Remote monitoring system for automatic packaging equipment

Abstract

The invention provides a remote monitoring system for automatic packaging equipment, and relates to the technical field of industrial equipment monitoring, and the system comprises a plurality of synchronous sensing units which are used for collecting synchronous original data of each branch of a pneumatic network; and a data processing device. The data processing device is internally provided with a digital twinborn model and can predict a theoretical pulse reference waveform according to process beat parameters obtained from an equipment controller; performing cancellation processing on the synchronous original data and the theoretical waveform to separate out a pure residual signal; and after the occurrence of the micro-leakage event is confirmed, further performing inversion calculation based on the time delay information of the multi-point residual signal and a pre-stored pneumatic network topology matrix so as to determine a branch identifier of leakage, and sending an alarm instruction. The reliability and the positioning precision of micro-leakage detection can be remarkably improved, and the production recovery efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial equipment monitoring, in particular to an automatic packaging equipment remote monitoring system. BACKGROUND

[0002] Automatic packaging equipment, such as automatic packing machines or carton sealing machines, usually includes a pneumatic network composed of multiple branches to drive different actuators to complete periodic operations. In order to ensure packaging quality and production efficiency, it is necessary to monitor the working state of the pneumatic network to prevent insufficient pressure caused by gas leakage. A common monitoring method is to set pressure or flow sensors on the pipelines of the pneumatic network to determine the integrity of the pneumatic network by monitoring the real-time changes of the signals.

[0003] However, during the operation of the automatic packaging equipment, the periodic action of the internal actuators will produce strong pressure pulsation. This normal business pulsation signal is highly similar in signal characteristics to the micro-leakage signal caused by aging of pipeline joints, wear of sealing rings, etc. This makes it difficult for the existing single-point sensor-based monitoring method to effectively distinguish between the two types of signals, often misreporting normal business pulsation as leakage, or missing the real micro-leakage due to pulsation interference, resulting in low detection reliability. In addition, even if an anomaly is detected, the single-point sensor-based solution cannot determine the specific branch where the leakage occurs due to the lack of spatial dimension information, resulting in a high dependence on manual inspection for fault diagnosis, and low efficiency for fault handling and production recovery.

[0004] Therefore, in the pneumatic network of the automatic packaging equipment, how to improve the detection accuracy and positioning efficiency of micro-leakage events to cope with the challenges brought by multi-branch complexity and dynamic working condition interference, so as to realize more reliable real-time monitoring and fault response, is a technical problem that needs to be solved by those skilled in the art. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides an automatic packaging equipment remote monitoring system, comprising:

[0006] A plurality of sensing units arranged at the inlet end and the terminal end of each branch of the target pneumatic network, for synchronously collecting the inlet end pressure value, the terminal end pressure value and the flow value of each branch under a unified clock reference, to generate synchronous raw data;

[0007] A data processing device in communication connection with the sensing units and the automatic packaging equipment controller, the data processing device storing a digital twin model and being configured to:

[0008] Receive the process beat parameters from the automatic packaging equipment controller;

[0009] based on the process beat parameter and the digital twin model, a theoretical pulsation reference waveform of a current working condition is calculated and obtained;

[0010] The synchronous raw data is differentially processed with the theoretical pulsation reference waveform to generate a residual signal;

[0011] According to the residual signal, the occurrence confidence of the micro-leakage event is determined, and in response to the occurrence confidence being greater than or equal to a preset threshold, micro-leakage event information is generated;

[0012] Based on the micro-leakage event information, the residual signals of different branches are cross-correlated to obtain time delay information, and inverse calculation is performed in combination with a pre-stored aerodynamic network topology matrix to determine the branch identifier of the micro-leakage event, and a micro-leakage event alarm instruction is sent to the automatic packaging equipment controller based on the branch identifier.

[0013] As an optional implementation, it further comprises an automatic time calibration device configured to start and manage the time calibration process of the sensing unit and the data processing device, and determine and store the time calibration value corresponding to each sensing unit respectively;

[0014] The sensing unit is further configured to record the actual arrival timestamp of the standard physical pulse received by it during the time calibration process, and provide the actual arrival timestamp to the automatic time calibration device;

[0015] The data processing device is further configured to calculate the theoretical arrival time of the standard physical pulse to each sensing unit during the time calibration process, and provide the theoretical arrival time to the automatic time calibration device;

[0016] The automatic time calibration device is further configured to compare the actual arrival timestamp with the theoretical arrival time to determine the time calibration value;

[0017] The sensing unit is further configured to correct the time information using the time calibration value when generating the synchronous raw data.

[0018] As an optional implementation, the automatic time calibration device is further configured to:

[0019] Obtain the process beat parameter from the automatic packaging equipment controller, and identify a preset idle process period based on the process beat parameter.

[0020] As an optional implementation, the automatic time calibration device is further configured to:

[0021] During the preset idle procedure period, a calibration control instruction is sent to the automated packaging equipment controller to drive an actuator in the target pneumatic network to generate the standard physical pulse.

[0022] As an optional implementation, the data processing apparatus is further configured to:

[0023] In response to the occurrence confidence being lower than a preset threshold, defining the residual error signal as a model error signal;

[0024] Adaptively correcting one or more model parameters in the digital twin model using the model error signal.

[0025] As an optional implementation, the adaptively correcting one or more model parameters in the digital twin model includes:

[0026] Resolving the procedure beat parameter into an actuator action event, and determining one or more target branches associated with the actuator action event using the pre-stored pneumatic network topology matrix;

[0027] Extracting a segmented error signal associated with the target branch in time from the model error signal;

[0028] Updating only dynamic physical parameters related to the target branch in the digital twin model using the segmented error signal.

[0029] As an optional implementation, the extracting a segmented error signal associated with the target branch in time from the model error signal includes:

[0030] Performing multi-scale wavelet transform on the model error signal to obtain wavelet coefficients distributed in the time-frequency domain;

[0031] Identifying one or more adjacent branches adjacent to the target branch using the pneumatic network topology matrix, and predicting time-frequency characteristics of a crosstalk signal caused by the actuator action event in the adjacent branches based on a topological relationship between the adjacent branches and the target branch;

[0032] Constructing a wavelet domain mask based on the time-frequency characteristics;

[0033] Filtering the wavelet coefficients using the wavelet domain mask, and then performing inverse wavelet transform to reconstruct and obtain a purified segmented error signal.

[0034] As an optional implementation, the data processing apparatus further stores a packaging process feature library, wherein a reference time-frequency feature corresponding to a plurality of preset packaging action types of the automatic packaging equipment is stored in the packaging process feature library;

[0035] The prediction of the time-frequency feature of the crosstalk signal caused by the actuator action event in the adjacent branch includes:

[0036] From the actuator action event, the corresponding packaging action type is identified;

[0037] From the packaging process feature library, the reference time-frequency feature corresponding to the identified packaging action type is extracted;

[0038] Based on the reference time-frequency feature and the topological relationship between the adjacent branch and the target branch, the time-frequency feature of the crosstalk signal is calculated and obtained.

