A device flow data monitoring system and method using an automatic plastic packaging system

Abstract

The application provides a device process data monitoring system and method using an automatic plastic packaging system, and relates to the technical field of data processing.The system comprises a time sequence analysis module, which is used for collecting real-time running data of a target device in the automatic plastic packaging system and performing runtime sequence analysis; a stage division module, which is used for dividing multidimensional time sequence data flow according to multiple plastic packaging process stages; a quality evaluation module, which is used for traversing multiple stage data to perform real-time state monitoring and plastic packaging quality evaluation and determine plastic packaging process deviation results; and a backtracking optimization module, which is used for backtracking adjustment and feeding back multi-stage process optimization data to a control terminal. The application can solve the technical problem that the automatic plastic packaging system has insufficient segmentation control granularity, it is difficult to perform full-stage fine control, and the plastic packaging quality fluctuates greatly in the prior art. Through quality evaluation and deviation adjustment of each process stage, the plastic packaging quality and production efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a device process data monitoring system and method applied to an automatic plastic packaging system. BACKGROUND

[0002] The automatic plastic packaging system is used for packaging articles. The existing automatic plastic packaging system often relies on fixed parameter settings and limited real-time feedback mechanisms, and lacks sufficient fine-grained control during the production process, especially in terms of adjustment and coordination between various process stages. Due to the lack of fine-grained segmented control and dynamic adjustment mechanisms, when facing some complex or subtle process problems, effective fine control cannot be performed, and inconsistent packaging quality often occurs, resulting in reduced production efficiency and rising costs.

[0003] In summary, the prior art has the technical problem of large plastic packaging quality fluctuations due to insufficient segmented control granularity of the automatic plastic packaging system, which makes it difficult to perform fine-grained control throughout the stages. SUMMARY

[0004] The purpose of the present application is to provide a device process data monitoring system and method applied to an automatic plastic packaging system to solve the technical problem of large plastic packaging quality fluctuations due to insufficient segmented control granularity of the automatic plastic packaging system, which makes it difficult to perform fine-grained control throughout the stages.

[0005] In view of the above problems, the present application provides a device process data monitoring system and method applied to an automatic plastic packaging system.

[0006] In a first aspect, the present application provides a device process data monitoring system applied to an automatic plastic packaging system, comprising: a time sequence analysis module for collecting real-time running data of a target device in the automatic plastic packaging system, performing runtime sequence analysis on the real-time running data based on a plastic packaging process flow, and obtaining a multi-dimensional time sequence data stream; a stage division module for dividing the multi-dimensional time sequence data stream according to a plurality of plastic packaging process stages to determine a plurality of stage data; a quality evaluation module for performing real-time state monitoring by traversing the plurality of stage data, generating a process state index set, performing plastic packaging quality evaluation according to the process state index set, and determining a plastic packaging process deviation result; a backtracking optimization module for backtracking adjustment of the plurality of stage data according to the plastic packaging process deviation result, constructing multi-stage process optimization data of the target device, and feeding back the multi-stage process optimization data to a control terminal of the automatic plastic packaging system.

[0007] Optionally, a synchronous acquisition unit is configured to acquire synchronous data of a target device by deploying multiple types of sensors in an automatic plastic packaging system, to obtain multiple sets of real-time running data; a data labeling unit is configured to label the multiple sets of real-time running data according to the multiple types of sensors, to generate data type labels; a data alignment unit is configured to extract a process timestamp based on a plastic packaging process time sequence, to align the multiple sets of real-time running data according to the process timestamp, to generate an original time sequence data set; and a data fusion unit is configured to match the original time sequence data set with the data type labels, to generate a multi-dimensional signal, to map the multi-dimensional signal to a same time reference for fusion, and to determine the multi-dimensional time sequence data stream.

[0008] Optionally, a signal processing channel matching subunit is configured to match a signal processing channel based on the data type labels, to determine a temperature space difference channel, a pressure waveform filtering channel, a trajectory smoothing channel, and a frequency spectrum extraction channel; a temperature analysis subunit is configured to map the original time sequence data set to the temperature space difference channel to perform full-mold temperature analysis on the target device, to construct a temperature gradient matrix; a pressure filtering subunit is configured to map the original time sequence data set to the pressure waveform filtering channel to perform pressure filtering on the target device, to obtain pressure distribution waveform data; a displacement measurement subunit is configured to map the original time sequence data set to the trajectory smoothing channel to perform displacement measurement on the target device, to obtain a displacement trajectory vector; a current energy calculation subunit is configured to map the original time sequence data set to the frequency spectrum extraction channel to perform current energy calculation on the target device, to obtain current frequency spectrum data; and a reference fusion subunit is configured to map the temperature gradient matrix, the pressure distribution waveform data, the displacement trajectory vector, and the current frequency spectrum data to a same time reference for fusion, to determine the multi-dimensional time sequence data stream.

[0009] Optionally, a dynamic segmentation unit is configured to load plastic packaging process full-process parameters, to dynamically segment the plastic packaging process full-process parameters, to obtain multiple plastic packaging process stages; a dynamic slicing unit is configured to map the multi-dimensional time sequence data stream to the multiple plastic packaging process stages for dynamic slicing, to generate multiple continuous subsequence stage data; a time sequence analysis unit is configured to perform running time sequence analysis based on the multiple continuous subsequence stage data, to generate a process stage running state feature vector set; a stage boundary detection unit is configured to perform stage boundary detection based on the process stage running state feature vector set, to determine multiple stage labels; and a label synchronization unit is configured to synchronize the multiple stage labels to the multi-dimensional time sequence data stream, to determine the multiple stage data.

[0010] Optionally, a first slicing subunit is configured to map the temperature gradient matrix to the multiple plastic packaging process stages to dynamically slice the multi-dimensional time sequence data stream, to generate first sub-sequence stage data; a second slicing subunit is configured to map the pressure distribution waveform data to the multiple plastic packaging process stages to dynamically slice the multi-dimensional time sequence data stream, to generate second sub-sequence stage data; a third slicing subunit is configured to map the displacement trajectory vector to the multiple plastic packaging process stages to dynamically slice the multi-dimensional time sequence data stream, to generate third sub-sequence stage data; a fourth slicing subunit is configured to map the current spectrum data to the multiple plastic packaging process stages to dynamically slice the multi-dimensional time sequence data stream, to generate fourth sub-sequence stage data; and a stage arrangement subunit is configured to arrange the first sub-sequence stage data, the second sub-sequence stage data, the third sub-sequence stage data, and the fourth sub-sequence stage data according to the plastic packaging process whole-process parameters, to generate the multiple continuous sub-sequence stage data.

[0011] Optionally, a parallel analysis unit is configured to traverse the multiple stage data to perform parallel analysis, to extract features according to parallel results of stages, to obtain multiple stage parallel feature parameters; a stage state monitoring unit is configured to perform stage state monitoring based on the multiple stage parallel feature parameters in combination with the plastic packaging process whole-process parameters, to generate a process state index set; a state reasoning unit is configured to perform online reasoning on a plastic packaging quality of a target device according to the process state index set, to generate a plastic packaging quality score; a defect analysis unit is configured to perform defect analysis according to the plastic packaging quality score, to determine a defect root cause, to perform positioning according to the defect root cause, to obtain a plastic packaging defect source; and a deviation determination unit is configured to match the plastic packaging defect source with the process state index set, to obtain a deviation process type, to map the deviation process type with the process state index set, and to extract the plastic packaging process deviation result.

[0012] Optionally, a cross-stage influence analysis subunit is configured to perform cross-stage influence analysis based on the multiple stage parallel feature parameters, to obtain multiple influence factors; an influence factor fusion subunit is configured to fuse the multiple influence factors according to the plastic packaging process whole-process parameters, to obtain a fused influence parameter; a time dimension analysis subunit is configured to call device spatial dimension convolution data of an automatic plastic packaging system, to perform time dimension analysis on the device of the automatic plastic packaging system according to the fused influence parameter, to determine time dimension convolution data; a stage state monitoring subunit is configured to construct a space-time convolution network according to the spatial dimension convolution data and the time dimension convolution data, to perform stage state monitoring through the space-time convolution network, to obtain multiple stage process state features; and a cross-stage correlation subunit is configured to correlate the multiple stage process state features across stages, to set the process state index set according to a correlation result.

[0013] Optionally, a defect mapping subunit is configured to map the plastic packaging defect source based on the process state indicator set to define a defect process mapping matrix; a gradient updating subunit is configured to perform gradient dynamic updating according to the defect process mapping matrix to construct a plastic packaging defect knowledge base; a defect retrieval subunit is configured to retrieve the process state indicator set as an index to the plastic packaging defect knowledge base for retrieval, perform distributed calculation according to the retrieval result to obtain process defect probability distribution data; a cluster analysis subunit is configured to perform cluster analysis according to the process defect probability distribution data according to the plastic packaging process type to obtain a deviation process type; and a reverse feature mapping subunit is configured to perform reverse feature mapping on the process state indicator set according to the deviation process type to determine the plastic packaging process deviation result.

