, the propagation errors caused by factors such as medium changes and reflection paths can be corrected, thereby improving the positioning accuracy of metal waste. This module overcomes the influence of environmental factors such as multipath effects and attenuation by compensating for propagation errors, enabling ultrasonic signals to be transmitted more precisely to the target area and ensuring the reliability of the detection results. By calculating the propagation error compensation value of the response beam, the positioning of the acoustic beam is ensured to be more accurate, enhancing the detection accuracy of metal waste. Using triangulation positioning technology, through multiple known sensor nodes and a propagation model, precise spatial positioning measurement can be carried out, enabling the efficient calculation of the three-dimensional coordinates of metal waste. Combining the propagation error compensation value, the accuracy of triangulation positioning measurement is optimized, ensuring the accuracy of waste positioning. Through precise three-dimensional coordinates, it supports the three-dimensional modeling of the metal waste area, providing a scientific basis for further analysis, recycling, or environmental assessment. Combining three-dimensional positioning information and multi-spectral images, the chemical composition and physical properties of metal waste can be accurately extracted for quantitative analysis. The multi-spectral image acquisition provides the reflection information of the metal surface in different bands, which can help the system reveal different components of metal waste and then conduct component deduction to identify valuable metals. Precise metal component deduction enables the recycling system to more effectively identify and sort different types of metals, improving the recycling efficiency and resource utilization rate of metal waste. By analyzing the component information of metal waste, the waste can be automatically classified, providing intelligent support for recycling, reuse, and environmental management. The introduction of reinforcement learning algorithms enables the system to continuously learn and optimize in practical applications, automatically adjusting the detection strategy according to new data, improving the flexibility and accuracy of detection. The intelligent detection model can adaptively adjust detection parameters in a complex environment, effectively identifying various different forms and types of metal waste, enhancing the robustness and universality of the system. Through intelligent optimization, the detection speed and accuracy of metal waste can be improved, and manual intervention can be reduced.Improve the overall system efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic flow chart of the steps of a method for detecting and identifying metal waste of the present invention; Figure 2 It is a schematic detailed implementation step flow chart of step S1; Figure 3 It is a schematic detailed implementation step flow chart of step S2; Figure 4 It is a schematic detailed implementation step flow chart of step S3. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] The embodiments of the present application provide a method and a system for detecting and identifying metal waste. The execution subjects of the method and system for detecting and identifying metal waste include but are not limited to: mechanical equipment, data processing platforms, cloud server nodes, network uploading devices, etc. that carry the system, which can be regarded as general computing nodes of the present application. The data processing platform includes but is not limited to: at least one of an audio and image management system, an information management system, and a cloud data management system.

[0026] Please refer to Figures 1 to 4 , the present invention provides a method. The method for detecting and identifying metal waste includes the following steps: Step S1: Apply high-frequency ultrasonic waves to the area of waste to be detected, and collect high-frequency ultrasonic response signals; mine the metal vibration mode parameters of the high-frequency ultrasonic response signals to obtain the metal vibration mode characteristics in the area; Step S2: Identify the ultrasonic excitation path of the high-frequency ultrasonic response signals, and perform ultrasonic spatial propagation modeling to construct a spatial ultrasonic trajectory propagation model; Step S3: Perform acoustic beam positioning detection on the high-frequency ultrasonic response signals based on the metal vibration mode characteristics in the area, and perform propagation error compensation calculation to obtain the response beam propagation error compensation value; Step S4: Perform ultrasonic triangulation positioning measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value to obtain the accurate three-dimensional coordinates of the metal waste; Step S5: Collect multi-spectral images based on the accurate three-dimensional coordinates of the metal waste, and then perform quantitative deduction of metal components to obtain the quantitative information of the metal waste components; Step S6: Perform metal waste feature classification and enhanced detection learning optimization based on the quantitative information of the metal waste components to construct an intelligent detection model for metal waste.

[0027] By applying high-frequency ultrasonic waves to the waste area, the present invention can effectively stimulate the vibration modes on and inside the metal surface. Different vibration modal characteristics of the metal exhibit different signal characteristics in the acoustic wave reflection, which can help identify the physical and structural properties of the waste. By mining the metal vibration modal parameters, characteristic information of the metal can be obtained, thereby improving the accuracy of identifying the types and properties of metals in the waste. The propagation path of ultrasonic waves is an important factor affecting the signal accuracy. Path recognition and modeling help clarify the propagation path of acoustic waves from the source point to the detection point, reducing signal attenuation and offset errors during propagation. The spatial ultrasonic trajectory propagation model can accurately describe the propagation characteristics of acoustic waves in the waste area, thereby enhancing the accuracy of subsequent positioning and composition analysis. Through acoustic beam positioning detection, the accurate position of the waste in space can be effectively determined. Especially in a complex environment, precise acoustic beam positioning can help avoid positioning deviations caused by propagation path errors. By compensating for the propagation errors, the signal accuracy can be greatly improved, ensuring the reliability of the measurement results, and thus providing more accurate data support for subsequent positioning and composition analysis. Through ultrasonic triangulation positioning technology, three-dimensional spatial precise positioning of metal waste can be achieved, providing a basis for subsequent classification and composition analysis. By further optimizing the error compensation, the accuracy of the positioning result can be significantly improved, avoiding errors caused by environmental factors or other interferences, thereby ensuring the high precision of the entire detection process. Through multi-spectral image acquisition, spectral data of metal waste can be obtained. Combining with the composition deduction technology, the specific composition of metal waste can be deduced. This not only helps with the quantitative analysis of metals but also helps determine the type, purity, and state of metals. The combination of spectral data and ultrasonic data provides comprehensive information support for the identification of metal waste, further improving the accuracy of waste composition analysis. By combining the precise three-dimensional coordinates of metal waste with the composition data, more intelligent and efficient waste characteristic classification can be achieved. The system can automatically identify different types of waste based on the learned characteristics. Based on big data and deep learning technologies, the waste classification model can be continuously optimized to make the detection of different types of waste more precise and efficient, further improving the overall detection performance. The intelligent detection model can gradually improve the waste identification and detection capabilities according to environmental changes and its own detection experience, achieving autonomous optimization and adaptation.

