Textile equipment real-time monitoring system based on circulating robot and stroboscopic control

Abstract

The invention discloses a spinning equipment real-time monitoring system based on an itinerant robot and stroboscopic control, particularly relates to the field of spinning equipment monitoring, is used for solving the problems of low monitoring precision and insufficient inspection efficiency, and is used for accurately evaluating the state of spinning equipment and generating a score based on historical operation data and real-time feedback. And then planning and dynamically adjusting an inspection path by adopting an ant colony optimization and reinforcement learning algorithm to ensure that high-risk equipment is preferentially monitored. Resource allocation is optimized and inspection efficiency is improved through accurate simulation of path time consumption and a coverage range; meanwhile, handshake connection is encrypted through multiple protocols, equipment operation parameters are extracted in real time, stroboscopic control is dynamically adjusted, and the definition of image collection and data safety are guaranteed; high-frame-rate high-definition camera shooting and self-adaptive stroboscopic control are combined, high-quality image data are collected and preprocessed, deep analysis is carried out by using a convolutional neural network and a multivariable correlation model, abnormal early warning is generated in real time, and an inspection path is optimized.

Description

Technical Field

[0001] The invention relates to the field of monitoring of spinning equipment, and more specifically, to a real-time monitoring system for textile equipment based on a patrol robot and stroboscopic control. Background Art

[0002] As the core link in the textile production line, the operating status of spinning equipment directly affects the quality and production efficiency of yarn. With the development of intelligent manufacturing and industrial automation, the monitoring technology of spinning equipment has gradually developed towards high precision, real-time and intelligent. Traditional monitoring methods mainly rely on manual inspections and simple sensor data collection, which is difficult to fully and accurately reflect the real-time operating status of the equipment. In addition, key parameters such as the rotation speed, shape consistency, hairiness distribution and cleanliness of the yarn tube during the spinning process have an important impact on the quality of the yarn. How to achieve efficient monitoring and accurate analysis of these parameters has become the key to improving production quality and efficiency.

[0003] However, the existing technology still has many shortcomings in the monitoring of spinning equipment. First, manual inspection is not only time-consuming and labor-intensive, but also easily leads to inaccurate and inconsistent monitoring results due to human factors, which makes it difficult to meet the real-time monitoring needs of large-scale production lines. Secondly, the traditional sensor data acquisition method lacks high-resolution image analysis of the surface state of the yarn tube, and cannot fully capture the slight changes of the yarn tube during high-speed operation, resulting in frequent lags and missed detections in abnormal detection. Furthermore, the existing monitoring system lacks intelligent means in data processing and analysis, making it difficult to achieve dynamic evaluation and adaptive adjustment of the equipment's operating status, which limits the response speed and optimization capabilities of the monitoring system. Therefore, there is an urgent need for an integrated and intelligent monitoring system that can realize comprehensive, real-time, and accurate monitoring and optimization management of the operating status of spinning equipment through high-precision image acquisition and deep learning analysis, combined with intelligent inspection robots, to improve the quality control and production efficiency of spinning production.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiment of the present invention provides a real-time monitoring system for textile equipment based on patrol robots and stroboscopic control. First, based on historical operation data and real-time feedback, the equipment status is accurately evaluated and a score is generated. Then, advanced ant colony optimization and reinforcement learning algorithms are used to plan and dynamically adjust the inspection path to ensure that high-risk equipment is monitored first. Through accurate simulation of path time consumption and coverage, resource allocation is optimized and inspection efficiency is improved. At the same time, through multi-protocol encrypted handshake connection, equipment operating parameters are extracted in real time and stroboscopic control is dynamically adjusted to ensure the clarity of image acquisition and data security. High-frame rate high-definition video is combined with adaptive stroboscopic control to collect and pre-process high-quality image data, and then convolutional neural networks and multivariate association models are used for in-depth analysis to generate abnormal warnings in real time and optimize the inspection path. The entire closed-loop optimization mechanism is formed to ensure the real-time and intelligent nature of monitoring, significantly improve the operating stability, yarn quality and production efficiency of spinning equipment, meet the needs of modern spinning workshops for efficient and accurate monitoring, and solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: The real-time monitoring system of textile equipment based on patrol robot and strobe control includes the following steps: Scoring calculation module: Based on the historical operation data of spinning equipment, the gradient boosting decision tree algorithm is used to perform weighted scoring on multi-factor features to generate equipment scores, and the abnormal warning data from the deep analysis module is received to dynamically adjust the scoring weights.

[0007] Path planning module: Based on the device scores and setting the score difference threshold, the ant colony optimization algorithm is used to prioritize the devices and plan the optimal inspection path for the patrol robot. The system will re-sort the devices only when the score difference exceeds the threshold, and dynamically adjust the inspection path based on the path optimization suggestions of the deep analysis module.

[0008] Path simulation module: evaluates the effectiveness of the current inspection path through simulation of path time consumption and coverage. If the inspection frequency of high-risk equipment is insufficient or the delay is too long, the path re-planning is triggered to shorten the inspection interval, and the adjustment results are fed back to the path planning module.

[0009] Handshake communication module: The patrol robot establishes an encrypted handshake connection with the spinning equipment through a multi-protocol wireless communication module, extracts the equipment operating parameters in real time and dynamically adjusts the strobe control parameters according to the extracted data to ensure that the yarn tube is stationary under high-frequency flash.

[0010] Image acquisition module: Using adaptive stroboscopic control system and high frame rate HD camera technology, high-resolution images of the bobbin are collected and pre-processed in real time to ensure high quality and consistency of the data, and the processed image data is passed to the in-depth analysis module.

[0011] Deep analysis module: Use convolutional neural networks to perform deep analysis on pre-processed images and multivariate sensor data, generate anomaly scores, and optimize inspection paths through reinforcement learning algorithms to achieve intelligent anomaly detection and dynamic path adjustment. At the same time, the anomaly warning data is fed back to the score calculation and path planning modules to form a closed-loop optimization mechanism.

[0012] In a preferred embodiment, the processing process of the score calculation module is as follows: Step 1: Extract multi-factor features from the operating data of the spinning equipment; all features are normalized and standardized to eliminate the dimensional differences between different features and make them uniformly comparable.

[0013] Step 2: After feature extraction is completed, the feature data is mapped to a score through a multi-layer nonlinear model; if the score is higher than the classification threshold, it is marked as a high-scoring device; conversely, if the score is not higher than the classification threshold, it is marked as a low-scoring device.

[0014] Step 3: Combine the abnormal feedback results of the deep analysis module to dynamically adjust the score. The specific logic is as follows: b1. If the deep analysis module detects that the device is a high-risk device, the risk level is used as an additional input for the score to increase the weight of features related to the anomaly.

