An automated bolt cleaning, inspection, and control system and method

Abstract

This invention relates to the field of intelligent detection and control technology, specifically disclosing an automated bolt cleaning and detection control system and method. The system includes an intelligent control and dispatch center, which is communicatively connected to the following modules: a rust self-adaptive cleaning module, used to integrate an intelligent sensing system and a parameter optimization ant colony algorithm to identify bolt rust levels in real time and dynamically optimize cleaning parameters to clean bolts with different rust levels; and a residual liquid interference prediction module, used to combine wireless networking and twin simulation to collect drying process data from multiple angles at high frequency, simulate the residual liquid evaporation path, and predict interference areas. This invention uses multi-modal sensing fusion technology to identify bolt rust levels in real time and combines it with an intelligent optimization algorithm to dynamically match cleaning parameters. It can output differentiated operation instructions for bolts with different rust levels in the same batch, achieving differentiated adaptive cleaning of rust, avoiding excessive damage and residue, and significantly improving cleaning quality and consistency.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection and control technology, and in particular to an automated bolt cleaning and detection control system and method. Background Technology

[0002] Bolts are key connecting components in mechanical structures, and their surface condition directly affects the connection strength and overall safety. In traditional manual cleaning and inspection processes, due to inconsistent operating standards and limited inspection accuracy, incomplete cleaning or omissions in inspection are prone to occur, which in turn affects product reliability. With the improvement of industrial automation, using automated equipment to clean and inspect bolts can achieve standardized and controllable operation processes, improving cleaning quality and inspection accuracy.

[0003] In existing technologies, when bolts with different degrees of corrosion are mixed in the same batch, a single cleaning parameter can cause residual corrosion or excessive damage to the thread root. Furthermore, residual cleaning fluid and microparticles on the bolt surface can interfere with visual inspection, leading to misjudgments of incomplete cleaning or rust recurrence. At the same time, bolts that have been cleaned and inspected are mixed into the cleaning queue, causing the risk of reprocessing qualified parts or reassembling missed parts. Ultimately, this restricts the reliability of the closed-loop control of the entire process of automated bolt cleaning. To address these issues, an automated bolt cleaning and inspection control system and method are proposed. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, the present invention provides an automated bolt cleaning and inspection control system and method, which can effectively solve the problems involved in the prior art.

[0005] The objective of this invention can be achieved through the following technical solution: Firstly, this invention provides an automated bolt cleaning and inspection control system, including an intelligent control and dispatch center, which is communicatively connected to the following modules: The rust self-adaptive cleaning module integrates an intelligent sensing system with a parameter optimization ant colony algorithm to identify the bolt rust level in real time and dynamically optimize cleaning parameters to clean bolts with different rust levels, effectively avoiding rust residue at the thread root and excessive damage to the substrate. The residual liquid interference prediction module is used to combine wireless networking and twin simulation to collect drying process data from multiple angles and at high frequency, simulate the evaporation path of residual liquid and predict the interference area, dynamically adjust the purging angle and hot air temperature, and eliminate the misjudgment interference of residual liquid on visual detection. The batch logistics anti-mixing module is used to assign a unique RFID batch code to each cleaning material box. The position of each bolt in the material box is fixed according to the matrix coordinates. A matrix positional status map is constructed using a digital identity strategy containing batch code and coordinate position. Through the unique RFID batch code of the material box and the matrix coordinates of the bolt in the box, the status of each bolt in the batch can be traced, preventing qualified parts from being mixed with unwashed parts in the cleaning and inspection process. The intelligent sorting and reconfiguration module, based on a matrix-style positional state graph, introduces a multi-agent system to coordinate the action sequence and load balancing of multiple sorting stations. At the same time, it uses a path reconstruction ant colony algorithm to dynamically reconstruct the sorting order and grasping path of the coordinate positions, minimizing the idle travel of the robotic arm and the waiting time in the queue. After sorting is completed, the state matrix is ​​cleared for each coordinate position, achieving strong consistency between physical sorting and data closure. This prevents qualified parts from being mixed into the waiting queue for cleaning or the return of missed parts, and achieves accurate diversion of qualified, rework, and scrapped parts.

[0006] Preferably, the rust self-adaptive cleaning module includes a rust layer multi-mode sensing unit and a cleaning parameter optimization unit; The rust layer multi-mode sensing unit is used to acquire the rust layer thickness, density and texture features of the bolt to be cleaned in real time using an intelligent sensing system based on laser contour scanning and impedance spectrum detection, identify the bolt corrosion level and generate a corrosion level map, so as to achieve accurate quantification and classification of the degree of corrosion. The cleaning parameter optimization unit, based on the rust level map, runs a parameter optimization ant colony algorithm to iteratively optimize the sandblasting pressure, cleaning time, and abrasive flow rate of the automated bolt cleaning equipment, and outputs differentiated cleaning instructions to the first-stage actuator to adaptively clean bolts with different rust levels, ensuring that each bolt is cleaned with the optimal parameters.

[0007] Preferably, the rust layer multi-mode sensing unit specifically includes: In response to the bolts to be cleaned entering the detection station of the automated bolt cleaning equipment, the laser profile scanner in the deployed intelligent sensing system is triggered to perform multi-section spiral scanning along the bolt axis, acquire three-dimensional point cloud data of the thread surface, extract rust layer thickness and texture feature parameters, and achieve high-precision three-dimensional reconstruction of rust layer morphology. The impedance spectrum detection probe is simultaneously activated to contact the bolt surface under test, a frequency conversion AC excitation signal is applied, and the amplitude frequency and phase frequency response curves are collected. The density and porosity distribution of the rust layer are calculated by inversion to reveal the compactness and porosity of the rust layer. By integrating laser profilometry and impedance spectroscopy detection results, the bolt corrosion level is identified according to the preset bolt corrosion classification threshold, and a corrosion level map of each bolt is generated. The map data is then packaged and uploaded to the cleaning parameter optimization unit to provide accurate level basis for differentiated cleaning.

[0008] Preferably, the cleaning parameter optimization unit specifically includes: Receive the rust level map, analyze the rust layer thickness, density and texture feature value corresponding to each bolt, construct the initial solution space with sandblasting pressure, cleaning time and abrasive flow rate as decision variables, and realize the accurate matching of cleaning parameters and rust state; The pheromone matrix of the parameter optimization ant colony algorithm is initialized, and the corrosion level map is mapped as a path heuristic factor to drive the artificial ants to iteratively search in the solution space. The parameter optimization is carried out with cleaning residue rate and matrix damage degree as multi-objective constraints to ensure that the optimal cleaning combination is quickly approached under multi-objective constraints. The optimal combination of cleaning parameters is converged and output, generating differentiated cleaning commands and sending them to the first-stage actuator. These commands control the sandblasting valve opening, the cleaning timer, and the abrasive supply pump speed, respectively, thus achieving automated closed-loop control of the differentiated cleaning commands.