[0028] In the embodiment of the present invention, refer to Figure 1 , which is a schematic diagram of the step flow of a method for detecting and identifying metal waste according to the present invention. In this example, the steps of the method include: Step S1: Apply high-frequency ultrasonic waves to the waste area to be detected, and collect high-frequency ultrasonic response signals; mine the metal vibration modal parameters from the high-frequency ultrasonic response signals to obtain the metal vibration modal characteristics in the area. In this embodiment, select a suitable high-frequency ultrasonic generator and sensor. The ultrasonic generator should have an adjustable frequency characteristic for testing in different frequency bands; the sensor should be able to accurately receive and record ultrasonic signals. Ensure that the environment of the waste area to be detected is suitable for ultrasonic experiments, minimize external noise and interference, and ensure the stability of the test environment. In addition, confirm that the surface of the waste to be detected is clean to improve the signal transmission quality. Start the ultrasonic generator, select a suitable frequency range (usually between 20 kHz and 200 kHz), and apply high-frequency ultrasonic signals to the area to be detected. According to needs, a single frequency or frequency scanning mode can be selected to cover a wider spectrum. While applying ultrasonic waves, use an array of sensors to monitor the signal propagation in real time. Multiple sensors should be evenly arranged in the area to be detected to capture response signals at different angles and positions to ensure the comprehensiveness of the data. Use a high-frequency data acquisition system to record the ultrasonic response signals received by the sensors. The acquisition system should have a high sampling rate (such as above 1 MHz) to ensure that it can capture subtle signal changes and responses. Store the collected ultrasonic signal data in a computer or data storage device in a suitable format (such as WAV or TXT files) for subsequent analysis and processing. Preprocess the collected ultrasonic response signals, including denoising, filtering, and time-domain to frequency-domain conversion. Analyze the signals using the Fast Fourier Transform (FFT) to identify spectral characteristics and modal information. By analyzing the spectral characteristics of the ultrasonic signals, extract the relevant parameters of the metal vibration mode. These parameters include natural frequency, damping ratio, and vibration mode, etc. Modal analysis software (such as the System Identification Toolbox of MATLAB) can be used for automated processing. Based on the extracted modal parameters, generate the feature vectors of the metal vibration mode. These features will be used for subsequent classification and analysis.

[0029] Step S2: Identify the ultrasonic excitation path for the high-frequency ultrasonic response signals, and perform ultrasonic spatial propagation modeling to construct a spatial ultrasonic trajectory propagation model. In this embodiment, a preliminary environmental model is established according to the specific environmental characteristics of the area to be detected (such as material type, geometric shape, and obstacle distribution). This model will serve as the basis for ultrasonic wave propagation and affect the propagation path and characteristics of the signal. Perform time-frequency analysis on the collected high-frequency ultrasonic response signals, using techniques such as short-time Fourier transform (STFT) or wavelet transform to extract the spectral characteristics of the signals. By analyzing the spectrogram, identify the main frequency components and response patterns of the signals. Use beamforming technology to process the signals to identify the propagation path of the ultrasonic waves. Reconstruct the propagation direction and path of the sound waves through the phase difference and time delay of multi-channel signals. Common methods include delay-and-sum and least squares method. According to the results of beamforming, extract the excitation paths of the ultrasonic waves. These paths indicate the propagation routes of the sound waves from the emission source to each receiving sensor and provide information about multipath effects. Select a suitable propagation model, commonly including ray tracing model, finite element model, or wave equation model. The selection should consider the complexity of the area to be detected, material properties, and actual application requirements. Based on the environmental model and the identified ultrasonic excitation paths, construct a spatial ultrasonic trajectory propagation model. It is necessary to define the boundary conditions, medium properties (such as density, sound speed, etc.), and propagation parameters (such as attenuation coefficient, etc.) of the model. Verify the constructed propagation model by comparing the results predicted by the model with the actually collected signals. If there are significant differences, adjust the model parameters or re-evaluate the accuracy of the excitation path identification. Store the identified ultrasonic excitation paths and the constructed propagation model in the database to ensure the integrity and traceability of the data. Form a structured data set for subsequent analysis and model optimization. Conduct a preliminary analysis of the ultrasonic propagation model to evaluate the effectiveness and applicability of the model. According to the needs, perform further optimization and adjustment to ensure that the model can accurately reflect the actual ultrasonic propagation characteristics.

[0030] Step S3: Perform acoustic beam localization detection on the high-frequency ultrasonic response signals based on the metal vibration mode characteristics in the area, and perform propagation error compensation calculation to obtain the response beam propagation error compensation value; In this embodiment, a suitable acoustic beam positioning detection method is selected. Commonly used methods include Time Difference of Arrival (TDOA), Phase Difference of Arrival (PDA), and beamforming. These methods can determine the exact position of the acoustic source through the response signals of multiple sensors. The collected high-frequency ultrasonic response signals are preprocessed, including denoising, filtering, and enhancement, to improve the signal quality. Wavelet transform or filter design can be used to remove background noise. The metal vibration mode characteristics in the area are matched with the ultrasonic response signals to identify the mode characteristics corresponding to a specific metal type. The Cross-Correlation method is used to determine the similarity between signals. Using the selected positioning method, by comparing the signals received by different sensors, the position of the acoustic source is calculated. For Time Difference of Arrival positioning, the time difference of the signal arriving at different sensors needs to be measured and calculated using the known propagation speed. During the positioning process, factors that may cause propagation errors are identified, including environmental factors (such as temperature, humidity), sensor accuracy, and signal attenuation. The influence of these factors on signal propagation is recorded. A propagation error model is established, usually using linear regression or machine learning models to describe the main factors affecting signal propagation. By analyzing historical data, the key variables affecting propagation are determined. The identified error factors are input into the error model to calculate the propagation error compensation value for each sensor. C = Δt ⋅ v, where C is the compensation value, Δt is the time deviation, and v is the acoustic wave propagation speed. Similar calculations are performed for all sensors to obtain a comprehensive propagation error compensation value.

[0031] Step S4: Perform ultrasonic triangulation measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the accurate three-dimensional coordinates of the metal waste; In this embodiment, it is ensured that the acoustic beam positioning detection in step S3 has been completed, and the propagation error compensation value is obtained. Prepare the three-dimensional coordinates of the sensors and their corresponding propagation distance data as the basis for triangulation. Conduct a preliminary verification of the spatial ultrasonic trajectory propagation model to ensure that the model can accurately reflect the propagation characteristics of sound waves in the area to be measured. This includes verifying the boundary conditions and material parameters of the model. Select a suitable triangulation algorithm. Commonly used algorithms include the Least Squares method, the non-linear minimization method, or simple geometric methods. These methods can calculate the precise coordinates of the metal waste through the distances and positions between the sensors. Solve the system of equations using the selected algorithm to obtain the preliminary three-dimensional coordinates (x, y, z) of the metal waste. Numerical methods such as the Newton method or the gradient descent method can be used to ensure the convergence and accuracy of the results. Apply the previously calculated propagation error compensation value to the preliminary positioning result. According to the compensation model, adjust the distance measurement values of each sensor to reduce the positioning deviation caused by environmental changes or equipment errors. Select a suitable optimization algorithm (such as the weighted least squares method) to optimize the adjusted distances through an iterative process. After optimization, update the three-dimensional coordinates of the metal waste to ensure that they reflect the precise position after compensation. Record each step in the optimization process for subsequent analysis and verification.