[0015] b2. For low-scoring devices, analyze the contribution values ​​of the main features that have caused the score to drop, and mark the influencing factors of the dominant features as high-priority features, so as to improve the priority ranking of the devices corresponding to the relevant features in path planning.

[0016] b3. When the feedback from the deep analysis module shows that a high-scoring device is a low-risk device, the feature weight corresponding to the high score of the device will be reduced by a preset ratio, and a penalty factor will be applied to the score of the device with lower risk to prevent the high-scoring device from masking potential problems due to insufficient risk assessment.

[0017] In a preferred embodiment, the processing process of the path planning module is as follows: Step 1: Prioritize devices based on the score difference threshold. The specific logic is as follows: c1. Calculate the score difference matrix, sort the device scores from high to low, and build a matrix based on the score difference between any two devices to filter and sort sensitive device pairs.

[0018] c2. Exclude device pairs whose score difference is less than the score difference threshold, and only retain device pairs with significant score difference for priority sorting, thereby reducing frequent path adjustments caused by score fluctuations.

[0019] c3. Generate an initial inspection path based on the priority sorting results to ensure priority inspection rights for high-risk equipment.

[0020] Step 2: Path planning uses the ant colony optimization algorithm, combined with equipment priority and scoring data, to generate the initial inspection path of the patrol robot. The path generation process is as follows: d1. Pheromone initialization: Initialize the pheromone strength of the connection paths between all devices. The pheromone value is proportional to the score difference.

[0021] d2. Path construction rules: When constructing a path, the ant individuals select the next device through a probability function based on the score difference matrix and the time taken by the current path.

[0022] d3. Path evaluation and pheromone update: After each round of iteration, the paths are scored according to their total time consumption and score matching. The pheromone concentration corresponding to the path with the highest score is enhanced, while the pheromone concentrations of other paths are attenuated.

[0023] Step 3: Simulate the time consumption and coverage of the path generated by ant colony optimization, combine the equipment distribution characteristics and inspection resource constraints, and evaluate the execution feasibility and coverage efficiency of the path. The simulation process includes: e1. Calculate the total time of the path, add the inspection time of each device and the navigation time between devices, and evaluate the time cost of the path.

[0024] e2. Simulate the path coverage, analyze whether the path focuses on covering high-scoring devices, and verify whether the detection frequency of low-scoring devices is reduced.

[0025] e3. If it is found that the inspection frequency of high-scoring equipment is insufficient or the delay is too long, it will be marked as a key optimization target and the path optimization module will be triggered.

[0026] Step 4: Combine the abnormal warnings and inspection suggestions generated by the deep analysis module to dynamically optimize the path. The specific logic is as follows: f1. If the deep analysis module marks a device as a high-risk device, its inspection interval will be shortened to the current maximum inspection interval multiplied by the ratio of the device anomaly score to the highest score. At the same time, the priority of the corresponding device in the path will be adjusted to the top position to ensure that the high-risk device is detected at least once in each inspection cycle.

[0027] f2. If the in-depth analysis results show that some low-scoring devices are low-risk devices, and the detection time accounts for more than the rated proportion of the total inspection time, the inspection interval of the corresponding device will be extended to the rated multiple of the current inspection cycle, and its priority ranking in the path will be lowered. The released inspection resources will be preferentially allocated to the detection tasks of high-risk devices.

[0028] In a preferred embodiment, the processing process of the path simulation module is as follows: Step 1: Use the geographical distribution of devices and inspection resource constraints to simulate the path time. The specific processing logic is as follows: g1. Calculation of navigation time between devices: Based on the spatial coordinates of the devices, a nonlinear path model is used to calculate the navigation time of the patrol robot between devices.

[0029] g2. Calculation of equipment inspection time: Based on the equipment score and priority, set the inspection dwell time for each equipment.

[0030] g3. Calculation of total path time: The total path time is the sum of navigation time and detection time.

[0031] Step 2: Analyze the monitoring coverage of the path for the device status, and identify the conditions for increasing or decreasing the device detection frequency based on the following metrics: h1. High-risk device detection interval: Define the expected detection interval of high-risk devices based on the maximum allowable detection interval and score; if the current detection interval exceeds the rated multiple of the expected detection interval, it is determined that the high-risk device detection coverage is insufficient and the detection frequency needs to be increased.

[0032] h2. Resource occupancy of low-risk devices: The ratio of detection time of low-scoring devices to total path time is counted. If the ratio of detection time of low-scoring devices to total path time exceeds the rated percentage, it is determined that the low-risk devices occupy too many resources and their detection frequency needs to be reduced.

[0033] h3. Define the coverage quality index as the ratio of the sum of detection intervals of high-risk devices to the sum of detection time of low-scoring devices. When the coverage quality index is greater than the preset standard, it indicates that high-risk devices are under-detected and low-risk devices occupy too many resources, and the path strategy needs to be adjusted.

[0034] Step 3: Based on the simulation results of time consumption and coverage, make the following adjustments to the path: i1. If the detection interval of a high-risk device exceeds the set standard of its target detection interval, the detection time of the low-scoring device will be reduced or the number of inspections of the low-scoring device will be reduced.

[0035] i2. If the proportion of the detection time of low-scoring devices to the total path time exceeds the set standard, the detection residence time of low-scoring devices will be reduced or low-scoring devices in the path will be deleted first.

[0036] i3. If the total path time exceeds the set standard, the high-time path segments are reordered and the device pairs whose navigation time exceeds the threshold are replaced with suboptimal connections.

[0037] In a preferred embodiment, the processing process of the handshake communication module is as follows: Step 1: The patrol robot establishes a communication connection with the device using the wireless communication protocol according to the target device list provided by the path planning module.

[0038] Step 2: The patrol robot initiates an authentication request to the target device, verifies the legitimacy of the connection through the device's unique identification code and digital certificate, and negotiates to generate a symmetric encryption key.

[0039] Step 3: After the encrypted connection is established, the patrol robot sends a parameter acquisition request to the target device through the communication interface to extract operation data in real time.

[0040] Step 4: Dynamically adjust the frequency and brightness of the strobe light according to the extracted operating parameters; the frequency adjustment is combined with the bobbin speed to calculate the strobe frequency to ensure that the bobbin is stationary during shooting; the brightness adjustment is based on the equipment vibration amplitude and operating load to optimize the strobe brightness to adapt to different lighting requirements.

[0041] In a preferred embodiment, the processing process of the image acquisition module is as follows: Step 1: Use a high-speed camera with the aid of a strobe light to capture a high-resolution image of the bobbin surface.

[0042] Step 2: Perform denoising on the collected images in time and space domains.

[0043] Step 3: Based on denoising, perform multi-dimensional enhancement processing on the image to improve the recognizability of key features.

[0044] Step 4: Standardize the size, color and value of the denoised and enhanced image data to ensure the consistency and adaptability of the data.