[0014] Optionally, a stage matching unit is configured to perform stage matching based on the plastic packaging process deviation result to determine target associated stage information; a single-stage optimization unit is configured to activate a backtracking module to backtrack multiple stage data to locate a target stage for single-stage optimization to obtain single-stage process optimization data when the target associated stage information is single-stage; and a cross-stage collaborative optimization unit is configured to activate the backtracking module to backtrack the multiple stage data to locate multiple target stages for cross-stage collaborative optimization to obtain multi-stage process optimization data when the target associated stage information is multi-stage.

[0015] In a second aspect, the present application further provides a device process data monitoring method for an automatic plastic packaging system, comprising: collecting real-time running data of a target device in the automatic plastic packaging system, performing runtime sequence analysis on the real-time running data based on a plastic packaging process flow to obtain multi-dimensional time sequence data flow; dividing the multi-dimensional time sequence data flow according to multiple plastic packaging process stages to determine multiple stage data; performing real-time state monitoring by traversing the multiple stage data to generate a process state indicator set, performing plastic packaging quality evaluation according to the process state indicator set to determine a plastic packaging process deviation result; performing backtracking adjustment on the multiple stage data according to the plastic packaging process deviation result to construct multi-stage process optimization data of the target device, and feeding back the multi-stage process optimization data to a control terminal of the automatic plastic packaging system.

[0016] The one or more technical solutions provided in the application have at least the following beneficial effects: the timing analysis module is used to collect real-time running data of a target device in an automatic plastic packaging system, perform runtime timing analysis on the real-time running data based on a plastic packaging process flow, and obtain a multi-dimensional timing data stream; the stage division module is used to divide the multi-dimensional timing data stream according to a plurality of plastic packaging process stages, and determine a plurality of stage data; the quality evaluation module is used to perform real-time state monitoring on the plurality of stage data, generate a process state index set, perform plastic packaging quality evaluation according to the process state index set, and determine a plastic packaging process deviation result; the backtracking optimization module is used to backtrack and adjust the plurality of stage data according to the plastic packaging process deviation result, construct multi-stage flow optimization data of the target device, and feed back the multi-stage flow optimization data to a control terminal of the automatic plastic packaging system. That is, by collecting running data of a target device, dividing the data according to a process flow, obtaining stage data, performing plastic packaging quality evaluation on real-time state indicators, determining a process deviation result, and backtracking and adjusting a plurality of stage data, the plastic packaging quality and production efficiency are improved.

[0017] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the application, nor is it used to limit the scope of the application. Other features of the application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 FIG. 1 is a structural schematic diagram of a device flow data monitoring system applied to an automatic plastic packaging system according to an embodiment of the application.

[0019] Figure 2 FIG. 2 is a flowchart of a device flow data monitoring method applied to an automatic plastic packaging system according to an embodiment of the application.

[0020] Reference signs: timing analysis module 11, stage division module 12, quality evaluation module 13, backtracking optimization module 14. DETAILED DESCRIPTION

[0021] The application provides a device flow data monitoring system and method applied to an automatic plastic packaging system, which solves the technical problem of large plastic packaging quality fluctuation due to insufficient segmentation control granularity of the automatic plastic packaging system and difficulty in performing full-stage fine control in the prior art. By collecting running data of a target device, dividing the data according to a process flow, obtaining stage data, performing plastic packaging quality evaluation on real-time state indicators, determining a process deviation result, and backtracking and adjusting a plurality of stage data, the plastic packaging quality and production efficiency are improved.

[0022] Below, the technical solutions in the present application will be described clearly and completely with reference to the drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application. In addition, it should be noted that, for the convenience of description, only parts related to the present application are shown in the drawings, not all.

[0023] Embodiment one, please refer to the attached Figure 1 The present application provides a device process data monitoring system applied to an automatic plastic packaging system, wherein the device process data monitoring system applied to the automatic plastic packaging system is used to implement the steps of a device process data monitoring method applied to the automatic plastic packaging system, and the device process data monitoring system applied to the automatic plastic packaging system comprises:

[0024] a time sequence analysis module 11, configured to collect real-time running data of a target device in the automatic plastic packaging system, perform runtime sequence analysis on the real-time running data based on a plastic packaging process, and obtain a multi-dimensional time sequence data stream.

[0025] Further, the time sequence analysis module 11 in the device process data monitoring system applied to the automatic plastic packaging system is further configured to: a synchronous collection unit, configured to perform synchronous collection of the target device by laying multiple types of sensors in the automatic plastic packaging system, and obtain multiple groups of real-time running data; a data labeling unit, configured to label the multiple groups of real-time running data according to multiple types of sensors, and generate data type labels; a data alignment unit, configured to extract a process timestamp based on a plastic packaging process time sequence, align the multiple groups of real-time running data according to the process timestamp, and generate an original time sequence data set; and a data fusion unit, configured to match the original time sequence data set with the data type labels, generate a multi-dimensional signal, map the multi-dimensional signal to the same time reference, and fuse the multi-dimensional signal to determine the multi-dimensional time sequence data stream.

[0026] Further, the timing analysis module 11 in the device process data monitoring system of the automatic plastic packaging system is also used for: a signal processing channel matching subunit for matching signal processing channels based on the data type tags, determining a temperature space difference channel, a pressure waveform filtering channel, a trajectory smoothing channel, and a spectrum extraction channel; a temperature analysis subunit for mapping the original timing data set to the temperature space difference channel to perform full-mold temperature analysis on the target device and constructing a temperature gradient matrix; a pressure filtering subunit for mapping the original timing data set to the pressure waveform filtering channel to perform pressure filtering on the target device and obtaining pressure distribution waveform data; a displacement ranging subunit for mapping the original timing data set to the trajectory smoothing channel to perform displacement ranging on the target device and obtaining a displacement trajectory vector; a current energy calculation subunit for mapping the original timing data set to the spectrum extraction channel to perform current energy calculation on the target device and obtaining current spectrum data; and a reference fusion subunit for mapping the temperature gradient matrix, the pressure distribution waveform data, the displacement trajectory vector, and the current spectrum data to the same time reference for fusion to determine the multi-dimensional timing data stream.

[0027] Specifically, the automatic plastic packaging system is used to package and heat seal articles through a plastic film. The packaging process is completed by controlling multiple links such as preheating, mold closing and pressurizing, pressure holding and curing, mold opening and unloading, etc. In the automatic plastic packaging system, multiple types of sensors are arranged to realize multi-dimensional and synchronous data acquisition of the target device. For example, temperature sensors are used to measure temperature, pressure sensors are used to measure pressure, and speed sensors are used to measure the speed of the conveying belt, etc. All sensors need to be synchronously acquired, i.e., their sampling times must be consistent to ensure that the data collected by all sensors at the same time have the same time stamp.

[0028] Through synchronous acquisition of the target device by multiple types of sensors, multiple sets of real-time running data are obtained, i.e., including device running data collected by multiple types of sensors. According to the multiple types of sensors, the multiple sets of real-time running data are marked by data type, i.e., each set of collected data is labeled to indicate its type (such as temperature data, pressure data, speed data, etc.).

[0029] According to the plastic packaging process timing, the process time stamp is extracted. The plastic packaging process timing refers to the time sequence and time point of each process link in the plastic packaging process. Different stages (such as preheating, pressurizing, pressure holding and curing, mold opening and unloading) occur at different times, and timing extraction is to obtain the time stamp according to the time nodes of these stages. The process time stamp is used to mark the specific time point of each step in the plastic packaging process. According to each process step (such as preheating, pressurizing, pressure holding and curing, mold opening and unloading, etc.) in the plastic packaging process, the time stamp of each process is extracted, i.e., the start and end time of each process stage is marked.

[0030] Aligning multiple sets of real-time operation data according to process timestamps forms the original time series dataset, that is, sorting all collected data according to the time sequence of the process flow forms a preliminary dataset. For example, assuming that a certain timestamp represents the moment when sealing starts, all data collected before or after this time point (such as temperature, pressure, speed, etc.) will be aligned according to this timestamp, ensuring that these data represent the state of the same time period.

[0031] Matching the original time series dataset and the data type label generates a multi-dimensional signal. Each dimension represents a different type of data (for example, temperature, pressure, speed). The multi-dimensional signal is fused according to a unified time reference, that is, all multi-dimensional signals are mapped to the same time axis, ensuring that the data of different sensors are compared at the same time point. The multi-dimensional signal after time reference alignment and fusion will form a multi-dimensional time series data stream containing all key data (such as temperature, pressure, speed, etc.) in the entire plastic packaging process, and each data point has an accurate timestamp, which can fully reflect each link in the production process.