[0009] Preferably, the residual liquid interference prediction module includes a drying process twin acquisition unit and a thermal field dynamic correction unit; The drying process twin acquisition unit is used to deploy multi-point temperature and humidity sensors and high-frequency cameras to collect bolt surface drying process data from multiple angles through wireless networking, construct a dynamic twin data body of residual liquid evaporation path, and capture residual liquid distribution and evaporation dynamics in real time. The thermal field dynamic correction unit is used to combine the dynamic twin data of the residual liquid evaporation path to simulate and analyze the residual liquid accumulation area and the interference shadow area, and drive the second-stage actuator to adaptively adjust the purging angle, hot air temperature and action time to accurately remove all residual liquid accumulation areas and ensure that the bolt surface is clean and dry.

[0010] Preferably, the drying process twin acquisition unit specifically includes: Multiple temperature and humidity sensors and high-frequency cameras are deployed around the drying station of the automated bolt cleaning equipment. Through wireless networking, each sensor is synchronously triggered to collect multi-source heterogeneous data on the bolt surface, including temperature, humidity and liquid film distribution images of different areas on the bolt surface, so as to achieve accurate synchronous collection of multi-dimensional data in the drying process. The collected multi-source heterogeneous data is mapped to a preset three-dimensional bolt digital model. The residual liquid evaporation rate of each coordinate point is marked according to the time series, and a dynamic twin data body of the residual liquid evaporation path is constructed to dynamically restore the spatiotemporal evolution of the entire residual liquid evaporation process. The dynamic twin data volume is continuously updated iteratively to identify the spatial coordinates of the residual liquid accumulation area and the evaporation lag point, generate the residual liquid evaporation path map and push it to the thermal field dynamic correction unit, so as to provide accurate residual liquid distribution basis for thermal field regulation.

[0011] Preferably, the dynamic thermal field correction unit specifically includes: Receive the residual liquid evaporation path map, import it into the heat flow simulation model, simulate the temperature field and airflow field distribution of the residual liquid accumulation area under different combinations of purging angle and hot air temperature, effectively locate the residual liquid accumulation area and avoid drying blind spots; Based on the simulation results, the boundary conditions of the interference shadow area are identified, and the minimum hot air temperature, optimal purging angle and action time parameters required to eliminate residual liquid accumulation are calculated in reverse to accurately eliminate interference shadows and prevent visual misjudgment. The calculated thermal field correction parameters are converted into drive commands for the second-stage actuator, which adjust the swing angle of the hot air nozzle, the duty cycle of the heating wire, and the purge timer to achieve adaptive drying control and ensure surface cleanliness.

[0012] Preferably, the batch logistics anti-mixing module specifically includes: At the feeding station of the automated bolt cleaning equipment, a unique RFID batch code is bound to each cleaning basket. At the same time, the presence or absence of bolts at each matrix coordinate position in the basket is identified by a visual positioning system, and an initial mapping relationship between the batch code and the coordinate position is established to ensure that each bolt obtains a unique traceable identity from the source. Fixed RFID readers and position sensors are deployed at each processing station to read the batch code of the material frame in real time and associate it with the processing result of the current station. The status is updated one by one according to the coordinate position as qualified, rework or scrap, so that the status of a single bolt is dynamically updated with the process and information lag and misalignment are eliminated. The status markers reported by each workstation are written into a matrix-style positional status map according to the time sequence. The status of each bolt is traceable and anti-confusion is achieved through the combination key of batch code and coordinate position. This prevents the status of bolts from being confused due to similar appearance and ensures that the data of the whole process is consistent and reliable.

[0013] Preferably, the intelligent sorting and reconfiguration module specifically includes: Read the matrix-style positional state map, extract the state markers and physical coordinates of each coordinate position, introduce a multi-agent system based on multiple sorting stations, allocate the target coordinate list of bolts to be sorted according to the load capacity of each sorting station, ensure that the load of each station is balanced, and avoid overload or idleness. The pheromone matrix of the ant colony algorithm is initialized and the Euclidean distance between each coordinate position is used as the path cost factor. The grasping order and the movement path of the robotic arm that minimize the empty stroke are dynamically iteratively searched, which significantly shortens the empty stroke of the robotic arm and improves sorting efficiency. The sorting process is executed sequentially along the optimized path, transferring qualified, reworked, or scrapped parts to their respective silos. After sorting is completed, the corresponding records in the matrix-style positional status map are cleared at each coordinate position, and the bolt cleaning results are displayed on the visualization interface of the intelligent control and scheduling center. This achieves strong consistency between physical sorting and data closed loop, eliminating the risk of mixed materials and missed inspections.

[0014] Secondly, the present invention also provides an automated bolt cleaning and detection control method, which is based on the aforementioned automated bolt cleaning and detection control system, and includes the following steps: S1. The bolts to be cleaned enter the automated bolt cleaning equipment, triggering laser profile and impedance spectrum fusion detection to extract thickness, texture, density and texture features. The bolt corrosion level is classified according to the bolt corrosion classification threshold and a corrosion level map is generated to achieve accurate quantification and graded identification of the degree of corrosion. S2. Analyze the rust level map, construct the solution space of sandblasting pressure, duration and flow rate, run the parameter optimization ant colony algorithm for iterative optimization, output differentiated cleaning instructions to the first stage actuator, realize graded adaptive cleaning, and avoid residual material at the root of the thread and over-cleaning damage to the substrate. S3. After cleaning, the bolts enter the drying station. The wireless network triggers the temperature and humidity sensor and the high-frequency camera to collect multi-source heterogeneous data and map it to the preset three-dimensional bolt digital model to construct a dynamic twin data body of the residual liquid evaporation path and predict the residual liquid accumulation area and evaporation lag point in advance. S4. Import the dynamic twin data volume into the thermal flow simulation model, traverse the nozzle angle and air temperature combination, reverse calculate the optimal purging parameters, drive the second-stage actuator to adjust the thermal field, and ensure that the bolt surface is clean and dry. S5. Bind RFID batch codes to the material frame, establish coordinate mapping for visual positioning, and link the processing results of readers and writers at each station in real time to construct a matrix-style positional status map, realize full-process traceability of single bolts, and prevent qualified parts from being confused with unwashed parts in the process. S6. Based on the matrix-style positional state graph, a multi-agent system is introduced to allocate sorting tasks. The path reconstruction ant colony algorithm reconstructs the grasping path, and qualified, reworked and scrapped parts are sorted in parallel. The matrix-style positional state graph is cleared and the database is synchronized to complete the closed-loop control of the whole process and eliminate the risk of repeated processing of qualified parts and repackaging of missed parts.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This automated bolt cleaning and inspection control system and method uses multi-modal sensing fusion technology to identify the bolt corrosion level in real time and combines intelligent optimization algorithm to dynamically match cleaning parameters. It can output differentiated operation instructions for bolts with different corrosion levels in the same batch, realize differentiated adaptive cleaning of corrosion, avoid excessive damage and residue, and significantly improve cleaning quality and consistency.