[0045] Step 5: Synchronously store the pre-processed high-quality images and equipment operating parameters according to timestamps to ensure the temporal consistency of image data and parameter data.

[0046] In a preferred embodiment, the processing of the deep analysis module is as follows: Step 1: Use convolutional neural network to perform multi-level feature extraction on the preprocessed image data, obtain the local texture features of the yarn tube surface through convolution operation, and compress the feature map through pooling operation to extract global features.

[0047] Step 2: Extract various features of the device from the sensor data, dynamically fuse them with the image features, and construct a comprehensive feature vector.

[0048] Step 3: Model the comprehensive feature vector through the Gaussian mixture model to quantify the anomaly score of each device; the anomaly score is based on the probability calculation of the feature distribution to identify abnormal features that deviate from the normal distribution; combine the anomaly score with the preset threshold. If the anomaly score is greater than the preset threshold, it is classified as a high-risk device, otherwise it is classified as a low-risk device. High-risk devices directly trigger an early warning and are marked as key inspection objects.

[0049] Step 4: Optimize the inspection path based on the reinforcement learning algorithm combined with the anomaly detection results. The state represents the real-time status and score of the equipment. The action is defined as the inspection path adjustment strategy. The reward function takes the high-risk equipment detection interval and the low-risk equipment detection time as the optimization target.

[0050] Step 5: Feedback the anomaly detection results and path optimization suggestions to the scoring calculation and path planning module.

[0051] The technical effects and advantages of the textile equipment real-time monitoring system based on patrol robots and stroboscopic control of the present invention are as follows: The present invention builds a highly intelligent spinning equipment monitoring system by integrating multi-level scoring calculation, optimized path planning, path time and coverage simulation, encrypted handshake communication, high-quality image acquisition and deep data analysis. First, based on historical operation data and real-time feedback, the equipment status is accurately evaluated and the score is generated. Then, the advanced ant colony optimization and reinforcement learning algorithms are used to plan and dynamically adjust the inspection path to ensure that high-risk equipment is monitored first. Through the accurate simulation of path time and coverage, resource allocation is optimized and inspection efficiency is improved. At the same time, through multi-protocol encrypted handshake connection, the equipment operation parameters are extracted in real time and the strobe control is dynamically adjusted to ensure the clarity of image acquisition and the security of data. High-frame rate high-definition video is combined with adaptive strobe control to collect and pre-process high-quality image data. Then, convolutional neural network and multivariate association model are used for deep analysis to generate abnormal warning in real time and optimize the inspection path. The whole closed-loop optimization mechanism is formed to ensure the real-time and intelligent monitoring, significantly improve the operation stability, yarn quality and production efficiency of spinning equipment, and meet the needs of modern spinning workshops for efficient and accurate monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1The present invention is a schematic structural diagram of a real-time monitoring system for textile equipment based on a patrol robot and stroboscopic control. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Embodiment 1: Figure 1 The present invention provides a real-time monitoring system for textile equipment based on a patrol robot and stroboscopic control, comprising: Scoring calculation module: Based on the historical operation data of spinning equipment, the gradient boosting decision tree algorithm is used to perform weighted scoring on multi-factor features to generate equipment scores, and the abnormal warning data from the deep analysis module is received to dynamically adjust the scoring weights.

[0055] Path planning module: Based on the device scores and setting the score difference threshold, the ant colony optimization algorithm is used to prioritize the devices and plan the optimal inspection path for the patrol robot. The system will re-sort the devices only when the score difference exceeds the threshold, and dynamically adjust the inspection path based on the path optimization suggestions of the deep analysis module.

[0056] Path simulation module: evaluates the effectiveness of the current inspection path through simulation of path time consumption and coverage. If the inspection frequency of high-risk equipment is insufficient or the delay is too long, the path re-planning is triggered to shorten the inspection interval, and the adjustment results are fed back to the path planning module.

[0057] Handshake communication module: The patrol robot establishes an encrypted handshake connection with the spinning equipment through a multi-protocol wireless communication module, extracts the equipment operating parameters in real time and dynamically adjusts the strobe control parameters according to the extracted data to ensure that the yarn tube is stationary under high-frequency flash.

[0058] Image acquisition module: Using adaptive stroboscopic control system and high frame rate HD camera technology, high-resolution images of the bobbin are collected and pre-processed in real time to ensure high quality and consistency of the data, and the processed image data is passed to the in-depth analysis module.

[0059] Deep analysis module: Use convolutional neural networks to perform deep analysis on pre-processed images and multivariate sensor data, generate anomaly scores, and optimize inspection paths through reinforcement learning algorithms to achieve intelligent anomaly detection and dynamic path adjustment. At the same time, the anomaly warning data is fed back to the score calculation and path planning modules to form a closed-loop optimization mechanism.

[0060] During the long-term operation of spinning equipment, the dynamic characteristics of its bobbin status directly affect the quality and production efficiency of the yarn. In order to achieve an accurate assessment of the equipment's operating status, it is necessary to integrate multiple characteristics, including bobbin speed fluctuations, shape consistency, hairiness distribution, and cleanliness characteristics. The scoring calculation module integrates these multi-factor characteristics, combines nonlinear models to establish a mapping relationship between equipment status and scores, and dynamically adjusts the scores based on real-time feedback, ultimately providing a priority basis for subsequent inspection path optimization.

[0061] The processing process of the score calculation module is as follows: Step 1: Extract multi-factor features from the operating data of the spinning equipment, including bobbin speed fluctuation features, shape consistency features, hairiness distribution features and cleanliness features. These features are defined as follows: Bobbin speed fluctuation characteristics: By analyzing the bobbin speed data during the equipment operation cycle, the dynamic change amplitude and frequency of the speed are calculated. The fluctuation trend of the speed during the entire operation cycle is integrated to reflect the stability of the equipment operation.

[0062] Shape consistency feature: The image analysis algorithm is used to extract the cross-sectional parameters of the yarn tube geometry, including the diameter variation and ellipticity, and the degree of shape deviation is quantified by the coefficient of variation of these parameters.

[0063] Hairiness distribution characteristics: record the total amount and spatial distribution of hairiness on the surface of the yarn tube, and evaluate its distribution uniformity by comparing the density of areas where hairiness is concentrated with that of sparse areas.

[0064] Cleanliness characteristics: By analyzing the dirt coverage and particle density in the yarn tube surface image, combined with the signal characteristics of the vibration sensor, the cleanliness of the equipment operating environment is calculated.

[0065] All features are normalized and standardized to eliminate the dimensional differences between different features and make them uniformly comparable.

[0066] Step 2: After feature extraction is completed, the feature data is mapped to scores through a multi-layer nonlinear model. The model is based on the following logic: a1. Integrate the variation amplitude and frequency of the bobbin speed fluctuation characteristics to reflect the operation stability of the equipment. If the speed fluctuation frequency of the equipment is high and the amplitude is large, the score will be significantly reduced, reflecting the instability of the equipment operation status.