[0039] As an optional implementation, the time delay information includes:

[0040] From the plurality of sensing units, the sensing unit that receives the strongest energy or the earliest time of the residual signal is determined as a reference sensing unit;

[0041] For at least one other sensing unit in addition to the reference sensing unit, the residual signal of the reference sensing unit and the residual signal of the other sensing unit are respectively subjected to Fourier transform to obtain their respective frequency spectrum signals;

[0042] Based on the obtained frequency spectrum signals, the cross power spectrum is calculated, and the cross power spectrum is subjected to phase transformation weighting to obtain a whitened spectrum;

[0043] The whitened spectrum is subjected to inverse Fourier transform to obtain a cross-correlation function;

[0044] The peak position of the cross-correlation function is identified to determine the time delay between the reference sensing unit and the other sensing unit;

[0045] One or more time delays thus determined are combined as the time delay information.

[0046] As an optional implementation, the combination of the pre-stored pneumatic network topological matrix for inversion calculation to determine the branch identification of the micro-leakage event includes:

[0047] Based on the pneumatic network topological matrix, the target pneumatic network is discretized into a plurality of candidate leakage points;

[0048] For each of the candidate leakage points, a theoretical time delay when the leakage occurs at the candidate leakage point is calculated by using the pneumatic network topology matrix;

[0049] The calculated theoretical time delay is compared with the time delay information to calculate a leakage position likelihood value corresponding to each of the candidate leakage points;

[0050] A candidate leakage point with the largest leakage position likelihood value is determined, and a branch in which the candidate leakage point is located is determined as the branch identification of the micro-leakage event.

[0051] Compared with the prior art, the remote monitoring system provided by the application can accurately predict and separate the dynamic pressure pulsation disturbance generated by the normal operation of the equipment by constructing and using a digital twin model associated with the equipment process beat, thereby clearly highlighting the micro-leakage signal that was previously masked by strong noise background, and fundamentally improving the accuracy and reliability of leakage detection. On this basis, the system uses the spatial and temporal information collected by the multiple synchronous sensing units deployed in each branch of the network, and combines the pre-set network topology structure, to realize rapid and accurate positioning of the leakage source. This combination of high-reliability detection and high-efficiency positioning effectively avoids production interruptions caused by false positives or false negatives, and replaces time-consuming manual inspection with automated remote diagnosis, significantly enhancing the stability and economy of the operation of the automated packaging equipment, and achieving more intelligent fault response and predictive maintenance. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 A schematic diagram of an automatic packaging equipment remote monitoring system provided by an embodiment of the application;

[0053] Figure 2 A flowchart of a remote monitoring method executed by a data processing device in an embodiment of the application;

[0054] Figure 3 A flowchart of an adaptive correction method provided by an embodiment of the application;

[0055] Figure 4 A flowchart of a method for extracting a segmented error signal associated with the target branch in time provided by an embodiment of the application. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the application will be described in detail below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all embodiments of the application.

[0057] The "and / or" mentioned herein is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent three cases of existence of A alone, existence of A and B together, and existence of B alone.

[0058] Referring to Figure 1 As shown in the figure, the system comprises:

[0059] A plurality of sensing units 100 are arranged at the inlet end and the terminal end of each branch of the target pneumatic network, for synchronously collecting the inlet end pressure value, the terminal end pressure value and the flow value of each branch under the unified clock reference, and generating the synchronous raw data.

[0060] A data processing device 200 is in communication connection with the sensing units 100 and the automatic packaging equipment controller 300, and the data processing device 200 stores a digital twin model.

[0061] In a specific implementation, the plurality of sensing units 100 are respectively arranged at the inlet end and the terminal end of each parallel branch in the target pneumatic network. In this embodiment, each sensing unit 100 can be an integrated intelligent sensing node, which internally encapsulates a pressure sensing chip, a flow sensing chip, a microcontroller (MCU) and a synchronization clock slave module supporting IEEE 1588 PTP protocol. Under the coordination of a unified master clock, all sensing units 100 can achieve microsecond-level synchronization under the unified clock reference, and synchronously collect the inlet end pressure value, the terminal end pressure value and the instantaneous flow value of each monitoring point to generate the synchronous raw data with high-precision time stamp.

[0062] Exemplarily, the sensing unit 100 can be implemented as an integrated intelligent sensing node to meet the requirements of high precision and high synchronism in industrial field. The hardware composition of the node can include a MEMS pressure sensing chip for measuring the differential pressure between the two branches and a flow sensing chip based on the hot film principle, and a signal conditioning circuit is supplemented to amplify and filter the weak analog signals output by the sensors. The conditioned signals are sent to a high-precision analog-to-digital converter, such as an ADC with a sampling rate of not less than 10 kS / s and a resolution of 24 bits, to realize high-fidelity digitization. The core processing module of the node is used to perform data processing and communication protocols. The microcontroller (MCU) inside the sensing unit 100 can execute a pre-set firmware program. The firmware program first exchanges messages with the network master clock through the Ethernet physical layer chip supporting the IEEE 1588 PTPv2 protocol, calibrates the error of the local clock and maintains it within 1 microsecond. Under the driving of this synchronized clock, the MCU generates a sampling interrupt at a fixed high frequency, for example, every 200 microseconds. In response to each interrupt, the MCU drives the ADC to perform analog-to-digital conversion on the pressure and flow signals, and immediately reads the precise time of the current synchronized clock as the timestamp of this set of sampling data. Finally, the MCU encapsulates the pressure value, flow value, high-precision timestamp, and unique ID of the sensing unit 100 into a UDP data packet, and sends it in real time to the data processing device 200 through the Ethernet interface.

[0063] The data processing device 200, which can be an edge computing gateway or an industrial computer (IPC) deployed in the production field in this embodiment, is connected in high-speed communication with all sensing units 100 through industrial Ethernet to real-time aggregate the synchronized raw data. At the same time, the data processing device 200 is also connected in communication with the automation packaging equipment controller 300 through industrial communication protocols such as OPC UA or Modbus-TCP. The automation packaging equipment controller 300 can be a PLC or an IPC, for example.

[0064] The digital twin model is pre-stored in the internal memory of the data processing device 200. The model is a fluid dynamics simulation model pre-established according to the physical structure of the target pneumatic network, such as the length, diameter of the pipeline, valve model, cylinder parameters, etc., and is used to describe the pressure and flow response of the pneumatic network to the action of various actuators in the ideal case without leakage.

[0065] Exemplarily, the data processing device 200 can adopt an embedded industrial computer (IPC) to meet the real-time operation requirements of complex models and algorithms. A Linux operating system with real-time patches can be run on the industrial computer, and the core monitoring application is designed as a modular architecture, mainly including a data access module, a controller interface module, a digital twin simulation engine, and a signal processing and diagnosis module, etc.