[0032] According to the type of each sensor data and the required processing method, the data type label is matched to the corresponding signal processing channel. Temperature data will enter the temperature space difference channel, pressure data will enter the pressure waveform filtering channel, displacement data will enter the trajectory smoothing channel, and current signal will enter the spectrum extraction channel. The temperature space difference channel is used for spatial difference calculation of the temperature data of the device; the pressure waveform filtering channel is used for waveform filtering processing of the collected pressure data, removing noise or high frequency components, and extracting effective pressure change signals; the trajectory smoothing channel is used for smoothing processing of displacement data, removing oscillations caused by noise or errors, and generating a smoother displacement trajectory; the spectrum extraction channel is used for converting current signals to the frequency domain, extracting the frequency spectrum data of the current signal, and revealing the frequency components of the current.

[0033] Mapping the original time series dataset to the temperature spatial difference channel, i.e. inputting the temperature data in the original time series dataset to the temperature spatial difference channel, performing full mold temperature analysis on the target device, which refers to detailed study on the temperature distribution inside the mold to understand how the temperature changes in different parts of the mold. Through the temperature spatial difference channel, the temperature of the entire target device (mold) is comprehensively analyzed. For example, if there are temperature sensors placed in the four corners on the surface of the device, the collected temperature data are 50℃, 55℃, 58℃ and 60℃ respectively. Through difference calculation, the temperature distribution of other areas on the surface of the device is inferred, and a complete temperature map is obtained. The temperature difference between adjacent positions in the mold is calculated, such as the difference between the second element of the first row and the first element of the first row, and the difference between the third element of the first row and the second element of the first row, and so on, to obtain a temperature gradient matrix. The temperature gradient matrix is a data matrix representing the distribution of temperature changes, which is usually used to describe the temperature changes of the device at different positions or time points.

[0034] Mapping the original time series dataset to the pressure waveform filtering channel, i.e. inputting the pressure data in the original time series dataset to the pressure waveform filtering channel for pressure filtering. The original pressure data may include some high-frequency noise or transient fluctuations, and the pressure waveform filtering channel is used to process the pressure data, filter the pressure data, and remove the noise or high-frequency components in the data. The filtering operation can extract a more stable and accurate pressure change trend, and remove short-term abnormal fluctuations caused by device vibration, electromagnetic interference or other factors.

[0035] In the pressure waveform filtering channel, the pressure data is processed by a filtering algorithm, such as a low-pass filter, to remove high-frequency noise in the pressure data and make the data smoother. For example, the waveform of the original pressure data may have sharp fluctuations (such as ±500 Pa) at certain times due to instability of the sensor or external interference. After processing by the filtering algorithm, the fluctuation amplitude of the data is significantly reduced (such as ±100 Pa), making the data more stable. The filtered pressure data will become pressure distribution waveform data, accurately reflecting the pressure change trend of the device at different time points, and not affected by external noise. For example, the pressure change of the mold monitored by multiple pressure sensors is as follows: the pressure at 1s is 3100 Pa, the pressure at 2s is 3400 Pa, the pressure at 3s is 3500 Pa, the pressure at 4s is 3700 Pa, and the pressure at 5s is 4000 Pa; after removing noise using a low-pass filter, the filtered pressure data is as follows: the pressure at 1s is 3120 Pa, the pressure at 2s is 3300 Pa, the pressure at 3s is 3380 Pa, the pressure at 4s is 3580 Pa, and the pressure at 5s is 3820 Pa. The pressure data after filtering is more stable, with a significantly reduced fluctuation amplitude, forming pressure distribution waveform data, ensuring the stability of the pressure during production.

[0036] The pressure distribution waveform data refers to the curve of the pressure of each part of the device changing with time, i.e., the pressure change curve data at different positions of the device during operation. After waveform filtering, more stable and denoised pressure data is obtained, which can clearly reflect the pressure change trend of the device during the entire operation process and reveal the non-uniformity, fluctuation amplitude, etc. of the pressure. The pressure distribution waveform data can reflect the change trend and distribution of the device pressure, helping to determine whether the pressure is uniform and whether there are abnormalities during the plastic packaging process.

[0037] The original time series data set is mapped to the trajectory smoothing channel, i.e., the original displacement data (such as displacement coordinates or speed of the device at different time points) collected by the original time series data set is transmitted to the trajectory smoothing channel for displacement measurement, and the motion trajectory of the device is tracked by measuring the position change of the device at different time points. Trajectory smoothing is a signal processing technique used to reduce noise and unnecessary fluctuations in displacement data. For example, the position information of the device at different time points is obtained by a displacement sensor, and the displacement data is as follows: (0, 0), (0.1, 0.2), (0.2, 0.5), (0.5, 0.8). After trajectory smoothing, it becomes (0, 0), (0.09, 0.19), (0.19, 0.48), (0.48, 0.78), the fluctuation of the displacement data is reduced, and a more accurate trajectory vector is obtained. In the trajectory smoothing channel, a smoothing algorithm is applied to process the original displacement data, which smoothes the data by calculating the average value near the data points, thereby reducing short-term fluctuations.

[0038] Displacement measurement refers to measuring the displacement of a device from one position to another, usually expressed as distance or coordinate change. In an automatic plastic packaging system, displacement measurement is used to track the movement state of the device and evaluate its operation accuracy. The displacement trajectory vector is a vector containing information about the position change of the device, representing the motion trajectory of the device within a certain time, usually composed of time series of displacement data, each data point representing the position or motion direction of the device at a certain time. The displacement trajectory vector after smoothing processing can more accurately reflect the motion trajectory of the device, removing noise interference. The displacement trajectory vector represents the position change of the device or workpiece at each time during the motion process, used to describe the motion trajectory of the workpiece, helping to adjust the motion control strategy of the device to ensure the accuracy of the plastic packaging process.

[0039] Mapping the original time series dataset into the spectrum extraction channel, i.e. transmitting the original time series current data in the original time series dataset to the spectrum extraction channel. The original time series current data is a signal of current changing with time, which may have certain fluctuations or noises. Through the spectrum extraction channel, the current signal is converted from time domain to frequency domain, revealing the frequency components of the current. By applying Fourier transform, the current signal in time domain is converted to frequency domain signal. Spectrum analysis distributes the amplitude of current signal in different frequency ranges, revealing various frequency components contained in the current. After converting the current signal to spectrum data, the power components in different frequency ranges are identified.

[0040] After Fourier transform, a complex array called frequency domain data (i.e. current spectrum data) is obtained, which contains the amplitude and phase information of each frequency point. By taking the absolute value, the amplitude of each frequency point is obtained, representing the energy of the signal at that frequency. For example, suppose the spectrum data obtained after Fourier transform is as follows: the amplitude at frequency 0 is 2.8, the amplitude at frequency 1 Hz is 1.0, and the amplitude at frequency 2 Hz is 0.5, indicating the energy distribution of the device current signal at each frequency. After obtaining the current spectrum data, current energy calculation is performed. The energy of current is usually calculated by squaring the spectrum amplitude. For each frequency point, the energy is the square of the amplitude. Current spectrum data is the frequency domain representation of current signal, reflecting the intensity or amplitude of current signal at different frequency components, obtaining the energy distribution of current signal at each frequency.

[0041] The current spectrum data is used to evaluate the load condition of the equipment. For example, if the amplitude of a certain frequency point is large, it means that the equipment has large energy consumption at this frequency, which may be heavy load. The current spectrum data is the frequency domain representation of the current signal obtained by Fourier transform, which reflects the energy distribution of the current at different frequency components, and can reveal the running state, load condition and energy efficiency level of the equipment. By analyzing the current spectrum data, it can be judged whether the equipment has abnormal load or energy efficiency problem.

[0042] For the temperature gradient matrix, pressure distribution waveform data, displacement trajectory vector, and current spectrum data, the collection time points may not be completely consistent, so it is necessary to align these data to the same time reference and use timestamps to synchronize the time information of different data streams. For example, assuming that the collection time intervals of temperature, pressure, displacement, and current data are 1s, 1s, 1s, and 2s respectively, the data with different timestamps are aligned by interpolation method (such as linear interpolation or spline interpolation) to ensure their timestamps are consistent. For temperature data, the temperature at each time point will be interpolated according to the corresponding timestamp to synchronize it with other data streams; similarly, for current spectrum data, if the collection frequency is 2s and other data is 1s, the current data will be interpolated to each 1s time point to ensure time alignment.

