Method for analyzing production state data of tinplate box based on internet of things

Abstract

The application provides a tin box production state data analysis method based on the Internet of Things. First, with the help of the Internet of Things sensor nodes deployed at the edge of the production equipment, a multi-source real-time monitoring data set containing equipment temperature monitoring sequences and the like is collected. Then, cross-dimension standardization processing is performed, covering time domain synchronization alignment and dimension unified conversion, to obtain a standardized monitoring data set. Then, a state feature set containing thermal distribution features and the like is extracted therefrom. Then, based on a preset abnormal state discrimination rule library, the state feature set is dynamically matched with a historical normal production feature template to generate an equipment health state score and an abnormal positioning identifier set. Finally, a production parameter optimization strategy set is generated and issued to an equipment control terminal to execute real-time regulation and control instructions, thereby achieving effective analysis and real-time regulation and control of the tin box production state.

Description

Technical Field

[0001] This application relates to the field of Internet of Things (IoT) technology, and more specifically, to a method for analyzing the production status data of tin boxes based on IoT. Background Technology

[0002] In the current tin box manufacturing industry, with the expansion of production scale and the increasing complexity of production processes, accurate monitoring and efficient management of the operating status of production equipment have become crucial for improving production quality and efficiency. However, existing production technologies have many limitations in this regard.

[0003] On the one hand, most existing methods are based on simple threshold judgments or empirical rules, lacking systematicity and dynamism. These methods cannot comprehensively evaluate based on the equipment's historical operating conditions and real-time status, making it difficult to accurately determine the equipment's health status, let alone pinpoint the specific location and type of anomalies. Once an equipment malfunctions, manual troubleshooting and diagnosis are often required, which not only consumes a lot of time and manpower but may also lead to production interruptions, affecting production efficiency and product quality.

[0004] On the other hand, existing production systems typically employ fixed production parameter settings or adjust parameters based on simple feedback mechanisms. These methods cannot automatically generate targeted production parameter optimization strategies based on the real-time health status of the equipment and anomaly location results. When the equipment's operating status changes, production parameters cannot be adjusted in a timely and effective manner to adapt to the actual operating needs of the equipment, thus affecting production stability and product quality consistency. Summary of the Invention

[0005] In view of the aforementioned problems, and in conjunction with the first aspect of this application, embodiments of this application provide a method for analyzing the production status data of tin boxes based on the Internet of Things, the method comprising:

[0006] The system collects a multi-source real-time monitoring data set by deploying IoT sensor nodes at the edge of production equipment. The multi-source real-time monitoring data set includes equipment temperature monitoring sequence, stamping pressure monitoring waveform, conveyor belt vibration spectrum, and ambient humidity change curve.

[0007] The multi-source real-time monitoring data set is subjected to cross-dimensional standardization processing to obtain a standardized monitoring data set, wherein the cross-dimensional standardization processing includes time-domain synchronization alignment operation and unit unification conversion operation;

[0008] Feature extraction is performed on the standardized monitoring data set to obtain a state feature set of the production equipment. The state feature set includes thermal distribution features, pressure fluctuation pattern features, mechanical resonance frequency point features, and environmental interference coupling features.

[0009] Based on a preset abnormal state discrimination rule base, the set of state features is dynamically matched with historical normal production feature templates to generate equipment health status scores and abnormal location identifier sets.

[0010] A set of production parameter optimization strategies is generated based on the equipment health status score and abnormal location identifier set, and the set of production parameter optimization strategies is sent to the equipment control terminal to execute real-time control commands.

[0011] In another aspect, embodiments of this application also provide a tin box production monitoring system, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to run the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0012] Based on the above, this application embodiment constructs an IoT-based data analysis method for tin box production status, achieving closed-loop control throughout the entire process, from multi-source heterogeneous data perception to equipment health status assessment and autonomous optimization of production parameters, significantly improving the intelligence level and operational efficiency of the production system. Specifically, by deploying IoT sensor nodes at the edge of production equipment, comprehensive real-time monitoring data on equipment temperature, stamping pressure, conveyor belt vibration, and environmental humidity are collected, effectively overcoming the one-sidedness of status perception caused by traditional single-source data analysis, and providing a rich data foundation for accurately assessing equipment operating status. Furthermore, cross-dimensional standardization processing technology is adopted, eliminating differences in time and spatial scales of different monitoring data through time-domain synchronization alignment and unified dimensional conversion, ensuring the comparability and fusion of multi-source data during feature extraction. This enables the extraction of multi-dimensional coupled features such as thermal distribution, pressure fluctuations, mechanical resonance, and environmental interference, significantly improving the accuracy and completeness of status feature extraction. Furthermore, based on a pre-defined rule base for identifying abnormal states, real-time state characteristics are dynamically matched with historical normal production characteristic templates. This not only enables quantitative scoring of equipment health status but also accurately locates the position and type of anomalies, providing strong support for subsequent fault warnings and maintenance. Finally, based on the equipment health status score and anomaly location results, targeted production parameter optimization strategies are automatically generated and sent to the equipment control terminal in real time to execute control commands, forming an intelligent closed-loop control system. This effectively improves the adaptability and stability of the production process, reduces the risk of production interruptions and quality fluctuations caused by equipment anomalies, and simultaneously reduces manual intervention and trial-and-error costs. Attached Figure Description

[0013] Figure 1This is a schematic diagram of the execution flow of the IoT-based tin box production status data analysis method provided in the embodiments of this application.

[0014] Figure 2 This is a schematic diagram of the hardware architecture of the tin box production monitoring system provided in the embodiments of this application. Detailed Implementation

[0015] The present application will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an embodiment of the IoT-based tin box production status data analysis method provided in this application. The following is a detailed description of the IoT-based tin box production status data analysis method.

[0016] Step S110: Collect a set of multi-source real-time monitoring data by deploying IoT sensor nodes at the edge of the production equipment. The set of multi-source real-time monitoring data includes equipment temperature monitoring sequence, stamping pressure monitoring waveform, conveyor belt vibration spectrum and ambient humidity change curve.

[0017] In the production process of tin boxes, in order to fully and accurately grasp the operating status of the production equipment, a series of Internet of Things (IoT) sensor nodes are reasonably deployed at the edge of the production equipment. These sensor nodes can capture various data information closely related to the production process in real time.

[0018] Specifically, the temperature sensor is responsible for collecting temperature monitoring data from the equipment. Taking a real-world production scenario as an example, this temperature sensor can be installed on critical parts of the production equipment, such as stamping dies and motor housings. Assuming the temperature sensor collects data at fixed time intervals, such as every 5 seconds, over a continuous 10-minute period, 120 temperature data points will be obtained. These data points are arranged in chronological order of collection time, forming a temperature monitoring sequence containing 120 values. Each value in this sequence represents the temperature of a specific part of the equipment at that corresponding moment. By analyzing this temperature monitoring sequence, the trend of temperature changes over time can be understood, and it can be determined whether the equipment is experiencing overheating or other abnormalities.

[0019] Stamping pressure sensors are used to acquire stamping pressure monitoring waveforms. In the stamping process, pressure changes are crucial to the forming quality of tin boxes. The stamping pressure sensor continuously records pressure changes over time during each stamping operation. For example, in a complete stamping cycle, the pressure gradually rises from an initial value to a peak value, and then decreases to the final value. Connecting the pressure change data recorded from multiple stamping operations forms a continuous stamping pressure monitoring waveform. This waveform visually reflects the dynamic changes in pressure during the stamping process. Analysis of this waveform can assess the operational stability of the stamping equipment and the rationality of the stamping process.

[0020] The function of a conveyor belt vibration sensor is to collect the vibration spectrum of the conveyor belt. Conveyor belts vibrate during operation, and different frequencies of vibration may reflect different operating states. The vibration sensor collects and analyzes the vibration signals of the conveyor belt to obtain the distribution of vibration energy at different frequencies, i.e., the conveyor belt vibration spectrum. For example, if an abnormal increase in vibration energy is found at a specific frequency through analysis of the conveyor belt vibration spectrum, it may indicate problems such as loosening or wear in the components corresponding to that frequency.

[0021] The ambient humidity sensor monitors changes in humidity within the production workshop. Humidity also affects the production of tin boxes; for example, excessive humidity can cause tinplate to rust, impacting product quality. The humidity sensor collects humidity data at set time intervals, such as every 10 seconds. Arranging this data chronologically creates a humidity variation curve. Analyzing this curve reveals the fluctuations in humidity over time, allowing for appropriate measures to control the humidity level.

[0022] Step S120: Perform cross-dimensional standardization processing on the multi-source real-time monitoring data set to obtain a standardized monitoring data set, wherein the cross-dimensional standardization processing includes time-domain synchronization alignment operation and unit unification conversion operation.

[0023] Since the different data in the multi-source real-time monitoring dataset have different time scales and dimensions, it is necessary to perform cross-dimensional standardization on these data to ensure that different types of data correspond in time and have a unified dimension, thereby facilitating subsequent feature extraction and state analysis.

[0024] Step S121: Perform timestamp alignment processing on the temperature monitoring sequence of the equipment, extract the temperature gradient change rate at each sampling time, and map the original temperature value to the normalized temperature range based on the preset industrial temperature measurement range.

