Remote monitoring platform and method for spinning foreign fiber removing equipment based on 5g and internet of things

Abstract

The application belongs to the technical field of production management, and provides a spinning foreign fiber removing equipment remote monitoring platform and method based on 5G and the Internet of Things. The method comprises the following steps: using a joint evaluation model to evaluate operation related data transmitted by the spinning foreign fiber removing equipment, and obtaining a fault trend type of the spinning foreign fiber removing equipment; if the fault trend type is a mild fault trend, the remote monitoring platform sends a first regulation and control command to an edge server, and the edge server regulates and controls the spinning foreign fiber removing equipment based on the first regulation and control command; if the fault trend type is an obvious fault trend, the remote monitoring platform sends a second regulation and control command containing regulation and control parameters to the edge server, and the edge server regulates and controls the spinning foreign fiber removing equipment based on the second regulation and control command. The application can realize accurate fault prediction and hierarchical regulation and control, reduce the risk of fault deterioration, improve the stability of the equipment and the spinning quality, and reduce the loss caused by shutdown.

Description

Technical Field

[0001] The present invention relates to the field of production management technology, and in particular to a remote monitoring platform and method for spinning foreign fiber removal equipment based on 5G and the Internet of Things. Background Art

[0002] In the textile industry, the quality of the spinning process directly determines the quality of subsequent fabrics and finished products. The presence of foreign matter (non-cotton fibers such as polypropylene, hair, and dyed yarn) mixed into the raw cotton is a key factor affecting spinning quality. Failure to effectively remove these foreign matter can lead to yarn defects, uneven dyeing, and even, in severe cases, the rejection of an entire batch of products, resulting in significant financial losses for the company. Therefore, as a core piece of equipment in a spinning production line, the stable and efficient operation of spinning foreign matter removal equipment is crucial to ensuring production quality.

[0003] To monitor spinning foreign fiber removal equipment, the industry has begun introducing automated monitoring technologies. These sensors, installed on the equipment, collect operating parameters (such as speed, temperature, and foreign fiber identification accuracy) and transmit the data to a local monitoring terminal via a wired or wireless network. However, these technologies still have significant shortcomings. For one thing, traditional networks (such as 4G and wired networks) rely on limited bandwidth and high latency, making them inadequate for the real-time transmission of large-scale equipment data. Furthermore, local monitoring terminals have limited data analysis capabilities, enabling only simple fault alarms and failing to predict equipment failure trends, making it even more difficult to formulate precise control strategies based on the severity of the problem. When equipment shows signs of a minor failure, remote intervention is impossible. However, when a significant failure trend emerges, the lack of in-depth root cause analysis often necessitates downtime and maintenance, resulting in reduced production efficiency.

[0004] In recent years, the rapid development of 5G communication technology and the Internet of Things (IoT) has provided new technical support for remote monitoring of spinning equipment. 5G technology, with its high bandwidth, low latency, and wide connectivity, enables real-time transmission of massive amounts of device data. The IoT, through sensors, embedded systems, and other technologies, establishes intelligent connections between devices and networks, laying the foundation for data collection and remote control. Remote monitoring platforms based on 5G and the IoT are becoming a hot topic in industry research, promising full lifecycle management of spinning foreign fiber removal equipment, including real-time status monitoring, fault warnings, and remote control.

[0005] However, existing 5G- and IoT-based remote monitoring technologies for spinning foreign fiber removal equipment still suffer from insufficiently refined control strategies. This is primarily reflected in the significant room for improvement in the control decisions between edge servers and remote monitoring platforms. Therefore, leveraging 5G and IoT technologies to develop a remote monitoring method capable of dynamically determining control strategies is a key requirement for improving equipment operational stability, reducing failure rates, and ensuring spinning quality. This is also the technical problem addressed by this invention. Summary of the Invention

[0006] In this regard, the present invention provides a remote monitoring method, remote monitoring platform, electronic device, computer storage medium and computer program product for spinning foreign fiber removal equipment based on 5G and the Internet of Things to solve at least one of the above-mentioned technical problems.

[0007] According to a first aspect of the present invention, a remote monitoring method for a spinning foreign fiber removal device based on 5G and the Internet of Things is provided, comprising the following method steps: a remote monitoring platform receives operation-related data transmitted by the spinning foreign fiber removal device on the field side, and uses a joint evaluation model to evaluate the operation-related data to obtain the fault trend type of the spinning foreign fiber removal device, including a mild fault trend and an obvious fault trend; wherein, the joint evaluation model is constructed based on an adaptive threshold algorithm and an LSTM algorithm; if the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to the edge server, and the edge server controls the spinning foreign fiber removal device based on the first control command; if the fault trend type is an obvious fault trend, the remote monitoring platform sends a second control command containing control parameters to the edge server, and the edge server controls the spinning foreign fiber removal device based on the second control command; wherein, the control parameters are generated at least based on the operation-related data.

[0008] According to a second aspect of the present invention, a remote monitoring platform for spinning foreign fiber removal equipment based on 5G and the Internet of Things is provided, comprising a receiving unit, a fault trend evaluation unit, and a fault handling decision unit; the receiving unit receives operation-related data transmitted by the spinning foreign fiber removal equipment on the field side; the fault trend evaluation unit uses a joint evaluation model to evaluate the operation-related data to obtain the fault trend type of the spinning foreign fiber removal equipment, including a mild fault trend and an obvious fault trend; wherein, the joint evaluation model is constructed based on an adaptive threshold algorithm and an LSTM algorithm; the fault handling decision unit, if the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to the edge server, and the edge server controls the spinning foreign fiber removal equipment based on the first control command; if the fault trend type is an obvious fault trend, the remote monitoring platform sends a second control command containing control parameters to the edge server, and the edge server controls the spinning foreign fiber removal equipment based on the second control command; wherein, the control parameters are generated at least based on the operation-related data.