[0016] 2. This automated bolt cleaning and inspection control system and method uses wireless networking and digital twin technology to perform high-frequency acquisition and dynamic simulation of the drying process. It can accurately predict the residual liquid accumulation area and adaptively adjust the thermal field parameters to ensure that the bolt surface is thoroughly dry and clean, eliminate the interference of residual liquid on visual inspection, improve the detection accuracy, and significantly reduce the probability of misjudging clean bolts as unclean or rusted bolts.

[0017] 3. This automated bolt cleaning and inspection control system and method binds a unique electronic identification to each material box and constructs a matrix-style status map by combining a coordinate position solidification strategy. At each workstation, the processing results are correlated in real time through automatic identification technology, so as to realize accurate recording and incremental updating of the status of each bolt in the batch, prevent material confusion, and reduce the risk of confusion between qualified and unwashed parts in the cleaning and inspection process.

[0018] 4. This automated bolt cleaning and inspection control system and method is based on the collaborative control of a matrix-type positional state map and a multi-agent system. It performs precise sorting according to the final state of the bolts and simultaneously clears the state records after sorting, ensuring a high degree of consistency between physical disposal and information data. This avoids qualified parts being processed repeatedly or missed parts re-entering the repackaging process, thus enhancing the reliability of the entire process. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the workflow of an automated bolt cleaning and inspection control system and method according to the present invention. Figure 2 This is a schematic diagram of the system architecture of an automated bolt cleaning and detection control system according to the present invention; Figure 3 This is a cloud map showing the optimization and fitness analysis of the pressure-time parameter for bolt sandblasting cleaning in an automated bolt cleaning and detection control system and method of the present invention. Figure 4 This invention provides a surface temperature field distribution and residual liquid accumulation zone location cloud map for the bolt hot air drying process in an automated bolt cleaning and inspection control system and method. Figure 5 This is a timing diagram of the control logic for an automated bolt cleaning and inspection control system and method according to the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0021] Example 1, please refer to Figures 1 to 5This invention provides a technical solution: an automated bolt cleaning and inspection control system, including an intelligent control and dispatch center, which is connected to the following modules for communication: The rust self-adaptive cleaning module integrates an intelligent sensing system and a parameter optimization ant colony algorithm to identify the bolt rust level in real time and dynamically optimize cleaning parameters to clean bolts with different rust levels. This effectively avoids rust residue at the thread root and excessive damage to the substrate. The rust self-adaptive cleaning module includes a rust layer multi-mode sensing unit and a cleaning parameter optimization unit. The rust layer multi-mode sensing unit employs an intelligent sensing system based on laser profile scanning and impedance spectroscopy detection to collect real-time data on the rust layer thickness, density, and texture features of the bolts to be cleaned. It identifies the bolt corrosion level and generates a rust level map, achieving precise quantification and classification of the corrosion degree. In response to the bolts entering the detection station of the automated bolt cleaning equipment, the deployed intelligent sensing system triggers the laser profile scanner to perform a multi-section spiral scan along the bolt axis, acquiring three-dimensional point cloud data of the thread surface. It extracts rust layer thickness and texture feature parameters, achieving high-precision three-dimensional reconstruction of the rust layer morphology. Simultaneously, the impedance spectroscopy detection probe contacts the bolt surface to be tested, applies a variable-frequency AC excitation signal, collects amplitude and phase frequency response curves, and inversely calculates the rust layer density and porosity distribution, revealing the internal density and porosity of the rust layer. By fusing the laser profile and impedance spectroscopy detection results, it identifies the bolt corrosion level according to a preset bolt corrosion classification threshold, generating a rust level map for each bolt. The map data is then encapsulated and uploaded to the cleaning parameter optimization unit, providing accurate level criteria for differentiated cleaning. It should be noted that when the bolts to be cleaned are conveyed into the inspection station of the automated bolt cleaning equipment by the conveyor chain, the laser profile scanner installed on one side of the inspection station is triggered and started. This scanner performs multi-section spiral scanning along the bolt axis at a scanning step of 0.5 mm, with a scanning speed of 20 mm / s. Each section collects 1200 profile points, obtaining three-dimensional point cloud data of the thread crest, flank, and root regions. Then, the rust layer thickness characteristic parameters are extracted from the three-dimensional point cloud data, with the thickness extraction accuracy set at ±0.02 mm. At the same time, the rate of change of the point cloud normal vector is calculated to characterize the roughness of the rust layer texture. Simultaneously, the resistance... The impedance spectroscopy probe is pressed against the bolt surface with a contact force of 0.3 to 0.5 N. An electrochemical workstation is used to apply a frequency-converted AC excitation signal with an excitation voltage amplitude of 10 mV. After acquiring the amplitude-frequency and phase-frequency response curves, the rust density and porosity distribution are calculated by inverting the equivalent circuit model. The rust thickness and texture feature parameters obtained from laser contour scanning are fused with the rust density and porosity distribution data obtained from impedance spectroscopy. Specifically, the third to eighth threads in the middle section of the bolt are used as the feature extraction window, and the average value of five measurement sections within this window is taken as the rust characterization vector of the bolt, with a vector dimension of four. The system uses four dimensions, corresponding to thickness, texture, density, and porosity, respectively. Four preset classification threshold ranges for bolt corrosion levels are defined: Slight corrosion corresponds to a thickness threshold of less than 0.10 mm and a porosity of less than 15%; moderate corrosion corresponds to a thickness threshold of 0.10 mm to 0.30 mm and a porosity of 15% to 35%; severe corrosion corresponds to a thickness threshold of 0.31 mm to 0.60 mm and a porosity of 36% to 55%; and extremely severe corrosion corresponds to a thickness threshold greater than 0.60 mm or a porosity greater than 55%. The fused feature vector is compared with these classification thresholds to determine the corrosion level of each bolt. The system generates a corrosion level map with the bolt number as the primary key. The map records the level identifier of each bolt and the corresponding four characteristic parameter values. The generated corrosion level map data is encapsulated according to a preset communication protocol. The encapsulated data packet includes a frame header, data length, total number of bolts, timestamp, number of each bolt and its corrosion level code, and the four characteristic parameter values. A CRC16 checksum is appended to the end of the frame. The encapsulated data packet is uploaded to the cleaning parameter optimization unit through the Ethernet interface. After receiving and parsing the data packet, the cleaning parameter optimization unit stores the corrosion level map in a local cache queue, waiting for the parameter optimization ant colony algorithm to call it. The cleaning parameter optimization unit, based on the rust level map, runs a parameter optimization ant colony algorithm to iteratively optimize the sandblasting pressure, cleaning time, and abrasive flow rate of the automated bolt cleaning equipment. It outputs differentiated cleaning commands to the first-stage actuator, adaptively cleaning bolts with different rust levels to ensure that each bolt is cleaned with the optimal parameters. It receives the rust level map, analyzes the rust layer thickness, density, and texture feature values ​​corresponding to each bolt, and constructs an initial solution space with sandblasting pressure, cleaning time, and abrasive flow rate as decision variables to achieve accurate matching between cleaning parameters and rust state. It initializes the pheromone matrix of the parameter optimization ant colony algorithm, maps the rust level map as a path heuristic factor, and drives artificial ants to iteratively search in the solution space. It optimizes parameters with cleaning residue rate and matrix damage as multi-objective constraints to ensure rapid approximation of the optimal cleaning combination under multi-objective constraints. It converges and outputs the optimal cleaning parameter combination, generates differentiated cleaning commands, and sends them to the first-stage actuator to adjust the sandblasting valve opening, cleaning timer, and abrasive supply pump speed, respectively, to achieve automated closed-loop control of differentiated cleaning commands. It should be noted that after analyzing the rust level map, for each bolt, an initial solution space is established with sandblasting pressure, cleaning time, and abrasive flow rate as decision variables, based on the rust thickness, density, and texture characteristics. The sandblasting pressure range is set to 0.3 MPa to 0.8 MPa, the cleaning time range is set to 5 seconds to 25 seconds, and the abrasive flow rate range is set to 2 kg / min to 8 kg / min. The pheromone matrix of the parameter optimization ant colony algorithm is initialized, mapping the level codes and feature parameters in the rust level map to path heuristic factors. The higher the level, the greater the weight of the heuristic factor. The number of artificial ants is set to 50, the maximum number of iterations is set to 200, the pheromone evaporation coefficient is 0.15, and the heuristic factor weight coefficient is 1.2, driving the artificial ants to... The solution space is traversed and searched, with multiple objectives including a cleaning residue rate of less than 5% and a matrix damage degree of less than 0.01 mm. After each ant completes a path search, the local pheromone is updated. The global optimal path pheromone increment coefficient is set to 0.8. The optimal solution combination is gradually approximated through iteration. During the iterative search, the fitness function value corresponding to each candidate cleaning parameter combination is calculated in real time. The fitness function is composed of a weighted average of the cleaning residue rate and the matrix damage degree, with the cleaning residue rate weight coefficient set to 0.6 and the matrix damage degree weight coefficient set to 0.4. For bolts with slight corrosion, low blasting pressure parameters are prioritized. The blasting pressure converges to the range of 0.35 MPa to 0.45 MPa, the cleaning time converges to the range of 6 seconds to 9 seconds, and the abrasive flow rate... The algorithm converges within the range of 2.5 kg / min to 3.5 kg / min. For severely corroded bolts, the sandblasting pressure converges within the range of 0.65 MPa to 0.75 MPa, the cleaning time converges within the range of 18 to 22 seconds, and the abrasive flow rate converges within the range of 6.5 kg / min to 7.5 kg / min. When the rate of change of the global optimal fitness function value is less than 0.5% over 20 consecutive iterations, the algorithm is considered converged. The current optimal cleaning parameter combination is output, and the corresponding sandblasting pressure, cleaning time, and abrasive flow rate are recorded as the differentiated cleaning command for that bolt. The converged optimal cleaning parameter combination is encapsulated into a differentiated cleaning command according to the equipment communication protocol and sent to the first-stage actuator via the Ethernet interface. The first-stage actuator includes the sandblasting... The system includes a valve controller, a