[0067] a2. Compare the degree of deviation of the shape consistency feature with the mean of the reference shape. When the deviation is large, the corresponding score decreases; if the shape remains stable over multiple operation cycles, it forms a positive gain for the score.

[0068] a3. The uniformity of the hairiness distribution is measured by the complexity of the spatial distribution. The more uniform the hairiness distribution, the higher its contribution to the score; if the hairiness is concentrated in a local area, the score will decrease according to the unevenness of the distribution.

[0069] a4. The cleanliness characteristic is evaluated based on the combined results of dirt coverage and particle density. If both dirt coverage and particle density are low, the score will reflect the positive contribution of cleanliness; when the cleanliness is below the set threshold, the score will be significantly reduced.

[0070] For example, the scoring module uses a three-layer association analysis model to establish a nonlinear mapping relationship between the above feature data and the scoring results. The calculation formula of the model can be: in: : Rating of the device.

[0071] :feature The basic weight of the network is generated by statistical analysis of historical data, and the weight value is adjusted dynamically. : The comprehensive evaluation value of each feature, the calculation process is as follows: For the bobbin speed fluctuation characteristics :Combined with the change of bobbin speed and frequency , through the formula Calculate its eigenvalue. For shape consistency features :According to the deviation of the bobbin geometric parameters, the construction characteristic value is ,in is the standard deviation of the shape, For the distribution characteristics of hairiness : Using spatial distribution entropy Total hairiness ,definition For cleanliness characteristics : Based on pollution coverage and particle density , the eigenvalue is calculated as . : Dynamic feedback adjustment item, where Indicates the abnormal feedback value obtained from the deep analysis module (such as abnormal frequency, equipment risk level, etc.), is the amplification factor of the feedback item. If the score is higher than the classification threshold, it is marked as a high-scoring device; otherwise, if the score is not higher than the classification threshold, it is marked as a low-scoring device.

[0072] Step 3: The score calculation module combines the abnormal feedback results of the deep analysis module to dynamically adjust the score. The specific logic is as follows: b1. If the deep analysis module detects that the device is a high-risk device, the risk level is used as an additional input for scoring, and the weight of features related to the anomaly is increased. For example, the weight of the speed fluctuation feature is increased to twice the original weight, so that the scoring model is more sensitive to changes in this feature.

[0073] b2. For low-scoring equipment, analyze the contribution values ​​of the main features that caused the score drop. For example, if the contribution of the hairiness distribution feature to the score drop exceeds 50%, or the abnormal score of the cleanliness feature is higher than the set threshold, mark the influencing factor of the dominant feature as a high-priority feature, and increase the priority ranking of the equipment corresponding to the relevant features in path planning to ensure that resources are tilted towards key problem equipment.

[0074] b3. When the feedback from the deep analysis module shows that a high-scoring device is a low-risk device, the score is partially adjusted by introducing a risk compensation mechanism. Specifically, the weight of the feature corresponding to the high score of the device is reduced by a preset ratio, and a penalty factor is used for the score of the device with lower risk, so as to avoid the high-scoring device from covering up potential problems due to insufficient risk assessment, and ensure that the score reflects the status of the device more objectively.

[0075] Finally, the generated equipment scores are output in the form of a list, providing the core data input for priority sorting for the path planning module.

[0076] Step 4: The final generated equipment score is output to the path planning module as the core basis for priority sorting. At the same time, the score calculation module compares the score trend with the historical score record to form a trend chart of the long-term status of the equipment, providing data support for the optimization of subsequent inspection paths. For example, when the equipment score drops significantly and the feedback frequency related to the dominant feature increases, the score calculation module can mark the equipment as a high-priority inspection object in the output.

[0077] The scoring calculation module extracts the multi-factor characteristics of the spinning equipment, including bobbin speed fluctuation, shape consistency, hairiness distribution and cleanliness, and combines the nonlinear multi-layer model to comprehensively evaluate the characteristics to generate a score that reflects the equipment's operating status and potential risks. The module uses nonlinear mapping between multidimensional features to capture feature correlation and dynamic changes, and adjusts feature weights in real time in combination with abnormal feedback from the deep analysis module to improve the dynamic response capability and accuracy of the score. The final output score results provide the path planning module with a basis for equipment priority sorting, and through the linkage of scoring and feedback, dynamic monitoring and precise management of spinning equipment are achieved.

[0078] The operating conditions of multiple spinning equipment vary significantly. High-risk equipment needs to be inspected first, while the inspection frequency of low-risk equipment can be appropriately reduced to optimize resource allocation. However, the volatility of equipment scores and the stability of path planning need to be balanced; overly sensitive scores will lead to frequent path adjustments, affecting inspection efficiency, while the lag in scores may result in high-risk equipment not being monitored in a timely manner. The path planning module integrates equipment scores and path time simulation, and uses the ant colony optimization algorithm combined with a dynamic feedback mechanism to achieve priority sorting and optimal generation of inspection paths. At the same time, the path is optimized in real time based on the feedback from the deep analysis module to ensure efficient monitoring coverage and reasonable allocation of inspection resources.

[0079] The processing of the path planning module is as follows: Step 1: First, receive the device scores generated by the score calculation module and prioritize the devices according to the score difference threshold. The specific logic is as follows: c1. Calculate the score difference matrix, sort the device scores from high to low, and build a matrix based on the score difference between any two devices to filter and sort sensitive device pairs.

[0080] c2. Exclude device pairs whose score difference is less than the score difference threshold, and only retain device pairs with significant score difference for priority sorting, thereby reducing frequent path adjustments caused by score fluctuations.

[0081] c3. Generate an initial inspection path based on the priority sorting results to ensure priority inspection rights for high-risk equipment.

[0082] Step 2: Path planning uses an improved ant colony optimization algorithm, combined with equipment priority and scoring data, to generate the initial inspection path of the patrol robot. The path generation process is as follows: d1. Pheromone initialization: Initialize the pheromone strength of the connection paths between all devices. The pheromone value is proportional to the score difference. The greater the score difference, the higher the initial pheromone strength of the corresponding path. d2. Path construction rules: When constructing a path, the ant individual selects the next device through a probability function based on the score difference matrix and the time consumption of the current path. The design of the probability function comprehensively considers the score difference and the path time consumption. The formula is: ;in: : ant slave device To device The probability of selection.

[0083] :path To device pheromone concentration.

[0084] :equipment To device 's rating differences.

[0085] : The impact weight of the score difference is fixed at 2. d3. Path evaluation and pheromone update: After each round of iteration, the paths are scored according to the total time consumed by the paths and the score matching degree. The pheromone concentration corresponding to the path with the highest score is enhanced, while the pheromone concentration of other paths is attenuated.