[0009] In a third aspect of the present invention, an electronic device is provided, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program implements any of the methods described above when executed by the processor.

[0010] According to a fourth aspect of the present invention, a computer storage medium is provided, wherein the computer storage medium stores a computer program executable by a processor to implement any of the methods described above.

[0011] According to a fifth aspect of the present invention, a computer program product is provided, which comprises a computer program executable by a processor to implement any of the methods described above.

[0012] This system uses 5G and the Internet of Things (IoT) to transmit real-time data, combining adaptive thresholds with an LSTM algorithm to assess fault trends. Minor faults are quickly regulated by edge servers, while significant faults are handled by edge servers after parameters are generated by a remote platform. This system enables precise fault prediction and hierarchical regulation, reducing the risk of fault deterioration, improving equipment stability and spinning quality, and minimizing downtime losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 is a flowchart of a remote monitoring method of a spinning foreign fiber removing device based on 5G and Internet of Things disclosed by the embodiments of the present application.

[0015] Figure 2 is a schematic diagram of a remote monitoring system disclosed by the embodiments of the present application.

[0016] Figure 3 is a structural schematic diagram of a gated neural network disclosed by the embodiments of the present application.

[0017] Figure 4 is a structural schematic diagram of a remote monitoring platform of a spinning foreign fiber removing device based on 5G and Internet of Things disclosed by the embodiments of the present application. DETAILED DESCRIPTION

[0018] The implementation manners of the present application are described below by specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the present specification. Obviously, the described embodiments are part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0019] In addition, the technical features involved in different embodiments of the present application described below can be combined with each other as long as there is no conflict.

[0020] As shown in Figure 1 , Figure 2 The embodiments of the present application disclose a remote monitoring method of a spinning foreign fiber removing device based on 5G and Internet of Things, which comprises the following method steps: S10, a remote monitoring platform receives running related data transmitted by a spinning foreign fiber removing device on a field side, uses a joint evaluation model to evaluate the running related data, and obtains a fault trend type of the spinning foreign fiber removing device, including a mild fault trend and an obvious fault trend; wherein the joint evaluation model is constructed based on a self-adaptive threshold algorithm and an LSTM algorithm.

[0021] Various sensors (such as temperature sensors, speed sensors, and surveillance cameras) are deployed on spinning foreign fiber removal equipment to collect key indicators of the equipment's operating status. For example, equipment operating status parameters directly reflect the equipment's mechanical and electrical operation and serve as the basis for determining whether the equipment is functioning properly. These parameters include speed (such as the speed of the rollers used to transport raw cotton and the speed of the actuators in the foreign fiber removal device), temperature (such as the temperature of the motor and the operating temperature of the equipment's core components to prevent performance degradation due to overheating), vibration frequency (collected through vibration sensors to monitor for mechanical looseness, component wear, and other issues), and current and voltage (reflecting the load status of the equipment's electrical system; abnormal fluctuations may indicate circuit failure).

[0022] Foreign fiber removal performance parameters are key indicators for evaluating the core functionality of the equipment and are directly related to spinning quality. These parameters primarily include foreign fiber recognition accuracy (the ratio of correctly identified foreign fibers to actual foreign fibers present; a decrease in accuracy may indicate a failure in the recognition algorithm or a malfunction in the optical detection component); foreign fiber removal efficiency (the ratio of successfully removed foreign fibers to identified foreign fibers, reflecting the effectiveness of the removal mechanism); and false positive rate (the proportion of normal fibers identified as foreign fibers; a high false positive rate can lead to raw material waste).

[0023] Environmental parameters: Equipment operation is significantly affected by the environment. This type of data is used to assist in analyzing the causes of fault trends. These parameters include ambient temperature (temperature fluctuations within the workshop can affect equipment heat dissipation and the physical properties of raw cotton), humidity (high humidity can cause raw cotton to stick together, hindering conveying and foreign fiber identification; low humidity can generate static electricity, interfering with equipment operation), and dust concentration (spinning workshops are often dusty, and excessive dust adhesion to equipment sensors or components can affect their performance).

[0024] Communication and interaction parameters: Due to reliance on 5G and IoT technologies, this type of data is used to ensure the stability of data transmission and command execution. Examples include data transmission latency (which reflects the communication quality of the 5G network; excessive latency can cause monitoring lags) and command response time (the time it takes for a device or edge server to receive and execute a control command, reflecting the real-time nature of the system).

[0025] The spinning foreign fiber removal equipment is also equipped with an Internet of Things communication module. Based on the connection to the 5G network, the operation-related data collected by the above sensors can be aggregated and transmitted to the remote monitoring platform.

[0026] like Figure 3As shown, the remote monitoring platform is pre-configured with a joint evaluation model that integrates an adaptive threshold algorithm and an LSTM network to enable a multi-dimensional assessment of the operational data related to spinning foreign fiber removal equipment. Specifically, the adaptive threshold algorithm dynamically determines the critical points between mild and significant fault trends based on the equipment's historical operating data and fault records throughout its lifecycle. For example, by statistically analyzing the correlation between temperature parameters and faults in historical data, and combining variables such as the equipment's current operating time, ambient temperature and humidity, the temperature threshold is adjusted in real time to ensure that the critical point is adapted to the equipment's actual operating status. The LSTM algorithm leverages its ability to process time series data to perform trend modeling on real-time operating data, extracting features such as parameter change rate and fluctuation frequency, and outputting a quantitative score between 0 and 1 to indicate the severity of the fault trend.