cleaning timer module, and an abrasive supply pump frequency converter. After receiving the pressure setpoint, the sandblasting valve controller adjusts the opening of the proportional solenoid valve to ensure that the output pressure tracks the target value within 0.2 seconds, with a control accuracy of ±0.02 MPa. The cleaning timer module starts a countdown based on the received duration parameter, with a timing error of less than 0.1 seconds. When the countdown ends, it automatically triggers the sandblasting valve to close. The abrasive supply pump frequency converter adjusts the motor speed based on the flow setpoint, with a flow control accuracy of ±0.2 kg / min and a response delay of less than 0.3 seconds. The three sets of actuators work in parallel and adjust independently in their respective closed-loop control loops to ensure that each bolt completes the cleaning operation with optimal parameters. After cleaning, the actuators automatically reset and wait for the differentiated instruction input for the next bolt. The residual liquid interference prediction module is used to combine wireless networking and twin simulation to collect drying process data from multiple angles and at high frequency, simulate the evaporation path of residual liquid and predict the interference area, dynamically adjust the purging angle and hot air temperature, and eliminate the misjudgment interference of residual liquid on visual detection. The residual liquid interference prediction module includes a drying process twin acquisition unit and a thermal field dynamic correction unit. The drying process twin acquisition unit is used to deploy multi-point temperature and humidity sensors and high-frequency cameras. It collects data on the bolt surface drying process from multiple angles through wireless networking, constructs a dynamic twin data body of the residual liquid evaporation path, and captures the residual liquid distribution and evaporation dynamics in real time. Multi-point temperature and humidity sensors and high-frequency cameras are deployed around the drying station of the automated bolt cleaning equipment. Through wireless networking, the sensors are synchronously triggered to collect multi-source heterogeneous data on the bolt surface, including temperature, humidity and liquid film distribution images of different areas of the bolt surface. This achieves accurate synchronous acquisition of multi-dimensional data of the drying process. The collected multi-source heterogeneous data is mapped to a preset three-dimensional bolt digital model. The residual liquid evaporation rate of each coordinate point is marked according to the time sequence, and a dynamic twin data body of the residual liquid evaporation path is constructed. The spatiotemporal evolution of the entire residual liquid evaporation process is dynamically restored. The dynamic twin data body is continuously iterated and updated. The spatial coordinates of residual liquid accumulation areas and evaporation lag points are identified, and a residual liquid evaporation path map is generated and pushed to the thermal field dynamic correction unit to provide accurate residual liquid distribution basis for thermal field control. It should be noted that six sets of temperature and humidity sensors and four high-frequency cameras are evenly deployed circumferentially at the drying station of the automated bolt cleaning equipment. The sensor measurement accuracy is ±0.3℃ and ±1.5%RH, the camera sampling frequency is set to 30 frames / second, and the image resolution is 5 megapixels. After the cleaned bolts enter the drying station via the conveyor chain, the wireless network controller triggers all sensors and cameras to start collecting data at a synchronization cycle of 50 milliseconds. This acquires the temperature, relative humidity, and liquid film reflectivity images of eight characteristic areas distributed along the thread axis on the bolt surface. The collected data packets are aggregated to the local data buffer via the 2.4GHz wireless protocol. Each set of data is accompanied by a collection timestamp and bolt number label to ensure the temporal consistency and traceability of subsequent mappings. The preset three-dimensional bolt digital model is established based on the bolt's nominal diameter, pitch, and length parameters, containing 1200 surface node coordinates. The residual liquid evaporation rate at each node is marked sequentially according to the time sequence. The evaporation rate is determined by combining the liquid film area change rate between adjacent frames with the temperature and humidity correction coefficient. The calculation shows that, with a time interval of 0.5 seconds, the dynamic twin data body of the constructed residual liquid evaporation path is stored in the form of a structured array. Each array element contains node coordinates, evaporation rate value, and time tag. The complete drying process within 8 seconds is continuously recorded. During the real-time update of the dynamic twin data body, the difference in evaporation rate between adjacent time slices is automatically compared, and nodes with a rate lower than 0.02% / second are identified as candidate points for evaporation lag. Each time data acquisition and mapping is completed, the identification algorithm is triggered to calculate the average evaporation rate of each node within the most recent 1-second time window. Nodes with an average rate lower than 40% of the overall average and a liquid film thickness greater than 5 micrometers are identified as residual liquid accumulation areas, and their spatial coordinates are recorded. At the same time, nodes with an evaporation rate lower than 0.01% / second for three consecutive time slices are marked as evaporation lag points. A residual liquid evaporation path map is generated with a period of 0.5 seconds. The map marks the center coordinates of the accumulation area, the distribution density of lag points, and the residual liquid migration direction vector, and is pushed to the thermal field dynamic correction unit through the Ethernet interface as the basis for adjusting the thermal field parameters. The thermal field dynamic correction unit is used to combine the dynamic twin data of the residual liquid evaporation path to simulate and analyze the residual liquid accumulation area and the interference shadow area. It drives the second-stage actuator to adaptively adjust the purging angle, hot air temperature and action time to accurately remove all residual liquid accumulation areas and ensure that the bolt surface is clean and dry. It receives the residual liquid evaporation path map, imports it into the heat flow simulation model, and simulates the temperature field and airflow field distribution of the residual liquid accumulation area under different combinations of purging angle and hot air temperature. It effectively locates the residual liquid accumulation area and avoids drying blind spots. Based on the simulation results, it identifies the boundary conditions of the interference shadow area and reverse calculates the minimum hot air temperature, optimal purging angle and action time parameters required to eliminate residual liquid accumulation. It accurately eliminates interference shadows and prevents visual misjudgment. The calculated thermal field correction parameters are converted into drive commands for the second-stage actuator to adjust the swing angle of the hot air nozzle, the duty cycle of the heating wire and the purging timer respectively to achieve adaptive drying control and ensure surface cleanliness. It should be noted that after receiving the residual liquid evaporation path map, it is imported into the embedded thermal flow simulation model. This model is based on the finite element analysis method and is established according to the actual geometric dimensions of the bolts and the thermal properties of the materials. For the center coordinates of the residual liquid accumulation area and the distribution density of hysteresis points marked in the residual liquid evaporation path map, the thermal flow simulation model performs a combined traversal simulation with a step size of 1 degree, within the nozzle swing angle range of 30 to 90 degrees, combined with the hot air temperature range of 40℃ to 80℃. During the simulation process, the surface temperature field distribution and airflow velocity vector field under each combination are simultaneously solved. Focusing on the shear stress distribution on the surface of the liquid film in the accumulation zone, and comparing the theoretical time required for the liquid film in the accumulation zone to achieve complete evaporation under different combinations, several effective parameter pairs that meet the evaporation requirements are selected. The system then enters the reverse calculation stage to determine a set of optimal thermal field correction parameters. This process aims to eliminate all residual liquid accumulation zones and introduces energy consumption and efficiency constraints, namely, seeking the parameter combination that minimizes the hot air temperature and the shortest action time. The calculation logic first identifies the boundary conditions of the interference shadow area, that is, the coordinates of the hidden area that the airflow cannot effectively flush. Based on this, the gradient descent method is used for iterative calculation to find the nozzle angle that makes the liquid film in the accumulation zone completely evaporate within 8 seconds and has the lowest hot air temperature from the candidate parameters that meet the basic evaporation requirements. The final output includes the parameter combination containing the optimal nozzle swing angle, the target duty cycle of the heating wire, and the set duration of the purge timer. In the drive commands that the second-stage actuator can execute, the swing angle command of the hot air nozzle is sent to the servo motor driver with a resolution of 0.5 degrees. The motor drives the nozzle to reciprocate within the set angle range to ensure that the airflow covers all hysteresis points. The heating wire duty cycle command is passed through... The pulse width modulation signal adjusts the on / off ratio of the solid-state relay to stabilize the hot air temperature within ±2℃ of the target value. The purge timer instruction sets a countdown. For example, when the calculation requires continuous purge for 7 seconds, the timer automatically triggers the hot air blower shutdown signal after reaching zero. The three sets of instructions are sent to their respective controllers in parallel. During the execution process, the real-time temperature and humidity feedback of the drying station is continuously monitored. If the actual evaporation rate is found to deviate from the predicted trajectory of the twin model, the fine-tuning instruction is triggered to perform closed-loop correction of the duty cycle or angle to ensure that the actual operation effect is consistent with the simulation expectation. The batch logistics anti-mixing module assigns a unique RFID batch code to each cleaning material box. The position of each bolt in the material box is fixed according to matrix coordinates. A matrix-style positional status map is constructed using a digital identity strategy containing batch code and coordinate position. By using the unique RFID batch code of the material box and the matrix coordinates of the bolt in the box (when feeding, visual positioning records whether there is a bolt at each coordinate position, establishing a mapping relationship between "batch code and coordinate position" and the physical position of the bolt, and deploying position sensors at each station to read the RFID of the material box in real time, associating it with the processing result of the current station, and marking each coordinate position as qualified / rework / scrap), the status of each bolt in the batch can be traced, preventing qualified parts from being mixed with unwashed parts in the cleaning and inspection process. The intelligent sorting and reconfiguration module, based on a matrix-style positional state graph, introduces a multi-agent system to coordinate the action sequence and load balancing of multiple sorting stations. At the same time, it uses a path reconstruction ant colony algorithm to dynamically reconstruct the sorting order and grasping path of the coordinate positions, minimizing the idle travel of the robotic arm and the waiting time in the queue. After sorting is completed, the state matrix is ​​cleared for each coordinate position, achieving strong consistency between physical sorting and data closure. This prevents qualified parts from being mixed into the waiting queue for cleaning or the return of missed parts, and achieves accurate diversion of qualified, rework, and scrapped parts.