[0025] First, timestamp alignment is performed. Since sampling times may differ between different sensors, the timestamps of the temperature monitoring sequence need to be uniformly adjusted to ensure accurate temporal correspondence between the equipment temperature monitoring sequence and other data. For example, the timestamps of the equipment temperature monitoring sequence are aligned using the sampling time of the stamping pressure monitoring waveform as a benchmark. Assuming the sampling interval of the stamping pressure monitoring waveform is 5 seconds, while the sampling interval of the equipment temperature monitoring sequence is 3 seconds, the temperature data needs to be filtered and interpolated to match the timestamps of the temperature data with those of the stamping pressure data.

[0026] Next, the rate of change of the temperature gradient at each sampling time is extracted. The rate of change of the temperature gradient reflects how quickly the temperature changes over time. Specifically, for the temperature values ​​at two adjacent sampling times, the difference is calculated and divided by the time interval. For example, if the temperature at the nth sampling time is Tn, the temperature at the (n+1)th sampling time is Tn+1, and the time interval is Δt, then the rate of change of the temperature gradient at the nth sampling time is (Tn+1-Tn) / Δt. By calculating this rate of change for each sampling time in the equipment temperature monitoring sequence, the rate of change of the temperature gradient at each sampling time can be obtained.

[0027] Finally, the original temperature values ​​are mapped to a normalized temperature range based on a preset industrial temperature measurement range. The preset industrial temperature measurement range is determined according to the normal operating temperature range of the production equipment. For example, assuming the normal operating temperature range of the production equipment is 0℃ to 100℃, then the industrial temperature measurement range is 0℃ to 100℃. The normalized temperature range is usually set between 0 and 1. The specific method for mapping the original temperature values ​​to the normalized temperature range is as follows: for each original temperature value T, its relative position within the industrial temperature measurement range is calculated, i.e., (T-Tmin) / (Tmax-Tmin), where Tmin is the minimum value of the industrial temperature measurement range (here, 0℃), and Tmax is the maximum value of the industrial temperature measurement range (here, 100℃). Through this mapping, the original temperature values ​​are converted into normalized temperature values ​​between 0 and 1, forming a normalized temperature range.

[0028] Step S122: Perform peak detection processing on the stamping pressure monitoring waveform, identify the pressure extreme points of each stamping cycle, and perform dimensionless transformation on the pressure extreme points according to the rated pressure parameters of the stamping equipment to generate a standard pressure fluctuation sequence.

[0029] Peak detection processing of the stamping pressure monitoring waveform is performed to accurately identify the pressure extreme points within each stamping cycle. This can be achieved using a local maximum value detection method. Specifically, each data point on the stamping pressure monitoring waveform is traversed, and the current data point is compared with its neighboring data points. If the value of the current data point is greater than the value of its neighboring data points, and a certain peak condition is met (such as the peak amplitude exceeding a certain threshold), then the data point is determined to be a peak point. In this way, the pressure extreme points within each stamping cycle, including the maximum and minimum pressure values, can be identified.

[0030] The pressure extreme points are transformed into dimensionless values ​​based on the rated pressure parameters of the stamping equipment. The rated pressure parameters of the stamping equipment are the normal operating pressure range specified in the equipment design; for example, the rated pressure range of a certain stamping equipment is 0 MPa to 20 MPa. For each identified pressure extreme point P, it is divided by the maximum rated pressure Pmax (here, 20 MPa) to obtain the dimensionless pressure value P' = P / Pmax. Arranging all the dimensionless transformed pressure extreme points in the order of the stamping cycle generates a standard pressure fluctuation sequence. Each value in this standard pressure fluctuation sequence represents the proportional relationship between pressure and rated pressure within the corresponding stamping cycle, eliminating the influence of differences in rated pressure between different equipment and facilitating subsequent comparison and analysis.

[0031] Step S123: Perform frequency band decomposition processing on the vibration spectrum of the conveyor belt, extract the energy integral value within the preset mechanical resonance frequency band, and convert the energy integral value into a standard vibration intensity index based on the sensitivity coefficient of the vibration sensor.

[0032] A preset bandpass filter is used to segment the original vibration spectrum into multiple sub-bands, resulting in energy distribution sequences for each sub-band. The cutoff frequency of the bandpass filter is dynamically adjusted based on the inherent vibration characteristics of the conveyor belt material. The bandpass filter allows only signals within a specific frequency range to pass through, thus dividing the original vibration spectrum into multiple sub-bands. The inherent vibration characteristics of the conveyor belt material affect its vibration response at different frequencies; therefore, the cutoff frequency of the bandpass filter needs to be dynamically adjusted according to the specific characteristics of the conveyor belt material. For example, for a conveyor belt made of a specific material, its inherent vibration frequencies are mainly concentrated between 50Hz and 200Hz. Therefore, the cutoff frequency of the bandpass filter can be set to 50Hz and 200Hz, thus dividing the original vibration spectrum into multiple sub-bands, each corresponding to a specific frequency range, and simultaneously obtaining the energy distribution sequence within each sub-band.

[0033] The target resonant frequency band corresponding to the current production batch is matched from the multiple sub-band energy distribution sequences. This target resonant frequency band is obtained through learning from historical vibration data. By analyzing and learning from a large amount of historical vibration data, the target resonant frequency band of the conveyor belt for different production batches can be determined. For example, when producing a batch of tin boxes, by analyzing the vibration data from previous batches, it is found that the conveyor belt is prone to resonance in the frequency range of 100Hz to 150Hz. This frequency range is then the target resonant frequency band corresponding to the current production batch. The sub-band energy distribution sequence corresponding to the target resonant frequency band is then selected from the multiple sub-band energy distribution sequences.

[0034] An initial energy integral value is obtained by integrating the energy distribution sequence within the target resonant frequency band. This initial energy integral value is then calibrated using speed compensation based on real-time monitoring of the conveyor belt's operating speed. Integrating the energy distribution sequence within the target resonant frequency band involves summing the energy values ​​at each frequency point within that band to obtain the initial energy integral value. Since the conveyor belt's operating speed affects its vibration, the initial energy integral value needs to be calibrated using real-time monitoring of the conveyor belt's operating speed. For example, when the conveyor belt's operating speed increases, its vibration energy may increase accordingly, requiring appropriate adjustment of the initial energy integral value. Specific compensation calibration methods can be determined based on experimental data and empirical formulas.

[0035] The calibrated energy integral value is compared with a preset vibration safety threshold. If the calibrated energy integral value exceeds the vibration safety threshold, a resonant band recalibration operation is triggered to update the target resonant band. The preset vibration safety threshold is determined based on the safety operation requirements of the production equipment. If the calibrated energy integral value exceeds this threshold, it indicates that the vibration of the conveyor belt may be abnormal, requiring recalibration of the target resonant band. The recalibration process includes re-analyzing and learning from historical vibration data, and determining a new target resonant band based on the current production situation.

[0036] The calibrated energy integral value is converted into a standard vibration intensity index based on the sensitivity coefficient of the vibration sensor. The sensitivity coefficient of the vibration sensor represents the sensor's response capability to vibration signals. Dividing the calibrated energy integral value by the sensitivity coefficient of the vibration sensor yields the standard vibration intensity index. This standard vibration intensity index more accurately reflects the vibration intensity of the conveyor belt, facilitating subsequent comprehensive analysis with other data.

[0037] Step S124: Perform sliding window smoothing on the environmental humidity change curve, calculate the humidity variation coefficient in each window, and perform ratio calculation between the humidity variation coefficient and the environmental reference humidity parameter to generate a standard humidity influence factor sequence.

[0038] The purpose of applying sliding window smoothing to the ambient humidity variation curve is to reduce noise interference in the data, making the humidity data smoother and more stable. The specific method of sliding window smoothing is to select a window of a fixed size, for example, a window with 5 data points, and then slide this window sequentially across the ambient humidity variation curve. For the data within each window, their average value is calculated as the smoothed humidity value at the center point of that window. By processing the entire ambient humidity variation curve sequentially, the smoothed humidity curve is obtained.

[0039] Calculate the coefficient of variation (CV) of humidity within each window. The CV reflects the dispersion of humidity data within a window. The calculation method is to first calculate the standard deviation of the humidity data within the window, and then divide it by the average value of the humidity data within the window. For example, if there are 5 humidity data values ​​in a window, H1, H2, H3, H4, and H5, first calculate their average value H_avg = (H1 + H2 + H3 + H4 + H5) / 5, then calculate the standard deviation σ, and finally obtain the CV = σ / H_avg. This calculation is performed for each window sequentially to obtain the CV of humidity within each window.

[0040] The humidity coefficient of variation (CV) is calculated by ratioing it to the environmental baseline humidity parameter. The environmental baseline humidity parameter is determined based on the normal humidity requirements of the production environment; for example, the environmental baseline humidity parameter for a production workshop is 60%. For each window, the humidity coefficient of variation (CV) is divided by the environmental baseline humidity parameter H_base (here, 60%), yielding the ratio CV' = CV / H_base. Arranging the ratios of all windows in chronological order generates a standard humidity influence factor sequence. Each value in this sequence represents the proportion of humidity variation within the corresponding window relative to the environmental baseline humidity, reflecting the degree of impact of changes in environmental humidity on the production process.