[0043] After the timestamps of all data streams are aligned, data fusion is performed to integrate the data in each dimension (such as temperature, pressure, displacement, current, etc.) into a multi-dimensional data stream in chronological order. For example, at time 5s, the temperature is 55℃, the pressure is 3740Pa, the displacement is 2.8mm, and the amplitude of the current spectrum is 2A. By fusing the data at each time point, a multi-dimensional time series data stream is obtained, which contains multiple dimensions of data of the equipment during the entire running process, such as temperature, pressure, displacement, and current spectrum, etc.

[0044] By performing runtime sequence analysis on the real-time running data of the target equipment through the plastic packaging process, a multi-dimensional time series data stream is obtained, which obtains comprehensive and multi-dimensional real-time running data of the target equipment in the automatic plastic packaging system, and comprehensively reflects the performance of the equipment during the running process. By processing different types of data through signal processing channels respectively, the processing accuracy of each data type is ensured, and errors caused by data mismatch are avoided. The multi-dimensional time series data stream can simultaneously display the temperature, pressure, displacement and current of the equipment, and help to comprehensively monitor the running state of the target equipment.

[0045] The stage division module 12 is used to divide the multi-dimensional time series data stream according to a plurality of plastic packaging process stages, and determine a plurality of stage data.

[0046] Further, the stage division module 12 in the equipment process data monitoring system for automatic plastic packaging system is also used for: a dynamic segmentation unit for loading plastic packaging process whole process parameters, dynamically segmenting the plastic packaging process whole process parameters, and obtaining a plurality of plastic packaging process stages; a dynamic slicing unit for mapping the multi-dimensional time sequence data stream to the plurality of plastic packaging process stages for dynamic slicing, generating a plurality of continuous sub-sequence stage data; a time sequence analysis unit for performing runtime sequence analysis based on the plurality of continuous sub-sequence stage data, generating a process stage running state feature vector set; a stage boundary detection unit for performing stage boundary detection based on the process stage running state feature vector set, determining a plurality of stage labels; and a label synchronization unit for synchronously identifying the plurality of stage labels to the multi-dimensional time sequence data stream, determining the plurality of stage data.

[0047] Further, the stage division module 12 in the equipment process data monitoring system for automatic plastic packaging system is also used for: a first slicing sub-unit for mapping the temperature gradient matrix to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream, generating first sub-sequence stage data; a second slicing sub-unit for mapping the pressure distribution waveform data to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream, generating second sub-sequence stage data; a third slicing sub-unit for mapping the displacement trajectory vector to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream, generating third sub-sequence stage data; a fourth slicing sub-unit for mapping the current spectrum data to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream, generating fourth sub-sequence stage data; and a stage arrangement sub-unit for arranging the first sub-sequence stage data, the second sub-sequence stage data, the third sub-sequence stage data, and the fourth sub-sequence stage data according to the plastic packaging process whole process parameters, generating the plurality of continuous sub-sequence stage data.

[0048] Specifically, the plastic packaging process whole process parameters refer to process parameters involved in each stage of the whole plastic packaging process, including temperature, pressure, displacement, time, etc., which will be different in different process stages. The parameters of the whole plastic packaging process are loaded to determine the requirements of different process stages. According to the actual running conditions or requirements, the whole process parameters of the plastic packaging process are dynamically segmented, and the whole process is divided into a plurality of plastic packaging process stages. The dynamic segmentation process is based on the change of real-time data, and the whole process is divided into multiple stages according to the changes of temperature, pressure, etc. The plastic packaging process stage refers to different stages in the plastic packaging process according to time sequence or process requirements. Each stage has specific targets and parameters, including a preheating stage, a mold closing and pressurizing stage, a pressure holding and curing stage, and a mold opening and unloading stage.

[0049] The multi-dimensional time series data stream is mapped to multiple encapsulation process stages for dynamic slicing to obtain multiple continuous sub-sequence stage data, i.e., continuous data segments extracted from the multi-dimensional time series data stream and divided according to the encapsulation process stages. Specifically, the temperature data in the temperature gradient matrix is cut according to different encapsulation process stages, and the temperature data of each stage is extracted to obtain first sub-sequence stage data. Similarly, the pressure distribution waveform data is mapped to multiple encapsulation process stages for dynamic slicing, and the pressure waveform data corresponding to each stage is extracted to obtain second sub-sequence stage data. The displacement trajectory vector is mapped to multiple encapsulation process stages for dynamic slicing, and the equipment displacement corresponding to each stage is extracted to obtain third sub-sequence stage data. The current spectrum data is mapped to multiple encapsulation process stages for dynamic slicing, and the current spectrum data corresponding to each stage is extracted to obtain fourth sub-sequence stage data. All generated sub-sequence stage data (temperature, pressure, displacement, current, etc.) are arranged according to the encapsulation process full-process parameters, and based on the encapsulation process full-process parameters (such as sequence, time), the data of each stage is arranged in the order of the process to form a complete data set containing data of all process stages. In other words, the temperature, pressure, displacement, and current of each stage in the first sub-sequence stage data, the second sub-sequence stage data, the third sub-sequence stage data, and the fourth sub-sequence stage data are arranged to obtain multiple continuous sub-sequence stage data, ensuring that the data of each stage is effectively integrated together. Each sub-sequence represents the data of a specific process stage, such as temperature, pressure, displacement, and current, etc.

[0050] The multiple continuous sub-sequence stage data is subjected to run-time sequence analysis to extract useful features or information from the time series data for evaluating the state of each process stage and identifying potential abnormalities and optimization directions. By performing time sequence analysis on each sub-sequence stage data, useful feature information is extracted from the data of each process stage, including temperature change rate, pressure fluctuation amplitude, displacement stability, and current energy change. The process stage running state feature vector set refers to the set of feature vectors extracted from the multiple continuous sub-sequence stage data, representing certain key information of each process stage.

[0051] The analysis process phase running state feature vector set is analyzed, the starting and ending positions of different process phases are automatically identified, the boundaries of each process phase are determined, so as to accurately distinguish the data of each phase. For example, when the parameter change in the process is relatively direct and easy to identify, according to the set threshold, if the feature vector (such as temperature, pressure, etc.) changes greatly, a new phase can be identified as starting. When the data change is gradual rather than sudden, the state transition of the process is determined by analyzing the change trend of the feature vector. When the process includes multiple variables, and the change of these variables includes not only sudden change but also gradual trend change, the clustering algorithm (such as K-means clustering) is used to automatically identify the boundaries of the process phase by monitoring the change of the feature vector.

[0052] Based on the results of the phase boundary detection, a phase label is assigned to each process phase, i.e. a preheating phase label, a mold closing and pressing phase label, a pressure holding and curing phase label, and a mold opening and unloading phase label, to obtain multiple phase labels. The multiple phase labels are synchronized and identified to the multi-dimensional time series data stream, and the phase labels of each process phase are synchronized and identified to the corresponding multi-dimensional time series data stream, so as to realize the phase division of the data. For example, it is assumed that between time t1 and time t3, the data is identified as belonging to the preheating phase, and the data in this period of time is identified as the preheating phase label. Subsequently, if the temperature and pressure parameters change, the start of the mold closing phase is identified, and the data in this period of time is identified as the mold closing phase label. The data of each phase includes all sensor data from the start to the end of the phase. For example, the preheating phase data: temperature = [50℃, 55℃, 60℃], pressure = [3100Pa, 3150Pa, 3210Pa], displacement = [3.0mm, 3.1mm, 3.2mm], current = [1.0A, 1.1A, 1.2A]; mold closing phase data: temperature = [60℃, 62℃, 64℃], pressure = [3210MPa, 3275MPa, 3310MPa], displacement = [3.1mm, 3.2mm, 3.3mm], current = [1.1A, 1.2A, 1.3A].

[0053] By synchronizing the multi-dimensional time series data stream with the process phase label, the data of each process phase is clearly identified, and the data of each phase is independently processed and identified, which helps to avoid confusion of data in different phases and improves the efficiency of data management. According to the running state feature vector set of each phase, the optimization analysis between phases is carried out to help find the bottlenecks or problems in the production process and improve the efficiency of the process flow.

[0054] The quality evaluation module 13 is used to traverse the multiple phase data for real-time state monitoring, generate a process state index set, perform plastic packaging quality evaluation according to the process state index set, and determine a plastic packaging process deviation result.

[0055] Further, the quality evaluation module 13 in the equipment process data monitoring system applying the automatic plastic packaging system is further used for: a parallel analysis unit for traversing the multiple stage data for parallel analysis, performing feature extraction according to stage parallel results, and obtaining multiple stage parallel feature parameters; a stage state monitoring unit for performing stage state monitoring based on the multiple stage parallel feature parameters in combination with the plastic packaging process whole-process parameters, generating a process state index set; a state reasoning unit for performing online reasoning on the plastic packaging quality of a target equipment according to the process state index set, generating a plastic packaging quality score; a defect analysis unit for performing defect analysis according to the plastic packaging quality score, determining a defect root cause, positioning according to the defect root cause, and obtaining a plastic packaging defect source; and a deviation determination unit for matching the plastic packaging defect source with the process state index set, obtaining a deviation process type, mapping according to the deviation process type and the process state index set, and extracting the plastic packaging process deviation result.