[0022] Example 2, as Figures 1 to 5 As shown, based on Embodiment 1, the present invention provides a technical solution: the batch logistics anti-mixing module specifically includes: binding a unique RFID batch code to each cleaning material box at the feeding station of the automated bolt cleaning equipment; simultaneously identifying the presence or absence of bolts at each matrix coordinate position within the material box through a visual positioning system; establishing an initial mapping relationship between the batch code and the coordinate position to ensure that each bolt obtains a unique traceable identity from the source; deploying fixed RFID readers and position sensors at each processing station to read the batch code of the material box in real time and associate it with the processing result of the current station; updating the status mark to qualified, rework, or scrap according to the coordinate position one by one, realizing the dynamic update of the single bolt status with the process, eliminating information lag and misalignment; writing the status marks reported by each station into a matrix positional status map according to the time sequence; and realizing full-process single bolt status traceability and anti-mixing through the combination key of batch code and coordinate position, preventing bolts from being confused due to similar appearance, and ensuring consistent and reliable data throughout the process. It should be noted that at the feeding station of the automated bolt cleaning equipment, the operator first places the bolts to be cleaned one by one according to the coordinate positions of the matrix-style material frames. A passive UHF RFID tag is pre-embedded at the bottom of each material frame, with a unique 24-digit batch code written inside. The visual positioning system uses two 5-megapixel industrial cameras to capture the entire material frame from an orthogonal direction. Image processing algorithms identify the presence of a bolt at each coordinate position with a positioning accuracy of ±0.5 mm. The system associates the read batch code with the visually recognized coordinate position's presence / absence status, establishing an initial mapping relationship between "batch code and coordinate position" and the bolt's physical location. This mapping relationship is stored in a local database in the form of a structured data table. Fixed UHF RFID readers are deployed at the entrance of each processing station. The reader antennas are installed on both sides of the conveyor chain, with a reading distance set to 30 cm to ensure stable reading of the batch code as the material frame passes. Each station is also equipped with a photoelectric position sensor. When the material frame is positioned within the station's working area, it triggers the processing... After the visual inspection is completed, the status of each bolt is marked according to its coordinate position. The status codes are defined as follows: 00 represents qualified, 01 represents rework, and 02 represents scrap. Each status mark is accompanied by the current station number and timestamp. The database record is updated by the combination key of batch code and coordinate position to ensure that the status of a single bolt is traceable and tamper-proof throughout the entire process. After the status marks reported by each station are written in time sequence, a matrix-style positional status map is automatically constructed. The matrix-style positional status map is based on the material frame as a unit. The status code corresponding to each coordinate position is stored in the form of a two-dimensional matrix. The matrix dimension is consistent with the actual number of rows and columns of the material frame. When the material frame enters the next station, the RFID reader reads the batch code again, calls the current matrix-style positional status map and superimposes the processing result of the new station to realize incremental update of status information. The combination key of batch code and coordinate position realizes the traceability of the status of a single bolt throughout the entire process, preventing confusion and errors caused by similar appearance of bolts, and ensuring the consistency of the identity of each bolt from feeding to sorting. The intelligent sorting and reconfiguration module specifically includes: reading the matrix-style positional state map, extracting the state markers and physical coordinates of each coordinate position, introducing a multi-agent system based on multiple sorting stations, allocating the target coordinate list of bolts to be sorted according to the load capacity of each sorting station to ensure balanced load of each station and avoid overload or idleness, initializing the pheromone matrix of the path reconstruction ant colony algorithm, using the Euclidean distance between each coordinate position as the path cost factor, dynamically iteratively searching for the grasping order and robotic arm movement path that minimizes the idle stroke, significantly shortening the idle stroke of the robotic arm and improving sorting efficiency, executing sorting actions sequentially according to the optimized path, transferring qualified parts, rework parts, or scrapped parts to the corresponding silos, clearing the corresponding records in the matrix-style positional state map position by position after sorting, and presenting the bolt cleaning results on the visualization interface of the intelligent control and scheduling center, achieving strong consistency between physical sorting and data closed loop, and eliminating the risk of mixed materials and missed detection; It should be noted that in the sorting station area, the matrix-style positional state map corresponding to the current material frame is first read from the database. The state marker (qualified, rework, scrap) of each coordinate position and its physical coordinates in the material frame coordinate system are analyzed. Based on the three sorting stations arranged on site, a multi-agent system collaborative control strategy is introduced. The target coordinate list of bolts to be sorted is dynamically allocated according to the rated load capacity of the robotic arm of each station (maximum gripping frequency of 15 times / minute per arm), avoiding overload or idleness of any station. After allocation, the pheromone matrix of the path reconstruction ant colony algorithm is initialized, and the Euclidean distance between each coordinate position is used as the path cost factor. The algorithm uses 30 ants, has a maximum of 100 iterations, a pheromone evaporation coefficient of 0.1, and a heuristic factor weighting coefficient of 1.0. It dynamically iterates through the solution space, minimizing the idle travel of the robotic arms and queue waiting time, and converges to output the optimal grasping order and movement path. When performing sorting actions according to the optimized path, the first sorting arm handles qualified parts, the second handles rework parts, and the third handles scrap parts. The three arms operate in parallel without interference. The robotic arms use vacuum suction cup end effectors, achieving a grasping and positioning accuracy of ±0.3 mm, and a single grasping cycle time of 3.5 seconds. Within the specified timeframe, qualified parts are transferred to the clean storage silo, rework parts to the rework buffer zone, and scrapped parts to the waste collection bin. During the transfer process, photoelectric sensors verify the gripping status. If an empty grip or detachment is detected, a repeat gripping command is immediately triggered, with a maximum of two retries. After sorting, the corresponding record in the matrix-style positional status map is cleared position by position, and the clearing mark is synchronized to the local database to ensure strong consistency between physical sorting and data closure, preventing qualified parts from being mixed into the cleaning queue or the risk of missed parts being repacked. The sorting results are displayed in real time on the visualization interface of the intelligent control and dispatch center. The interface displays the final status of each coordinate position in the form of a material frame matrix diagram, with qualified parts showing... The color is green, rework parts are yellow, and scrapped parts are red. The interface simultaneously displays the total number of bolts processed, the pass rate, the rework rate, and the scrap rate for the current batch. The data refresh cycle is 0.5 seconds. Operators can click on any coordinate position on the touch screen to bring up the full process status record of that bolt, including the corrosion level, cleaning parameters, test results, and sorting timestamp. The empty material box that has been sorted is conveyed to the feeding station by the conveyor chain, automatically unlocking the pattern lock of the batch code, releasing the database storage space, and waiting for the loading of the next batch of bolts. At this point, the closed-loop control of the entire process from feeding, cleaning, drying, testing to sorting is completed, and all data can be traced back to a single bolt.