[0027] The joint evaluation model combines the quantitative scores output by the LSTM with the critical points defined by the adaptive threshold algorithm to complete the classification. For example, when the score exceeds the minor fault threshold, it is judged as a minor fault trend; when the score reaches the obvious fault threshold, it is judged as a obvious fault trend, achieving accurate prediction of equipment failure status.

[0028] S20: If the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to the edge server, and the edge server controls the spinning foreign fiber removing device based on the first control command.

[0029] If the remote monitoring platform determines that the spinning foreign fiber removal equipment is showing signs of a minor malfunction, such as a temperature slightly above the normal range but rising slowly or a slight decrease in foreign fiber identification accuracy, it generates a first control command. This first control command, which contains only the control command and no specific control parameters, is sent to the edge server via the low-latency 5G network.

[0030] The edge server, deployed at the edge of the local network close to the equipment, provides low-latency data processing and device control capabilities. Upon receiving the first control command, it communicates directly with the spinning contamination removal equipment to execute the control operation. For example, in the event of a slight speed fluctuation, the edge server can automatically generate control parameters based on the operating data transmitted by the spinning contamination removal equipment. It then adjusts the device driver module parameters in real time to restore the speed to a stable state. This localized control mode reduces remote transmission latency, ensuring that minor faults are quickly addressed in the early stages and ensuring continuous equipment operation.

[0031] Of course, if the edge server has already discovered a mild fault trend and even performed regulatory measures when receiving the first regulation command, it may not respond to the first regulation command of the remote monitoring platform involving the same mild fault trend.

[0032] S30, if the failure trend type is an obvious failure trend, the remote monitoring platform sends a second regulation command containing regulation parameters to the edge server, and the edge server regulates the spinning foreign fiber removing device based on the second regulation command; wherein the regulation parameters are generated based on at least the operation-related data.

[0033] When the remote monitoring platform determines that the spinning foreign fiber removing device has an obvious failure trend, regulation parameters are generated based on the operation-related data of the device, and the edge server directly executes precise regulation of the spinning foreign fiber removing device based on these regulation parameters. For example, when the temperature of the spinning foreign fiber removing device rapidly rises and approaches a critical value (obvious failure trend), combined with real-time temperature data, historical heat dissipation efficiency curves, etc., the specific parameters of the cooling system power, operation interval, etc. that need to be adjusted are calculated to ensure the pertinence of the regulation.

[0034] The remote monitoring platform sends a second regulation command containing regulation parameters to the edge server. Compared with the first regulation command, the second command information is more detailed, and the specific values and operation logic of the regulation are specified. After receiving the command, the edge server regulates the device in depth according to the parameters in the command, such as adjusting the running state of the heat dissipation device according to the calculated cooling system power parameter, or optimizing the identification model threshold based on the foreign fiber recognition accuracy reduction data. Through the local execution of the edge server, the accuracy and timeliness of the regulation are taken into account, avoiding the expansion of the failure due to the delay of remote processing, and ensuring the controllable operation of the device under obvious failure trend.

[0035] The present application transmits data in real time through 5G and the Internet of Things, evaluates the failure trend by combining adaptive threshold and LSTM algorithm, and regulates the light failure quickly by the edge server, and executes the obvious failure by the edge after generating parameters by the remote platform. The present application can realize precise prediction and hierarchical regulation of failure, reduce the risk of failure deterioration, improve the stability of the device and the spinning quality, and reduce the loss of downtime.

[0036] As an example, the use of a joint evaluation model to evaluate the operation-related data to obtain the failure trend type of the spinning foreign fiber removing device includes: constructing a parameter-failure probability distribution model based on historical operation data of the device to calculate the baseline threshold of each operation parameter; collecting real-time operation-related data through a sliding window algorithm to generate a threshold correction coefficient; multiplying the baseline threshold and the threshold correction coefficient to dynamically generate a light failure threshold and an obvious failure threshold under the current working condition; inputting the continuously collected real-time operation data into an LSTM network to extract time sequence features, weighting the time sequence feature vectors through an attention mechanism, and mapping the weighted feature vectors to normalized failure trend quantitative scores through a fully connected layer; comparing the failure trend quantitative scores with the light failure threshold and the obvious failure threshold to determine the failure trend type of the spinning foreign fiber removing device.

[0037] First, based on the historical operating data of the spinning foreign fiber removal equipment (such as temperature, speed, and fault records over the past year), a statistical analysis (such as maximum likelihood estimation) was used to establish a mapping between each parameter and the probability of failure. For example, the failure probability increases sharply when the temperature exceeds 80°C, thereby determining a baseline safety range for the temperature parameter. Simultaneously, a sliding window algorithm was used to capture the spinning foreign fiber removal equipment's real-time operating data (such as ambient temperature and humidity, equipment operating time, etc.) at fixed intervals (e.g., 5 minutes). This data was then normalized (e.g., using min-max standardization) to convert the data into a correction factor ranging from 0 to 1.

[0038] For example, if the ambient temperature threshold is 15°C (lower limit) - 35°C (upper limit), exceeding this range may affect device operation. When the current ambient temperature is 30°C, using min-max normalization, the correction factor is (30-15) / (35-15) = 15 / 20 = 0.75. If the temperature is 15°C, the factor is 0; if the temperature is 35°C, the factor is 1. If real-time data exceeds the threshold range (for example, the temperature is 38°C), the correction factor is forcibly set to 1.2 (for exceeding the upper limit) or -0.2 (for falling below the lower limit) to enhance response to extreme conditions.