[0032] Step S5: Based on the precise three-dimensional coordinates of the metal waste, collect multi-spectral images, and then perform quantitative deduction of metal components to obtain the quantitative information of the metal waste components; In this embodiment, a suitable multispectral imaging device is selected to ensure that it has sufficient spectral resolution and spatial resolution to capture the details of metal waste. The device should be able to cover the bands of interest, usually including the visible and near-infrared spectra. According to the accurate three-dimensional coordinates of the metal waste obtained in step S4, the acquisition position is confirmed to ensure that the device can accurately align with these coordinates and avoid the degradation of image quality caused by position deviation. Ensure that the illumination in the acquisition environment is uniform and avoid strong direct light and shadows. At the same time, check the status of the acquisition device to ensure its normal function and clean lens to obtain the best image quality. Develop a detailed image acquisition plan to determine the acquisition bands and the exposure time for each band. The acquisition can be carried out band by band or synchronously to ensure that each image has a good signal-to-noise ratio. After confirming the device and environmental conditions, start the multispectral imaging device, align it with the three-dimensional coordinates of the metal waste for image acquisition, and ensure clear images are obtained in each band. Record the relevant parameters of each image (such as band, exposure time, etc.), and store the acquired multispectral image data in a computer in a suitable file format (such as TIFF or HDF5) for subsequent processing and analysis. Ensure the integrity of the data and the standardization of file naming for easy management. Select a suitable component analysis method, usually including spectral matching, regression analysis, or machine learning algorithms. According to the characteristics of the metal waste, select an algorithm that can extract key spectral features. Prepare a training data set of known metal components, including their corresponding multispectral images and component information, which will be used to train the model so that it can learn the relationship between spectral features and metal components. Preprocess the acquired multispectral images, including denoising, standardization, and spectral feature extraction. Use principal component analysis (PCA) or other feature extraction methods to extract the main spectral features related to metal components. Use the prepared training set to train the selected component analysis model, and evaluate the accuracy of the model through methods such as cross-validation to ensure that the model can effectively map spectral features to metal components. Input the extracted spectral features into the trained model for quantitative deduction of metal components, record the concentration or proportion of each metal component, and generate a detailed component analysis report.

[0033] Step S6: Based on the quantitative information of the metal waste components, conduct metal waste feature classification and enhanced detection learning optimization, thereby constructing an intelligent detection model for metal waste.

[0034] In this embodiment, effective features are selected from the integrated data for classification. These features may include the concentration of metal components, modal parameters, ultrasonic response features, etc. Feature selection techniques (such as recursive feature elimination (RFE) or LASSO regression) are used to screen the features that have the greatest impact on the classification results. According to the nature and features of the data, a suitable classification model is selected. Commonly used models include support vector machine (SVM), random forest, decision tree, and neural network, etc. The model selection should consider its accuracy, interpretability, and computational efficiency. The integrated dataset is randomly divided into a training set and a test set. Usually, 70% of the data is used for training and 30% for testing to ensure that the divided data can represent the overall situation for obtaining a reliable model performance evaluation. The selected classification model is trained using the training set, the model parameters (such as learning rate, tree depth, etc.) are adjusted, and cross-validation techniques are used to evaluate the model performance to ensure that the model does not overfit. The trained model is evaluated on the test set, and metrics such as confusion matrix, accuracy, recall, and F1 score are used to measure the performance of the model. According to the evaluation results, the model parameters are adjusted or other models are selected to improve the classification accuracy. A suitable reinforcement learning framework (such as TensorFlow, PyTorch, or OpenAI Gym) is selected to support the construction and optimization of the intelligent detection model. Reinforcement learning will be used to improve the performance of the model in practical applications. An environment for training the intelligent detection model is created. In this environment, the agent can learn the best strategy for detecting metal waste through interaction with the environment. The environment should include the features of metal waste, detection targets, and reward mechanisms. The agent is designed to be responsible for performing detection tasks in the environment. Using reinforcement learning algorithms (such as deep Q-network (DQN) or policy gradient method), the detection strategy of the agent is optimized through continuous trial and error learning. The agent obtains rewards based on the detection results and thus adjusts its behavior to improve the detection efficiency.

[0035] In this embodiment, refer to Figure 2 , which is a schematic diagram of the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: Apply high-frequency ultrasonic waves to the waste area to be detected through a microarray ultrasonic sensor, and collect high-frequency ultrasonic response signals; Step S12: Perform noise filtering and denoising on the high-frequency ultrasonic response signals to extract filtered and denoised ultrasonic signals; Step S13: Perform multi-scale frequency decomposition on the filtered and denoised ultrasonic signals to obtain acoustic frequency domain diagrams of different frequencies; Step S14: Based on the acoustic frequency domain diagrams of different frequencies, perform regional vibration distribution evolution to generate vibration distribution characteristics of the area to be detected; Step S15: Mine the metal vibration mode parameters of the vibration distribution characteristics of the area to be detected, so as to obtain the metal vibration mode characteristics within the area.

[0036] In this embodiment, appropriate microarray ultrasonic sensors are selected. These sensors can emit high-frequency ultrasonic waves and receive echo signals, ensuring that the frequency range of the sensors meets the requirements of the characteristics of the waste to be detected. An array of sensors is arranged in the area of the waste to be detected, ensuring that the spacing between the sensors is appropriate to cover the entire detection area. The sensors should be fixed in a stable position to avoid external interference. The ultrasonic emission system is started, and a high-frequency ultrasonic signal with a preset frequency is applied. Usually, a frequency range of 40 kHz to 1 MHz is selected, and the specific frequency is adjusted according to the characteristics of the material to be detected. The ultrasonic response signal is collected in real time by the sensors. The collected signal will contain echo information from the waste. The time stamps during the recording process are recorded for subsequent analysis. The collected ultrasonic signals are analyzed to identify the noise sources, including environmental noise, equipment noise, and signal interference. Understanding the noise characteristics helps to select appropriate noise reduction methods. Appropriate signal processing techniques are selected for noise filtering. Common methods include low-pass filters, band-pass filters, and wavelet transforms, etc. For high-frequency ultrasonic signals, band-pass filters can effectively retain important signal components. The selected filtering method is applied to process the ultrasonic response signal. By setting a reasonable cutoff frequency, the noise components above or below a specific frequency are filtered out, ensuring that the filtered signal retains the main characteristics of the original signal. The signals before and after the filtering process are visually compared to confirm the noise reduction effect. By plotting the time-domain graph and frequency-domain graph, it is checked whether the filtered signal is clear and free from obvious noise interference. A suitable multi-scale frequency decomposition method is selected, such as Wavelet Transform or Short-Time Fourier Transform (STFT), to extract different frequency components of the signal. The selected frequency decomposition method is applied to process the filtered ultrasonic signal. Wavelet transform can decompose the signal into sub-signals of different frequencies and retain the time information, generating the acoustic frequency-domain graph of each decomposed frequency, showing the amplitude and phase information of different frequency components. Visualization tools such as Matplotlib can be used to plot the frequency-domain graph. The characteristic parameters of each frequency component are extracted, such as the center frequency, bandwidth, and energy distribution, providing basic data for subsequent vibration analysis. A suitable vibration analysis method is selected, such as vibration mode analysis, energy distribution analysis, or Power Spectral Density (PSD) analysis. Based on the acoustic frequency-domain graph of different frequencies, the energy distribution of each frequency component is calculated, and its change in the area to be detected is analyzed. The characteristic parameters of the regional vibration are extracted, including amplitude, frequency distribution, and phase information, to describe the vibration distribution characteristics within the area. By comparing the vibration characteristics of different frequencies, possible abnormal areas are identified. The vibration distribution characteristics are visualized to generate a vibration distribution map, intuitively showing the vibration situation in the area to be detected and helping to identify potential problems. The vibration mode parameters of the metal material are determined, including natural frequency, damping ratio, and vibration mode, etc. Understanding these parameters helps with subsequent feature mining. Appropriate modal analysis techniques are selected,Such as experimental modal analysis (EMA) or numerical modal analysis (FEM) to extract the vibration modal characteristics of the metal. Based on the detected vibration distribution characteristics, analyze the dynamic response of the metal material. By identifying the vibration frequency, judge the natural frequency of the metal and its corresponding modal characteristics, and record the mined metal vibration modal characteristics, including natural frequency, damping ratio, vibration mode, etc., to form a detailed report. These data will be used for subsequent structural health monitoring and evaluation.