[0086] Step 3: Simulate the time consumption and coverage of the path generated by ant colony optimization, and evaluate the execution feasibility and coverage efficiency of the path based on the characteristics of equipment distribution and inspection resource constraints. The simulation process includes: e1. Calculate the total time of the path, add the inspection time of each device and the navigation time between devices, and evaluate the time cost of the path.

[0087] e2. Simulate the path coverage, analyze whether the path focuses on covering high-scoring devices, and verify whether the detection frequency of low-scoring devices is reasonably reduced.

[0088] e3. If it is found that the inspection frequency of high-scoring equipment is insufficient or the delay is too long, it will be marked as a key optimization target and the path optimization module will be triggered.

[0089] Step 4: Combine the abnormal warnings and inspection suggestions generated by the deep analysis module to dynamically optimize the path. The specific logic is as follows: f1. If the deep analysis module marks a device as a high-risk device, its inspection interval will be shortened to the current maximum inspection interval multiplied by the ratio of the device anomaly score to the highest score. At the same time, the priority of the device in the path will be adjusted to the top position to ensure that the high-risk device is detected at least once in each inspection cycle.

[0090] f2. If the in-depth analysis results show that some low-scoring devices are low-risk devices, and the detection time accounts for more than the rated proportion of the total inspection time, the inspection interval of the corresponding device will be extended to the rated multiple of the current inspection cycle, and its priority ranking in the path will be lowered. The released inspection resources will be preferentially allocated to the detection tasks of high-risk devices.

[0091] Step 5: The optimized path includes device priority sorting, inspection time allocation and dynamic inspection interval. Output the path data to the handshake communication module, and feed back the path optimization results to the scoring calculation module for the next round of scoring weight adjustment and dynamic optimization.

[0092] In the inspection task of spinning equipment, the inspection efficiency after path planning directly determines the timely detection of high-risk equipment and the rationality of resource allocation. Since the inspection path involves the geographical distribution, priority sorting and time constraints between equipment, the planned path needs to be simulated to verify the effectiveness of the path. The path simulation module identifies potential bottlenecks of the inspection path (such as too long inspection intervals for high-risk equipment or too many inspection resources occupied by low-risk equipment) through simulation analysis of path time consumption and coverage, and combines dynamic inspection needs to make adjustment suggestions and feedback to the path planning module to optimize the inspection strategy.

[0093] The processing of the path simulation module is as follows: Step 1: Receive the initial inspection path output by the path planning module, and use the geographical distribution of devices and inspection resource constraints to accurately simulate the path time. The specific processing logic is as follows: g1. Calculation of navigation time between devices: According to the spatial coordinates of the devices, a nonlinear path model is used to calculate the navigation time of the patrol robot between devices. Assume that the device and The coordinates of and , navigation time Expressed as: ;in, is the average travel speed of the patrol robot. The navigation time takes into account the impact of obstacles and the actual complexity of the equipment layout. g2. Equipment inspection time calculation: Combine the equipment score and priority to set the inspection stay time for each equipment. The dwell time of high-scoring devices is based on their score ratio, and the formula is: in, The basic stay time, Rating of the device, The highest score. g3. Total path time calculation: Total path time is the sum of navigation time and detection time, and the formula is: ; Evaluate the time efficiency of the path by calculating the total time consumption and mark the key nodes of the high time-consuming path. Step 2: Analyze the monitoring coverage of the path to the device status, and combine the following metrics to clarify the conditions for increasing or decreasing the device detection frequency: h1. High-risk device detection interval: Define the expected detection interval for high-risk devices , with the maximum allowed detection interval And the ratings are based on: ;in, The detection frequency of high-risk devices in the path. For example, if the current detection interval exceeds the rated multiple of the expected detection interval, it is determined that the detection coverage of high-risk devices is insufficient and the detection frequency needs to be increased.

[0094] h2. Resource usage of low-risk devices: Count the proportion of detection time of low-scoring devices to the total path time. : For example, if the detection time of a low-scoring device accounts for more than the rated percentage of the total path time, it is determined that the low-risk device occupies too many resources and its detection frequency needs to be reduced. h3. Define coverage quality index It is the ratio of the sum of detection intervals of high-risk devices to the sum of detection time of low-scoring devices: in, Indicates high risk equipment The detection interval time refers to the time difference between two consecutive inspections of the device in the inspection path; Indicates low risk device The single detection time is reflected in the dwell time on the device. For example, when the coverage quality index is greater than the preset standard, it indicates that high-risk devices are under-detected and low-risk devices occupy too many resources, and the path strategy needs to be adjusted.

[0095] Step 3: Based on the simulation results of time consumption and coverage, make the following adjustments to the path: i1. Increase the detection frequency of high-risk devices: If the detection interval of a high-risk device exceeds the set standard of its target detection interval, the detection time of low-scoring devices will be reduced or the number of inspections of low-scoring devices will be reduced to increase the detection frequency of high-risk devices.

[0096] i2. Reduce the detection frequency of low-risk devices: If the detection time of low-scoring devices accounts for a proportion of the total path time that exceeds the set standard, the detection residence time of low-scoring devices will be reduced or low-scoring devices in the path will be deleted first.

[0097] i3. Optimization of total path time: If the total path time exceeds the set standard, the high-time path segments are reordered and the device pairs whose navigation time exceeds the threshold are replaced with suboptimal connections.

[0098] Step 4: Feedback the adjusted optimized path parameters and detection frequency recommendations to the path planning module, including: detection interval adjustment targets for high-risk devices; detection time reduction and inspection frequency adjustment recommendations for low-scoring devices; and device connection sequence and resource allocation ratio after total path time optimization.

[0099] The path simulation module simulates and analyzes the time consumption and coverage of the inspection path, evaluates the total time consumption of the path based on the accumulation of navigation time and detection time, and calculates the coverage quality index to quantify the path monitoring effect through the detection interval of high-risk equipment and the proportion of detection time of low-risk equipment. The module combines clear metrics and puts forward specific optimization suggestions when the detection interval exceeds the target range or the resource occupancy ratio deviates from the reasonable threshold, including adjusting the detection time allocation, optimizing the path sequence and the connection method of key nodes, to ensure the priority inspection of high-risk equipment and the reasonable allocation of inspection resources. The optimization results are fed back to the path planning module to form a closed-loop adjustment mechanism, which significantly improves the efficiency of path execution and the balance of monitoring coverage.

[0100] The key to real-time monitoring of the operating status of spinning equipment lies in stable communication and data interaction with the patrol robot. The patrol robot obtains the equipment operating parameters through the wireless communication module for subsequent data collection and analysis. Due to the variety of equipment and complex operating environment, the reliability, security and real-time performance of communication are particularly important. At the same time, the monitoring of the bobbin at high speed depends on the dynamic adjustment of the stroboscopic control parameters to ensure that the collected images are clear enough for subsequent analysis. The core of the handshake communication module is to establish a high-security connection through an encrypted handshake protocol, extract equipment parameters in real time, and adjust the stroboscopic control according to the equipment operating status to achieve accurate data collection support.