[0039] The baseline threshold is then multiplied by the correction factor to obtain the minor / major fault threshold adapted to the current operating conditions. For example, if the baseline temperature threshold is 70°C, the current ambient temperature is high, and the correction factor is 1.1, the dynamic minor fault threshold is 77°C, ensuring that the fault threshold adjusts in real time as the operating conditions change.

[0040] Then, continuously collected real-time data (such as rotation speed and recognition accuracy every 10 seconds) is sequenced in chronological order and fed into the LSTM network. The LSTM network uses a gating mechanism to capture long-term dependencies and extract features such as parameter change rate (such as a temperature increase of 5°C per hour) and fluctuation period (such as a rotation speed fluctuation every 2 minutes), thereby reflecting the development trend of the fault.

[0041] Based on a fault impact weight library (pre-labeled with high-weight parameters such as foreign fiber identification accuracy and motor temperature), sensitive parameters in the feature vector are given higher weights. For example, a sudden drop in foreign fiber identification accuracy is weighted 0.8, and a fluctuating ambient humidity is weighted 0.2 to emphasize the impact of critical information. The fully connected layer maps the weighted feature vectors to a score between 0 and 1, with higher scores indicating a more pronounced fault trend. For example, a score of 0.3 indicates a moderate trend, while a score of 0.8 indicates a near-obvious fault.

[0042] Next, the quantitative score is compared with the dynamic threshold. If the score is less than 0.5 and the parameter does not exceed the mild threshold, it is judged as normal; if 0.5≤score<0.8 and the parameter exceeds the mild threshold, it is judged as a mild fault trend; if the score is ≥0.8 or the parameter reaches the obvious threshold, it is judged as a obvious fault trend.

[0043] As an example, the vector weighting of the time series feature vector through the attention mechanism includes: retrieving the universal weight matrix of the spinning foreign fiber removal equipment from the fault impact weight library, and the universal weight matrix is ​​generated based on the fault statistics of similar equipment; collecting the historical fault records and maintenance logs of the spinning foreign fiber removal equipment, extracting high-frequency fault parameters, and calculating the specific correction coefficient of the high-frequency fault parameters; multiplying the universal weight matrix with the specific correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal equipment; performing a dot product operation on the time series feature vector extracted by the LSTM network and the dynamic weight matrix to obtain a weighted feature vector.

[0044] Although the universal weight matrix can cover the common needs of most devices, it cannot adapt to the specific failure modes of a single device due to its age, maintenance history, etc. For example, a certain device has frequent abnormal roller speeds due to long-term high-load operation, while high-frequency failures of similar devices may be concentrated in temperature anomalies. If universal weights are directly used, the influence of the device's unique high-frequency failure parameters will be weakened, resulting in a deviation in the attention mechanism's capture of key features. Therefore, the present invention adopts a method of dynamically adjusting the attention mechanism, integrating universal rules with device specificity, and improving the accuracy of feature weighting.

[0045] Specifically, a pre-set universal weight matrix is ​​first retrieved from the fault impact weight library. This matrix contains baseline weights for parameters such as foreign fiber recognition accuracy, motor temperature, and speed. These weights are determined based on statistical data from a large number of similar (e.g., same-model) spinning foreign fiber removal machines. For example, the baseline weights for foreign fiber recognition accuracy are set at 0.3, and for motor temperature at 0.25, to reflect their general impact on equipment failures.

[0046] The current spinning foreign fiber removal equipment's historical fault records (e.g., fault type and frequency over the past year) and maintenance logs are retrieved, and statistical analysis is performed to extract high-frequency fault parameters. For example, if the equipment experienced 12 roller speed fluctuation faults in the past year, significantly more than other fault types, roller speed is marked as a specific parameter. A parameter-specific correction factor is calculated for this parameter based on fault frequency and repair cost (e.g., for every 5 increases in frequency, the factor increases by 0.1). This resulting parameter-specific correction factor is used to strengthen the parameter's weighting in the weight matrix.

[0047] The universal weight matrix is ​​element-wise multiplied by the parameter-specific correction coefficient to produce a dynamic weight matrix tailored to the current device. For example, if the universal weight for roller speed is 0.2 and its parameter-specific correction coefficient is 1.5, the dynamic weight is adjusted to 0.3, significantly higher than the universal value. The dynamic weight matrix preserves the influence of common device characteristics while highlighting the importance of device-specific, high-frequency fault parameters, making the attention mechanism's weighting logic more aligned with the device's actual operating status.

[0048] The time series feature vector extracted by the LSTM network (including features such as the rate of change and fluctuation period of each parameter) is dot-producted with the dynamic weight matrix to produce a weighted feature vector. For example, the roller speed fluctuation period feature has a value of 0.6 in the time series vector. After multiplying it by the dynamic weight of 0.3, the weight of this feature is 0.18, which is higher than the original weight of 0.12. This operation increases the proportion of high-frequency fault features unique to the device in the vector, ensuring that the attention mechanism prioritizes fault signals that are more critical to the device.

[0049] It is understood that if the spinning foreign fiber removal equipment has no similar failure records in the past 30 days, the impact of the specific correction factor will be reduced by a certain percentage (for example, 5%) each day (for example, from 1.5 to 1.425, 1.35, etc.) until it returns to the general weight level, avoiding overweighting the improved fault parameters. If the same fault reappears during this period, the correction factor is recalculated (for example, if the fault frequency increases, the factor is adjusted to 1.6), and the dynamic weight matrix is ​​updated based on the new factor to ensure that the attention mechanism is always synchronized with the current fault characteristics of the equipment.