[0041] Step S125: Integrate the normalized temperature range, standard pressure fluctuation sequence, standard vibration intensity index, and standard humidity influence factor sequence into the standardized monitoring data set according to the time dimension.

[0042] After standardizing the equipment temperature monitoring sequence, stamping pressure monitoring waveform, conveyor belt vibration spectrum, and ambient humidity change curve, normalized temperature ranges, standard pressure fluctuation sequences, standard vibration intensity indices, and standard humidity influencing factor sequences were obtained. Next, these standardized data need to be integrated according to the time dimension to form a unified standardized monitoring data set.

[0043] The specific approach involves combining data from the same time point within the normalized temperature range, standard pressure fluctuation sequence, standard vibration intensity index, and standard humidity influence factor sequence, indexed by time. For example, at a specific time point t, the normalized temperature range contains a corresponding normalized temperature value T', the standard pressure fluctuation sequence contains a corresponding standard pressure value P', the standard vibration intensity index contains a corresponding standard vibration intensity value V', and the standard humidity influence factor sequence contains a corresponding standard humidity influence factor value H'. These four values ​​are combined to form a four-dimensional data vector [T', P', V', H']. This combination operation is performed sequentially for all time points to obtain a standardized monitoring data set containing multiple four-dimensional data vectors. Each data vector in this standardized monitoring data set represents the comprehensive operating status information of the production equipment at the corresponding time point, providing a unified data foundation for subsequent feature extraction and status analysis.

[0044] Step S130: Extract features from the standardized monitoring data set to obtain a set of state features of the production equipment. The set of state features includes thermal distribution features, pressure fluctuation pattern features, mechanical resonance frequency features, and environmental interference coupling features.

[0045] To gain a deeper understanding of the operational status of production equipment, it is necessary to extract key features that reflect the equipment's condition from standardized monitoring datasets. These features will help identify whether any abnormalities exist in the equipment, and determine the specific type and location of these abnormalities.

[0046] Step S131: Perform spatial interpolation on the normalized temperature range in the standardized monitoring data set to generate a thermal distribution map of the equipment surface, and extract the direction of the maximum temperature gradient change and the coordinates of the high temperature accumulation area from the thermal distribution map of the equipment surface.

[0047] Spatial interpolation of the normalized temperature range is performed to obtain more comprehensive temperature distribution information on the equipment surface. Since temperature sensors typically only measure the temperature of a limited number of points on the equipment surface, spatial interpolation is necessary to obtain the temperature distribution across the entire surface. Common spatial interpolation methods include linear interpolation and spline interpolation. For example, using linear interpolation, for two points A and B with known temperatures on the equipment surface, and an unknown point C between them, the temperature of point C can be calculated linearly based on the temperature values ​​of A and B and the relative position of C to A and B. By sequentially performing interpolation calculations for all unknown points on the equipment surface, a thermal distribution map of the equipment surface can be generated.

[0048] Extracting the direction of maximum temperature gradient change from the thermal distribution map of the equipment surface. The temperature gradient represents the rate of temperature change in space, and the direction of maximum temperature gradient change reflects the direction of the fastest temperature change. Specifically, on the thermal distribution map of the equipment surface, for each point, calculate its rate of temperature change in different directions, and select the direction with the largest rate of temperature change as the direction of maximum temperature gradient change for that point. By calculating for all points on the entire equipment surface, the distribution of the directions of maximum temperature gradient change on the equipment surface can be obtained.

[0049] Simultaneously, the coordinates of high-temperature clusters are identified from the thermal distribution map of the equipment surface. High-temperature clusters refer to areas on the equipment surface where the temperature is relatively high and concentrated; these areas may indicate abnormal conditions such as overheating. A temperature threshold can be set to mark areas where the equipment surface temperature exceeds this threshold as high-temperature areas. Then, using methods such as cluster analysis, adjacent high-temperature points are merged into high-temperature clusters, and the coordinates of the center point of each high-temperature cluster are determined; these coordinates are the coordinates of the high-temperature clusters.

[0050] Step S132: Perform time-frequency joint analysis on the standard pressure fluctuation sequence to extract the slope features of the pressure rise phase, the stability index of the pressure holding phase, and the decay rate of the pressure release phase, and construct a pressure fluctuation pattern encoding vector.

[0051] Joint time-frequency analysis of standard pressure fluctuation sequences aims to comprehensively understand the characteristics of pressure variation in time and frequency. There are many methods for joint time-frequency analysis, such as short-time Fourier transform and wavelet transform. Taking wavelet transform as an example, the standard pressure fluctuation sequence is decomposed using wavelet transform to obtain the coefficients of different frequency components at different time points. By analyzing these coefficients, the distribution of pressure in different times and frequencies can be obtained.

[0052] Extracting the slope characteristics of the pressure rise phase. First, identify the initial pressure point, peak pressure point, and final pressure point within the pressure fluctuation cycle. The initial pressure point is where the pressure begins to rise, the peak pressure point is where the pressure reaches its maximum value, and the final pressure point is where the pressure drops to its minimum value. Calculate the time interval from the initial pressure point to the peak pressure point as the duration of the rise phase. Perform linear fitting on the pressure sampling points within the rise phase to obtain an initial slope estimate. The linear fitting method uses the least squares method to find a straight line that minimizes the sum of squared errors between this line and the pressure sampling points within the rise phase. Simultaneously, calculate the root mean square error (RMSE) of the fitting residuals to evaluate linearity. The fitting residuals are the differences between the actual pressure sampling points and the corresponding points on the linearly fitted line; the RMSE is the square root of the average of the sum of squares of these differences. When the RMSE exceeds a preset linearity threshold, it indicates poor linear fitting performance. In this case, a piecewise polynomial fitting method is used to recalculate the dynamic slope change curve of the pressure rise phase. Piecewise polynomial fitting divides the rise phase into several small segments and performs polynomial fitting on each segment separately. Extract the maximum slope value, average slope value, and slope variation coefficient of the dynamic slope change curve, and combine the three into a slope feature.

[0053] Extract the stability index for the pressure holding phase. The pressure holding phase refers to the period after the pressure reaches its peak and remains relatively stable. The stability index can be measured by calculating the standard deviation of the pressure values ​​during this phase. The smaller the standard deviation, the more stable the pressure is during the holding phase.

[0054] Extract the decay rate during the pressure release phase. The pressure release phase refers to the period when the pressure decreases from its peak value to its final value. The decay rate can be obtained by calculating the slope of the pressure decrease during this phase. Specifically, a linear fit is performed on the pressure sampling points within the pressure release phase to obtain the slope of decrease; the absolute value of this slope is the decay rate during the pressure release phase.

[0055] By combining the slope characteristics of the pressure rise phase, the stability index of the pressure holding phase, and the decay rate of the pressure release phase, a pressure fluctuation pattern encoding vector is constructed. Each element in this vector represents a feature of the pressure fluctuation pattern, and different pressure fluctuation patterns can be identified through analysis of this vector.

[0056] Step S133: Perform wavelet packet decomposition on the standard vibration intensity index, extract the proportion of low-frequency energy related to mechanical loosening and the number of high-frequency energy abrupt change points related to bearing wear, and generate a mechanical resonance frequency feature matrix.

[0057] Wavelet packet decomposition is performed on the standard vibration intensity index. Wavelet packet decomposition is a more refined time-frequency analysis method than wavelet decomposition, capable of dividing the signal into more detailed frequency subbands. In practice, a suitable wavelet basis function is selected, such as the Daubechies wavelet basis, to perform multi-level wavelet packet decomposition on the standard vibration intensity index sequence. Each level of decomposition further decomposes the signal from the previous level into low-frequency and high-frequency subbands. After multiple levels of decomposition, multiple subbands with different frequency ranges can be obtained.

[0058] Extract the proportion of low-frequency energy associated with mechanical loosening. Mechanical loosening typically produces significant changes in vibration energy in the low-frequency range. From the multiple sub-band signals obtained through wavelet packet decomposition, determine the range of low-frequency sub-bands; for example, consider sub-bands with frequencies below 100Hz as those associated with mechanical loosening. Calculate the sum of the energies of these low-frequency sub-band signals by summing the squared values ​​of each data point in the sub-band signal. Simultaneously, calculate the total energy of all sub-band signals by summing the squared values ​​of all data points in all sub-band signals. Divide the sum of the energies of the low-frequency sub-band signals by the total energy of all sub-band signals; the resulting ratio is the proportion of low-frequency energy associated with mechanical loosening. For example, assuming the sum of the energies of the low-frequency sub-band signals is E_low and the total energy of all sub-band signals is E_total, then the proportion of low-frequency energy is E_low / E_total.

[0059] Extract the number of high-frequency energy abrupt changes related to bearing wear. Bearing wear often causes abrupt changes in vibration energy at high frequencies. Determine the range of high-frequency sub-bands related to bearing wear, for example, consider sub-bands with frequencies above 1000Hz as high-frequency sub-bands. Analyze the signals of these high-frequency sub-bands to detect energy abrupt changes. One method for detecting energy abrupt changes is to set an energy change threshold; that is, calculate the energy difference between adjacent data points, and if the difference exceeds the preset energy change threshold, the point is considered an energy abrupt change. Iterate through all data points in the high-frequency sub-band signals and count the number of energy abrupt changes.