[0037] In this embodiment, the specific steps of step S14 are as follows: Calculate the acoustic wave vibration frequency for the acoustic wave frequency domain diagrams of different frequencies to extract the vibration frequency of each frequency domain diagram; Calculate the vibration peak value for each frequency of the acoustic wave frequency domain diagrams of different frequencies to obtain the vibration peak value of each frequency domain diagram; Analyze the peak width of the vibration peak value of each frequency domain diagram to obtain the vibration peak width of each frequency domain diagram; Identify the vibration frequency offset according to the vibration peak width of each frequency domain diagram to generate the local frequency offset characteristics of each frequency domain diagram; Mine the time-series vibration changes of the local frequency offset characteristics of each frequency domain diagram and the vibration frequency of each frequency domain diagram, so as to generate the time-series vibration change characteristics of each frequency domain diagram; Generate the vibration distribution characteristics of the area to be detected according to the time-series vibration change characteristics of each frequency domain diagram for the evolution of the regional vibration distribution.

[0038] In this embodiment, first, ensure that the obtained acoustic frequency-domain map is the result of being processed by wavelet transform or Fourier transform, and can reflect the distribution characteristics of the acoustic signal at different frequencies. Use a signal processing tool (such as the find_peaks function in SciPy) to identify the peak positions in the frequency-domain map. Each peak corresponds to a vibration frequency, which can be obtained through the x-axis coordinate of the frequency-domain map. Extract the vibration frequencies of each frequency-domain map, and record the frequency of the highest peak in each frequency-domain map as the main vibration frequency of this map. This frequency usually reflects the main dynamic characteristics of the system. Store the extracted vibration frequencies in a database or data structure for subsequent analysis and comparison. Determine the criteria for calculating peaks, including peak thresholds, minimum spacing, etc., to ensure that the extracted peaks are significant and not noise interference. For each frequency-domain map, use a peak identification algorithm (such as find_peaks mentioned above) to calculate the vibration peaks one by one, and record the amplitude and corresponding frequency of each peak. Organize all the extracted peak information into structured data, including the frequency, amplitude of the peaks and their corresponding frequency-domain map numbers. The peak width is usually defined as the full width at half maximum (FWHM) of the peak, that is, the frequency range at half of the peak amplitude. For each extracted peak, calculate its corresponding half-height point, and use an interpolation method (such as linear interpolation) to accurately locate the corresponding frequency range in the frequency-domain map. Record the width of each peak and associate it with the frequency and amplitude of the peak to form a complete data set. Integrate the calculated peak width information into the previous peak data to form a comprehensive data set containing frequency, amplitude, and width. The vibration frequency offset refers to the change relative to the reference frequency, which may be caused by structural changes, material property changes, or external environmental influences. Select a reference frequency (usually the frequency in the normal state), compare the vibration frequency of each frequency-domain map with the reference frequency, calculate the frequency offset, and record the local frequency offset characteristics of each frequency-domain map, including the offset amount and its direction, to form an offset characteristic data set. Integrate the local frequency offset characteristics of all frequency-domain maps with the vibration frequencies into a time series data set for subsequent analysis. Conduct statistical analysis on the integrated time series data, calculate the time series change characteristics of each frequency-domain map, such as the change rate, fluctuation amplitude, etc., and apply time series analysis methods (such as autoregressive model, moving average model) to analyze the trends and periodic changes in the data and identify potential abnormal patterns. Select an appropriate model (such as spatial interpolation method, Kriging interpolation) to model the regional vibration distribution to draw the vibration distribution characteristics within the region. Apply the selected interpolation method to extend the time series vibration change characteristics of each frequency-domain map to the entire area to be detected to generate a regional vibration distribution map. Visualize the generated regional vibration distribution characteristics, and use heat maps or three-dimensional graphics to display the vibration characteristics to help identify abnormal vibration patterns within the region.

[0039] In this embodiment, refer to Figure 3, which is a schematic diagram of the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: Identify multi-sensor nodes based on a microarray ultrasonic sensor; Step S22: Calculate the acoustic wave reflection response of the multi-sensor nodes based on the high-frequency ultrasonic response signal, so as to obtain the acoustic wave propagation time of each node; Step S23: Identify the ultrasonic excitation path of the multi-sensor nodes, so as to generate the ultrasonic excitation trajectory of each node; Step S24: Perform ultrasonic spatial propagation modeling on the ultrasonic excitation trajectory based on each node according to the acoustic wave propagation time of each node, and construct a spatial ultrasonic trajectory propagation model.

[0040] In this embodiment, the microarray ultrasonic sensors are arranged in the area to be detected, ensuring that the distance between the sensors is small enough to capture the subtle characteristics of acoustic wave propagation. Usually, a uniform grid layout is selected. The sensor array is activated, a high-frequency ultrasonic signal is applied, and the response signals of each sensor are collected in real time. Through the array layout of the sensors, it is ensured that each sensor node can independently collect signals. Signal processing algorithms (such as the fast Fourier transform) are used to analyze the response signals of each sensor node to identify valid ultrasonic signals. By comparing the response characteristics of each node, it is determined which sensors can be regarded as valid nodes. A unique identifier is assigned to each identified sensor node for subsequent path identification and modeling analysis. A node database is established to record the position information and status of each node. The collected high-frequency ultrasonic response signals are denoised and filtered to ensure clear signals for accurate extraction of the acoustic wave propagation time. By analyzing the response signals of each node, the reflected echoes of the acoustic waves are identified. Signal processing tools (such as peak detection) are used to determine the time difference between the emission and reception of the acoustic waves. According to the propagation speed of the acoustic waves (usually about 343 m / s in air) and the measured propagation time, the propagation distance of the acoustic waves between each node is calculated. The acoustic wave propagation time of each node is recorded, and the correspondence between time and nodes is established. Methods for identifying the excitation path are determined. Commonly used ones include algorithms based on geometric distance and algorithms based on signal strength. The goal of path identification is to determine the specific path of the signal from emission to reception. According to the acoustic wave propagation time of each node and combined with the three-dimensional coordinate information of the nodes, the path of the acoustic wave propagation is drawn. Graphical tools (such as MATLAB or Matplotlib in Python) can be used for visual display to generate the acoustic wave excitation trajectory of each node. The path information of the acoustic wave propagation between each node is recorded to ensure that the path information accurately reflects the actual situation of the acoustic wave propagation, including any possible reflections or scattering. The generated ultrasonic excitation trajectory data is stored in the database for subsequent spatial propagation modeling. A suitable spatial propagation model is selected, such as the ray tracing model, the wave model, or the finite element model. The model selection should be based on the complexity and requirements of the system. According to the acoustic wave propagation time and excitation trajectory of each node, the model parameters are set, including the propagation speed, medium characteristics, and boundary conditions. Signal processing software or programming languages (such as MATLAB, Python) are used to implement the ultrasonic propagation model. By inputting the propagation time and excitation trajectory of each node, the propagation behavior of the acoustic wave in space is simulated. By comparing the simulation results with the actual measurement data, the accuracy of the model is verified, and necessary adjustments and optimizations are made to ensure that the model can effectively reflect the ultrasonic propagation characteristics.