[0023] Example 3, as Figures 1 to 5 As shown, based on Embodiments 1 and 2, the present invention also provides an automated bolt cleaning and detection control method, which is implemented based on the above-mentioned automated bolt cleaning and detection control system, and includes the following steps: S1. The bolts to be cleaned enter the automated bolt cleaning equipment, triggering laser profile and impedance spectrum fusion detection to extract thickness, texture, density and texture features. The bolt corrosion level is classified according to the bolt corrosion classification threshold and a corrosion level map is generated to achieve accurate quantification and graded identification of the degree of corrosion. S2. Analyze the rust level map, construct the solution space of sandblasting pressure, duration and flow rate, run the parameter optimization ant colony algorithm for iterative optimization, output differentiated cleaning instructions to the first stage actuator, realize graded adaptive cleaning, and avoid residual material at the root of the thread and over-cleaning damage to the substrate. S3. After cleaning, the bolts enter the drying station. The wireless network triggers the temperature and humidity sensor and the high-frequency camera to collect multi-source heterogeneous data and map it to the preset three-dimensional bolt digital model to construct a dynamic twin data body of the residual liquid evaporation path and predict the residual liquid accumulation area and evaporation lag point in advance. S4. Import the dynamic twin data volume into the thermal flow simulation model, traverse the nozzle angle and air temperature combination, reverse calculate the optimal purging parameters, drive the second-stage actuator to adjust the thermal field, ensure that the bolt surface is clean and dry, and eliminate the interference of residual liquid film on visual inspection. S5. Bind RFID batch codes to the material frame, establish coordinate mapping for visual positioning, and link the processing results of readers and writers at each station in real time to construct a matrix-style positional status map, realize full-process traceability of single bolts, and prevent qualified parts from being confused with unwashed parts in the process. S6. Based on the matrix-style positional state graph, a multi-agent system is introduced to allocate sorting tasks. The path reconstruction ant colony algorithm reconstructs the grasping path, and qualified, reworked and scrapped parts are sorted in parallel. The matrix-style positional state graph is cleared and the database is synchronized to complete the closed-loop control of the whole process and eliminate the risk of repeated processing of qualified parts and repackaging of missed parts.