[0060] A mechanical resonance frequency characteristic matrix is ​​generated. The proportion of low-frequency energy related to mechanical loosening and the number of high-frequency energy abrupt changes related to bearing wear are used as elements of the matrix. Assuming the above analysis was performed on standard vibration intensity indices for multiple different time periods, multiple values ​​for the proportion of low-frequency energy and the number of high-frequency energy abrupt changes were obtained. These values ​​are arranged in chronological order to form a two-dimensional matrix. Each row of the matrix corresponds to a time period, the first column represents the proportion of low-frequency energy in that time period, and the second column represents the number of high-frequency energy abrupt changes in that time period. This two-dimensional matrix is ​​the mechanical resonance frequency characteristic matrix, which comprehensively reflects the vibration characteristics related to mechanical loosening and bearing wear in different time periods.

[0061] Step S134: Perform lag correlation analysis on the standard humidity influencing factor sequence, calculate the time lag correlation coefficient between environmental humidity changes and equipment temperature fluctuations, and identify the duration of the impact of humidity mutation events on temperature stability.

[0062] A lag correlation analysis was performed on the standard humidity influencing factor series to explore the relationship between changes in ambient humidity and equipment temperature fluctuations. First, the equipment temperature fluctuation series needs to be obtained, which can be extracted from a normalized temperature range, reflecting the change in equipment temperature over time. Then, a series of lag times were set, for example, from 0 to 100 time steps. For each lag time k, the standard humidity influencing factor series was shifted forward by k time steps, and its correlation with the equipment temperature fluctuation series was calculated. The correlation calculation can use the Pearson correlation coefficient method, which measures the linear correlation between two series by calculating the ratio of the product of their covariance and standard deviation.

[0063] For each lag time k, a correlation coefficient is calculated. These correlation coefficients reflect the strength of the correlation between changes in ambient humidity and fluctuations in equipment temperature at different lag times. Among these correlation coefficients, the one with the largest absolute value is identified; the corresponding lag time is the time lag, and this largest correlation coefficient is the time-lag correlation coefficient. For example, when the lag time k = 20, the calculated correlation coefficient has the largest absolute value, which is 0.8. Therefore, the time lag is 20 time steps, and the time-lag correlation coefficient is 0.8.

[0064] Identify the duration of the impact of humidity abrupt events on temperature stability. Humidity abrupt events can be determined by setting a humidity change threshold. When the difference between a data point in the standard humidity influence factor sequence and the previous data point exceeds the preset humidity change threshold, a humidity abrupt event is considered to have occurred. For each humidity abrupt event, observe the changes in the equipment temperature fluctuation sequence after the event. Set a temperature stability threshold. When the fluctuation amplitude of the equipment temperature fluctuation sequence continuously exceeds the temperature stability threshold for a certain period after the humidity abrupt event, this period is considered the time period during which the humidity abrupt event affects temperature stability. Record the start and end times of this period, calculate the difference between the two, and obtain the duration of the humidity abrupt event's impact on temperature stability. For example, if a humidity abrupt event occurs at time t1, and the equipment temperature fluctuation continuously exceeds the temperature stability threshold during the time period from t1 to t2, then the duration of the humidity abrupt event's impact on temperature stability is t2-t1.

[0065] Step S135: Integrate the maximum change direction of the temperature gradient, the pressure fluctuation mode encoding vector, the mechanical resonance frequency feature matrix, and the time delay correlation coefficient into the state feature set.

[0066] After extracting features from each data point in the standardized monitoring dataset, we obtained the direction of maximum temperature gradient change, the pressure fluctuation pattern encoding vector, the mechanical resonance frequency feature matrix, and the time-delay correlation coefficient. Next, these features are integrated to form a set of state features for the production equipment.

[0067] Specifically, the direction of maximum temperature gradient change is represented as a vector, with elements corresponding to components along the length, width, and height of the equipment. The pressure fluctuation pattern encoding vector is itself a vector, containing information such as the slope characteristics of the pressure rise phase, the stability index of the pressure holding phase, and the decay rate of the pressure release phase. The mechanical resonance frequency feature matrix is ​​a two-dimensional matrix reflecting the vibration characteristics related to mechanical loosening and bearing wear over different time periods. The time-delay correlation coefficient is a scalar value reflecting the correlation between changes in ambient humidity and equipment temperature fluctuations.

[0068] The vector representing the direction of maximum temperature gradient change, the pressure fluctuation pattern encoding vector, each row of the mechanical resonance frequency feature matrix, and the time-delay correlation coefficient are concatenated in a predetermined order. For example, the vector representing the direction of maximum temperature gradient change is placed first, followed by the pressure fluctuation pattern encoding vector and each row of the mechanical resonance frequency feature matrix, and finally, the time-delay correlation coefficient is added as a separate element to the end of the vector. This concatenation operation forms a high-dimensional vector, which represents the set of state features of the production equipment. This set integrates the equipment's thermal distribution characteristics, pressure fluctuation pattern characteristics, mechanical resonance frequency characteristics, and environmental interference coupling characteristics, comprehensively reflecting the operating status of the production equipment.

[0069] Step S140: Based on the preset abnormal state discrimination rule library, the state feature set is dynamically similar to the historical normal production feature template to generate equipment health status score and abnormal location identifier set.

[0070] To assess the health status of production equipment and locate potential anomalies, it is necessary to perform dynamic similarity matching between the extracted set of status features and historical normal production feature templates, and generate equipment health status scores and anomaly location identifier sets based on a preset anomaly status discrimination rule library.

[0071] Step S141: Retrieve a set of reference features that are the same as the current production batch number from the historical normal production feature template. The set of reference features includes historical thermal distribution benchmark, pressure fluctuation pattern benchmark, mechanical resonance frequency benchmark, and humidity influence benchmark.

[0072] The historical normal production feature template is a collection of feature data accumulated during past normal production processes, containing reference features from different production batches. When performing similarity matching, the first step is to retrieve the reference feature set that matches the current production batch number from the historical normal production feature template. This reference feature set is a comprehensive set of features. The historical thermal distribution benchmark is the temperature distribution characteristics of the equipment surface during normal production of that batch, such as the standard value of the direction of maximum temperature gradient change and the standard coordinates of high-temperature accumulation areas. The pressure fluctuation pattern benchmark is the pressure fluctuation characteristics during normal production, such as the standard slope characteristics of the pressure rise phase, the standard stability index of the pressure holding phase, and the standard decay rate of the pressure release phase. The mechanical resonance frequency benchmark is the vibration characteristics related to mechanical loosening and bearing wear during normal production, such as the standard value of the proportion of low-frequency energy and the standard value of the number of high-frequency energy abrupt change points. The humidity influence benchmark is the correlation standard between changes in ambient humidity and equipment temperature fluctuations during normal production, such as the standard value of the time-delay correlation coefficient. By retrieving the reference feature set corresponding to the current production batch, an accurate comparison benchmark is provided for subsequent similarity matching.

[0073] Step S142: Calculate the cosine of the angle between the direction of the maximum change in the temperature gradient and the historical thermal distribution benchmark, and use it as the thermal anomaly index.

[0074] The direction of maximum temperature gradient change in the current production cycle is represented as a three-dimensional direction vector, where the three dimensions correspond to the length, width, and height of the equipment, respectively. A standard temperature gradient direction vector for the same production batch is extracted from historical thermal distribution data. The dot product of these two vectors is calculated by multiplying the elements of the corresponding dimensions and then summing them. Then, the magnitudes of the two vectors are calculated separately; the magnitude is the square root of the sum of the squares of the vector's elements. Dividing the dot product by the product of the magnitudes of the two vectors yields the cosine of the direction angle. For example, if the current maximum temperature gradient change direction vector is A = (a1, a2, a3), and the standard temperature gradient direction vector is B = (b1, b2, b3), then the dot product is A·B = a1*b1 + a2*b2 + a3*b3, and the magnitude of vector A is |A| = √(a1² + a2² + a3²). 2 The magnitude of vector B is |B| = √(b1² + b2² + b3²). 2 The cosine of the direction angle is cosθ = (A·B) / (|A|*|B|). This cosine value of the direction angle is used as the thermal anomaly index. The closer the value is to 1, the closer the direction of the current maximum change in temperature gradient is to the historical baseline, and the more normal the thermal state is. The closer the value is to -1, the greater the difference between the two directions, and the thermal state may be abnormal.

[0075] When the cosine value of the directional angle is lower than a preset directional consistency threshold, a thermal distribution anomaly alarm is triggered and the coordinates of the abnormal area are recorded. The preset directional consistency threshold is determined based on the normal operating requirements of the production equipment and experience, for example, it is set to 0.8. If the calculated cosine value of the directional angle is lower than 0.8, it is considered that the current maximum temperature gradient change direction differs significantly from the historical benchmark, triggering a thermal distribution anomaly alarm. Simultaneously, the coordinates of the high-temperature accumulation area are recorded from the thermal distribution map of the equipment surface; these coordinates are the coordinates of the abnormal area, facilitating further investigation and handling of the anomaly.