[0066] Exemplarily, the digital twin model can be a pre-established fluid dynamics simulation model based on the lumped parameter method.

[0067] In the model construction phase, first, based on the physical design drawings of the target pneumatic network including the information of pipeline length, diameter, connection relationship, etc. and the specification of the elements deployed thereon, such as valve model, cylinder parameters, etc., a mathematical framework describing the network topology is established. In this framework, each pipeline can be equivalent to an electrical circuit unit composed of gas resistance (R), gas inductance (I) and gas capacitance (C) elements to represent the flow friction loss, inertia effect and compressibility of the gas respectively; and the action of each actuator is modeled as a time-varying flow consumption source with a specific gas consumption curve.

[0068] In the model parameterization phase, in order to make the model accurately reflect the real physical characteristics, an offline system identification method can be used.

[0069] Specifically, on a reference pneumatic network confirmed to have no leakage, the automated packaging equipment is operated to perform all standard procedures, and a set of complete and high-fidelity synchronous raw data is collected by using the plurality of sensing units 100. Subsequently, an optimization algorithm, such as the least squares fitting algorithm, is started to repeatedly compare the output waveform of the simulation model under the same process excitation with the collected data, and automatically adjust the key physical parameters in the model, such as the gas resistance, gas capacitance coefficients of each pipe section and the gas consumption curve parameters of each actuator, until the root mean square error between the simulation output and the real data reaches a minimum. Through this process, a set of accurately identified model parameters that can highly reproduce the real physical process is obtained.

[0070] The complete model including the network topology, mathematical equations and accurately identified parameters is solidified and stored in the data processing device 200. When the process cycle parameters are received during system operation, the simulation engine drives the static model to perform forward calculation, thereby obtaining a highly realistic theoretical pulsation reference waveform under the current working condition.

[0071] Reference Figure 2 is a flowchart of the remote monitoring method performed by the data processing device in the embodiments of the present application. The data processing device is configured to perform the following steps S101-S105, wherein:

[0072] S101: receiving a process beat parameter from the automated packaging equipment controller;

[0073] S102: based on the process beat parameter and the digital twin model, calculating a theoretical pulsation reference waveform of a current working condition;

[0074] S103: differentially processing the synchronization raw data and the theoretical pulsation reference waveform to generate a residual signal;

[0075] S104: determining a confidence level of occurrence of a micro-leakage event according to the residual signal, and generating micro-leakage event information in response to the confidence level being greater than or equal to a preset threshold;

[0076] S105: based on the micro-leakage event information, performing cross-correlation calculation on the residual signals of different branches to obtain time delay information, and performing inversion calculation in combination with a pre-stored pneumatic network topology matrix to determine a branch identifier of the micro-leakage event, and sending a micro-leakage event alarm instruction to the automated packaging equipment controller based on the branch identifier.

[0077] In a specific implementation, the controller interface module of the data processing device 200 subscribes and receives the process beat parameter from the automated packaging equipment controller 300 in real time through the OPC UA protocol (S101). After receiving the parameter, the digital twin simulation engine immediately drives the pre-stored static digital twin model based on the principle of fluid dynamics to perform simulation and calculate the theoretical pulsation reference waveform (S102). At the same time, the data access module sends the synchronization raw data gathered from each sensing unit 100 to the signal processing and diagnosis module, which performs step S103 to subtract the real-time data from the theoretical waveform point by point in a vectorized manner to generate a residual signal.

[0078] Then, when performing step S104 to determine the micro-leakage event, the module can adopt a detection method based on short-time energy. For example, the root mean square (RMS) value of the residual signal is calculated within a sliding time window, and when the RMS value exceeds a noise threshold set according to historical data for a certain period of time, for example, for 1 second, it is determined that the confidence level meets the standard, and micro-leakage event information is generated. Once the micro-leakage event information is generated, the system starts the positioning program of step S105.

[0079] For example, for cross-correlation calculation, the signal processing and diagnosis module can use a standard cross-correlation function algorithm based on fast Fourier transform (FFT) in the art to calculate the delay between different residual signals and obtain time delay information. For inversion calculation, the module can use an iterative search algorithm based on the least squares method.

[0080] For example, the algorithm first virtually divides all the tubes into several 10cm-long line segments as candidate leak points based on the topology matrix, then iterates through each candidate point, calculates the multi-path time delay theoretically generated when a leak occurs at this point, and calculates the Euclidean distance between the theoretical delay set and the actual measured delay set as the error. Finally, the branch where the candidate point with the smallest error is located is determined as the branch identifier where the micro-leak event occurs. After determining the branch identifier, the data processing device 200 generates a structured alarm instruction and sends it to the automated packaging equipment controller 300 through the controller interface module to trigger subsequent response actions.

[0081] For example, the pneumatic network topology matrix can be a structured data file, such as an XML file or a JSON file, which records in the form of multiple information tables all the physical and logical relationships of the target pneumatic network in detail. This file is generated once during system deployment according to the actual equipment tube layout.

[0082] For example, the topology matrix file can include the following core information:

[0083] Node information table: This table is used to define the attributes of all key function points (nodes) in the network. Each row represents a node and includes at least the following information columns:

[0084] Node ID: a unique identifier, such as "Sensor-A1-Inlet", "Actuator-MainPress", or "Junction-Tee-01"; Node type: defines the nature of the point, such as "sensor unit", "actuator", or "tee / junction"; Branch ID: explicitly indicates which specific branch in the pneumatic network the node belongs to, such as "Branch-A" or "Branch-B"; 3D spatial coordinates: records the X, Y, Z coordinates of the node in the device coordinate system for more accurate distance calculation.

[0085] Pipe information table: This table is used to define the attributes of the physical pipes (edges) connecting various nodes. Each row represents a pipe and includes at least the following information columns:

[0086] Pipe ID: a unique pipe identifier; Start node ID: the node ID connected to one end of the pipe; End node ID: the node ID connected to the other end of the pipe; Pipe length: the actual physical length of the pipe, such as 1.5 meters; Pipe inner diameter: the internal diameter of the pipe, such as 8mm; Material sound speed: the preset pressure wave propagation speed according to the pipe material and working medium (compressed air).

[0087] Event mapping table: This table is used to establish the association between the upper control logic and the underlying physical entities. Each row represents a mapping relationship and includes at least the following information columns:

[0088] Process beat parameter / event name: a command from the automation packaging equipment controller with a clear meaning, such as "MainPress_Engage"; associated actuator ID: the node ID of the "actuator" type directly corresponding to the command, such as "Actuator-MainPress".