[0024] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An automated bolt cleaning and inspection control system, comprising an intelligent control and dispatch center, characterized in that, The intelligent control and dispatch center has the following communication connection modules: The rust self-adaptive cleaning module integrates an intelligent sensing system with a parameter optimization ant colony algorithm to identify the bolt rust level in real time and dynamically optimize cleaning parameters to clean bolts with different degrees of rust. The residual liquid interference prediction module is used to combine wireless networking and twin simulation to collect drying process data from multiple angles and high frequencies, simulate the residual liquid evaporation path and predict the interference area, and dynamically adjust the purging angle and hot air temperature. The batch logistics anti-mixing module is used to assign a unique RFID batch code to each cleaning material box. The position of each bolt in the material box is fixed according to the matrix coordinates. A matrix positional status map is constructed using a digital identity strategy containing batch code and coordinate position. The status of each bolt in the batch can be traced through the unique RFID batch code of the material box and the matrix coordinates of the bolt in the box. The intelligent sorting and reconfiguration module, based on a matrix-style positional state graph, introduces a multi-agent system to coordinate the action timing and load balancing of multiple sorting stations. At the same time, it uses a path reconstruction ant colony algorithm to dynamically reconfigure the sorting order and grasping path of the coordinate positions, so as to achieve accurate diversion of qualified, rework, and scrapped parts.

2. The automated bolt cleaning and inspection control system according to claim 1, characterized in that: The rust self-adaptive cleaning module includes a rust layer multi-mode sensing unit and a cleaning parameter optimization unit. The rust layer multi-mode sensing unit is used to acquire the rust layer thickness, density and texture features of the bolt to be cleaned in real time using an intelligent sensing system based on laser contour scanning and impedance spectrum detection, identify the bolt corrosion level and generate a corrosion level map. The cleaning parameter optimization unit, based on the rust level map, runs a parameter optimization ant colony algorithm to iteratively optimize the sandblasting pressure, cleaning time, and abrasive flow rate of the automated bolt cleaning equipment, and outputs differentiated cleaning instructions to the first-stage actuator to adaptively clean bolts with different degrees of rust.

3. The automated bolt cleaning and inspection control system according to claim 2, characterized in that: The rust layer multi-mode sensing unit specifically includes: In response to the bolts to be cleaned entering the detection station of the automated bolt cleaning equipment, the laser profile scanner in the deployed intelligent sensing system is triggered to perform a multi-section spiral scan along the bolt axis, acquire three-dimensional point cloud data of the thread surface, and extract rust layer thickness and texture feature parameters. Synchronously start the impedance spectrum detection probe to contact the bolt surface under test, apply a frequency-converted AC excitation signal, collect the amplitude frequency and phase frequency response curves, and invert and calculate the rust layer density and porosity distribution. By integrating the results of laser profilometry and impedance spectroscopy detection, the bolt corrosion level is identified according to the preset bolt corrosion classification threshold, and then a corrosion level map of each bolt is generated. The map data is then packaged and uploaded to the cleaning parameter optimization unit.