[0076] The weighting coefficient of the thermal anomaly index is dynamically adjusted based on the difference between the cosine of the directional angle and the directional consistency threshold. When the difference is large, it indicates a high degree of thermal anomaly; in this case, the weighting coefficient of the thermal anomaly index is increased to highlight the impact of thermal anomalies on the equipment health status score. Conversely, when the difference is small, it indicates a relatively low degree of thermal anomaly; in this case, the weighting coefficient of the thermal anomaly index is appropriately decreased. For example, a weighting adjustment function can be defined to calculate a new weighting coefficient based on the magnitude of the difference, and the adjusted weighting coefficient can be applied to subsequent equipment health status score calculations.

[0077] Step S143: Perform dynamic time warping matching between the pressure fluctuation pattern encoding vector and the pressure fluctuation pattern benchmark to obtain the pressure pattern deviation score.

[0078] Dynamic Time Warping (DTW) is a method for comparing the similarity of two time series. It can find the optimal matching path between them without requiring the two series to have the same length. This method involves dynamically warping the pressure fluctuation pattern encoding vector against a pressure fluctuation pattern benchmark. First, a distance matrix is ​​constructed, where each element represents the distance between each element in the pressure fluctuation pattern encoding vector and the corresponding element in the benchmark. Distances can be calculated using methods such as Euclidean distance; for example, for two elements x and y, the Euclidean distance is d = √((xy)²). Then, using dynamic programming, an optimal path is found in the distance matrix that minimizes the sum of distances along this path. The sum of distances along this optimal path is the dynamic time warping distance. The dynamic time warping distance is then normalized, for example, by dividing it by a preset maximum distance value, to obtain a pressure pattern deviation score. A larger score indicates a greater deviation between the current pressure fluctuation pattern and the benchmark pattern, suggesting a possible anomaly in the pressure pattern; a smaller score indicates a closer similarity and a more normal pressure pattern.

[0079] Step S144: Perform singular value decomposition on the mechanical resonance frequency feature matrix and the mechanical resonance frequency reference, and extract the maximum singular value ratio as a mechanical state degradation index.

[0080] Singular Value Decomposition (SVD) is a method that decomposes a matrix into a product of three matrices, revealing the matrix's intrinsic structure and characteristics. SVD is applied to the characteristic matrix and reference matrix of the mechanical resonance frequency point. Specifically, for the characteristic matrix M and the reference matrix N of the mechanical resonance frequency point, SVD yields M = U1*Σ1*V1^T and N = U2*Σ2*V2^T, where U1 and U2 are orthogonal matrices, Σ1 and Σ2 are diagonal matrices, and the elements on the diagonal are the singular values. V1^T and V2^T are the transposes of V1 and V2, respectively.

[0081] The maximum singular values ​​in Σ1 and Σ2 are extracted and denoted as σ1_max and σ2_max, respectively. The ratio of the maximum singular values, σ1_max / σ2_max, is calculated. This ratio serves as an indicator of mechanical degradation, reflecting the degree of difference between the current mechanical resonance frequency characteristics and historical benchmarks. The closer the ratio is to 1, the more similar the current mechanical condition is to the historical normal condition, indicating a better mechanical condition. The greater the deviation of the ratio from 1, the more likely the mechanical condition has degraded, and the greater the possibility of problems such as mechanical loosening or bearing wear.

[0082] Step S145: Perform sliding window correlation analysis on the time-delay correlation coefficient and the humidity influence benchmark to calculate the rate of change of humidity interference intensity.

[0083] Set a sliding window size, for example, 10 time steps. Arrange the time-delay correlation coefficient sequence and the humidity influence baseline sequence in chronological order, and slide the sliding window sequentially across these two sequences. Within each window, calculate the correlation between the time-delay correlation coefficient sequence and the humidity influence baseline sequence; the correlation can be calculated using the Pearson correlation coefficient method. After obtaining the correlation coefficient within each window, calculate the difference in correlation coefficients between adjacent windows. Divide these differences by the window time interval to obtain the rate of change of humidity interference intensity. For example, if the correlation coefficient in the i-th window is ri, the correlation coefficient in the (i+1)-th window is ri+1, and the window time interval is Δt, then the rate of change of humidity interference intensity is (ri+1-ri) / Δt. The rate of change of humidity interference intensity reflects how the intensity of the interference of environmental humidity changes on equipment temperature fluctuations changes over time. The larger the rate of change, the more unstable the humidity interference, which may have a greater impact on equipment operation.

[0084] Step S146: Generate the equipment health status score based on the weighted combination of the thermal anomaly index, pressure mode deviation score, mechanical condition degradation index, and humidity interference intensity change rate.

[0085] Weighting coefficients are assigned to the thermal anomaly index, pressure mode deviation score, mechanical condition degradation index, and humidity disturbance intensity change rate, respectively. These weighting coefficients are determined based on the importance of each feature to the equipment's health status. For example, the weight of the thermal anomaly index is set to w1 = 0.3, the weight of the pressure mode deviation score to w2 = 0.2, the weight of the mechanical condition degradation index to w3 = 0.3, and the weight of the humidity disturbance intensity change rate to w4 = 0.2. Each feature value is multiplied by its corresponding weighting coefficient, and these products are then added together to obtain the equipment health status score. That is, Equipment Health Status Score = w1 * Thermal Anomaly Index + w2 * Pressure Mode Deviation Score + w3 * Mechanical Condition Degradation Index + w4 * Humidity Disturbance Intensity Change Rate. This weighted combination comprehensively considers multiple factors such as the equipment's thermal distribution, pressure fluctuations, mechanical resonance, and humidity disturbance, resulting in a comprehensive score that reflects the equipment's health status. A higher score indicates better equipment health; a lower score indicates potential abnormalities requiring further attention and action.

[0086] Simultaneously, based on the comparison results of each feature value with its corresponding benchmark value, an anomaly location marker set is generated. For example, if the thermal anomaly index is lower than the directional consistency threshold, the thermal distribution is marked as abnormal; if the pressure pattern deviation score exceeds the preset pressure anomaly threshold, the pressure fluctuation pattern is marked as abnormal; if the mechanical condition degradation index deviates significantly from 1, the mechanical resonance frequency point is marked as abnormal; if the humidity interference intensity change rate exceeds the preset humidity interference threshold, the humidity influence is marked as abnormal. These anomaly markers are combined to form an anomaly location marker set, which can clearly indicate the specific aspects of potential equipment anomalies, providing accurate location information for subsequent production parameter optimization.

[0087] Step S150: Generate a set of production parameter optimization strategies based on the equipment health status score and abnormal location identifier set, and send the set of production parameter optimization strategies to the equipment control terminal to execute real-time control commands.

[0088] After obtaining the equipment health status score and abnormal location identifier set, it is necessary to generate a corresponding set of production parameter optimization strategies based on this information, and then send these strategies to the equipment control terminal to achieve real-time control of the production process, ensuring the normal operation of the production equipment and product quality.

[0089] Step S151: When the health status score of the equipment is lower than the first health threshold, a stamping pressure reduction strategy is generated. The stamping pressure reduction strategy includes a pressure peak reduction ratio and a pressure holding time adjustment scheme.

[0090] The first health threshold is a scoring limit set based on the normal operating requirements of production equipment and experience, for example, 70 points. When the equipment health status score is below 70 points, it indicates that the overall health status of the equipment is poor, and there may be problems such as abnormal pressure fluctuations. At this time, it is necessary to generate a stamping pressure reduction strategy.

[0091] Determine the pressure peak reduction percentage. By analyzing the pressure pattern deviation score and historical production data, assess the impact of the current stamping pressure on the equipment's health. For example, a high pressure pattern deviation score indicates significant pressure fluctuations, potentially placing a heavy burden on the equipment, necessitating an appropriate reduction in the pressure peak. Based on the assessment results, determine the pressure peak reduction percentage, for example, a 10% reduction. That is, if the current stamping pressure peak is P, the adjusted pressure peak will be P*(1-10%).

[0092] Develop a pressure holding time adjustment plan. Pressure holding time refers to the length of time the pressure remains at its peak during the stamping process. Analyze the stability index during the pressure holding phase and the actual operating conditions of the equipment to determine the direction and magnitude of the pressure holding time adjustment. If the stability index during the pressure holding phase is low, it indicates that the pressure fluctuates significantly during this phase, and the pressure holding time may need to be appropriately shortened. For example, adjust the pressure holding time from the original t seconds to t*(1-5%) seconds. Combine the pressure peak reduction ratio with the pressure holding time adjustment plan to form a stamping pressure reduction strategy.

[0093] Step S152: When the mechanical condition degradation index exceeds the preset mechanical wear threshold, a vibration suppression strategy is generated. The vibration suppression strategy includes conveyor belt tension adjustment parameters and lubricant replenishment cycle.

[0094] The preset mechanical wear threshold is a limit value determined based on the mechanical performance and service life of the production equipment, for example, set to 1.2. When the mechanical condition degradation index exceeds 1.2, it indicates that the mechanical condition of the equipment has deteriorated significantly, and there may be problems such as mechanical loosening and bearing wear, which leads to increased vibration. At this time, it is necessary to generate a vibration suppression strategy.