[0089] Through such structured information organization, the data processing device 200 can efficiently perform various complex queries and calculations. For example, the device first queries the event mapping table according to the received process beat parameter to find the corresponding actuator ID; then, by querying the node information table through the ID, the target branch ID is determined, i.e. the target branch; finally, by querying the pipeline information table and the node information table, all other branches directly or indirectly connected to the target branch, i.e. the adjacent branches, can be easily found, providing accurate and structured data support for subsequent calculations.

[0090] In this way, by constructing and utilizing the digital twin model associated with the device process beat, the dynamic pressure pulsation disturbance generated by the normal operation of the device can be accurately predicted and separated, thereby clearly highlighting the trace leakage signal that was previously masked by strong noise background, fundamentally improving the accuracy and reliability of leakage detection. On this basis, the system utilizes the spatial and temporal information collected by the multiple synchronous sensing units deployed in each branch of the network, and combines the pre-set network topology, to realize fast and accurate positioning of the leakage source. This combination of high reliability detection and high efficiency positioning effectively avoids production interruptions caused by false positives or false negatives, and replaces time-consuming manual inspection with automated remote diagnosis, significantly enhancing the stability and economy of the operation of the automation packaging equipment, and achieving more intelligent fault response and predictive maintenance.

[0091] In order to further improve the positioning accuracy of the system described in the present application and eliminate the static timestamp offset caused by the difference in physical wiring length of each sensing unit 100, the inherent response delay difference of the internal circuit and sensing element, as an optional embodiment, the present application further comprises: an automatic time calibration device configured to start and manage the time calibration process of the sensing unit and the data processing device, and determine and store the time calibration values corresponding to each sensing unit respectively;

[0092] The sensing unit is further configured to record the actual arrival timestamp of the standard physical pulse received during the time calibration process and provide the actual arrival timestamp to the automatic time calibration device;

[0093] The data processing device is further configured to, in the time calibration process, calculate a theoretical arrival time of the standard physical pulse to each of the sensing units and provide the theoretical arrival time to the automatic time calibration device;

[0094] The automatic time calibration device is further configured to compare the actual arrival time stamp with the theoretical arrival time to determine the time calibration value;

[0095] The sensing unit is further configured to, in generating the synchronized raw data, correct the time information by using the time calibration value.

[0096] In a specific implementation, the automatic time calibration device is not a separate hardware entity, but can be integrated and run as a dedicated software functional module in the data processing device 200. The automatic time calibration device is configured to start and manage a time calibration process involving the sensing units 100 and the data processing device 200, which aims to determine and store a unique time calibration value for each sensing unit 100.

[0097] After starting the calibration process, a standard physical pulse needs to be applied to the target pneumatic network.

[0098] An exemplary implementation can be achieved by temporarily connecting an external calibration pulse generator, for example, a high-speed switching valve driven by a signal generator, to a known reference point in the pipe network. The pulse generator generates a steep and clear pressure pulse at a specified time.

[0099] All sensing units 100 in the network are configured to monitor the pressure signal during this calibration process. When it detects that the rising edge slope of the signal of the standard physical pulse exceeds a preset threshold, the internal MCU will immediately latch and record the PTP synchronization time at the current time as the actual arrival time stamp of the pulse. Subsequently, each sensing unit 100 uploads the actual arrival time stamp recorded by itself to the data processing device 200, which is then forwarded to the automatic time calibration device.

[0100] After receiving the actual arrival time stamps of each sensing unit 100, the data processing device 200 calculates the theoretical arrival time of the pulse to each sensing unit according to the pre-stored pneumatic network topology matrix information.

[0101] Specifically, the data processing device 200 queries the known position coordinates of the calibration pulse generator and the position coordinates of each sensing unit 100 from the topology matrix, and calculates the straight-line physical distance between them. Then, according to the material sound speed stored in the topology matrix, the theoretical time required for the pressure pulse to propagate this distance is calculated.

[0102] The automatic time calibration device compares the actual arrival timestamp of each sensor unit received with the corresponding theoretical arrival time calculated by the data processing device 200, usually by performing a subtraction operation, for example, time calibration value = theoretical arrival time - actual arrival timestamp, to determine the time calibration value specific to each sensor unit 100.

[0103] After determining the time calibration values of all sensor units, the automatic time calibration device will distribute these calibration values to the corresponding sensor units 100 respectively. After receiving its own time calibration value, the sensor unit 100 will store it in the internal non-volatile memory.

[0104] After completing the above calibration process, the firmware configuration of the sensor unit 100 is updated. In its subsequent regular work, when it generates synchronous raw data, it will perform a final time correction step: that is, add the raw sampling PTP timestamp to the time calibration value stored locally to obtain a final timestamp that is more reflective of the true physical event occurrence time. This corrected synchronous raw data is then sent to the data processing device 200 for subsequent leak detection and positioning, thereby greatly improving the accuracy of the final positioning result.

[0105] As an optional implementation, the generation of the standard physical pulse can be fully automated without any external device access or manual intervention. This further utilizes the deep communication and collaboration capabilities between the system of the present application and the automated packaging equipment controller 300.

[0106] Specifically, in order to realize intelligent selection of calibration timing, the automatic time calibration device, as a functional module inside the data processing device 200, is further configured to continuously obtain and analyze the process beat parameters from the automated packaging equipment controller 300. By analyzing these parameters, it can clearly identify the preset idle process period in the equipment production process. For example, it can identify the interval after completing a complete packaging action to the start of the next action, or the standby state of the equipment waiting for materials.

[0107] After identifying this safe, disturbance-free pre-defined idle process period, the automatic time calibration device is further configured to actively send a pre-defined calibration control instruction to the automated packaging equipment controller 300. Upon receiving this instruction, the automated packaging equipment controller 300 will, according to pre-defined logic, drive one or more pre-defined, fast-responding actuators in the target pneumatic network, such as a large-bore solenoid valve or a fast-discharge valve of a main press cylinder, to perform a fast opening and closing action within a very short time, for example, within 50 milliseconds. This action will instantaneously consume or release a certain amount of compressed air, thereby generating an ideal, standard physical pulse in the entire pneumatic network for calibration purposes.

[0108] By way of example, this process can be implemented based on the client / server model of OPC UA (Open Platform Communications Unified Architecture). The automated packaging equipment controller 300 internally runs an OPC UA server. This server publishes its internal key variables, such as a status register for indicating the current production step number, for example, DB1.DINT10, as a node in the OPC UA address space. This node has a unique node ID, such as ns=2;s="MachineStatus.CycleStep".

[0109] The data processing device 200 runs an OPC UA client. Upon system initialization, this client connects to the OPC UA server of the controller and subscribes to the above-mentioned status node. Using the subscription mode instead of the polling mode ensures that as soon as the value of the status register in the controller changes, the server will immediately actively push the new value to the data processing device 200, thereby achieving efficient, real-time parameter acquisition.