[0041] In this embodiment, refer to Figure 4 , which is a schematic diagram of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: Perform acoustic beam localization detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics within the region to extract the acoustic response beam of the metal vibration mode; Step S32: Conduct acoustic attenuation analysis on the acoustic response beam of the metal vibration mode to generate the acoustic response beam attenuation characteristics; Step S33: Carry out diffusion multi-path reflection evolution on the acoustic response beam of the metal vibration mode to obtain the multi-path reflection evolution data of the acoustic response beam; Step S34: Perform propagation interference error analysis based on the acoustic response beam attenuation characteristics and the multi-path reflection evolution data of the acoustic response beam to obtain the response beam propagation interference error parameters; Step S35: Conduct propagation error compensation calculation on the response beam propagation interference error parameters to obtain the response beam propagation error compensation value.

[0042] In this embodiment, the characteristics of the metal vibration mode are clarified, including the natural frequency, damping ratio, and vibration mode. These characteristics will be used to guide the localization of the acoustic beam. A high-frequency ultrasonic signal is applied to the area to be detected, and the response signal is collected by a microarray ultrasonic sensor. During this process, ensure that the signal intensity and sampling frequency are high enough to capture the subtle vibration modes. Use time-frequency analysis techniques (such as short-time Fourier transform or wavelet transform) to process the collected ultrasonic response signal, extract the acoustic response beam that conforms to the metal vibration mode characteristics, and apply a beam localization algorithm (such as beamforming technology) to determine the specific position of the acoustic response beam. This process involves spatial filtering of the signal to improve the signal-to-noise ratio and directivity. Determine the relevant factors affecting acoustic attenuation, including propagation distance, medium characteristics, and frequency, etc., for subsequent analysis. Compare the intensities of the acoustic response beam at different receiving nodes, calculate the attenuation during the propagation of the acoustic wave, and the attenuation amount can be calculated using the logarithmic method. The formula is: Attenuation = , where p1 and p2 are the transmitted and received acoustic wave powers respectively. Record the attenuation characteristics of each node, including the attenuation amplitude and frequency characteristics, generate the attenuation characteristic data of the acoustic wave response beam, determine a suitable multipath reflection model. Commonly used ones are the ray tracing model or the finite element model. The selection should consider the environmental characteristics and material properties. According to the acoustic wave propagation characteristics, simulate the reflection path of the acoustic wave on the metal structure, record the propagation time and reflection characteristics of each path, use algorithms (such as the Monte Carlo method) to simulate the propagation and reflection of the acoustic wave on different paths, generate the multipath reflection evolution data of the acoustic wave response beam. This process needs to consider the propagation loss and phase change of different paths, determine the factors that may cause propagation interference, including environmental noise, equipment error, signal interference, and the complexity of the reflection path, integrate the attenuation characteristics and the multipath reflection evolution data, analyze the relationship between the two, identify the patterns of interference errors, use statistical analysis methods (such as regression analysis) to evaluate the impact of propagation interference on the acoustic wave response beam, calculate the interference error parameters. These parameters can include interference amplitude, frequency offset, etc., select a suitable error compensation algorithm. Commonly used ones are the Kalman filter, the least squares method, and machine learning algorithms. The selection should be based on the system requirements and data characteristics, apply the selected compensation algorithm to process the propagation interference error parameters, calculate the compensation value for each acoustic wave response beam, ensure that the compensated data reflects the true acoustic wave propagation characteristics, verify the compensation effect by comparing the compensated signal with the actual measurement data, and adjust the compensation parameters if necessary to optimize the result. Record the response beam propagation error compensation value in the database to provide support for subsequent analysis and application.

[0043] In this embodiment, step S4 includes the following steps: Step S41: Quantify the multi-sensor reflection distance difference of the acoustic wave response beam based on the metal vibration mode, so as to obtain the distance of the response beam reaching each sensor node; Step S42: Perform ultrasonic triangulation measurement on the spatial ultrasonic trajectory propagation model according to the distance of the response beam reaching each sensor node, so as to obtain the three-dimensional positioning coordinates of the metal waste; Step S43: Optimize the positioning error accuracy of the three-dimensional coordinates of the metal waste based on the response beam propagation error compensation value, so as to obtain the accurate three-dimensional coordinates of the metal waste.