[0101] The processing process of the handshake communication module is as follows: Step 1: The patrol robot dynamically selects the optimal wireless communication protocol (such as Wi-Fi, ZigBee or Bluetooth) to establish a communication connection with the device based on the target device list provided by the path planning module. By detecting the signal strength in real time, ensure that the communication signal strength is higher than the standard. If the signal strength is insufficient, it will automatically switch to other communication protocols. At the same time, the communication delay is calculated by the round-trip time of the handshake packet. If the delay exceeds the standard, the device is marked as a communication problem node, and the inspection order is adjusted to optimize the connection stability.

[0102] Step 2: The patrol robot initiates an authentication request to the target device, verifies the legitimacy of the connection through the device's unique identification code and digital certificate, and negotiates to generate a symmetric encryption key. The encryption key negotiation uses the elliptic curve encryption algorithm to ensure the security and efficiency of the key. After completing the key negotiation, both parties confirm the connection status and record the success or failure of the handshake connection. If the handshake fails, the retry logic is triggered until a stable encrypted connection is completed.

[0103] Step 3: After the encrypted connection is established, the patrol robot sends a parameter acquisition request to the target device through the communication interface to extract real-time operating data such as bobbin speed, equipment vibration amplitude, and operating load status. These parameters are recorded as time series and synchronously transmitted to the strobe control module. The frequency of data extraction is dynamically adjusted according to the priority of the target device to ensure the real-time and integrity of the operating data of high-risk equipment.

[0104] Step 4: Dynamically adjust the frequency and brightness of the strobe light according to the extracted operating parameters. The frequency adjustment is combined with the bobbin speed to calculate the strobe frequency to ensure that the bobbin is stationary when shooting. The brightness adjustment is based on the equipment vibration amplitude and operating load to optimize the strobe brightness to adapt to different lighting requirements. The adjustment of strobe frequency and brightness is controlled by a nonlinear model to ensure clear and stable image acquisition in a dynamic operating environment. For example, dynamically adjust the frequency of the strobe light according to the equipment operating parameters and brightness : Frequency adjustment logic: combined with bobbin speed , calculate the frequency of the strobe light to ensure that the bobbin appears stationary when photographing. The frequency calculation formula is: in, The number of rotating blades of the bobbin ensures that the strobe frequency is synchronized with the bobbin speed. Brightness adjustment logic: combined with the vibration amplitude of the equipment and load status , optimize the strobe light brightness to adapt to different shooting environments. The brightness adjustment logic is: ;in, is the basic brightness, and is the maximum value of vibration and load.

[0105] Step 5: The handshake communication module monitors the communication status in real time, records the signal strength, delay and number of failures of each handshake process, and marks abnormal communication devices. At the same time, the effect of the strobe control parameters is verified through image pre-capture. If the bobbin does not appear to be still or the image brightness is insufficient, the strobe parameters are readjusted and the number of adjustments is recorded. The communication status and strobe control results are passed to the path planning module as feedback data to further optimize the inspection strategy and strobe parameter configuration.

[0106] The handshake communication module dynamically selects the optimal wireless communication protocol and uses a high-security encrypted handshake mechanism to establish a stable connection with the equipment, and extracts key parameters such as bobbin speed, equipment vibration amplitude, and operating load in real time. The module dynamically adjusts the frequency and brightness of the strobe light based on the extracted data to ensure that the bobbin is stationary under high-frequency flash. At the same time, data feedback is formed through communication status monitoring and strobe effect verification, abnormal communication equipment is marked, and the strobe control strategy is optimized, ultimately providing reliable data support for the accuracy of image acquisition and subsequent analysis.

[0107] The processing of the image acquisition module is as follows: Step 1: Use a high-speed camera that supports thousands of frames per second to capture high-resolution images of the bobbin surface with the aid of a strobe light. The camera frame rate is kept consistent with the strobe frequency to avoid image overlap or blur. To ensure the integrity of image details, the camera exposure time and resolution are dynamically adjusted in combination with the bobbin surface feature requirements, and the image acquisition effect is optimized by recording and adjusting the matching exposure brightness.

[0108] Step 2: Perform denoising in the time and space domains on the captured images to reduce blur and surface texture noise caused by light fluctuations and vibrations. Temporal denoising eliminates random noise by averaging the brightness of the same pixel position in consecutive frames; spatial denoising combines adaptive median filtering to adjust the filter window size according to the complexity of image texture, thereby eliminating noise while retaining the detail integrity of the key feature areas of the bobbin.

[0109] Step 3: Based on denoising, the image is enhanced in multiple dimensions to improve the recognizability of key features. The surface contrast of the bobbin is enhanced by histogram equalization to highlight the shape and hairiness features, while the edge texture of the bobbin is enhanced by Laplace filtering to ensure clear geometric shape features. In addition, the brightness distribution of the entire image is adjusted by brightness uniformity processing to ensure the overall light balance of the image and adapt to the diverse acquisition environment requirements.

[0110] Step 4: Standardize the size, color, and value of the denoised and enhanced image data to ensure data consistency and adaptability. By cropping the image to a uniform size, the impact of resolution and scale differences on subsequent analysis is eliminated; the image tone is adjusted based on the reference grayscale template to eliminate the color deviation of ambient light; and the pixel values ​​are mapped to a fixed value range through normalization processing to provide standardized input data for the depth analysis module.

[0111] Step 5: Synchronize and store the pre-processed high-quality images and equipment operating parameters according to the timestamp to ensure the timing consistency of the image data and parameter data. The pre-processed data is transmitted to the deep analysis module through an efficient transmission mechanism, and the data transmission status is monitored in real time. If a delay or loss occurs, the retransmission logic is started and the problematic device is recorded to provide data support for subsequent analysis and feedback.

[0112] In the condition monitoring of spinning equipment, the surface condition of the yarn tube (such as hairiness distribution, shape consistency and cleanliness) and the equipment operating parameters (such as vibration amplitude and speed stability) directly reflect the health of the equipment operation. Traditional monitoring methods have difficulty processing these complex, multi-dimensional and dynamic data, so it is necessary to use deep learning technology to automatically extract abnormal features from high-resolution images and multivariate sensor data, and combine reinforcement learning to optimize the inspection path to achieve an efficient and intelligent monitoring closed loop. The core of the deep analysis module is to extract image features through convolutional neural networks, identify equipment anomalies in combination with multivariate association models, and use reinforcement learning to achieve path optimization and dynamic adjustment.