[0050] As an example, the general weight matrix is ​​multiplied by the specific correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device, including: separating non-fault maintenance data from the maintenance log, counting the occurrence frequency of each high-frequency fault parameter in the non-fault maintenance data, and generating an interference correction coefficient based on the occurrence frequency; multiplying the general weight matrix with the specific correction coefficient and the interference correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device.

[0051] Identify and extract non-fault maintenance data from maintenance logs, such as records of scheduled cleaning, calibration, and preventive maintenance. These operations involve parameter adjustments but are not directly related to equipment failures.

[0052] Maintenance logs contain non-fault-related maintenance data (such as scheduled cleaning and routine calibration), which can cause some parameters to appear frequently. However, these adjustments are not directly related to equipment failures. For example, the roller speed of a certain device may be recorded three times due to weekly calibration, far exceeding its actual failure frequency (once per quarter). Directly incorporating this into the specificity correction coefficient calculation would exaggerate the failure relevance of this parameter. Therefore, this embodiment further introduces an interference correction coefficient to eliminate the interference of non-fault maintenance on parameter weights, ensuring that the dynamic weight matrix only reflects actual failure characteristics and avoiding weight bias caused by maintenance activities.

[0053] Specifically, maintenance logs are filtered to identify non-fault-related maintenance records, including those for regular cleaning (e.g., monthly sensor cleaning), routine calibration (e.g., weekly speed sensor calibration), and preventive component replacement (e.g., semi-annual bearing replacement). Keyword matching (e.g., "calibration," "cleaning," and "preventive") is used to automatically identify and separate this data. For example, from 100 maintenance records, 30 non-fault records can be extracted, while the remaining 70 are fault records. This separated data is then used to analyze the extent to which parameters are affected by maintenance, avoiding confusion with fault data.

[0054] The frequency of occurrence of each high-frequency fault parameter in non-fault maintenance data is calculated. This is the ratio of the number of times a high-frequency fault parameter was adjusted during non-fault maintenance to the total number of non-fault maintenance events. For example, roller speed was adjusted 15 times out of 30 non-fault records, representing a 50% frequency; motor temperature was adjusted 6 times, representing a 20% frequency.

[0055] The interference correction coefficient is set according to the frequency. For example, if the frequency is ≥40% (such as roller speed 50%), it is judged as strong maintenance interference and the coefficient is set to 0.8 (weakening weight); if the frequency is <40% (such as motor temperature 20%), it is judged as weak interference and the coefficient is set to 1.0 (no adjustment).

[0056] Finally, the universal weight matrix, the specific correction factor, and the interference correction factor are element-by-element multiplied to generate the dynamic weight matrix. For example, if the universal weight is 0.2 for roller speed, the specific correction factor is 1.5 (due to high-frequency failures), and the interference correction factor is 0.8 (due to high maintenance frequency), then the dynamic weight is 0.2 × 1.5 × 0.8 = 0.24. If the universal weight is 0.25 for motor temperature, the specific correction factor is 1.2, and the interference correction factor is 1.0, then the dynamic weight is 0.25 × 1.2 × 1.0 = 0.3.

[0057] The final dynamic weight matrix not only retains the influence of equipment-specific fault parameters (such as the fault correlation of roller speed), but also eliminates the inflated weights caused by maintenance behavior, allowing the attention mechanism to focus more accurately on the real fault characteristics.

[0058] As an example, the control parameters are generated based at least on the operation-related data, including: collecting the operation data of the upstream equipment of the spinning foreign fiber removal equipment within a preset time window, the upstream equipment operation data at least including the raw material conveying speed, opening degree, and impurity removal rate; performing spatiotemporal correlation analysis on the operation-related data and the upstream equipment operation data, and screening out the upstream equipment operation data that are strongly correlated; constructing a multivariate regression model based on the strongly correlated upstream equipment operation data, taking the abnormal parameters in the operation-related data as the dependent variable, and the upstream equipment operation data and the equipment's own operation parameters as the independent variables, and outputting the contribution weight of each parameter to the abnormality; sorting according to the contribution weight, selecting several parameters with the highest weight as key influencing factors, and combining the optimal control threshold in the equipment's historical maintenance record to generate control parameters including parameter adjustment amplitude and execution timing.

[0059] Traditional control methods generate parameters solely based on the equipment's own operating data, ignoring the interconnectedness of upstream and downstream equipment in the spinning process. For example, insufficient opening in the upstream cotton opener can cause foreign fibers to become entrapped in the cotton clump. Even if the foreign fiber removal equipment itself is functioning properly, this can create the illusion of decreased recognition accuracy. Adjusting only the equipment's own parameters (such as increasing recognition sensitivity) has limited effectiveness and can also increase the rate of false positives. This embodiment, by incorporating upstream equipment operating data, addresses issues at the source of the raw materials, achieving collaborative optimization throughout the entire process.

[0060] Specifically, operating data from upstream equipment (such as cotton openers and cleaners) is collected within a preset time window (e.g., the first 30 minutes). This data includes raw material delivery speed (the amount of raw cotton delivered per unit time), opening degree (the degree to which the raw cotton is loosened), and impurity removal rate (the proportion of impurities removed by upstream equipment to the total impurities in the raw material). This operating data reflects the state of the raw material entering the foreign fiber removal equipment. For example, excessively fast raw material delivery speeds may result in inadequate foreign fiber identification, while insufficient opening degree may make foreign fibers encased in cotton clumps difficult to detect.