[0095] First, determine the conveyor belt tension adjustment parameters. Insufficient conveyor belt tension may cause slippage and vibration during operation, while excessive tension may increase wear on the conveyor belt and transmission components. By analyzing information such as the proportion of low-frequency energy and the number of high-frequency energy abrupt changes in the mechanical resonance frequency characteristic matrix, the relationship between conveyor belt vibration and tension is assessed. If the proportion of low-frequency energy is too high, it may indicate insufficient conveyor belt tension, requiring an appropriate increase in tension; if the number of high-frequency energy abrupt changes is too high, it may indicate excessive tension or other abnormalities, requiring adjustment.

[0096] Based on historical data and experimental results, a tension adjustment model is established. For example, when the proportion of low-frequency energy exceeds a set threshold, the tension adjustment amount is increased accordingly for each increase in the proportion of low-frequency energy. Assume that for every 5% increase in the proportion of low-frequency energy, the conveyor belt tension will increase by 10N. By analyzing the characteristic matrix of the current mechanical resonance frequency, the required tension value is calculated, thus obtaining the conveyor belt tension adjustment parameters.

[0097] Next, the lubricant replenishment cycle is determined. Wear on mechanical parts leads to increased friction, which in turn generates more vibration. Lubricants can reduce friction between mechanical parts, thus reducing wear and vibration. The wear level and lubrication requirements of mechanical parts are assessed by analyzing the mechanical resonant frequency characteristic matrix and factors such as equipment operating time and load conditions.

[0098] Based on historical data and the equipment's instruction manual, a lubricant replenishment cycle model is established. For example, when the mechanical degradation index exceeds a set threshold, the lubricant replenishment cycle is shortened accordingly for each additional set value of the mechanical degradation index. Assuming that for every 0.1 increase in the mechanical degradation index, the lubricant replenishment cycle is shortened from 30 days to 25 days. By analyzing the current mechanical degradation index, the adjusted lubricant replenishment cycle is calculated, resulting in an adjustment scheme for the lubricant replenishment cycle.

[0099] By combining the conveyor belt tension adjustment parameters and the lubricant replenishment cycle adjustment scheme, a vibration suppression strategy is formed.

[0100] Step S153: When the thermal anomaly index continuously exceeds the preset temperature rise threshold, a cooling system optimization strategy is generated. The cooling system optimization strategy includes a fan speed increase gradient and a coolant flow rate adjustment curve.

[0101] The preset temperature rise threshold is a limit value set based on the normal operating temperature range and heat dissipation requirements of the production equipment, for example, 0.8. When the thermal anomaly index exceeds 0.8 multiple times consecutively, it indicates that the thermal state of the equipment is continuously abnormal, and overheating may have occurred. At this time, it is necessary to generate a cooling system optimization strategy.

[0102] The base rotational speed increase is determined based on the difference between the thermal anomaly index and the temperature rise threshold, and a dynamic compensation coefficient is calculated based on the current equipment load rate. First, the difference between the thermal anomaly index and the temperature rise threshold is calculated. For example, if the current thermal anomaly index is 0.9 and the temperature rise threshold is 0.8, the difference is 0.1. Based on historical data and experimental results, a relationship model between the base rotational speed increase and the difference is established. For example, for every 0.01 increase in the difference, the base rotational speed increase increases by 100 rpm; therefore, for a difference of 0.1, the base rotational speed increase is 1000 rpm.

[0103] The current load rate of the equipment reflects its workload; a higher load rate means more heat is generated and a greater cooling capacity is required. The current load rate is calculated by monitoring parameters such as the equipment's current and power. For example, if the equipment's rated power is P0 and its current actual power is P1, then the current load rate is P1 / P0. Based on historical data and experimental results, a model is established to model the relationship between the dynamic compensation coefficient and the current load rate. For example, when the current load rate is between 50% and 70%, the dynamic compensation coefficient is 1.2; when the current load rate is between 70% and 90%, the dynamic compensation coefficient is 1.5. The dynamic compensation coefficient is then determined based on the current equipment load rate.

[0104] The actual speed increase gradient is obtained by multiplying the base speed increase by the dynamic compensation coefficient, and the speed increase rate is limited to not exceeding the motor's maximum acceleration. Multiplying the previously calculated base speed increase of 1000 rpm by the dynamic compensation coefficient 1.5 yields an actual speed increase gradient of 1500 rpm. Simultaneously, to prevent motor damage due to excessive speed increase, the speed increase rate needs to be limited to the motor's maximum acceleration. Assuming the motor's maximum acceleration is 2000 rpm², the time interval for speed increase is reasonably adjusted based on the motor's start-up time and the actual speed increase gradient to ensure the speed increase rate remains within the motor's maximum acceleration range.

[0105] The direction of maximum temperature gradient change determines the priority coolant supply area, generating zoned flow control commands. The direction of maximum temperature gradient change reflects the direction of the fastest temperature change on the equipment surface; areas along this direction often have higher temperatures and require priority cooling. The direction of maximum temperature gradient change is determined by analyzing the thermal distribution map of the equipment surface. For example, if the direction of maximum temperature gradient change points to a specific corner of the equipment, then the area containing that corner is designated as the priority coolant supply area.

[0106] Based on the priority coolant supply area, zoned flow control commands are generated. For example, the cooling system is divided into multiple zones. For the priority coolant supply area, the coolant flow rate in that zone is increased; for other zones, the coolant flow rate is adjusted appropriately according to the temperature distribution. The specific flow rate adjustment values ​​can be calculated based on the temperature distribution on the equipment surface and the cooling requirements.

[0107] The coolant demand density distribution is calculated based on the coordinates of the high-temperature accumulation region, and the coolant flow rate adjustment curve is obtained by fitting the curve. The coordinates of the high-temperature accumulation region are determined by analyzing the thermal distribution map of the equipment surface. The coolant demand density for each region is calculated based on factors such as its size and temperature. For example, the higher the temperature and the larger the area of ​​the high-temperature accumulation region, the higher the coolant demand density for that region.

[0108] The coolant demand density of each region is arranged in a predetermined order to obtain the coolant demand density distribution. Then, a fitting method, such as polynomial fitting, is used to fit a coolant flow rate adjustment curve based on the coolant demand density distribution. This curve shows how the coolant flow rate should be adjusted at different locations and times to meet the cooling needs of the equipment.

[0109] A coordinated control timing sequence is established between the fan speed increase gradient and the coolant flow rate adjustment curve to ensure phase synchronization between fan acceleration and coolant supply increment. To improve the efficiency of the cooling system, it is necessary to ensure that fan acceleration and coolant supply increment are synchronized in time. A coordinated control timing sequence is developed based on the actual fan speed increase gradient and coolant flow rate adjustment curve. For example, as the fan begins to accelerate, the coolant flow rate begins to increase according to the coolant flow rate adjustment curve, ensuring that the coolant flow rate reaches its corresponding maximum value when the fan reaches its maximum speed. This coordinated control enables the cooling system to more effectively reduce equipment temperature and resolve thermal anomalies.

[0110] Step S154: When the rate of change of the humidity interference intensity exceeds the environmental adaptation threshold, a humidity compensation strategy is generated. The humidity compensation strategy includes the sealed chamber air pressure adjustment parameters and the desiccant replacement frequency.

[0111] The environmental adaptation threshold is a limit value set based on the production equipment's ability to adapt to changes in environmental humidity, for example, set to 0.05. When the rate of change of humidity interference intensity exceeds 0.05, it indicates that changes in environmental humidity have a significant impact on the equipment and may affect its normal operation. In this case, a humidity compensation strategy needs to be generated.

[0112] Determine the air pressure adjustment parameters for the sealed chamber. The sealed chamber can isolate the influence of external environmental humidity to a certain extent. By adjusting the air pressure inside the sealed chamber, the humidity inside can be controlled. Analyze the rate of change of humidity interference intensity and the environmental humidity change curve to assess the degree and trend of the current humidity's impact on the equipment. If the rate of change of humidity interference intensity continues to increase, it indicates that the interference of environmental humidity changes on the equipment is intensifying, and the air pressure inside the sealed chamber needs to be appropriately increased to reduce the entry of external humidity.

[0113] Based on historical data and experimental results, a model was established to show the relationship between the air pressure adjustment parameters of the sealed chamber and the rate of change of humidity interference intensity. For example, for every 0.01 increase in the rate of change of humidity interference intensity, the air pressure inside the sealed chamber increased by 100 Pascals. By analyzing the current rate of change of humidity interference intensity, the required air pressure adjustment value of the sealed chamber was calculated, thus obtaining the air pressure adjustment parameters.

[0114] Determine the desiccant replacement frequency. The desiccant absorbs moisture from the sealed chamber, reducing humidity. Analyze the rate of change in humidity interference intensity and the humidity changes within the sealed chamber to assess the desiccant's moisture absorption effect and lifespan. If the rate of change in humidity interference intensity is large, it indicates that the desiccant's moisture absorption rate may not keep up with the humidity changes, requiring a shorter desiccant replacement frequency.

[0115] Based on historical data and desiccant performance parameters, a model is established to model the relationship between desiccant replacement frequency and the rate of change in humidity interference intensity. For example, for every 0.01 increase in the rate of change in humidity interference intensity, the desiccant replacement frequency is adjusted from once a week to once every three days. By analyzing the current rate of change in humidity interference intensity, the adjusted desiccant replacement frequency is calculated, thus obtaining an adjustment scheme for the desiccant replacement frequency.