[0110] Upon receiving this raw process beat parameter, for example, an integer value "20", the data processing device 200 needs to interpret it. For this purpose, the data processing device 200 internally pre-stores an event definition file, which can be a configuration file in JSON or XML format, for example. This file defines the mapping between each integer value and a specific, well-defined physical meaning of an actuator action event. For example, the file can include the following mapping: {"20": "MainPress_Engage", "35": "SideClamp_Release", "99": "Cycle_End"}. Upon receiving the raw parameter "20", the data processing device 200 can interpret it as "MainPress_Engage" (main press action closing) by querying the event definition file, which is an event that can be understood by subsequent logic modules.

[0111] Further, after obtaining and parsing the meaningful event sequence, the automatic time calibration device can clearly identify the preset idle procedure period in the equipment production process based on the events through a state machine logic.

[0112] For example, the state machine can include at least three states: a production running state, a potential idle state, and a confirmed idle state. The workflow is as follows:

[0113] When the system is normally running, the state machine is in the production running state.

[0114] When the automatic time calibration device parses an event representing the end of a complete packaging process, such as a Cycle_End event, it starts an internal timer and switches the state machine to the potential idle state.

[0115] In the potential idle state, the device continues to wait for the next event representing the start of a new process, such as a Cycle_Start event.

[0116] Case 1: If the Cycle_Start event is received before the timer reaches a preset continuous production threshold, it means that the equipment is in high-frequency continuous production. At this time, the device resets the timer and switches the state machine back to the production running state, and does not perform calibration this time.

[0117] Case 2: If the timer runs out of the continuous production threshold and still does not receive the Cycle_Start event, the system determines that the equipment has entered a truly stable pause or waiting phase. At this time, the device switches the state machine to the confirmed idle state.

[0118] The duration of this confirmed idle state is identified as the preset idle procedure period. Only in this state, the automatic time calibration device is authorized to perform the action of sending calibration control instructions to the controller to ensure that the calibration process does not conflict with any normal production actions. When the Cycle_Start event is received again, the state machine switches back to the production running state, ending the idle period.

[0119] In this way, the system can use the equipment's own tooling and running intervals to achieve fully automatic, non-intrusive online time calibration, greatly improving the system's intelligence level and maintenance convenience.

[0120] To further improve the long-term running accuracy of the system described in the present application, to cope with the slow drift of equipment characteristics caused by physical wear, aging, or environmental changes, as an optional implementation, the data processing device is further configured to:

[0121] in response to the occurrence confidence being lower than a preset threshold, defining the residual signal as a model error signal;

[0122] using the model error signal, adaptively correcting one or more model parameters in the digital twin model.

[0123] The core idea of this process is the intelligent reuse and dual-role definition of the residual signal. In the conventional leakage diagnosis process, a significant residual signal is considered as evidence of the existence of leakage. However, in this embodiment, the data processing device 200 is configured to perform a more refined judgment logic:

[0124] When the micro-leakage event occurrence confidence calculated from the residual signal is lower than a preset threshold, the system determines that it is currently in a healthy or no-leakage state. In this case, that non-zero, weak residual signal is no longer a leakage signal in its physical sense, but is redefined as a model error signal. This error signal exactly reflects the subtle deviation between the current digital twin model and its corresponding real physical entity, i.e. the degree of model drift.

[0125] Once the residual signal is defined as a model error signal, the data processing device 200 will start a background running adaptive correction algorithm. This algorithm uses this model error signal as input to continuously, slightly adjust and correct one or more model parameters in the digital twin model. Through this online, actual running data based feedback correction, the digital twin model can continuously improve itself, automatically compensate for the effects of equipment aging and other factors, and always maintain its high accuracy in predicting the real physical process.

[0126] This adaptive correction mechanism makes the monitoring system of the present application evolve from a static diagnosis system to a dynamic intelligent system that can co-evolve with the monitored equipment, thereby ensuring extremely high monitoring reliability and accuracy throughout the entire equipment life cycle.

[0127] Further, in order to solve the problem of non-uniform aging of physical characteristics of different branches in the target pneumatic network due to uneven work load, and to achieve a more accurate and efficient model parameter correction, the adaptive correction process can be an event-driven targeted correction method.

[0128] As an optional implementation, see Figure 3 A flowchart of an adaptive correction method provided by the embodiment of the present application includes steps S201-S203, wherein:

[0129] S201: parse the process beat parameter into an actuator action event, and determine one or more target branches associated with the actuator action event by using the pre-stored pneumatic network topology matrix;

[0130] S202: extract a segment error signal from the model error signal, which is associated with the target branch in time;

[0131] S203: update only the dynamic physical parameters in the digital twin model that are related to the target branch, by using the segment error signal.

[0132] In a specific implementation, the data processing device 200 precisely focuses each model correction on the specific branch that has just performed a physical action. The specific implementation process is as follows:

[0133] First, after receiving a process beat parameter, the data processing device 200 does not generally correct the entire digital twin model, but first performs a target recognition step. The received process beat parameter, for example, “MainPress_Engage”, is parsed into a specific actuator action event. Then, by using the event mapping table and node information table pre-stored in the pneumatic network topology matrix, a precise mapping query is completed: the event mapping table is used to find the actuator ID associated with the event, for example, “Actuator-MainPress”, and then the node information table is used to query the branch ID to which it belongs, for example, “Branch-A”. In this way, the system binds an upper-level process instruction to a specific physical target branch “Branch-A”.

[0134] Next, error extraction is performed. In the signal processing and diagnosis module of the data processing device 200, a data segment that is closely associated with the target branch identified in time is extracted from the continuous, global model error signal stream.

[0135] Specifically, according to the occurrence timestamp of the actuator action event, a time window of a predetermined length is intercepted after the timestamp, and the error signal parts from each sensing unit on the target branch “Branch-A” within the window are combined into a segment error signal dedicated to this correction.

[0136] Finally, targeted updating is performed. This segment error signal is input into an adaptive correction algorithm, for example, a recursive least squares algorithm module.

[0137] Importantly, the output of this algorithm, i.e. the calculated parameter correction, will only act on those dynamic physical parameters in the digital twin model that are directly related to the identified target branch "Branch-A". For example, the algorithm will update the pipe flow resistance coefficients of each pipe segment in "Branch-A", or update the air consumption model parameters of "Actuator-MainPress", while all model parameters related to other irrelevant branches such as "Branch-B", "Branch-C", etc. will remain unchanged in this update cycle.

[0138] In this way, through this event-driven, topology-linked targeted correction mechanism, the present application can accurately track and compensate for the unique aging trajectory of each independent branch, improving the efficiency and accuracy of model adaptive correction, so that the digital twin model can maintain a very high fidelity in long-term operation.