[0061] Through spatiotemporal correlation analysis, operational data related to the spinning contamination removal equipment (such as contamination identification accuracy and processing efficiency) is matched with operational data from upstream equipment. Temporally, the acquisition timestamps of the two are aligned to ensure that the analysis focuses on the processing of the same batch of raw materials. Spatially, the focus is on the raw material transmission path from the upstream equipment to the contamination removal equipment. Statistical methods such as the Pearson correlation coefficient are used to calculate correlation. If the absolute value of the correlation coefficient is ≥0.6, it is determined to be a strong correlation (for example, a strong positive correlation between raw material delivery speed and contamination removal equipment processing load). This upstream data is then incorporated into the basis for generating control parameters, eliminating interfering data with weak correlations.

[0062] A multivariate regression model was constructed using abnormal parameters in operational data (e.g., a 10% drop in foreign fiber identification accuracy) as the dependent variable, and screened out strongly correlated upstream equipment operational data (e.g., fluctuations in raw material delivery speed) and the equipment's own operating parameters (e.g., identification module sensitivity) as independent variables. This model fitted the data using the least squares method and outputted the contribution weight of each variable to the abnormal parameter. For example, the contribution weight of raw material delivery speed was 0.4, and the contribution weight of opening degree was 0.3, reflecting the degree of influence of each parameter on the current abnormality.

[0063] Sort by contribution weight from high to low, and select the top 3-5 parameters (e.g., raw material conveying speed, opening degree, and equipment identification module sensitivity) as key influencing factors. Query the equipment's historical maintenance records to obtain the optimal control thresholds for these parameters under similar abnormal conditions (e.g., the optimal range for raw material conveying speed is 8-10m / min). Combined with the extent to which the current parameters deviate from the threshold, calculate the specific adjustment value (e.g., from 12m / min to 9m / min). At the same time, based on the linkage between parameters (e.g., adjusting the conveying speed first and then optimizing the identification sensitivity), determine the execution sequence, and ultimately generate control parameters that include the parameter adjustment range, execution order, and effective time to ensure the targeted and effective control.

[0064] As an example, the control parameters are generated at least based on the operation-related data, and also include: if the upstream equipment operation data with strong correlation is not screened out, the control parameters are generated only based on the abnormal parameters and historical control cases in the operation-related data.

[0065] When temporal and spatial correlation analysis was performed between the operating data of the spinning foreign fiber removal equipment and the operating data of upstream equipment, if the absolute value of the correlation coefficient of all upstream equipment operating data was less than 0.6 (indicating no strong correlation), it was determined that the abnormal parameters of the current equipment (such as abnormal temperature increase and decreased foreign fiber detection efficiency) were unrelated to the upstream equipment, eliminating the possibility of upstream factors interfering with the generation of control parameters. For example, if the temperature of the equipment motor suddenly rose, but the parameters of the upstream cotton opener, such as the raw material delivery speed and opening degree, remained within the stable range, and the correlation coefficient between the two was 0.2, it indicated that the abnormality was caused by factors within the equipment itself.

[0066] At this point, abnormal parameters are located and extracted from the spinning foreign fiber removal equipment's operational data to determine the type of abnormality and the degree of deviation. For example, if the current motor temperature is 85°C, exceeding the normal threshold (50-70°C) by 15°C, the foreign fiber identification accuracy rate drops to 82% (normal threshold ≥ 95%). These parameters are marked as core abnormality indicators and serve as a direct basis for generating control parameters.

[0067] At the same time, the system accesses a library of historical device control cases to identify cases with similar parameter types and deviations to the current abnormality. The case library contains fault types, abnormal parameter values, control measures implemented, and feedback on their effectiveness over the past three years. For example, a historical case with a motor temperature of 83°C and an identification accuracy of 80% was identified. The corresponding control measures were "reducing the motor load by 10% and cleaning the heat sink." After implementation, the temperature returned to 68°C, and the accuracy rate rose to 96%.

[0068] By combining the current abnormal parameters with the control measures of the matching cases, adaptive control parameters are generated. For example, if there is a completely matching abnormal scenario in the historical case (such as the same deviation in temperature and accuracy), the parameter adjustment range (such as reducing the load by 10%) and execution sequence (such as cleaning the heat sink first and then adjusting the load) in the case are directly reused. If there are only partial matching cases, the control experience of multiple cases is integrated through weighted calculation. For example, in Case A, the load is reduced by 8% when the temperature exceeds 10°C, and in Case B, the load is reduced by 15% when the temperature exceeds 20°C. If the current temperature exceeds 15°C, the calculated load reduction is 11.5%. The generated control parameters must include specific operation values ​​(such as increasing the cooling fan speed to 2000r / min), execution priority (such as completing heat sink cleaning within 30 seconds), and effect verification indicators (such as the temperature must drop below 75°C after 5 minutes) to ensure that the control is executable and traceable.

[0069] like Figure 4 As shown, an embodiment of the present invention further provides a remote monitoring platform 100 for a spinning foreign fiber removal device based on 5G and the Internet of Things, comprising a receiving unit 1010, a fault trend evaluation unit 1020, and a fault handling decision unit 1030; the receiving unit 1010 receives operation-related data transmitted by the spinning foreign fiber removal device on the field side; the fault trend evaluation unit 1020 uses a joint evaluation model to evaluate the operation-related data to obtain a fault trend type of the spinning foreign fiber removal device, including a mild fault trend and a significant fault trend; wherein the joint evaluation model is constructed based on an adaptive threshold algorithm and an LSTM algorithm; the fault handling decision unit 1030, if the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to an edge server, and the edge server controls the spinning foreign fiber removal device based on the first control command; if the fault trend type is a significant fault trend, the remote monitoring platform sends a second control command containing control parameters to the edge server, and the edge server controls the spinning foreign fiber removal device based on the second control command; wherein the control parameters are generated at least based on the operation-related data.