[0056] Further, the quality evaluation module 13 in the equipment process data monitoring system applying the automatic plastic packaging system is further used for: a cross-stage influence analysis subunit for performing cross-stage influence analysis based on the multiple stage parallel feature parameters, obtaining multiple influence factors; an influence factor fusion subunit for fusing the multiple influence factors according to the plastic packaging process whole-process parameters, obtaining a fused influence parameter; a time dimension analysis subunit for calling equipment space dimension convolution data of an automatic plastic packaging system, performing time dimension analysis on the automatic plastic packaging system according to the fused influence parameter, and determining time dimension convolution data; a stage state monitoring subunit for constructing a space-time convolution network according to the space dimension convolution data and the time dimension convolution data, performing stage state monitoring through the space-time convolution network, and obtaining multiple stage process state features; and a cross-stage correlation subunit for correlating the multiple stage process state features across stages, and setting the process state index set according to a correlation result.

[0057] Further, the quality evaluation module 13 in the device flow data monitoring system of the automatic plastic packaging system is also used for: a defect mapping subunit, which is used for mapping based on the plastic packaging defect source in combination with the process state index set to define a defect process mapping matrix; a gradient updating subunit, which is used for gradient dynamic updating according to the defect process mapping matrix to construct a plastic packaging defect knowledge base; a defect retrieval subunit, which is used for synchronously retrieving the process state index set to the plastic packaging defect knowledge base as an index to perform distributed calculation according to a retrieval result to obtain process defect probability distribution data; a clustering analysis subunit, which is used for clustering analysis according to a plastic packaging process type based on the process defect probability distribution data to obtain a deviation process type; and a reverse feature mapping subunit, which is used for performing reverse feature mapping on the process state index set according to the deviation process type to determine the plastic packaging process deviation result.

[0058] Specifically, multiple stage data are traversed to perform parallel analysis, that is, independent and parallel analysis is performed on the data of each plastic packaging process stage to obtain stage parallel results. Parallel analysis refers to simultaneous analysis of the data of multiple stages, rather than individual analysis, which can improve processing efficiency, especially in the case of complex and large data.

[0059] By combining multiple stage parallel feature parameters and full-flow parameters to perform stage state monitoring, it is possible to evaluate in real time whether each stage is running according to the predetermined process requirements. Specifically, cross-stage influence analysis is performed on the multiple stage parallel feature parameters, that is, it is evaluated how the process parameters (such as temperature and pressure) of a stage affect other stages. For example, the temperature in the preheating stage can affect the pressure and clamping time in the clamping stage. Cross-stage influence analysis refers to analysis of the influence of the running state of one stage on other stages. Multiple influence factors refer to key factors identified in the cross-stage influence analysis, which can affect multiple stages in the plastic packaging process.

[0060] The multiple influence factors are combined with the plastic packaging process whole-process parameters to obtain a fusion influence parameter, and the comprehensive influence of multiple factors on the whole plastic packaging process is evaluated. The device spatial dimension convolution data of the automatic plastic packaging system is called, and the device spatial dimension convolution data is the data after convolution operation on each part of the device in space, which usually involves the running state of different physical parts (such as top, bottom, side, etc.) of the device in space, such as temperature, pressure, etc. Different running states may exist at different positions (such as top, bottom, side) of the device in space, so the state of each part of the device needs to be analyzed. Convolution operation can help extract feature information of each part in the spatial dimension of the device.

[0061] The fusion influence parameter is combined with the running state of the device to perform time dimension analysis on the device. The running data of the device changes with time, and the time dimension convolution data can extract the state change characteristics of the device in time to determine the change of the running state of the device at different time points, such as the evolution of temperature, pressure, etc. with time. Based on the fusion influence parameter, such as temperature, pressure, etc., dynamic change data of the device at different time points is obtained through convolution analysis in the time dimension.

[0062] The spatial dimension convolution data is obtained by convolution operation on the sensor data of the device at different physical positions, which reflects the state of each part (such as top, bottom, side, etc.) of the device in space, and is used to extract feature information of the device in space, such as temperature, pressure, displacement, etc. The time dimension convolution data is obtained by convolution processing of the device running data in the time dimension, which reflects the change of the device state with time, including the trend of temperature, pressure, etc. changing with time, and is used to identify the dynamic change of the device at different time points.

[0063] The spatial dimension convolution data and the time dimension convolution data are combined to construct a space-time convolution network. The space-time convolution network processes data in two dimensions of space and time through convolution operations to extract features of the device in space and time. The space-time convolution network processes data through a series of convolution layers to fuse spatial and temporal information and obtain rich space-time features. In the space-time convolution network, the processed data will be used for stage state monitoring. The space-time convolution network can monitor the running state of the device in different stages in real time and determine whether an abnormality occurs. By extracting the state features of each stage, potential problems in the operation of the device are identified. The space-time convolution network analyzes the data of each stage and evaluates the state of the stage according to the extracted spatial and temporal features. If the temperature fluctuation of a stage is too large or the pressure change is abnormal, the space-time convolution network will immediately detect it and mark it as an abnormal state. For example, in the preheating stage, the temperature of the device suddenly rises from 50°C to 100°C, and the rising rate exceeds the normal range. The space-time convolution network detects this change and marks it as abnormal, triggering an alarm.

[0064] Through the stage state monitoring of the space-time convolution network, the process state features of each process stage are extracted, and a plurality of stage process state features are obtained, reflecting the running state of the device in each stage, which can help analyze whether there is deviation or abnormality. The plurality of stage process state features are cross-stage correlated to analyze the mutual relationship between different process stages, and a process state index set is obtained. The process state features extracted from each process stage are analyzed to identify the mutual influence between different stages, i.e., how the running state of a stage affects the process performance of the subsequent stage. After cross-stage correlation analysis, according to the correlation results, a process state index set is constructed, which contains process state features from multiple stages and takes into account the mutual influence and dependence between stages. Based on the cross-stage correlation analysis, the weight and priority of each stage are set. For example, if the temperature in the preheating stage has a great influence on the pressure fluctuation in the mold closing stage, the weight of the temperature change influencing factor may be higher.

[0065] The process state index set is a key index extracted from the state features of multiple process stages, which is used to evaluate the state of the overall plastic packaging process and can reflect the health status of each stage. Through cross-stage influence analysis, the mutual relationship between different stages can be identified and quantified to help optimize the process flow and avoid adverse interactions between stages.

[0066] According to the process state index set, the plastic packaging quality of the target device is inferred online, that is, according to the current process state index set, rapid data analysis and inference are performed, the result is obtained immediately and fed back to the operator, the current running state of the device is evaluated in real time, it is judged whether it meets the process requirements, and immediate adjustment is made. By collecting and processing the process state data of each stage in real time, analyzing these data and making inferences, it is judged whether the current process state is normal and whether it meets the predetermined quality standard. First, real-time data (such as temperature, pressure, displacement, etc.) from each stage will be collected and processed, and these data will be matched with the process state index set. Using the online inference model, the differences between these data and historical data, standard thresholds are analyzed to determine whether the current process meets the standard. If some data in the process state index set is abnormal (for example, the temperature is too high or the pressure fluctuation is too large), the inference model will automatically identify the problem and provide feedback. For example, assuming that in the mold closing stage, the real-time collected pressure data is 4300Pa, and the predetermined standard is 3500Pa, the temperature rises too fast in the preheating stage (from 50℃ to 80℃), the online inference system will analyze the gap between these data and the standard value, infer that there may be a problem of insufficient pressure in the mold closing stage, and generate an alarm.

[0067] The plastic packaging quality score is a comprehensive evaluation result based on online inference, according to the state of each process stage and the feedback information obtained by online inference, a final quality score is calculated. The plastic packaging quality score is within a certain range (such as 0 to 100), the higher the value, the better the plastic packaging quality, the lower the value, the more serious the quality problem. Each factor is calculated according to its weight in the index set to obtain the corresponding score, and these scores are weighted and averaged to obtain a total plastic packaging quality score. For example, assuming that the scores of each stage are as follows: temperature 95 points, pressure 85 points, displacement 90 points, current frequency spectrum 80 points, the final plastic packaging quality score is 95*0.4+85*0.3+90*0.2+80*0.1=89.5. It shows that the current plastic packaging quality is close to the standard, and the quality is good.