[0139] In order to solve the technical problem that the dynamic coupling effect between the branches of the pneumatic network may cause the segmented error signal to mix with crosstalk components from adjacent branches, thereby affecting the correction accuracy, as an optional implementation, see Figure 4 The flowchart of a method for extracting a segmented error signal associated with the target branch in time provided by the embodiment of the present application includes steps S301-S304, wherein:

[0140] S301: performing multi-scale wavelet transform on the model error signal to obtain wavelet coefficients distributed in the time-frequency domain;

[0141] S302: using the pneumatic network topology matrix to identify one or more adjacent branches adjacent to the target branch, and based on the topological relationship between the adjacent branches and the target branch, predicting the time-frequency characteristics of the crosstalk signal caused by the actuator action event in the adjacent branches;

[0142] S303: constructing a wavelet domain mask based on the time-frequency characteristics;

[0143] S304: filtering the wavelet coefficients using the wavelet domain mask, and then performing inverse wavelet transform to reconstruct and obtain the purified segmented error signal.

[0144] In specific implementation, the signal processing and diagnosis module of the data processing device 200 adopts more refined signal processing techniques when performing error extraction steps. First, the module will perform a multi-scale wavelet transform on the model error signal or its segment in the target time window.

[0145] Exemplarily, a Complex Morlet wavelet, which is suitable for analyzing transient impact signals, is used as the base function to decompose the one-dimensional time-domain error signal into a two-dimensional time-frequency matrix, i.e., a wavelet coefficient matrix, which can simultaneously display time and frequency information.

[0146] Next, the system will perform a physical model-based crosstalk prediction process guided by the topology information.

[0147] Specifically, the data processing device 200 performs a graph theory search algorithm, such as the Dijkstra algorithm, using the pneumatic network topology matrix to identify the shortest physical path from the target branch where the actuator action event occurs to the sensor on each adjacent branch. Subsequently, by accumulating the path length and combining the preset sound speed information, the theoretical propagation delay of the pressure wave is calculated, and based on a simplified attenuation and dispersion model, the energy attenuation and main frequency band of the crosstalk signal after propagation can be estimated. By synthesizing these calculation results, the system can generate a complete time-frequency feature prediction containing time, frequency and amplitude information for each adjacent branch about its crosstalk signal.

[0148] On this basis, the wavelet domain mask can be constructed. The mask is essentially a two-dimensional weight matrix of the same size as the wavelet coefficient matrix. First, initialize the mask matrix to all values of "1", then iterate through all the predicted crosstalk signal time-frequency features, and modify the weight values in the time-frequency region of the mask matrix corresponding to these features to the suppression value "0".

[0149] Finally, by performing element-wise point multiplication between the original wavelet coefficient matrix and the constructed wavelet domain mask, all predicted crosstalk signal components can be accurately suppressed. Then, performing an inverse wavelet transform on the filtered new wavelet coefficient matrix can reconstruct and obtain the final purified segmented error signal. This high-purity signal will be used as the input for the subsequent targeted updating step to ensure the accuracy of the model parameter correction.

[0150] Exemplarily, the complete calculation process of this procedure is described in detail taking the A action event as an example.

[0151] The A is deployed on the target branch Branch-A, and its action will produce dynamic coupling crosstalk to its adjacent branch Branch-B.

[0152] After the data processing device 200 has extracted the time series data, e.g. comprising 1024 sample points, from the global model error signal that is relevant to the action event, it invokes its internal digital signal processing library to perform a continuous wavelet transform on the one-dimensional sequence. In a specific embodiment, the mother wavelet selected can be a Complex Morlet wavelet with good time-frequency localization, e.g. cmor1.5-1.0, where 1.5 is the bandwidth parameter and 1.0 is the center frequency. The sequence of scales for the transform can be set to a sequence of logarithmically equidistant values from 2 to 128, which covers the complete frequency band from the high-frequency impact of the actuator action to the low-frequency oscillations in the secondary network. The result of this transform is a complex matrix of 128 rows (scale axis) x 1024 columns (time axis), i.e. the wavelet coefficients. Taking the modulus of the matrix elements yields the time-frequency spectrum for analysis and visualization.

[0153] The data processing device 200 queries the event mapping table and the node information table in the pneumatic network topology matrix according to the A action event, and confirms that the action source is located on the target branch Branch-A. Then, by querying the connection relationship in the pipe information table, it identifies that the adjacent branch of Branch-A is Branch-B.

[0154] Subsequently, the system initiates a path search algorithm, e.g. Dijkstra's algorithm, to calculate the shortest physical path from the action source node on Branch-A to the sensor node on Branch-B in the network graph defined by the topology matrix, and accumulates the pipe lengths of all pipe sections on this path to obtain the total propagation distance, e.g. 2.5 meters. Based on the sound speed preset for the pipe material in the topology matrix, e.g. 340 meters / second, the theoretical propagation delay is calculated to be 2.5 / 340 ≈ 7.4 milliseconds.

[0155] At the same time, the loss of signal energy and the change in frequency band can also be estimated based on a decay and dispersion model. For example, the model can predict that the energy of the crosstalk signal will decay to less than 5% of the original main signal energy, and since the high-frequency components decay faster in the fluid, their main energy will be concentrated in the lower frequency band of 100-300 Hz. Finally, a specific time-frequency feature prediction is synthesized for the crosstalk of Branch-B: "a weak crosstalk signal is expected to appear in the frequency range of 100-300 Hz within a time window of 7.4 milliseconds ± 2 milliseconds after the main action occurs".

[0156] For example, the data processing device 200 creates a wavelet domain mask matrix in memory with the same size (128 x 1024) as the aforementioned wavelet coefficient matrix, and initializes all its element values to 1.0 (representing complete passage).

[0157] Next, the system converts the predicted crosstalk time-frequency characteristics from the previous step into index coordinates of a mask matrix. For example, a frequency range of [100 Hz, 300 Hz] can correspond to rows 60 to 80 of the scale axis, while a time window of [5.4 ms, 9.4 ms] can correspond to columns 27 to 47 of the time axis. Then, the system modifies all element values in the rectangular region defined by [rows 60:80, columns 27:47] in the mask matrix from 1.0 to 0.0 (representing complete suppression). If there are multiple crosstalk predictions from adjacent branches, this process is repeated to carve multiple suppression regions on the mask.

[0158] Further, the signal processing and diagnosis module of the data processing device 200 performs an element-wise matrix multiplication between the original wavelet coefficient matrix obtained in the first step and the wavelet domain mask matrix constructed in the previous step. After this operation, the energy of the wavelet coefficients in the crosstalk region has been substantially zeroed out, while the coefficients in the main signal region remain unchanged. Finally, the module calls an inverse continuous wavelet transform (Inverse CWT) function with the purified new wavelet coefficient matrix as input to reconstruct the signal and finally outputs a one-dimensional time series, which is the desired purified segmented error signal.

[0159] By way of example, the goal of the attenuation and dispersion model is to quickly calculate how a disturbance signal generated in a target branch will evolve in its waveform characteristics after propagating to an adjacent branch, based on known physical principles.