[0070] As an example, the fault trend evaluation unit 1020 is used to: construct a parameter-fault probability distribution model based on the historical operation data of the equipment, and calculate the baseline threshold range of each operating parameter; collect real-time operation related data through a sliding window algorithm to generate a threshold correction coefficient; multiply the baseline threshold by the threshold correction coefficient to dynamically generate a mild fault threshold and an obvious fault threshold under the current working conditions; input the continuously collected real-time operation data into the LSTM network to extract time series features, vector weight the time series feature vector through the attention mechanism, and map the weighted feature vector into a normalized fault trend quantization score through a fully connected layer; compare the fault trend quantization score with the mild fault threshold and the obvious fault threshold to determine the fault trend type of the spinning foreign fiber removal equipment.

[0071] As an example, the fault trend assessment unit 1020 is used to: retrieve a universal weight matrix for the spinning foreign fiber removal device from the fault impact weight library, where the universal weight matrix is ​​generated based on fault statistical data of similar equipment; collect historical fault records and maintenance logs of the spinning foreign fiber removal device, extract high-frequency fault parameters, and calculate specific correction coefficients of the high-frequency fault parameters; multiply the universal weight matrix by the specific correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device; perform a dot product operation on the time series feature vector extracted by the LSTM network and the dynamic weight matrix to obtain a weighted feature vector.

[0072] As an example, the fault trend assessment unit 1020 is used to: separate non-fault maintenance data from the maintenance log, count the occurrence frequency of each high-frequency fault parameter in the non-fault maintenance data, and generate an interference correction coefficient based on the occurrence frequency; multiply the general weight matrix with the specific correction coefficient and the interference correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device.

[0073] As an example, the fault handling decision unit 1030 is used to: collect the operating data of the upstream equipment of the spinning foreign fiber removal equipment within a preset time window, and the upstream equipment operating data at least includes the raw material conveying speed, opening degree, and impurity removal rate; perform spatiotemporal correlation analysis on the operation-related data and the upstream equipment operating data, and screen out the upstream equipment operating data that are strongly correlated; construct a multivariate regression model based on the strongly correlated upstream equipment operating data, with the abnormal parameters in the operation-related data as the dependent variable, and the upstream equipment operating data and the equipment's own operating parameters as the independent variables, and output the contribution weight of each parameter to the abnormality; sort according to the contribution weight, select several parameters with the highest weight as key influencing factors, and combine the optimal control threshold in the equipment's historical maintenance record to generate control parameters including parameter adjustment amplitude and execution timing.

[0074] As an example, the fault handling decision unit 1030 is configured to: if no strongly correlated upstream device operation data is screened out, generate control parameters based only on abnormal parameters in the operation-related data and historical control cases.

[0075] An embodiment of the present invention further provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program implements any of the aforementioned methods when executed by the processor.

[0076] An embodiment of the present invention further provides a computer storage medium storing a computer program that can be executed by a processor to implement any of the methods described above.

[0077] An embodiment of the present invention further provides a computer program product, which includes a computer program that can be executed by a processor to implement any of the methods described above.

[0078] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0079] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A remote monitoring method for spinning foreign fiber removal equipment based on 5G and the Internet of Things, characterized by: The method comprises the following steps: The remote monitoring platform receives operational data transmitted by on-site spinning foreign fiber removal equipment and evaluates the data using a joint evaluation model to determine the fault trend type of the spinning foreign fiber removal equipment, including mild fault trends and significant fault trends. The severity of the fault trend is represented by a quantitative score from 0 to 1. The joint evaluation model is constructed based on an adaptive threshold algorithm and an LSTM algorithm. If the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to the edge server, and the edge server controls the spinning foreign fiber removal device based on the first control command; wherein the first control command only includes the control command but does not include specific control parameters; If the fault trend type is an obvious fault trend, the remote monitoring platform sends a second control command containing control parameters to the edge server, and the edge server controls the spinning foreign fiber removal device based on the second control command; wherein the control parameters are generated based on at least the operation-related data; The operation-related data are evaluated using a joint evaluation model to obtain the failure trend types of the spinning foreign matter removal equipment, including: A parameter-fault probability distribution model is constructed based on historical equipment operating data to calculate the baseline threshold range for each operating parameter. A sliding window algorithm is used to collect real-time operating data and generate threshold correction coefficients. Multiply the baseline threshold by the threshold correction coefficient to dynamically generate the mild fault threshold and obvious fault threshold under the current working condition; The continuously collected real-time operation data is input into the LSTM network to extract the time series feature vector. The time series feature vector is vector-weighted through the attention mechanism, and the weighted feature vector is mapped into a normalized fault trend quantification score through the fully connected layer. Comparing the fault tendency quantified score with a slight fault threshold and a significant fault threshold to determine the fault tendency type of the spinning foreign matter removal device; The vector weighting of the time series feature vector by the attention mechanism includes: Retrieving a universal weight matrix of a spinning foreign fiber removal device from a fault impact weight library, wherein the universal weight matrix is ​​generated based on fault statistical data of similar devices; Collect historical fault records and maintenance logs of the spinning foreign fiber removal equipment, extract high-frequency fault parameters, and calculate specific correction coefficients for the high-frequency fault parameters; Multiplying the universal weight matrix by the specific correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device; Perform dot product operation on the time series feature vector extracted by the LSTM network and the dynamic weight matrix to obtain the weighted feature vector; The method of multiplying the universal weight matrix by the specific correction coefficient to generate a dynamic weight matrix adapted to the spinning foreign fiber removal device includes: Separating non-fault maintenance data from the maintenance log, counting the occurrence frequency of each high-frequency fault parameter in the non-fault maintenance data, and generating an interference correction coefficient based on the occurrence frequency; Multiplying the universal weight matrix with the specificity correction coefficient and the interference correction coefficient to generate a dynamic weight matrix suitable for the spinning foreign fiber removal device; The control parameters are generated based at least on the operation-related data, including: Collecting operation data of an upstream device of the spinning foreign fiber removal device within a preset time window, wherein the upstream device operation data at least includes a raw material conveying speed, an opening degree, and an impurity removal rate; Performing spatiotemporal correlation analysis on the operation-related data and the upstream equipment operation data, and screening out upstream equipment operation data with strong correlation; A multivariate regression model is constructed based on the strongly correlated upstream equipment operation data, with the abnormal parameters in the operation-related data as the dependent variable, the upstream equipment operation data and the equipment's own operating parameters as the independent variables, and the contribution weight of each parameter to the abnormality is output; According to the contribution weight sorting, several parameters with higher weights are selected as key influencing factors. Combined with the optimal control threshold in the historical maintenance record of the equipment, the control parameters including parameter adjustment range and execution timing are generated.