[0068] The results of the plastic packaging quality score are analyzed to determine whether defect analysis is needed. If the quality score is low, indicating problems in the production process, the defect analysis module is activated to analyze each stage of the plastic packaging process (such as preheating, mold closing, pressure holding, etc.). By comparing the differences between real-time process data and standard values, possible quality deviations are found. For example, if the temperature fluctuation in the preheating stage exceeds the standard range and the pressure value in the mold closing stage is also lower than the standard requirement, these two stages may be the cause of the low plastic packaging quality score, and further analysis of the data in these stages is recommended. Defect analysis refers to analyzing process data and feedback information when the plastic packaging process quality score is low to find the root cause of the quality problem and the key factors affecting the quality of plastic packaging, thereby providing a basis for solving the problem. Defect root cause refers to the most fundamental cause of the plastic packaging process or final product quality problem, including process parameters (such as temperature, pressure) not meeting standards, equipment failure, raw material quality problems, etc. Defect root cause may be improper process control in a certain stage (such as too high temperature or too low pressure), or equipment failure or raw material quality problems.

[0069] According to the defect root cause, the defect source is located to determine the problem in which process stage or equipment. By combining data with device configuration, sensor location, etc., the specific defect source is accurately located, such as a specific temperature control device, pressure sensor or a specific operation link. For example, through sensor data, it is found that the temperature sensor in the preheating stage has a large error for a period of time, and the temperature exceeds the standard range, while other parts of the device are working normally. Through positioning, it is found that a component of the temperature control device has a fault, causing the temperature fluctuation to be too large, so it is inferred that the temperature control system is the defect source.

[0070] After completing the defect source positioning, the detailed information of the plastic packaging defect source is output, and improvement suggestions are provided. The defect source may be improper process control, equipment failure, material problem or personnel operation error. After the defect source is determined, the operator can take appropriate repair measures, such as adjusting process parameters, repairing equipment, checking raw material quality, etc.

[0071] By mapping the plastic packaging defect source and process state indicator set, a matrix structure, namely the defect process mapping matrix, is established to describe the relationship between different process parameters and defects. Each element of the defect process mapping matrix represents the mapping degree between a certain process state (such as temperature, pressure, displacement, etc.) and a defect source. Each defect source (such as excessive temperature, insufficient pressure, displacement deviation, etc.) is associated with a specific process state indicator (such as the readings of temperature sensors, pressure sensors). Based on new data, the defect process mapping matrix is adjusted by optimization algorithms such as gradient descent to reduce the error of model prediction. Through continuous training and updating, learn from new data, gradually improve the recognition accuracy of defect process. Gradient dynamic update is to continuously update the defect process mapping matrix through optimization algorithm to improve the accuracy of process deviation detection. Each time new process data or defect information is input, the matrix is adjusted by calculating the gradient to ensure the accuracy and real-time performance of the model. For example, if it is found that the pressure in the clamping stage in the historical data is lower than the standard value and the probability of causing defects is higher, the weight in the dynamic updating mapping matrix is adjusted to adjust the correlation between pressure and defects. Gradient dynamic update refers to the adjustment and optimization of the defect process mapping matrix, which is continuously updated and improved with the addition of new data. By dynamically adjusting the matrix, the recognition accuracy of different process defects is gradually improved. The plastic packaging defect knowledge base is a data warehouse that stores information about various defects in the plastic packaging process, including defect sources, process parameters, quality scores, and other information obtained from historical data and process feedback.

[0072] The process state indicator set is used as an index to search in the plastic packaging defect knowledge base, and the search results are obtained. According to the search results, the probability distribution data of defects under each process state is calculated, that is, the process defect probability distribution data. Process defect probability distribution data refers to the probability distribution of various possible defects under different process conditions. According to the type of plastic packaging process, the process defect probability distribution data is subjected to cluster analysis, and different process states are divided into several types, and the deviation degree of these process types is determined according to the defect probability distribution data. Through cluster analysis, the deviation process type under different process conditions is identified, so as to carry out targeted process optimization. The deviation process type refers to the type of process deviation caused by the fact that some process parameters do not meet the expected standards or appear abnormally in the production process.

[0073] According to the known deviation process type, the key parameters in the process state indicator set are inferred in reverse to determine which specific process parameters (such as temperature, pressure, displacement, etc.) cause the generation of deviation process. The plastic packaging process deviation result refers to the deviation state or process difference finally obtained by analyzing and mapping the changes of process parameters, which helps to evaluate whether the process meets the standard and provides a basis for adjustment.

[0074] By analyzing stages in parallel, the data of multiple process stages are processed simultaneously, reducing the time of analysis one by one. By combining these characteristic parameters with the full-process parameters of the plastic packaging process, a process state index set is generated for online reasoning of the plastic packaging quality score, reflecting the quality state of the product in real time, which helps to discover and solve problems in production in a timely manner. Through defect analysis and positioning, the root cause of the plastic packaging defect is determined, and corresponding measures are taken for correction.

[0075] The backtracking optimization module 14 is used to backtrack and adjust the multiple stage data according to the plastic packaging process deviation result, construct the multi-stage process optimization data of the target device, and feed back the multi-stage process optimization data to the control terminal of the automatic plastic packaging system.

[0076] Further, the backtracking optimization module 14 in the device process data monitoring system of the automatic plastic packaging system is also used for: a stage matching unit for matching stages based on the plastic packaging process deviation result to determine target associated stage information; a single-stage optimization unit for activating the backtracking module to backtrack the multiple stage data when the target associated stage information is a single stage, positioning the target stage for single-stage optimization, and obtaining single-stage process optimization data; and a cross-stage collaborative optimization unit for activating the backtracking module to backtrack the multiple stage data when the target associated stage information is multiple stages, positioning multiple target stages for cross-stage collaborative optimization, and obtaining multi-stage process optimization data.

[0077] Specifically, according to the plastic packaging process deviation result, stages are matched, the plastic packaging process deviation result is compared with the data of each stage, and it is found out which stage has a problem with the process parameters. Through stage matching, it is clear that process deviation occurs in which specific stage, which helps to accurately locate the problem. For example, if the analysis shows that the lack of pressure in the clamping stage leads to defects in the final product, the data of the clamping stage is matched with the deviation result to identify the stage as the problem.

[0078] When the target associated stage information points to a process stage, the backtracking module is activated to backtrack the historical data of the stage, find out the specific deviation reason, and perform single-stage optimization. The backtracking module is a functional module for tracing historical data, which analyzes whether each key parameter deviates from the standard by reviewing the historical data and operation of the target stage. For example, when the pressure value is found to be too low in the clamping stage, the pressure change trend of the stage is traced back to see if the deviation is caused by equipment failure, control system problems or other reasons. Single-stage process optimization data is provided for the root cause found, such as adjusting the pressure setting, upgrading the equipment control system, etc. Single-stage optimization refers to adjusting and optimizing only the stage when the problem only occurs in one process stage, the purpose of which is to improve the process quality of the stage and avoid affecting other stages.

[0079] When the target correlation stage information is multi-stage, the backtracking module will be activated to analyze multiple stages, identify possible mutual influences between multiple stages, and perform cross-stage collaborative optimization. The data of multiple stages are comprehensively analyzed. For example, the preheating stage and the mold closing stage may be mutually related, and an abnormality in one stage may affect the process of the next stage. The backtracking module will analyze the dependency between these stages, identify the influence chain, and perform collaborative optimization between multiple stages. For example, the impact of insufficient pressure in the mold closing stage can be reduced by adjusting the temperature of the preheating stage. Multi-stage optimization refers to the cross-stage collaborative optimization when the data of multiple process stages deviate, considering the mutual influence of multiple stages to improve the production process as a whole. For example, if the temperature of the preheating stage is too high, causing pressure fluctuations in the mold closing stage, it is recommended to adjust the temperature setting of the preheating stage and the pressure control strategy of the mold closing stage to reduce the deviation of the two stages.

[0080] Single-stage process optimization data and multi-stage process optimization data respectively represent the adjustment data required after optimization in a single stage or multiple stages. The adjustment data not only includes adjusted process parameters, but also may involve equipment upgrades, changes in operating methods, etc. The multi-stage process optimization data is fed back to the control terminal of the automatic plastic packaging system. The control terminal will automatically adjust the process parameters according to these optimization data, thereby optimizing the production process in real time. Backtracking adjustment helps to accurately locate and correct process deviations, and multi-stage process optimization data helps to improve the stability of the overall process flow. After feeding these optimization data back to the control terminal, the process parameters can be automatically adjusted in real time, thereby realizing intelligent and precise production optimization.