[0160] Regarding the attenuation model, it can be implemented as a frequency-dependent energy attenuation function. The core of this model is that it considers the attenuation coefficient a as a function of frequency f, i.e. a(f). According to the theory of fluid acoustics in pipes, attenuation is mainly due to viscous and thermal conduction losses of the pipe wall. A simplified relationship that can be used is that the attenuation coefficient is approximately proportional to the square root of the frequency.

[0161] When performing the prediction, the data processing device 200 performs the following calculation: First, it knows the original disturbance signal, for example, the initial frequency band of a pulse generated by an actuator action. Then, according to the aerodynamic network topology matrix, the physical distance L that the signal needs to propagate is obtained. For each frequency component in the original frequency band, an attenuated amplitude can be calculated. Since the attenuation coefficient a(f) of high frequency components is much larger than that of low frequency components, after propagation over a long distance L, the high frequency components in the original signal will be disproportionately and more severely attenuated.

[0162] Regarding the dispersion model, it can be implemented as a phenomenological description of the consequences of the above-mentioned attenuation model. That is, the system does not have to calculate the complex frequency-dependent phase velocity, but directly models its final effect.

[0163] Taking into account both the attenuation and dispersion effects, the data processing device 200 can make the following prediction: a sharp impulse signal with its energy mainly concentrated in the 500-1000 Hz frequency band at the source (the target branch), after propagating a distance of 2.5 meters to the adjacent branch, its main energy will shift and concentrate in a lower frequency band, for example, 100-300 Hz. At the same time, the model can also include a simplified pulse broadening parameter to describe the signal's dispersion effect in time, i.e., a sharp impulse signal with a duration of 10 milliseconds, after propagation, can be broadened to a more gentle waveform with a duration of 25 milliseconds.

[0164] Finally, the model outputs a complete time-frequency feature of the crosstalk signal, which includes the theoretical propagation delay, the predicted energy attenuation ratio, the predicted central frequency band, and the predicted time width. This specific and quantitative feature description will serve as the precise basis for subsequent construction of the wavelet domain mask.

[0165] As an optional implementation, the data processing device also stores a packaging process feature library, which stores reference time-frequency features corresponding to multiple preset packaging action types of the automated packaging equipment.

[0166] The time-frequency feature of the crosstalk signal predicted by the actuator action event in the adjacent branch includes:

[0167] From the actuator action event, the corresponding packaging action type is identified;

[0168] From the packaging process feature library, the reference time-frequency feature corresponding to the identified packaging action type is extracted;

[0169] Based on the reference time-frequency feature and the topological relationship between the adjacent branch and the target branch, the time-frequency feature of the crosstalk signal is calculated and obtained.

[0170] Further, in order to further improve the prediction accuracy of the crosstalk signal, the data processing device 200 not only understands the physical layout of the pipe network, but also deeply understands the physical "individuality" of each packaging action.

[0171] To this end, a structured library of packaging process signatures is also pre-stored in the internal memory of the data processing device 200. This library is generated in an offline calibration phase before system deployment, and its purpose is to establish a precise signal fingerprint for each unique packaging motion. In the calibration process, each pre-defined packaging motion type, such as a high-speed "tension motion" or a slow-speed "corner protection motion", is individually and repeatedly triggered on a reference pneumatic network, and a reference-level sensor close to the motion source is used to capture the purest original pressure pulsation waveform of the motion. After signal averaging, wavelet transform and other processing, the key reference time-frequency features of the captured signal are extracted and parameterized. For example, the reference signature of a "tension motion" can be stored as a set of parameters, specifically {motion type: "TENSION", reference center frequency: 850 Hz, reference bandwidth: 300 Hz, reference energy: 1.0, reference duration: 12 ms}. The entire signature library is composed of a plurality of parameterized fingerprints of such motion types.

[0172] When the system enters the online monitoring state and performs crosstalk prediction, its internal process is optimized into a highly intelligent three-step deduction:

[0173] The first step is to identify the motion type. After receiving the process cycle parameters and parsing them into actuator motion events, the data processing device 200 will further query the event mapping table in the pneumatic network topology matrix or a special process knowledge base. The purpose of this query is to identify the specific packaging motion type corresponding to the current motion event. For example, the event "MainPress_Engage" is identified as a "TENSION" type.

[0174] The second step is to extract the reference signature. Once the motion type is identified as "TENSION", the data processing device 200 will immediately access the packaging process signature library and extract the corresponding reference time-frequency signature parameter set containing center frequency, bandwidth, energy and duration information.

[0175] The third step is to make a propagation prediction based on the benchmark. The data processing device 200 takes the highly realistic benchmark time-frequency feature extracted in the previous step as the initial disturbance source for the simulation calculation, instead of a general, hypothetical standard pulse. Then it starts the aforementioned physical propagation model and performs the calculation in combination with the information provided by the aerodynamic network topology matrix. For example, based on the benchmark center frequency 850 Hz and the pipe flow resistance coefficient on the propagation path, it applies a frequency-dependent attenuation model to calculate that after propagating to the adjacent branch, the main energy of the signal will be attenuated and concentrated to a new, lower center frequency, such as 180 Hz. At the same time, a dispersion model is also applied, for example, to calculate that the benchmark pulse with a duration of 12 ms will be spread to 30 ms after propagation. Finally, instead of a vague estimate, a specific, quantified prediction of the time-frequency feature of the crosstalk signal is obtained.

[0176] In this way, through this deep integration of the packaging process knowledge base and the physical propagation model, the wavelet domain mask constructed by the system can achieve unprecedented accuracy. The target it wants to suppress is no longer any possible noise, but a crosstalk signal that has a specific fingerprint, appears at a specific time, and in a specific frequency band, and can be accurately predicted. This makes the purification effect of the segmented error signal optimal, provides the highest quality data input for subsequent model adaptive correction, and improves the intelligence and reliability of the entire system to a new level.

[0177] As an optional implementation, the obtaining of the time delay information comprises:

[0178] From the plurality of sensing units, a sensing unit that receives the strongest energy or the earliest time of the residual error signal is determined as a reference sensing unit;

[0179] For at least one other sensing unit in addition to the reference sensing unit, the residual error signal of the reference sensing unit and the residual error signal of the other sensing unit are respectively subjected to Fourier transform to obtain their respective frequency spectrum signals;

[0180] Based on the obtained frequency spectrum signals, a cross-power spectrum is calculated, and the cross-power spectrum is subjected to phase transform weighting to obtain a whitened frequency spectrum;

[0181] The whitened frequency spectrum is subjected to inverse Fourier transform to obtain a cross-correlation function;

[0182] The peak position of the cross-correlation function is identified to determine the time delay between the reference sensing unit and the other sensing unit;

[0183] One or more time delays thus determined are combined as the time delay information.