[0116] By combining the pressure adjustment parameters of the sealed chamber with the desiccant replacement frequency adjustment scheme, a humidity compensation strategy is formed.

[0117] Step S155: The stamping pressure reduction strategy, vibration suppression strategy, cooling system optimization strategy and humidity compensation strategy are sorted by priority and integrated into the production parameter optimization strategy set.

[0118] Based on the degree of impact and urgency of each strategy on the equipment's health status, the following priorities are established: stamping pressure reduction strategy, vibration suppression strategy, cooling system optimization strategy, and humidity compensation strategy. Generally, if the equipment has overheating issues, the cooling system optimization strategy has the highest priority, as overheating may lead to equipment damage or even safety accidents. If the equipment has abnormal mechanical vibration issues, the vibration suppression strategy has the next highest priority, as excessive vibration can affect the equipment's stability and lifespan. If the equipment experiences abnormal pressure fluctuations, the stamping pressure reduction strategy has the next lowest priority. If ambient humidity significantly interferes with the equipment, the humidity compensation strategy has a relatively low priority.

[0119] For example, the priority of the cooling system optimization strategy is set to 1, the vibration suppression strategy to 2, the stamping pressure reduction strategy to 3, and the humidity compensation strategy to 4. These strategies are then arranged sequentially from highest to lowest priority, forming a set of production parameter optimization strategies. This set of strategies includes optimization measures for different abnormal equipment conditions, enabling comprehensive adjustments to production parameters to ensure normal equipment operation.

[0120] Step S156: Send the set of production parameter optimization strategies to the equipment control terminal to execute real-time control commands.

[0121] The press pressure reduction strategy is converted into a hydraulic system control signal, which includes pressure sensor calibration parameters and servo valve opening adjustment step size. The pressure peak reduction ratio and holding time adjustment scheme in the press pressure reduction strategy need to be implemented through the hydraulic system. The pressure peak reduction ratio is converted into pressure sensor calibration parameters, for example, by adjusting the pressure sensor's measurement range accordingly to accurately measure the adjusted pressure value. The holding time adjustment scheme is converted into a servo valve opening adjustment step size, for example, by determining the adjustment range and time interval of the servo valve opening based on the shortening or lengthening of the holding time. These parameters are combined to form the hydraulic system control signal.

[0122] The vibration suppression strategy is converted into mechanical transmission system control commands, which include a tension roller displacement setpoint and automatic lubrication device triggering conditions. The conveyor belt tension adjustment parameters and lubricant replenishment cycle in the vibration suppression strategy need to be implemented through the mechanical transmission system. The conveyor belt tension adjustment parameters are converted into a tension roller displacement setpoint, for example, calculating the distance the tension roller needs to move based on the required increase or decrease in tension. The lubricant replenishment cycle is converted into automatic lubrication device triggering conditions, for example, determining the triggering time interval of the automatic lubrication device based on a shortened or extended lubricant replenishment cycle. These parameters are combined to form the mechanical transmission system control commands.

[0123] The cooling system optimization strategy is converted into PID parameter adjustment instructions for the temperature control module. These instructions include a proportional gain correction and an update value for the integral time constant. The fan speed increase gradient and coolant flow rate adjustment curve in the cooling system optimization strategy need to be implemented through the temperature control module. The fan speed increase gradient and coolant flow rate adjustment curve are converted into PID parameter adjustment instructions; for example, based on the changes in the speed increase gradient and flow rate adjustment curves, the proportional gain correction and the update value for the integral time constant are calculated. By adjusting the PID parameters, the temperature control module can more accurately control the fan speed and coolant flow rate, achieving effective regulation of the equipment temperature.

[0124] The humidity compensation strategy is converted into an environmental control system operation sequence, which includes a valve opening and closing timing diagram and a desiccant dispensing mechanism trigger frequency. The sealed chamber pressure adjustment parameters and desiccant replacement frequency in the humidity compensation strategy need to be implemented through the environmental control system. The sealed chamber pressure adjustment parameters are converted into a valve opening and closing timing diagram, for example, determining the valve opening and closing times based on the required increase or decrease in sealed chamber pressure. The desiccant replacement frequency is converted into a desiccant dispensing mechanism trigger frequency, for example, determining the trigger time interval of the desiccant dispensing mechanism based on a shortened or extended desiccant replacement frequency. These parameters are combined to form the environmental control system operation sequence.

[0125] The control signals, control commands, adjustment commands, and operation sequences are encapsulated into real-time data packets using an industrial bus protocol. These packets are then distributed to the control terminals of each device according to a preset priority queue, and execution status feedback is received to verify the effectiveness of the strategy. An industrial bus protocol, such as Modbus or PROFIBUS, is a standard protocol used for communication between devices in industrial automation systems. Hydraulic system control signals, mechanical transmission system control commands, PID parameter adjustment commands from the temperature control module, and environmental control system operation sequences are encapsulated according to the format of the industrial bus protocol to form real-time data packets.

[0126] Based on the previously set policy priorities, real-time data packets are sent to each device control terminal in order of priority queue. Upon receiving the data packets, each device control terminal executes the corresponding control commands. Simultaneously, the device control terminal sends back execution status feedback information. By analyzing this feedback information, the effectiveness of the production parameter optimization strategy is verified. If the strategy is found to be ineffective, it needs to be re-evaluated and adjusted to ensure the equipment can return to normal operating status.

[0127] Figure 2This illustration shows the hardware structure of a tin box production monitoring system 100 provided in an embodiment of this application for implementing the above-described IoT-based tin box production status data analysis method. Figure 2 As shown, the tin box production monitoring system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0128] In one possible design, the tin box production monitoring system 100 can be a single server or a group of servers. The server group can be centralized or distributed (e.g., the tin box production monitoring system 100 can be a distributed system). In some embodiments, the tin box production monitoring system 100 can be local or remote. For example, the tin box production monitoring system 100 can access information and / or data stored in machine-readable storage medium 120 via a network. Alternatively, the tin box production monitoring system 100 can be directly connected to machine-readable storage medium 120 to access stored information and / or data. In some embodiments, the tin box production monitoring system 100 can be implemented on a tin box production monitoring system. By way of example only, the tin box production monitoring system can include a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-layered cloud, etc., or any aggregation thereof.

[0129] Machine-readable storage medium 120 may store data and / or instructions. In some embodiments, machine-readable storage medium 120 may store data acquired from an external terminal. In some embodiments, machine-readable storage medium 120 may store data and / or instructions used by the tin box production monitoring system 100 to perform or use in order to accomplish the exemplary methods described in this application.

[0130] In the specific implementation process, one or more processors 110 execute computer-executable instructions stored in machine-readable storage medium 120, so that processor 110 can execute the IoT-based tin box production status data analysis method as described in the above method embodiment. Processor 110, machine-readable storage medium 120 and communication unit 140 are connected through bus 130. Processor 110 can be used to control the sending and receiving actions of communication unit 140.

[0131] The specific implementation process of processor 110 can be found in the various method embodiments executed by the tin box production monitoring system 100 described above. The implementation principle and technical effect are similar, and will not be repeated here.

[0132] Furthermore, this application embodiment also provides a readable storage medium containing computer-executable instructions. When the processor executes the computer-executable instructions, the above-mentioned IoT-based tin box production status data analysis method is implemented.

[0133] It should be noted that, in order to simplify the description disclosed in this application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of this application sometimes combines multiple features into a single embodiment, drawing, or description thereof. Similarly, it should be noted that, in order to simplify the description disclosed in this application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of this application sometimes combines multiple features into a single embodiment, drawing, or description thereof.

Claims

1. A tin box production state data analysis method based on Internet of Things, characterized in that, The method comprises: Collecting a multi-source real-time monitoring data set through an Internet of Things sensor node deployed at the edge of a production device, the multi-source real-time monitoring data set comprising a device temperature monitoring sequence, a stamping pressure monitoring waveform, a conveyor belt vibration frequency spectrum, and an environmental humidity change curve; Performing cross-dimension standardization processing on the multi-source real-time monitoring data set to obtain a standardized monitoring data set, wherein the cross-dimension standardization processing comprises time domain synchronization alignment operation and dimension unification conversion operation; Performing feature extraction on the standardized monitoring data set to obtain a state feature set of the production device, the state feature set comprising thermal distribution features, pressure fluctuation mode features, mechanical resonance frequency point features, and environmental interference coupling features; Based on a preset abnormal state discrimination rule library, performing dynamic similarity matching between the state feature set and a historical normal production feature template to generate a device health state score and an abnormal positioning identifier set; Retrieving a reference feature set identical to a current production batch number from the historical normal production feature template, the reference feature set comprising a historical thermal distribution benchmark, a pressure fluctuation mode benchmark, a mechanical resonance frequency point benchmark, and a humidity influence benchmark; Performing dynamic similarity matching to obtain a thermal abnormality index, a pressure mode deviation score, a mechanical state degradation index, and a humidity interference intensity change rate, and generating the device health state score according to a weighted combination of the thermal abnormality index, the pressure mode deviation score, the mechanical state degradation index, and the humidity interference intensity change rate; Generating a production parameter optimization strategy set according to the device health state score and the abnormal positioning identifier set, and issuing the production parameter optimization strategy set to a device control terminal to execute real-time regulation and control instructions, comprising: When the device health state score is lower than a first health threshold, generating a stamping pressure down-regulation strategy, the stamping pressure down-regulation strategy comprising a pressure peak value reduction ratio and a pressure maintaining time adjustment scheme; When the mechanical state degradation index exceeds a preset mechanical wear threshold, generating a vibration suppression strategy, the vibration suppression strategy comprising a conveyor belt tensioning force adjustment parameter and a lubricant replenishment period; When the thermal abnormality index continuously exceeds a preset temperature rise threshold, generating a cooling system optimization strategy, the cooling system optimization strategy comprising a fan rotating speed increase gradient and a coolant flow adjustment curve; When the humidity interference intensity change rate exceeds an environmental adaptation threshold, generating a humidity compensation strategy, the humidity compensation strategy comprising a sealed cabin air pressure adjustment parameter and a desiccant replacement frequency; After prioritizing the stamping pressure down-regulation strategy, the vibration suppression strategy, the cooling system optimization strategy, and the humidity compensation strategy, integrating them into the production parameter optimization strategy set.