[0081] In summary, the device process data monitoring system provided by the present application has the following technical effects:

[0082] The timing analysis module is used to collect real-time running data of a target device in an automatic plastic packaging system, perform runtime timing analysis on the real-time running data based on a plastic packaging process flow, and obtain a multi-dimensional timing data stream; the stage division module is used to divide the multi-dimensional timing data stream according to a plurality of plastic packaging process stages, and determine a plurality of stage data; the quality evaluation module is used to perform real-time state monitoring on the plurality of stage data, generate a process state index set, perform plastic packaging quality evaluation according to the process state index set, and determine a plastic packaging process deviation result; the backtracking optimization module is used to backtrack and adjust the plurality of stage data according to the plastic packaging process deviation result, construct multi-stage flow optimization data of the target device, and feed back the multi-stage flow optimization data to a control terminal of the automatic plastic packaging system. That is, by collecting running data of a target device, dividing the data according to a process flow, obtaining stage data, performing plastic packaging quality evaluation on real-time state indicators, determining a process deviation result, and backtracking and adjusting a plurality of stage data, the plastic packaging quality and production efficiency are improved.

[0083] In the second embodiment, based on the same inventive concept as the device flow data monitoring system of the first embodiment, the application also provides a device flow data monitoring method for an automatic plastic packaging system. Please refer to the attached Figure 2 The device flow data monitoring method for an automatic plastic packaging system includes the following steps:

[0084] S100: Collect real-time running data of a target device in an automatic plastic packaging system, perform runtime timing analysis on the real-time running data based on a plastic packaging process flow, and obtain a multi-dimensional timing data stream; S200: Divide the multi-dimensional timing data stream according to a plurality of plastic packaging process stages, and determine a plurality of stage data; S300: Perform real-time state monitoring on the plurality of stage data, generate a process state index set, perform plastic packaging quality evaluation according to the process state index set, and determine a plastic packaging process deviation result; S400: Backtrack and adjust the plurality of stage data according to the plastic packaging process deviation result, construct multi-stage flow optimization data of the target device, and feed back the multi-stage flow optimization data to a control terminal of the automatic plastic packaging system.

[0085] Further, the real-time running data of the target device in the automatic plastic packaging system is collected, and runtime sequence analysis is performed on the real-time running data based on a plastic packaging process flow to obtain a multi-dimensional time sequence data stream, including: synchronously collecting the target device by laying multiple types of sensors on the automatic plastic packaging system to obtain multiple sets of real-time running data; marking the multiple sets of real-time running data according to the data types of the multiple types of sensors to generate data type labels; extracting a process timestamp based on a plastic packaging process time sequence, aligning the multiple sets of real-time running data according to the process timestamp to generate an original time sequence data set; matching the original time sequence data set with the data type labels to generate a multi-dimensional signal, mapping the multi-dimensional signal to the same time reference for fusion to determine the multi-dimensional time sequence data stream.

[0086] Further, the matching of the original time sequence data set with the data type labels to generate a multi-dimensional signal, the mapping of the multi-dimensional signal to the same time reference for fusion, and the determination of the multi-dimensional time sequence data stream include: matching signal processing channels based on the data type labels to determine a temperature space difference value channel, a pressure waveform filtering channel, a trajectory smoothing channel, and a frequency spectrum extraction channel; mapping the original time sequence data set to the temperature space difference value channel to perform full-mold temperature analysis on the target device to construct a temperature gradient matrix; mapping the original time sequence data set to the pressure waveform filtering channel to perform pressure filtering on the target device to obtain pressure distribution waveform data; mapping the original time sequence data set to the trajectory smoothing channel to perform displacement ranging on the target device to obtain a displacement trajectory vector; mapping the original time sequence data set to the frequency spectrum extraction channel to perform current energy calculation on the target device to obtain current frequency spectrum data; and mapping the temperature gradient matrix, the pressure distribution waveform data, the displacement trajectory vector, and the current frequency spectrum data to the same time reference for fusion to determine the multi-dimensional time sequence data stream.

[0087] Further, the multi-dimensional time sequence data stream is divided according to multiple plastic packaging process stages to determine multiple stage data, including: loading plastic packaging process full-flow parameters, dynamically segmenting the plastic packaging process full-flow parameters to obtain multiple plastic packaging process stages; mapping the multi-dimensional time sequence data stream to the multiple plastic packaging process stages for dynamic slicing to generate multiple continuous subsequence stage data; performing runtime sequence analysis based on the multiple continuous subsequence stage data to generate a process stage running state feature vector set; performing stage boundary detection based on the process stage running state feature vector set to determine multiple stage labels; synchronously identifying the multiple stage labels to the multi-dimensional time sequence data stream to determine the multiple stage data.

[0088] Further, the mapping of the multi-dimensional time sequence data stream to the plurality of plastic packaging process stages for dynamic slicing generates a plurality of continuous sub-sequence stage data, including: mapping the temperature gradient matrix to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream to generate first sub-sequence stage data; mapping the pressure distribution waveform data to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream to generate second sub-sequence stage data; mapping the displacement trajectory vector to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream to generate third sub-sequence stage data; mapping the current frequency spectrum data to the plurality of plastic packaging process stages for dynamic slicing of the multi-dimensional time sequence data stream to generate fourth sub-sequence stage data; and arranging the first sub-sequence stage data, the second sub-sequence stage data, the third sub-sequence stage data, and the fourth sub-sequence stage data according to the plastic packaging process full-process parameters to generate the plurality of continuous sub-sequence stage data.

[0089] Further, the real-time state monitoring of the plurality of stage data generates a process state indicator set, and the plastic packaging quality evaluation is performed according to the process state indicator set to determine a plastic packaging process deviation result, including: performing parallel analysis on the plurality of stage data, extracting features according to stage parallel results to obtain a plurality of stage parallel feature parameters; performing stage state monitoring based on the plurality of stage parallel feature parameters in combination with the plastic packaging process full-process parameters to generate a process state indicator set; performing online inference on the plastic packaging quality of a target device according to the process state indicator set to generate a plastic packaging quality score; performing defect analysis according to the plastic packaging quality score to determine a defect root cause, positioning according to the defect root cause to obtain a plastic packaging defect source; matching the plastic packaging defect source with the process state indicator set to obtain a deviation process type, and mapping the deviation process type with the process state indicator set to extract the plastic packaging process deviation result.

[0090] Further, the stage state monitoring based on the plurality of stage parallel feature parameters in combination with the plastic packaging process full-process parameters to generate a process state indicator set includes: performing cross-stage influence analysis based on the plurality of stage parallel feature parameters to obtain a plurality of influence factors; fusing the plurality of influence factors according to the plastic packaging process full-process parameters to obtain a fused influence parameter; calling device spatial dimension convolution data of an automatic plastic packaging system, performing time dimension analysis on the device of the automatic plastic packaging system according to the fused influence parameter to determine time dimension convolution data; constructing a space-time convolution network according to the spatial dimension convolution data and the time dimension convolution data, performing stage state monitoring through the space-time convolution network to obtain a plurality of stage process state features; and correlating the plurality of stage process state features across stages, and setting the process state indicator set according to the correlation result.

[0091] Further, the matching the plastic packaging defect source with the process state indicator set, obtaining a deviation process type, mapping the deviation process type with the process state indicator set, and extracting the plastic packaging process deviation result, comprises: mapping the plastic packaging defect source based on the process state indicator set to define a defect process mapping matrix; performing gradient dynamic update according to the defect process mapping matrix to construct a plastic packaging defect knowledge base; synchronizing the process state indicator set to the plastic packaging defect knowledge base for retrieval, performing distributed calculation according to the retrieval result to obtain process defect probability distribution data; performing clustering analysis based on the process defect probability distribution data according to the plastic packaging process type to obtain the deviation process type; and performing reverse feature mapping on the process state indicator set according to the deviation process type to determine the plastic packaging process deviation result.

[0092] Further, the backtracking adjustment of the plurality of stage data according to the plastic packaging process deviation result to construct the multi-stage process optimization data of the target device, comprises: determining target associated stage information based on the plastic packaging process deviation result; when the target associated stage information is single stage, activating a backtracking module to backtrack the plurality of stage data, positioning the target stage for single stage optimization to obtain single stage process optimization data; when the target associated stage information is multi-stage, activating the backtracking module to backtrack the plurality of stage data, positioning the plurality of target stages for cross-stage collaborative optimization to obtain multi-stage process optimization data.

[0093] Each of the embodiments in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The foregoing Figure 1 The device flow data monitoring system for applying an automatic plastic packaging system in Embodiment One and the specific example are also applicable to the device flow data monitoring method for applying an automatic plastic packaging system in this embodiment. Those skilled in the art can clearly know the device flow data monitoring method for applying an automatic plastic packaging system in this embodiment through the foregoing detailed description of the device flow data monitoring system for applying an automatic plastic packaging system. Therefore, for the sake of brevity of the specification, it will not be described in detail here.