[0044] In this embodiment, first, by applying high-frequency ultrasonic excitation in the metal waste area, an acoustic wave response beam is generated. The metal vibration mode characteristics have a significant impact on the propagation path and characteristics of ultrasonic waves. Therefore, the reflected signals captured by the high-frequency ultrasonic sensor array can provide feedback information about the metal surface, physical properties, and metal waste. Each sensor node will receive different response signals reflected from the metal waste area, and the signal propagation time is related to factors such as the propagation path, reflection angle, and attenuation. Perform a timing analysis on the reflected signals received by each sensor node, calculate the time difference of the signals received by each sensor node. By accurately recording the arrival time of the signals, the time difference of the reflected signals reaching each sensor can be quantified. The propagation speed of the acoustic wave is known. Based on the propagation time difference and the sound speed, the distance from the metal waste to each sensor can be deduced. This method relies on the Time Difference of Arrival (Time Difference of Arrival (TDOA), this method uses the time difference of signals received by different sensor nodes to estimate the position of the source. By the time difference of signals received by each sensor, combined with the known data of the ultrasonic propagation speed, the differential positioning algorithm (such as the TDOA method) is used to calculate the distance difference of the acoustic response beam reaching each sensor node. The distance difference information of multiple sensor nodes will provide the necessary data support for subsequent triangulation positioning. Based on the propagation distance data between multiple sensor nodes and the metal waste obtained from step S41, it is first necessary to construct an ultrasonic triangulation positioning model. The triangulation positioning principle relies on three or more known sensor nodes. By measuring the distance between each node and the target (i.e., the metal waste), the precise position of the target can be calculated. Specifically, when implementing, by constructing a spatial coordinate system and inputting the known positions of each sensor node and the corresponding distance information into the positioning model, an ultrasonic positioning equation set is formed. By solving these equations, the three-dimensional spatial coordinates of the metal waste can be obtained. Numerical solution methods such as the least squares method or the weighted least squares method (WLS) are used for spatial positioning according to the distance data from multiple sensor nodes to the metal waste. The least squares method optimizes the solution result by minimizing the difference between the actual measurement value and the predicted value, thereby calculating the precise position of the metal waste. To ensure accuracy, it may be necessary to use multiple sensor nodes at different positions for multiple measurements and integrate the measurement data of all sensor nodes for three-dimensional positioning. Through the average value of multiple measurements, the errors in single measurements can be eliminated, thereby improving the overall positioning accuracy. During the positioning process, the positioning result may be affected by factors such as measurement errors and signal attenuation. Therefore, the system will further improve the accuracy through error analysis and data weighting. If the sensor distribution is uneven or the signal quality of some sensors is poor, the weighted average method can be used to assign higher weights to sensors with good signal quality, thereby improving the positioning accuracy. The three-dimensional positioning coordinates of the metal waste obtained in step S42 may be affected by various factors, such as the nonlinearity of the ultrasonic signal propagation path, environmental noise, and sensor errors. To improve the positioning accuracy, it is necessary to conduct error analysis on the positioning result. First, analyze the propagation time difference from each sensor to the metal waste, and combine environmental factors (such as temperature, humidity, etc.) and the noise characteristics of the system to quantify the error and calculate the positioning error that each sensor may cause. Based on the error analysis, error compensation algorithms such as Kalman filtering and least squares error optimization are used to correct the positioning result. The Kalman filtering algorithm estimates and compensates for the signal propagation error in real time by combining the measurement data of multiple sensors and the dynamic model of the system, thereby improving the positioning accuracy.,

[0045] In addition, the compensation coefficients for propagation errors can be determined through simulation or experiments, and these compensation coefficients are applied to correct the three-dimensional positioning coordinates. By effectively compensating for the errors, the deviations caused by environmental changes or sensor errors can be reduced. Once the error compensation is completed, the system will optimize the accuracy of the three-dimensional positioning coordinates of the metal waste. This can be achieved by introducing accuracy evaluation indicators (such as positioning accuracy, error range) to check the optimization effect. For scenarios with high-precision requirements, it may also be necessary to adopt methods of multiple measurements and comprehensive data to further improve the robustness and accuracy of positioning. For example, by repeatedly measuring and accumulating feedback, the positioning algorithm and parameters are gradually corrected to achieve the required accuracy.

[0046] In this embodiment, step S5 includes the following steps: Step S51: Collect multi-spectral images based on the accurate three-dimensional coordinates of the metal waste to obtain high-resolution multi-spectral images of the metal surface; Step S52: Perform image detail sharpening on the high-resolution multi-spectral images of the metal surface to obtain sharpened and enhanced multi-spectral images; Step S53: Decompose the pixel points of the sharpened and enhanced multi-spectral images to extract all the pixel points in the spectral images; Step S54: Conduct multi-spectral channel feature analysis based on all the pixel points in the spectral images to obtain multi-spectral channel features; Step S55: Perform quantitative deduction of metal components on the multi-spectral channel features to obtain quantitative information on the components of the metal waste.

[0047] In this embodiment, a suitable multispectral imaging device is selected to ensure that it can perform high-resolution imaging on multiple spectral channels. The device should have good spectral resolution and spatial resolution. Place the metal waste in an environment with uniform illumination, avoiding the interference of direct sunlight and shadows. During the acquisition process, ensure that the sample surface is clean to obtain the best image quality. According to the preset three-dimensional coordinates, use the multispectral imaging device to scan the metal waste and collect its multispectral image on the surface. During the acquisition process, record the spectral channel information of each image to ensure the integrity of the spectral data of each pixel. Store the collected multispectral image data in a computer and establish a data management system for convenient subsequent processing and analysis. Select a suitable image processing software (such as MATLAB, the OpenCV library in Python, or other professional image processing tools) for image sharpening. Adopt common sharpening algorithms, such as the Laplacian operator, enhanced edge detection, or high-pass filtering. The selected algorithm should be able to highlight the details in the image and improve the image clarity. Input the collected multispectral image into the image processing software, apply the selected sharpening algorithm, and pay attention to adjusting the sharpening parameters during the execution to avoid noise generated by oversharpening the image. Generate a sharpened and enhanced multispectral image to ensure that the image details are clear and important spectral information is retained for subsequent analysis. Select a suitable image processing tool that can perform pixel-level analysis on the multispectral image. Usually, use the NumPy and OpenCV libraries in Python for processing. Load the sharpened multispectral image into the processing program, obtain its pixel data, decompose the image into pixel points, and extract the spectral information of each pixel. This process usually involves looping through each pixel and recording its values in different spectral channels. Determine a suitable feature analysis method. Commonly used methods include principal component analysis (PCA), independent component analysis (ICA), or spectral feature extraction algorithms. Preprocess the extracted pixel point data, including denoising and standardization, to ensure the quality and consistency of the data. Apply the selected feature analysis algorithm to process the spectral data of the pixel points and extract the characteristic parameters of each spectral channel. These characteristics may include mean, variance, peak position, etc. Generate multispectral channel feature data, record the numerical value of each feature and its corresponding channel information to form a feature matrix. Save the extracted multispectral channel feature data to a database or file for subsequent component analysis and quantitative deduction. Select a suitable quantitative analysis model. Commonly used models include machine learning models such as linear regression, support vector machine (SVM), or random forest. Prepare a training dataset, including metal samples with known components and their corresponding spectral characteristics. These data will be used to train the model. Use the training data to train the selected deduction model, adjust the model parameters to improve the prediction accuracy, and ensure that the model can effectively map the relationship between spectral characteristics and metal components. Apply the trained model to deduce the extracted multispectral channel features.Calculate the quantitative information of the components of the metal waste, and record the concentration or proportion of each component.

[0048] In this embodiment, step S6 includes the following steps: Step S61: Classify the characteristics of the metal waste based on the quantitative information of the components of the metal waste, so as to obtain the type of the metal waste; Step S62: Dynamically visualize and mark the spatial ultrasonic trajectory propagation model based on the type of the metal waste and the precise three-dimensional coordinates of the metal waste to obtain a dynamic visualization model; Step S63: Optimize the dynamic visualization model through enhanced detection learning to construct an intelligent detection model for metal waste.