4. The automated bolt cleaning and inspection control system according to claim 2, characterized in that: The cleaning parameter optimization unit specifically includes: Receive the rust level map, analyze the rust layer thickness, density and texture feature value corresponding to each bolt, and construct the initial solution space with sandblasting pressure, cleaning time and abrasive flow rate as decision variables; The pheromone matrix of the parameter optimization ant colony algorithm is initialized, the corrosion level map is mapped as a path heuristic factor, and the artificial ants are driven to iteratively search in the solution space. The parameter optimization is carried out with cleaning residue rate and matrix damage degree as multi-objective constraints. The optimal combination of cleaning parameters is converged and output, generating differentiated cleaning commands and sending them to the first-stage actuator to adjust the sandblasting valve opening, cleaning timer, and abrasive supply pump speed, respectively.

5. An automated bolt cleaning and inspection control system according to claim 2, characterized in that: The residual liquid interference prediction module includes a drying process twin acquisition unit and a thermal field dynamic correction unit; The drying process twin acquisition unit is used to deploy multi-point temperature and humidity sensors and high-frequency cameras to collect bolt surface drying process data from multiple angles through wireless networking, and construct a dynamic twin data body of the residual liquid evaporation path. The thermal field dynamic correction unit is used to combine the dynamic twin data of the residual liquid evaporation path to simulate and analyze the residual liquid accumulation area and the interference shadow area, and drive the second-stage actuator to adaptively adjust the purging angle, hot air temperature and action time.

6. An automated bolt cleaning and inspection control system according to claim 5, characterized in that: The twin acquisition unit for the drying process specifically includes: In the drying station of the automated bolt cleaning equipment, multiple temperature and humidity sensors and high-frequency cameras are deployed around the perimeter. Through wireless networking, each sensor is synchronously triggered to collect multi-source heterogeneous data on the bolt surface, including temperature, humidity and liquid film distribution images in different areas of the bolt surface. The collected multi-source heterogeneous data is mapped to a preset three-dimensional bolt digital model, and the residual liquid evaporation rate at each coordinate point is marked according to the time series to construct a dynamic twin data body of the residual liquid evaporation path; The dynamic twin data volume is continuously updated iteratively to identify the spatial coordinates of the residual liquid accumulation area and the evaporation lag point, generate the residual liquid evaporation path map and push it to the thermal field dynamic correction unit.

7. An automated bolt cleaning and inspection control system according to claim 5, characterized in that: The dynamic thermal field correction unit specifically includes: Receive the residual liquid evaporation path map, import it into the heat flow simulation model, and simulate the temperature field and airflow field distribution of the residual liquid accumulation area under different combinations of purging angle and hot air temperature. Based on the simulation results, the boundary conditions of the interference shadow area are identified, and the minimum hot air temperature, optimal purging angle and action time parameters required to eliminate residual liquid accumulation are calculated in reverse. The calculated thermal field correction parameters are converted into drive commands for the second-stage actuator, which adjust the swing angle of the hot air nozzle, the duty cycle of the heating wire, and the purge timer, respectively.

8. An automated bolt cleaning and inspection control system according to claim 5, characterized in that: The batch logistics anti-mixing module specifically includes: At the feeding station of the automated bolt cleaning equipment, a unique RFID batch code is bound to each cleaning material box. At the same time, the presence or absence of bolts at each matrix coordinate position in the material box is identified by the visual positioning system, and the initial mapping relationship between the batch code and the coordinate position is established. Fixed RFID readers and position sensors are deployed at each processing station to read the batch code of the material frame in real time and associate it with the processing result of the current station. The status is updated one by one according to the coordinate position as qualified, rework or scrap. The status markers reported by each workstation are written into a matrix-style positional status map according to the time sequence. The status of a single bolt throughout the entire process is traceable and prevented from being confused by the combination key of batch code and coordinate position.

9. An automated bolt cleaning and detection control system according to claim 8, characterized in that: The intelligent sorting and reconfiguration module specifically includes: Read the matrix-style positional state map, extract the state markers and physical coordinates of each coordinate position, introduce a multi-agent system based on multiple sorting stations, and allocate the target coordinate list of bolts to be sorted according to the load capacity of each sorting station. The pheromone matrix of the ant colony algorithm is initialized and the Euclidean distance between each coordinate position is used as the path cost factor. The grasping order and the robot arm movement path that minimize the empty stroke are dynamically iteratively searched. The sorting action is executed sequentially according to the optimized path, and qualified parts, rework parts or scrapped parts are transferred to the corresponding silos. After sorting is completed, the corresponding records in the matrix position status map are cleared at each coordinate position, and the bolt cleaning results are presented on the visualization interface of the intelligent control and scheduling center.

10. An automated bolt cleaning and inspection control method, implemented based on the automated bolt cleaning and inspection control system described in any one of claims 1-9, characterized in that, Includes the following steps: S1. The bolts to be cleaned enter the automated bolt cleaning equipment, triggering laser profile and impedance spectrum fusion detection to extract thickness, texture, density and texture features, classify bolt corrosion levels according to bolt corrosion classification thresholds and generate corrosion level maps. S2. Analyze the rust level map, construct the solution space of sandblasting pressure, duration and flow rate, run the parameter optimization ant colony algorithm for iterative optimization, and output differentiated cleaning instructions to the first stage actuator to achieve graded adaptive cleaning. S3. After cleaning, the bolts enter the drying station. The wireless network triggers the temperature and humidity sensor and the high-frequency camera to collect multi-source heterogeneous data and map it to the preset three-dimensional bolt digital model to construct a dynamic twin data body of the residual liquid evaporation path. S4. Import the dynamic twin data volume into the thermal flow simulation model, traverse the nozzle angle and air temperature combination, reverse calculate the optimal purging parameters, and drive the second-stage actuator to adjust the thermal field. S5. Bind RFID batch codes to the material frame, establish coordinate mapping through visual positioning, and link and process the results of readers and writers at each station in real time to construct a matrix-style positional status map to achieve full-process traceability of a single bolt. S6. Based on the matrix-style positional state graph, a multi-agent system is introduced to allocate sorting tasks. The path reconstruction ant colony algorithm reconstructs the grasping path, and qualified, reworked and scrapped parts are sorted in parallel. The matrix-style positional state graph is cleared and the database is synchronized to complete the closed-loop control of the entire process.

Images