[0094] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0095] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the application and its equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A process data monitoring system for an automated plastic sealing system, characterized in that, The method comprises the following steps: a time sequence analysis module is used to collect real-time running data of a target device in an automatic plastic packaging system, perform runtime sequence analysis on the real-time running data based on a plastic packaging process flow, and obtain a multi-dimensional time sequence data stream; a stage division module is used to divide the multi-dimensional time sequence data stream according to a plurality of plastic packaging process stages, and determine a plurality of stage data; a quality evaluation module is used to perform real-time state monitoring on the plurality of stage data, generate a process state index set, perform plastic packaging quality evaluation according to the process state index set, and determine a plastic packaging process deviation result; a backtracking optimization module is used to backtrack and adjust the plurality of stage data according to the plastic packaging process deviation result, construct multi-stage flow optimization data of the target device, and feed back the multi-stage flow optimization data to a control terminal of the automatic plastic packaging system; the time sequence analysis module comprises: a synchronous acquisition unit is used to acquire a plurality of groups of real-time running data of a target device by arranging a plurality of types of sensors in an automatic plastic packaging system; a data labeling unit is used to label data types of the plurality of groups of real-time running data according to the plurality of types of sensors, and generate data type labels; a data alignment unit is used to extract a process timestamp based on a plastic packaging process time sequence, align the plurality of groups of real-time running data according to the process timestamp, and generate an original time sequence data set; a data fusion unit is used to match the original time sequence data set with the data type labels, generate a multi-dimensional signal, map the multi-dimensional signal to the same time reference for fusion, and determine the multi-dimensional time sequence data stream; the data fusion unit comprises: a signal processing channel matching subunit is used to match a signal processing channel based on the data type labels, determine a temperature space difference channel, a pressure waveform filtering channel, a trajectory smoothing channel, and a frequency spectrum extraction channel; a temperature analysis subunit is used to map the original time sequence data set to the temperature space difference channel to perform full-mold temperature analysis on a target device, and construct a temperature gradient matrix; a pressure filtering subunit is used to map the original time sequence data set to the pressure waveform filtering channel to perform pressure filtering on a target device, and obtain pressure distribution waveform data; a displacement measurement subunit is used to map the original time sequence data set to the trajectory smoothing channel to perform displacement measurement on a target device, and obtain a displacement trajectory vector; a current energy calculation subunit is used to map the original time sequence data set to the frequency spectrum extraction channel to perform current energy calculation on a target device, and obtain current frequency spectrum data; a reference fusion subunit is used to map the temperature gradient matrix, the pressure distribution waveform data, the displacement trajectory vector, and the current frequency spectrum data to the same time reference for fusion, and determine the multi-dimensional time sequence data stream; the stage division module comprises: a dynamic segmentation unit is used to load plastic packaging process full-flow parameters, dynamically segment the plastic packaging process full-flow parameters, and obtain a plurality of plastic packaging process stages; a dynamic slicing unit is used to map the multi-dimensional time sequence data stream to the plurality of plastic packaging process stages for dynamic slicing, and generate a plurality of continuous subsequence stage data. a time series analysis unit configured to perform runtime sequence analysis based on the plurality of continuous sub-sequence stage data to generate a process stage operation state feature vector set; a stage boundary detection unit configured to perform stage boundary detection based on the process stage operation state feature vector set to determine a plurality of stage labels; a label synchronization unit configured to synchronize the plurality of stage labels to the multi-dimensional time series data stream to determine the plurality of stage data; the dynamic slicing unit comprises: a first slicing sub-unit configured to map the temperature gradient matrix to the multi-dimensional time series data stream of the plurality of plastic packaging process stages to perform dynamic slicing to generate first sub-sequence stage data; a second slicing sub-unit configured to map the pressure distribution waveform data to the multi-dimensional time series data stream of the plurality of plastic packaging process stages to perform dynamic slicing to generate second sub-sequence stage data; a third slicing sub-unit configured to map the displacement trajectory vector to the multi-dimensional time series data stream of the plurality of plastic packaging process stages to perform dynamic slicing to generate third sub-sequence stage data; a fourth slicing sub-unit configured to map the current spectrum data to the multi-dimensional time series data stream of the plurality of plastic packaging process stages to perform dynamic slicing to generate fourth sub-sequence stage data; a stage arrangement sub-unit configured to arrange the first sub-sequence stage data, the second sub-sequence stage data, the third sub-sequence stage data, and the fourth sub-sequence stage data according to the plastic packaging process full-process parameters to generate the plurality of continuous sub-sequence stage data.

2. The equipment flow data monitoring system using an automatic plastic sealing system according to claim 1, wherein, the quality evaluation module comprises: a parallel analysis unit configured to perform parallel analysis on the plurality of stage data, extract features according to stage parallel results, and obtain a plurality of stage parallel feature parameters; a stage state monitoring unit configured to perform stage state monitoring based on the plurality of stage parallel feature parameters in combination with the plastic packaging process full-process parameters to generate a process state index set; a state reasoning unit configured to perform online reasoning on the plastic packaging quality of a target device according to the process state index set to generate a plastic packaging quality score; a defect analysis unit configured to perform defect analysis according to the plastic packaging quality score, determine a defect root cause, and locate according to the defect root cause to obtain a plastic packaging defect source; a deviation determination unit configured to match the plastic packaging defect source with the process state index set to obtain a deviation process type, map the deviation process type with the process state index set, and extract the plastic packaging process deviation result.

3. The equipment flow data monitoring system using an automatic plastic sealing system according to claim 2, wherein, the stage state monitoring unit comprises: a cross-stage influence analysis sub-unit configured to perform cross-stage influence analysis based on the plurality of stage parallel feature parameters to obtain a plurality of influence factors; an influence factor fusion sub-unit configured to fuse the plurality of influence factors according to the plastic packaging process full-process parameters to obtain a fused influence parameter; and a time dimension analysis sub-unit configured to retrieve device spatial dimension convolution data of an automatic plastic packaging system, perform time dimension analysis on the device of the automatic plastic packaging system according to the fused influence parameter, and determine time dimension convolution data. The stage state monitoring subunit is configured to construct a space-time convolution network according to the space-dimension convolution data and the time-dimension convolution data, and to perform stage state monitoring through the space-time convolution network to obtain a plurality of stage process state features. The cross-stage association subunit is configured to associate the plurality of stage process state features across stages, and to set the process state index set according to an association result.

4. The equipment flow data monitoring system using an automatic plastic sealing system according to claim 2, wherein The bias determination unit includes: The defect mapping subunit is configured to map the process state index set based on the plastic package defect source, and to define a defect process mapping matrix. The gradient updating subunit is configured to perform gradient dynamic updating according to the defect process mapping matrix, and to construct a plastic package defect knowledge base. The defect retrieval subunit is configured to retrieve the process state index set as an index from the plastic package defect knowledge base, to perform distribution calculation according to a retrieval result, and to obtain process defect probability distribution data. The cluster analysis subunit is configured to perform cluster analysis according to a plastic package process type based on the process defect probability distribution data, and to obtain a bias process type. The reverse feature mapping subunit is configured to perform reverse feature mapping on the process state index set according to the bias process type, and to determine the plastic package process bias result.

5. The equipment flow data monitoring system using an automatic plastic sealing system according to claim 1, wherein, The backtracking optimization module includes: The stage matching unit is configured to perform stage matching based on the plastic package process bias result, and to determine target associated stage information. The single-stage optimization unit is configured to activate the backtracking module to backtrack a plurality of stage data when the target associated stage information is a single stage, to locate a target stage to perform single-stage optimization, and to obtain single-stage process optimization data. The cross-stage collaborative optimization unit is configured to activate the backtracking module to backtrack a plurality of stage data when the target associated stage information is a plurality of stages, to locate a plurality of target stages to perform cross-stage collaborative optimization, and to obtain multi-stage process optimization data.

6. A method for monitoring equipment process data using an automated plastic sealing system, characterized in that, The device process data monitoring system of an automatic plastic package system according to any one of claims 1 to 5 is executed, and the device process data monitoring method of the automatic plastic package system includes: Collecting real-time running data of a target device in an automatic plastic package system, performing runtime sequence analysis on the real-time running data based on a plastic package process flow, and obtaining a plurality of multi-dimensional time sequence data streams; Dividing the plurality of multi-dimensional time sequence data streams according to a plurality of plastic package process stages, and determining a plurality of stage data; Traversing the plurality of stage data to perform real-time state monitoring, generating a process state index set, performing plastic package quality evaluation according to the process state index set, and determining a plastic package process bias result; Performing backtracking adjustment on the plurality of stage data according to the plastic package process bias result, constructing multi-stage process optimization data of the target device, and feeding back the multi-stage process optimization data to a control terminal of the automatic plastic package system.

Images