2. The remote monitoring method for spinning foreign fiber removal equipment based on 5G and the Internet of Things according to claim 1 is characterized in that: The control parameters are generated at least based on the operation-related data, and also include: if the upstream equipment operation data with strong correlation is not screened out, the control parameters are generated only based on abnormal parameters and historical control cases in the operation-related data.

3. A remote monitoring platform for spinning foreign fiber removal equipment based on 5G and the Internet of Things, characterized by: It includes receiving unit, fault trend assessment unit and fault handling decision unit; The receiving unit receives the operation-related data transmitted by the spinning foreign fiber removal device on the field side; The fault trend evaluation unit evaluates the operation-related data using a joint evaluation model to determine the fault trend type of the spinning foreign fiber removal device, including a mild fault trend and a significant fault trend, where the severity of the fault trend is represented by a quantitative score ranging from 0 to 1; wherein the joint evaluation model is constructed based on an adaptive threshold algorithm and an LSTM algorithm; The fault handling decision unit, if the fault trend type is a mild fault trend, the remote monitoring platform sends a first control command to the edge server, and the edge server controls the spinning foreign fiber removal device based on the first control command; wherein the first control command only includes the control command and does not include specific control parameters; if the fault trend type is a significant fault trend, the remote monitoring platform sends a second control command including the control parameters to the edge server, and the edge server controls the spinning foreign fiber removal device based on the second control command; wherein the control parameters are generated at least based on the operation-related data; The fault trend assessment unit is configured to: construct a parameter-fault probability distribution model based on historical equipment operation data and calculate a baseline threshold interval for each operating parameter; collect real-time operation-related data using a sliding window algorithm to generate a threshold correction coefficient; multiply the baseline threshold by the threshold correction coefficient to dynamically generate a mild fault threshold and a significant fault threshold under the current operating condition; input the continuously collected real-time operation data into an LSTM network to extract time series features, perform vector weighting on the time series feature vectors using an attention mechanism, and map the weighted feature vectors into a normalized fault trend quantization score using a fully connected layer; and compare the fault trend quantization score with the mild fault threshold and the significant fault threshold to determine the fault trend type of the spinning foreign fiber removal device. The fault trend assessment unit is configured to: retrieve a universal weight matrix for a spinning foreign fiber removal device from a fault impact weight library, the universal weight matrix being generated based on fault statistics of similar devices; collect historical fault records and maintenance logs of the spinning foreign fiber removal device, extract high-frequency fault parameters, and calculate specific correction coefficients for the high-frequency fault parameters; multiply the universal weight matrix by the specific correction coefficients to generate a dynamic weight matrix adapted for the spinning foreign fiber removal device; and perform a dot product operation on a time series feature vector extracted by the LSTM network and the dynamic weight matrix to obtain a weighted feature vector. The fault trend assessment unit is configured to: separate non-fault maintenance data from the maintenance log, count the occurrence frequencies of various high-frequency fault parameters in the non-fault maintenance data, and generate interference correction coefficients based on the occurrence frequencies; and multiply a general weight matrix by the specific correction coefficient and the interference correction coefficient to generate a dynamic weight matrix adapted to the spinning foreign fiber removal device; The control parameters are generated based at least on the operation-related data, including: Collecting operation data of an upstream device of the spinning foreign fiber removal device within a preset time window, wherein the upstream device operation data at least includes a raw material conveying speed, an opening degree, and an impurity removal rate; Performing spatiotemporal correlation analysis on the operation-related data and the upstream equipment operation data, and screening out upstream equipment operation data with strong correlation; A multivariate regression model is constructed based on the strongly correlated upstream equipment operation data, with the abnormal parameters in the operation-related data as the dependent variable, the upstream equipment operation data and the equipment's own operating parameters as the independent variables, and the contribution weight of each parameter to the abnormality is output; According to the contribution weight sorting, several parameters with higher weights are selected as key influencing factors. Combined with the optimal control threshold in the historical maintenance record of the equipment, the control parameters including parameter adjustment range and execution timing are generated.

4. An electronic device, comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program implements the method according to claim 1 or 2 when executed by the processor.

5. A computer storage medium, characterized in that: The computer storage medium stores a computer program that can be executed by a processor to implement the method according to claim 1 or 2.

6. A computer program product, characterized in that: The computer program product comprises a computer program executable by a processor to implement the method according to claim 1 or 2 .

Images