[0113] The processing of the deep analysis module is as follows: Step 1: Use convolutional neural networks to perform multi-level feature extraction on the preprocessed image data, obtain local texture features of the bobbin surface through convolution operations, and compress feature maps through pooling operations to extract global features, such as bobbin shape, hairiness distribution, and cleanliness. Finally, the local texture and global features are fused into low-dimensional feature vectors through the fully connected layer to generate highly expressive image features, providing accurate descriptions for subsequent analysis.

[0114] Step 2: Extract the vibration characteristics, speed fluctuation characteristics and operating load characteristics of the equipment from the sensor data, dynamically fuse them with the image characteristics, and construct a comprehensive feature vector. The fusion process combines the weight ratio of the image and sensor characteristics, and dynamically adjusts the feature importance weight according to the equipment status to ensure that the comprehensive characteristics can reflect the state characteristics of the bobbin surface and cover the dynamic behavior of the equipment during operation.

[0115] Step 3: Model the comprehensive feature vector through the Gaussian mixture model to quantify the abnormal score of each device. The abnormal score is based on the probability calculation of the feature distribution to identify abnormal features that deviate from the normal distribution. Combining the abnormal score with the preset threshold, if the abnormal score is greater than the preset threshold, it is classified as a high-risk device, otherwise it is classified as a low-risk device. High-risk devices directly trigger an early warning and are marked as key inspection objects. For example, the abnormal score of the comprehensive feature vector is calculated through the Gaussian mixture model. , the formula is: ;in is the number of Gaussian distributions, is the weight, For the The mean and covariance of the distribution, is a comprehensive feature vector. Step 4: Optimize the inspection path based on the reinforcement learning algorithm combined with the anomaly detection results. The state represents the real-time state and score of the device. The action is defined as the inspection path adjustment strategy. The reward function takes the detection interval of high-risk devices and the detection time of low-risk devices as the optimization target. Through the deep Q network (DQN) learning path adjustment strategy, the detection interval of high-risk devices is shortened first, while the detection frequency of low-risk devices is reasonably reduced to achieve dynamic optimization allocation of resources.

[0116] Step 5: Feedback the anomaly detection results and path optimization suggestions to the scoring calculation and path planning module. Provide the scoring calculation module with the anomaly level data of high-risk equipment to dynamically adjust the scoring weight; pass the optimized inspection priority and path adjustment suggestions to the path planning module to form a closed-loop mechanism for real-time adjustment to ensure the dynamic adaptability and efficiency of the inspection strategy.

[0117] The deep analysis module extracts the local texture and global features of the bobbin image through a convolutional neural network, and builds a multivariate comprehensive feature vector based on sensor data such as vibration, speed fluctuation and load. It uses a Gaussian mixture model to quantify the equipment anomaly score and divide the risk level. The module optimizes the inspection path through a reinforcement learning algorithm, prioritizes shortening the inspection interval of high-risk equipment, reasonably reduces the inspection frequency of low-risk equipment, and realizes dynamic allocation of resources and path adjustment. At the same time, the anomaly detection results and path optimization suggestions are fed back to the score calculation and path planning modules in a closed-loop form to ensure the real-time and intelligent nature of the inspection strategy.

[0118] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0119] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0120] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0121] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A real-time monitoring system for textile equipment based on patrol robots and stroboscopic control, characterized in that: Includes steps: Scoring calculation module: Based on the historical operation data of spinning equipment, the gradient boosting decision tree algorithm is used to perform weighted scoring on multi-factor features, generate equipment scores, and receive abnormal warning data from the deep analysis module to dynamically adjust the scoring weights; Path planning module: Based on the device scores and the set score difference threshold, the ant colony optimization algorithm is used to prioritize the devices and plan the optimal inspection path for the patrol robot. The device will be re-sorted only when the score difference exceeds the threshold, and the inspection path will be dynamically adjusted according to the path optimization suggestions of the deep analysis module. Path simulation module: evaluates the effectiveness of the current inspection path through simulation of path time consumption and coverage. If the inspection frequency of high-risk equipment is insufficient or the delay is too long, it triggers path re-planning to shorten the inspection interval and feeds back the adjustment results to the path planning module. Handshake communication module: The patrol robot establishes an encrypted handshake connection with the spinning equipment through a multi-protocol wireless communication module, extracts the equipment operating parameters in real time and dynamically adjusts the stroboscopic control parameters according to the extracted data to ensure that the yarn tube is stationary under high-frequency flashing; Image acquisition module: using adaptive stroboscopic control system and high frame rate HD camera technology, collect high-resolution images of bobbins and perform real-time preprocessing to ensure high quality and consistency of data, and pass the processed image data to the in-depth analysis module; Deep analysis module: Use convolutional neural networks to perform deep analysis on pre-processed images and multivariate sensor data, generate anomaly scores, and optimize inspection paths through reinforcement learning algorithms to achieve intelligent anomaly detection and dynamic path adjustment. At the same time, the anomaly warning data is fed back to the score calculation and path planning modules to form a closed-loop optimization mechanism.

2. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 1 is characterized in that: The processing of the scoring calculation module is as follows: Step 1: Extract multi-factor features from the operating data of the spinning equipment; all features are normalized and standardized to eliminate the dimensional differences between different features and make them uniformly comparable; Step 2: After feature extraction is completed, the feature data is mapped to a score through a multi-layer nonlinear model; if the score is higher than the classification threshold, it is marked as a high-scoring device; conversely, if the score is not higher than the classification threshold, it is marked as a low-scoring device; Step 3: Combine the abnormal feedback results of the deep analysis module to dynamically adjust the score. The specific logic is as follows: b1. If the deep analysis module detects that the device is a high-risk device, the risk level is used as an additional input for the score, increasing the weight of features related to the anomaly; b2. For low-scoring devices, analyze the contribution values ​​of the main features that have caused the score to drop, mark the influencing factors of the dominant features as high-priority features, and improve the priority ranking of the devices corresponding to the relevant features in path planning; b3. When the feedback from the deep analysis module shows that a high-scoring device is a low-risk device, the feature weight corresponding to the high score of the device will be reduced by a preset ratio, and a penalty factor will be applied to the score of the device with lower risk to prevent the high-scoring device from masking potential problems due to insufficient risk assessment.

3. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 2 is characterized in that: The processing of the path planning module is as follows: Step 1: Prioritize devices based on the score difference threshold. The specific logic is as follows: c1. Calculate the score difference matrix, sort the device scores from high to low, and build a matrix based on the score difference between any two devices to screen and sort sensitive device pairs; c2. Exclude device pairs with a score difference less than the score difference threshold, and only retain device pairs with significant score differences for priority sorting, thereby reducing frequent path adjustments caused by score fluctuations; c3. Generate an initial inspection route based on the priority sorting results to ensure that high-risk equipment has priority for inspection; Step 2: Path planning uses the ant colony optimization algorithm, combined with equipment priority and scoring data, to generate the initial inspection path of the patrol robot. The path generation process is as follows: d1. Pheromone initialization: Initialize the pheromone strength of the connection paths between all devices. The pheromone value is proportional to the score difference. d2. Path construction rules: When constructing a path, the ant individuals select the next device through a probability function based on the score difference matrix and the time taken by the current path; d3. Path evaluation and pheromone update: After each round of iteration, the paths are scored according to their total time consumption and score matching degree. The pheromone concentration corresponding to the path with the highest score is increased, while the pheromone concentration of other paths is attenuated. Step 3: Simulate the time consumption and coverage of the path generated by ant colony optimization, combine the equipment distribution characteristics and inspection resource constraints, and evaluate the execution feasibility and coverage efficiency of the path. The simulation process includes: e1. Calculate the total time of the path, add the inspection time of each device and the navigation time between devices, and evaluate the time cost of the path; e2. Simulate the path coverage, analyze whether the path focuses on covering high-scoring devices, and verify whether the detection frequency of low-scoring devices is reduced; e3. If it is found that the inspection frequency of high-scoring equipment is insufficient or the delay is too long, it will be marked as a key optimization target and the path optimization module will be triggered.

4. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 3 is characterized in that: The processing of the path planning module also includes the following: Step 4: Combine the abnormal warnings and inspection suggestions generated by the deep analysis module to dynamically optimize the path. The specific logic is as follows: f1. If the deep analysis module marks a device as a high-risk device, its inspection interval will be shortened to the current maximum inspection interval multiplied by the ratio of the device anomaly score to the highest score, and the priority of the corresponding device in the path will be adjusted to the top position to ensure that the high-risk device is inspected at least once in each inspection cycle; f2. If the in-depth analysis results show that some low-scoring devices are low-risk devices, and the detection time accounts for more than the rated proportion of the total inspection time, the inspection interval of the corresponding device will be extended to the rated multiple of the current inspection cycle, and its priority ranking in the path will be lowered. The released inspection resources will be preferentially allocated to the detection tasks of high-risk devices.

5. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 4 is characterized in that: The processing of the path simulation module is as follows: Step 1: Use the geographical distribution of devices and inspection resource constraints to simulate the path time. The specific processing logic is as follows: g1. Calculation of navigation time between devices: Based on the spatial coordinates of the devices, a nonlinear path model is used to calculate the navigation time of the patrol robot between devices; g2. Equipment inspection time calculation: Combine equipment scores and priorities to set the inspection stay time for each equipment; g3. Calculation of total path time: The total path time is the sum of navigation time and detection time; Step 2: Analyze the monitoring coverage of the path for the device status, and identify the conditions for increasing or decreasing the device detection frequency based on the following metrics: h1. High-risk device detection interval: Define the expected detection interval of high-risk devices based on the maximum allowed detection interval and score; if the current detection interval exceeds the rated multiple of the expected detection interval, it is determined that the detection coverage of high-risk devices is insufficient and the detection frequency needs to be increased; h2. Resource usage of low-risk devices: Statistics on the proportion of detection time of low-scoring devices to the total path time; If the detection time of low-scoring devices accounts for more than the rated percentage of the total path time, it is determined that the low-risk devices occupy too many resources and need to reduce their detection frequency; h3. Define the coverage quality index as the ratio of the sum of detection intervals of high-risk devices to the sum of detection time of low-scoring devices. When the coverage quality index is greater than the preset standard, it indicates that high-risk devices are under-detected and low-risk devices occupy too many resources, and the path strategy needs to be adjusted.

6. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 5 is characterized in that: The processing of the path simulation module also includes the following process: Step 3: Based on the simulation results of time consumption and coverage, make the following adjustments to the path: i1. If the detection interval of a high-risk device exceeds the set standard of its target detection interval, the detection time of the low-scoring device will be reduced or the number of inspections of the low-scoring device will be reduced; i2. If the proportion of the detection time of low-scoring devices to the total path time exceeds the set standard, the detection stay time of low-scoring devices is reduced or low-scoring devices in the path are removed first; i3. If the total path time exceeds the set standard, the high-time path segments are reordered and the device pairs whose navigation time exceeds the threshold are replaced with suboptimal connections.

7. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 6 is characterized in that: The processing process of the handshake communication module is as follows: Step 1: The patrol robot establishes a communication connection with the device using the wireless communication protocol according to the target device list provided by the path planning module; Step 2: The patrol robot initiates an identity authentication request to the target device, verifies the legitimacy of the connection through the device's unique identification code and digital certificate, and negotiates to generate a symmetric encryption key; Step 3: After the encrypted connection is established, the patrol robot sends a parameter acquisition request to the target device through the communication interface to extract operation data in real time; Step 4: Dynamically adjust the frequency and brightness of the strobe light according to the extracted operating parameters; the frequency adjustment is combined with the bobbin speed, and the strobe frequency is calculated to ensure that the bobbin is stationary during shooting; Brightness adjustment is based on the vibration amplitude and operating load of the equipment, optimizing the brightness of the light to suit different lighting requirements.

8. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 7 is characterized in that: The processing process of the image acquisition module is as follows: Step 1: Use a high-speed camera to capture a high-resolution image of the bobbin surface with the aid of a strobe light; Step 2: Perform denoising on the collected images in time and space domains; Step 3: Perform multi-dimensional enhancement processing on the image based on denoising to improve the recognizability of key features; Step 4: Standardize the size, color and value of the denoised and enhanced image data to ensure the consistency and adaptability of the data; Step 5: Synchronously store the pre-processed high-quality images and equipment operating parameters according to timestamps to ensure the temporal consistency of image data and parameter data.

9. The textile equipment real-time monitoring system based on patrol robot and stroboscopic control according to claim 8 is characterized in that: The processing of the deep analysis module is as follows: Step 1: Use convolutional neural network to perform multi-level feature extraction on the preprocessed image data, obtain the local texture features of the yarn tube surface through convolution operation, and compress the feature map through pooling operation to extract global features; Step 2: Extract various features of the device from the sensor data, dynamically fuse them with the image features, and construct a comprehensive feature vector; Step 3: Model the comprehensive feature vector through the Gaussian mixture model to quantify the anomaly score of each device; the anomaly score is based on the probability calculation of the feature distribution to identify abnormal features that deviate from the normal distribution; Combined with the anomaly score and the preset threshold, if the anomaly score is greater than the preset threshold, it is classified as a high-risk device, otherwise it is classified as a low-risk device. High-risk devices directly trigger an early warning and are marked as key inspection objects; Step 4: Optimize the inspection path based on the reinforcement learning algorithm combined with the anomaly detection results. The state represents the real-time status and score of the device. The action is defined as the inspection path adjustment strategy. The reward function takes the high-risk device detection interval and the low-risk device detection time as the optimization target. Step 5: Feedback the anomaly detection results and path optimization suggestions to the scoring calculation and path planning module.

Images