[0049] In this embodiment, the previously obtained quantitative information on the composition of metal waste is sorted to ensure that the data is complete and in a consistent format. The composition information includes the concentration and ratio of different metal elements. Suitable features are selected for classification, usually including the concentration, specific gravity, physical and chemical properties of the metal components, etc., to ensure that the selected features can effectively distinguish different types of metal waste. Suitable machine learning classification models are selected, and commonly used ones include decision trees, support vector machines (SVMs), random forests or neural networks, etc. The model selection should be based on the complexity of the data and the classification requirements. A training data set of known types of metal waste is prepared, and the model is trained to learn the relationship between the composition features and the categories. The model performance is evaluated using methods such as cross-validation to ensure the accuracy of the classification. The quantitative information on the composition of the metal waste is input into the trained model to perform type prediction, and the classification results of each type of metal waste are recorded. According to the model output, each type of metal waste is marked, and suitable visualization tools and software (such as Unity, Blender, etc.) are selected. or ParaView) to build a dynamic visualization model, ensure that the tool can process three-dimensional data and support dynamic effects, integrate the precise three-dimensional coordinates and classification information of metal waste into the visualization software, build a three-dimensional model of metal waste, and mark each type of metal waste with a different color or shape for easy distinction. Implement dynamic effects in the model, such as state changes and position movements of metal waste. Use animation and interactive functions to show the dynamic behavior of metal waste in the environment, optimize visualization effects, ensure that the model performs well from different perspectives, and that the dynamic effects are smooth. Consider adding effects such as lighting and shadows to improve the realism and ornamental value of the model. Save the file of the dynamic visualization model in a format that can be displayed and analyzed to ensure its compatibility in subsequent use. Select a suitable reinforcement learning framework (such as TensorFlow, PyTorch, or OpenAI). Gym) to train and optimize the model, create a reinforcement learning environment, and integrate the dynamic visualization model into it. The environment should be able to simulate the detection scenario of metal waste, including sensor feedback and environmental changes. Design an intelligent agent to be responsible for performing detection tasks in the environment. The intelligent agent should be able to adjust its strategy based on environmental feedback to improve detection efficiency and accuracy. Use reinforcement learning algorithms (such as Q-learning, deep Q networks, etc.) to train the intelligent agent. Through continuous trial and error learning, optimize the detection strategy and improve the detection capability of metal waste. Evaluate the performance of the intelligent detection model, use indicators such as detection accuracy and response time for evaluation, adjust and improve the model based on the results, and ensure the effectiveness of the intelligent agent in practical applications.

[0050] In this embodiment, a metal waste detection and identification system is provided, which is used to execute the metal waste detection and identification method as described above, including: A metal vibration mode module, which is used to apply high-frequency ultrasonic waves to the area of waste to be detected, collect high-frequency ultrasonic response signals, and mine metal vibration mode parameters from the high-frequency ultrasonic response signals, so as to obtain the metal vibration mode characteristics in the area; An ultrasonic trajectory module, which is used to identify the ultrasonic excitation path of the high-frequency ultrasonic response signal, perform ultrasonic spatial propagation modeling, and construct a spatial ultrasonic trajectory propagation model; A propagation error compensation module, which is used to perform acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the area, and perform propagation error compensation calculation, so as to obtain the response beam propagation error compensation value; A triangulation positioning measurement module, which is used to perform ultrasonic triangulation positioning measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the accurate three-dimensional coordinates of the metal waste; A quantitative deduction module, which is used to collect multi-spectral images based on the accurate three-dimensional coordinates of the metal waste, and then perform quantitative deduction of the metal components, so as to obtain the quantitative information of the metal waste components; An intelligent detection module, which is used to classify the characteristics of the metal waste and optimize the enhanced detection learning based on the quantitative information of the metal waste components, so as to construct an intelligent detection model for the metal waste.

[0051] Through the extraction of the metal vibration mode, the physical properties of the metal can be accurately reflected, such as the metal surface state, microstructure and its deformation characteristics, which is of great significance for the early identification and qualitative analysis of metal waste. The metal vibration mode characteristics provide the "fingerprints" of different metal types or surface damages, which can quickly identify the regional characteristics of the waste and provide key data for subsequent ultrasonic wave propagation modeling and error compensation. By mining the metal vibration mode, the response of the ultrasonic sensor can be optimized, the signal quality can be enhanced, and a foundation for efficient positioning and detection can be laid. The identification of the ultrasonic wave propagation path can help to construct an accurate propagation model, thus avoiding the influence of errors during the waste detection process and ensuring that the signal can accurately cover the area to be detected. By establishing the propagation model, the propagation path and reflection of the ultrasonic wave signal can be predicted in advance, the detection path can be optimized, and the signal attenuation or error can be reduced. Through the spatial ultrasonic wave trajectory model, the propagation trajectory of the ultrasonic wave in space can be accurately described, which helps to provide accurate path data support for subsequent triangulation positioning measurement. Through propagation error compensation, the propagation errors caused by factors such as medium change and reflection path can be corrected, thereby improving the positioning accuracy of metal waste. This module overcomes the influence of environmental factors such as multipath effect and attenuation by compensating the propagation error, enabling the ultrasonic wave signal to be transmitted more accurately to the target area and ensuring the reliability of the detection result. By calculating the propagation error compensation value of the response beam, the positioning of the acoustic beam is ensured to be more accurate, enhancing the detection accuracy of metal waste. Using the triangulation positioning technology, through multiple known sensor nodes and the propagation model, accurate spatial positioning measurement can be carried out, and the three-dimensional coordinates of the metal waste can be efficiently calculated. Combining with the propagation error compensation value, the accuracy of the triangulation positioning measurement is optimized, ensuring the accuracy of the waste positioning. Through the accurate three-dimensional coordinates, the three-dimensional modeling of the metal waste area can be supported, providing a scientific basis for further analysis, recycling or environmental assessment. Combining the three-dimensional positioning information and the multispectral image, the chemical composition and physical properties of the metal waste can be accurately extracted for quantitative analysis. The multispectral image acquisition provides the reflection information of the metal surface in different bands, which can help the system to reveal different components of the metal waste and then conduct component deduction to identify valuable metals. The accurate metal component deduction enables the recycling system to more effectively identify and sort different types of metals, improving the recycling efficiency and resource utilization rate of metal waste. By analyzing the component information of the metal waste, the waste can be automatically classified, providing intelligent support for recycling, reuse and environmental management. The introduction of the reinforcement learning algorithm enables the system to continuously learn and optimize in practical applications, automatically adjust the detection strategy according to new data, and improve the flexibility and accuracy of detection. The intelligent detection model can adaptively adjust the detection parameters in a complex environment, effectively identify various metal wastes with different forms and types, and enhance the robustness and universality of the system. Through intelligent optimization, the detection speed and accuracy of metal waste can be improved, and manual intervention can be reduced.Improve the overall system efficiency.,

[0052] Therefore, from every point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0053] As described above, these are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A method for detecting and identifying metal waste, characterized in that: The following steps are involved: Step S1: applying high-frequency ultrasonic waves to the waste area to be inspected, collecting high-frequency ultrasonic wave response signals; mining the metal vibration modal parameters of the high-frequency ultrasonic wave response signals, thereby obtaining the metal vibration modal characteristics in the area; Step S2: performing ultrasonic excitation path identification on the high-frequency ultrasonic response signal, and performing ultrasonic spatial propagation modeling to construct a spatial ultrasonic trajectory propagation model; Step S3: performing acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration modal characteristics in the area, and performing propagation error compensation calculation to obtain a response beam propagation error compensation value; Step S4: performing ultrasonic triangulation positioning measurement on the spatial ultrasonic trajectory propagation model, and optimizing the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the precise three-dimensional coordinates of the metal waste; Step S5: multi-spectral image acquisition based on the precise three-dimensional coordinates of the metal waste, and then quantitative deduction of the metal components, so as to obtain quantitative information on the composition of the metal waste; Step S6: Based on the quantitative information of metal waste components, metal waste feature classification and enhanced detection learning optimization are performed to build a metal waste intelligent detection model.

2. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S1 are: Step S11: applying high-frequency ultrasonic waves to the waste area to be detected through a microarray ultrasonic sensor, and collecting high-frequency ultrasonic response signals; Step S12: performing noise filtering and noise reduction on the high-frequency ultrasonic response signal, and extracting the filtered and noise-reduced ultrasonic signal; Step S13: performing multi-scale frequency decomposition on the filtered and denoised ultrasonic signal, thereby obtaining frequency domain graphs of sound waves of different frequencies; Step S14: performing regional vibration distribution evolution based on the sound wave frequency domain diagrams of different frequencies to generate vibration distribution characteristics of the area to be detected; Step S15: mining metal vibration modal parameters on the vibration distribution characteristics of the area to be detected, thereby obtaining metal vibration modal characteristics in the area.

3. The method for detecting and identifying metal waste according to claim 2, characterized in that: The specific steps of step S14 are: Calculate the sound wave vibration frequency for the sound wave frequency domain graphs of different frequencies to extract the vibration frequency of each frequency domain graph; The vibration peak values ​​of the frequency domain graphs of sound waves of different frequencies are calculated one by one to obtain the vibration peak value of each frequency domain graph; Performing a peak width analysis on the vibration peak of each frequency domain graph to obtain the vibration peak width of each frequency domain graph; Vibration frequency offset is identified according to the vibration peak width of each frequency domain graph, and a local frequency offset feature of each frequency domain graph is generated; Mining the time-series vibration changes of the local frequency offset characteristics of each frequency domain graph and the vibration frequency of each frequency domain graph, thereby generating the time-series vibration change characteristics of each frequency domain graph; The regional vibration distribution is evolved according to the time-series vibration variation characteristics of each frequency domain graph to generate the vibration distribution characteristics of the area to be detected.

4. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: identifying multi-sensor nodes based on micro-array ultrasonic sensors; Step S22: calculating the acoustic wave reflection response of the multi-sensor nodes based on the high-frequency ultrasonic response signal, thereby obtaining the acoustic wave propagation time of each node; Step S23: performing ultrasonic excitation path identification on multiple sensor nodes, thereby generating an ultrasonic excitation trajectory for each node; Step S24: ultrasonic spatial propagation modeling is performed on the ultrasonic excitation trajectory based on each node according to the sound wave propagation time of each node, so as to construct a spatial ultrasonic trajectory propagation model.

5. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: performing acoustic wave beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the area to extract the acoustic wave response beam of the metal vibration mode; Step S32: performing acoustic wave attenuation analysis on the acoustic wave response beam of the metal vibration mode to generate an acoustic wave response beam attenuation feature; Step S33: performing diffuse multipath reflection evolution on the acoustic wave response beam of the metal vibration mode, thereby obtaining multipath reflection evolution data of the acoustic wave response beam; Step S34: performing propagation interference error analysis based on the acoustic wave response beam attenuation characteristics and the acoustic wave response beam multipath reflection evolution data, thereby obtaining the response beam propagation interference error parameters; Step S35: performing propagation error compensation calculation on the response beam propagation interference error parameter, thereby obtaining a response beam propagation error compensation value.

6. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: quantifying the multi-sensor reflection distance difference based on the acoustic wave response beam of the metal vibration mode, thereby obtaining the distance from the response beam to each sensor node; Step S42: performing ultrasonic triangulation positioning measurement on the spatial ultrasonic trajectory propagation model according to the distance from the response beam to each sensor node, thereby obtaining the three-dimensional positioning coordinates of the metal waste; Step S43: optimizing the positioning error accuracy of the three-dimensional positioning coordinates of the metal waste based on the response beam propagation error compensation value, thereby obtaining the precise three-dimensional coordinates of the metal waste.

7. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: performing multispectral image acquisition based on the precise three-dimensional coordinates of the metal waste to obtain a high-resolution multispectral image of the metal surface; Step S52: performing image detail sharpening processing on the high-resolution multispectral image of the metal surface, thereby obtaining a sharpened and enhanced multispectral image; Step S53: performing pixel decomposition on the sharpened and enhanced multi-spectral image to extract all the pixels in the spectrum image; Step S54: performing multi-spectral channel feature analysis based on all pixel points in the spectrum map to obtain multi-spectral channel features; Step S55: Quantitatively deduce the metal components of the multi-spectral channel features to obtain quantitative information on the metal waste components.

8. The method for detecting and identifying metal waste according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: classifying metal waste characteristics based on the quantitative information of metal waste components, thereby obtaining metal waste types; Step S62: dynamically visually marking the spatial ultrasonic trajectory propagation model based on the type of metal waste and the precise three-dimensional coordinates of the metal waste to obtain a dynamic visualization model; Step S63: Perform enhanced detection learning optimization on the dynamic visualization model to build a metal waste intelligent detection model.

9. A metal waste detection and identification system, characterized in that: The method for detecting and identifying metal waste according to claim 1 comprises: The metal vibration mode module is used to apply high-frequency ultrasonic waves to the waste area to be inspected and collect high-frequency ultrasonic response signals; the metal vibration mode parameters of the high-frequency ultrasonic response signals are mined to obtain the metal vibration mode characteristics in the area; Ultrasonic trajectory module, used to identify the ultrasonic excitation path of high-frequency ultrasonic response signals, model ultrasonic spatial propagation, and construct a spatial ultrasonic trajectory propagation model; A propagation error compensation module is used to perform acoustic beam positioning detection on the high-frequency ultrasonic response signal based on the metal vibration mode characteristics in the area, and perform propagation error compensation calculation to obtain a response beam propagation error compensation value; The triangulation positioning measurement module is used to perform ultrasonic triangulation positioning measurement on the spatial ultrasonic trajectory propagation model, and optimize the positioning error accuracy based on the response beam propagation error compensation value, so as to obtain the precise three-dimensional coordinates of the metal waste; The quantitative deduction module is used to collect multi-spectral images based on the precise three-dimensional coordinates of metal waste, and then perform quantitative deduction of metal components to obtain quantitative information on the composition of metal waste; The intelligent detection module is used to classify metal waste characteristics and enhance detection learning optimization based on the quantitative information of metal waste components, thereby building an intelligent detection model for metal waste.

Images