2. The Internet of Things-based tin box production status data analysis method according to claim 1, characterized in that, The cross-dimension standardization processing on the multi-source real-time monitoring data set to obtain a standardized monitoring data set comprises: Performing timestamp alignment processing on the device temperature monitoring sequence, extracting a temperature gradient change rate at each sampling time, and mapping an original temperature value to a normalized temperature interval based on a preset industrial temperature measurement range; Peak detection is performed on the stamping pressure monitoring waveform to identify pressure extreme points of each stamping cycle, and the pressure extreme points are converted into dimensionless form according to the rated pressure parameter of the stamping equipment to generate a standard pressure fluctuation sequence; The frequency band decomposition process is performed on the conveyor belt vibration frequency spectrum to extract the energy integral value in the preset mechanical resonance frequency band, and the energy integral value is converted into a standard vibration intensity index based on the sensitivity coefficient of the vibration sensor; The sliding window smoothing process is performed on the environmental humidity change curve to calculate the humidity variation coefficient in each window, and the humidity variation coefficient is subjected to ratio operation with the environmental reference humidity parameter to generate a standard humidity influence factor sequence; The normalized temperature interval, the standard pressure fluctuation sequence, the standard vibration intensity index and the standard humidity influence factor sequence are integrated into the standardized monitoring data set according to the time dimension. 3.The Internet of Things based tin box production status data analysis method according to claim 2, characterized in that, The frequency band decomposition process on the conveyor belt vibration frequency spectrum includes: A preset band-pass filter is called to segment the original vibration frequency spectrum into multiple sub-band energy distribution sequences, wherein the cutoff frequency of the band-pass filter is dynamically adjusted according to the inherent vibration characteristics of the conveyor belt material; The target resonance frequency band corresponding to the current production batch is matched from the multiple sub-band energy distribution sequences, and the target resonance frequency band is obtained through learning of historical vibration data; Integral operation is performed on the energy distribution sequence in the target resonance frequency band to obtain an initial energy integral value, and the initial energy integral value is calibrated based on the real-time monitoring value of the conveyor belt running speed; The calibrated energy integral value is compared with a preset vibration safety threshold, and if the calibrated energy integral value exceeds the vibration safety threshold, a resonance frequency band re-calibration operation is triggered to update the target resonance frequency band. 4.The Internet of Things based tin box production status data analysis method according to claim 2, characterized in that, The feature extraction on the standardized monitoring data set includes: The normalized temperature interval in the standardized monitoring data set is subjected to spatial interpolation processing to generate a device surface thermal distribution map, and the maximum temperature gradient change direction and the high-temperature aggregation area coordinates are extracted from the device surface thermal distribution map; by setting a temperature threshold, the area with a device surface temperature higher than the threshold is marked as a high-temperature area; then, through clustering analysis method, adjacent high-temperature points are merged into a high-temperature aggregation area; The standard pressure fluctuation sequence is subjected to time-frequency joint analysis to extract the slope feature of the pressure rising stage, the stability index of the pressure maintaining stage and the decay rate of the pressure release stage, and a pressure fluctuation mode coding vector is constructed; The standard vibration intensity index is subjected to wavelet packet decomposition to extract the low-frequency energy proportion related to mechanical looseness and the high-frequency energy mutation point number related to bearing wear, and a mechanical resonance frequency point feature matrix is generated; The standard humidity influence factor sequence is subjected to lag correlation analysis to calculate the time lag correlation coefficient between environmental humidity change and device temperature fluctuation, and the influence duration of humidity mutation event on temperature stability is identified; The temperature gradient maximum change direction, pressure fluctuation mode coding vector, mechanical resonance frequency point feature matrix and time lag correlation coefficient are integrated into the state feature set. 5.The Internet of Things based tin box production status data analysis method according to claim 4, characterized in that, The standard pressure fluctuation sequence is subjected to time-frequency joint analysis, and a slope feature of a pressure rising stage is extracted, including: identifying a starting pressure point, a peak pressure point and an ending pressure point in a pressure fluctuation period, and calculating a time interval from the starting pressure point to the peak pressure point as a rising stage duration; linear fitting is performed on pressure sampling points in the rising stage to obtain an initial slope estimation value, and a root mean square error of fitting residual is calculated to evaluate linearity; when the root mean square error exceeds a preset linearity threshold, a dynamic slope change curve of the pressure rising stage is recalculated by using a piecewise polynomial fitting method; the maximum slope value, the average slope value and the slope coefficient of variation of the dynamic slope change curve are extracted, and the three are combined into the slope feature. 6.The Internet-of-Things based tinplate box production status data analysis method according to claim 4, characterized in that, The dynamic similarity degree matching obtains a thermal anomaly index, a pressure pattern deviation score, a mechanical state degradation index and a humidity interference intensity change rate, including: calculating a direction angle cosine value between the temperature gradient maximum change direction and a historical thermal distribution benchmark as the thermal anomaly index; performing dynamic time warping matching on the pressure fluctuation mode coding vector and a pressure fluctuation mode benchmark to obtain the pressure pattern deviation score; performing singular value decomposition on the mechanical resonance frequency point feature matrix and a mechanical resonance frequency point benchmark to extract a maximum singular value ratio as the mechanical state degradation index; performing sliding window correlation analysis on the time lag correlation coefficient and a humidity influence benchmark to calculate the humidity interference intensity change rate.

7. The Internet of Things-based tin box production status data analysis method according to claim 6, characterized by, The direction angle cosine value between the temperature gradient maximum change direction and the historical thermal distribution benchmark is calculated as the thermal anomaly index, including: the temperature gradient maximum change direction of the current production period is represented as a three-dimensional direction vector, wherein the three dimensions correspond to the length direction, the width direction and the height direction of the equipment respectively; a standard temperature gradient direction vector of the same production batch is extracted from the historical thermal distribution benchmark; the dot product of the three-dimensional direction vector and the standard temperature gradient direction vector is calculated, and divided by the product of the lengths of the two, to obtain the direction angle cosine value; when the direction angle cosine value is lower than a preset direction consistency threshold, triggering a thermal distribution abnormality alarm and recording the abnormal region coordinates; the weight coefficient of the thermal anomaly index is dynamically adjusted according to the difference between the direction angle cosine value and the direction consistency threshold. 8.The Internet-of-Things based tinplate box production status data analysis method according to claim 1, characterized in that, When the thermal anomaly index continuously exceeds a preset temperature rise threshold, a cooling system optimization strategy is generated, including a fan speed increase gradient and a coolant flow adjustment curve, including: determining a basic speed increase amount according to the difference between the thermal anomaly index and the temperature rise threshold, and calculating a dynamic compensation coefficient based on the current load rate of the equipment; multiplying the basic speed increase amount and the dynamic compensation coefficient to obtain an actual speed increase gradient, and limiting the speed increase speed to be not more than the maximum acceleration of the motor; determining a coolant preferential supply area according to the temperature gradient maximum change direction, and generating a partitioned flow control instruction; The coolant demand density distribution is calculated based on the high-temperature aggregation area coordinates, and the coolant flow adjustment curve is fitted; The coordination control timing of the rotation speed promotion gradient and the flow adjustment curve is set to ensure the phase synchronization of the fan acceleration and the coolant supply increment. 9.The Internet-of-Things based production status data analysis method of tinplate box according to claim 1, characterized in that, The production parameter optimization strategy set is issued to the equipment control terminal to execute real-time regulation and control instructions, including: The stamping pressure down-regulation strategy is converted into a hydraulic system control signal, which contains pressure sensor calibration parameters and servo valve opening adjustment step size; The vibration suppression strategy is converted into a mechanical transmission system control instruction, which contains a tension roller displacement set value and an automatic lubrication device trigger condition; The cooling system optimization strategy is converted into a PID parameter adjustment instruction of a temperature control module, which contains a proportional coefficient correction amount and an integral time constant update value; The humidity compensation strategy is converted into an environmental regulation system operation sequence, which contains a gas valve opening and closing timing diagram and a desiccant feeding mechanism trigger frequency; The control signals, control instructions, adjustment instructions, and operation sequences are packaged into real-time data packets through an industrial bus protocol, issued to each equipment control terminal according to a pre-set priority queue, and receive execution state feedback for strategy effectiveness verification.

Images