[0184] As an optional implementation, the step of performing inversion calculation based on the pre-stored pneumatic network topology matrix to determine the branch identification where the micro-leakage event occurs comprises:

[0185] discretizing the target pneumatic network into a plurality of candidate leakage points based on the pneumatic network topology matrix;

[0186] for each of the candidate leakage points, calculating a theoretical time delay when a leakage occurs at the candidate leakage point by using the pneumatic network topology matrix;

[0187] comparing the calculated theoretical time delay with the time delay information to calculate a leakage position likelihood value corresponding to each of the candidate leakage points;

[0188] determining the candidate leakage point with the largest leakage position likelihood value, and determining the branch where the candidate leakage point is located as the branch identification where the micro-leakage event occurs.

[0189] To further improve the robustness and accuracy of the leakage positioning in the case of extremely low signal-to-noise ratio of the micro-leakage signal and possible minor inconsistencies in the measurement data, the positioning procedure in step S105 can be implemented by using a more advanced algorithm combination.

[0190] After the positioning procedure is started, a high-precision time delay information acquisition process is first performed. The first step of the process is to pre-process the multiple residual signals to determine a reference sensing unit.

[0191] Specifically, the data processing device 200 can calculate the total energy or peak amplitude of each residual signal in a short time window, and dynamically select the sensing unit corresponding to the signal with the strongest energy or the earliest signal arrival time as the reference sensing unit for the current positioning calculation.

[0192] Then, for the other relevant sensing units in the network except the reference sensing unit, the system pairs the residual signals of the sensing units with the residual signal of the reference sensing unit, and performs generalized cross-correlation-phase transform (GCC-PHAT) calculation for each pair. The core of the calculation is to abandon the amplitude information which is easily contaminated by noise, and focus on the most reliable phase relationship between signals.

[0193] In a specific implementation, the first step is to perform a Fast Fourier Transform (FFT) on both time series signals to obtain their complex spectra. Then, the cross-power spectrum is calculated and normalized by a phase transform weighting function, i.e., dividing the cross-power spectrum by its own modulus. The mathematical effect of this weighting operation is to whiten the spectrum, and the physical meaning is to greatly suppress the frequency components dominated by noise, and only retain the phase spectrum information that accurately reflects the time delay. Subsequently, an Inverse Fast Fourier Transform (IFFT) is performed on the weighted and whitened spectrum to obtain a cross-correlation function with an extremely sharp peak.

[0194] Finally, a peak search algorithm, such as maximum value finding or parabolic interpolation, is used to accurately identify the peak position of the function and convert it into the precise time delay between the two sensing units. After completing the pairing calculation for all relevant sensing units, the system obtains a set of high-precision time delay information relative to the same reference point.

[0195] After obtaining this set of high-precision time delay information, the system then starts the inversion calculation program based on maximum likelihood estimation to achieve high-robust positioning. The program first generates a series of discrete candidate leak points along all paths of the pipeline network with a preset physical step size, such as 5 centimeters, based on the aerodynamic network topology matrix, thereby converting the continuous pipeline network space into a discrete point set for searching.

[0196] Subsequently, the program enters a traversal calculation loop. For each candidate leak point, it uses the pipe length and sound speed information stored in the topology matrix to calculate the theoretical propagation time required for the pressure wave to propagate to the reference sensing unit and all other relevant sensing units when the leak source is precisely located at this point, and thereby obtains a set of theoretical time delays.

[0197] Next, the program compares this set of theoretical time delays with the actual time delay information calculated by the GCC-PHAT algorithm to evaluate the "credibility" of the current candidate point.

[0198] In a specific implementation, the leak location likelihood value can be obtained by calculating the Gaussian probability density between the two sets of time delay vectors; in short, the smaller the difference between the theoretical and actual values, the higher the likelihood value of the candidate point.

[0199] After traversing all the candidate leakage points and calculating the respective likelihood values, the program performs a maximum value search operation to find the candidate leakage point with the global maximum leakage position likelihood value. Finally, the system queries the node information table of the topology matrix to determine the branch to which the optimal candidate point belongs, and outputs the ID of the branch as the final branch identification of the micro-leakage event. By combining the robust time delay estimation algorithm of GCC-PHAT with the robust positioning algorithm of maximum likelihood, the embodiment can still achieve accurate and reliable positioning of the micro-leakage event even in extremely harsh noise environments.

[0200] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent replacements for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An automated packaging equipment remote monitoring system, characterized in that, The method comprises the following steps: a plurality of sensing units are arranged at the inlet ends and the terminal ends of each branch of the target pneumatic network, and are used to synchronously collect the inlet end pressure value, the terminal end pressure value and the flow value of each branch under a unified clock reference, and generate synchronous raw data; a data processing device is in communication connection with the sensing units and the automatic packaging equipment controller, the data processing device stores a digital twin model, and is configured to: receive a process beat parameter from the automatic packaging equipment controller; based on the process beat parameter and the digital twin model, calculate a theoretical pulsation reference waveform of the current working condition; differentially process the synchronous raw data and the theoretical pulsation reference waveform to generate a residual signal; determine the occurrence confidence of a micro-leakage event according to the residual signal, and generate micro-leakage event information in response to the occurrence confidence being greater than or equal to a preset threshold; based on the micro-leakage event information, perform cross-correlation calculation on the residual signals of different branches to obtain time delay information, and perform inversion calculation in combination with a pre-stored pneumatic network topology matrix to determine the branch identifier of the micro-leakage event, and send a micro-leakage event alarm instruction to the automatic packaging equipment controller based on the branch identifier.

2. The remote monitoring system for an automated packaging apparatus of claim 1, wherein, Further comprising: an automatic time calibration device configured to start and manage a time calibration process of the sensing units and the data processing device, and determine and store time calibration values corresponding to each sensing unit respectively; wherein the sensing units are further configured to record an actual arrival time stamp of a standard physical pulse received by the sensing units during the time calibration process, and provide the actual arrival time stamp to the automatic time calibration device; the data processing device is further configured to calculate a theoretical arrival time of the standard physical pulse to each sensing unit during the time calibration process, and provide the theoretical arrival time to the automatic time calibration device; the automatic time calibration device is further configured to compare the actual arrival time stamp and the theoretical arrival time to determine the time calibration value; the sensing units are further configured to correct time information using the time calibration value when generating the synchronous raw data.

3. The remote monitoring system for an automated packaging apparatus of claim 2, wherein, The automatic time calibration device is further configured to: obtain a process beat parameter from the automatic packaging equipment controller, and identify a preset idle process period based on the process beat parameter.

4. The remote monitoring system for an automated packaging apparatus of claim 3, wherein, The automatic time calibration device is further configured to: send a calibration control instruction to the automatic packaging equipment controller to drive an actuator in the target pneumatic network to generate the standard physical pulse during the preset idle process period.

5. The remote monitoring system for an automated packaging apparatus of claim 1, wherein, The data processing device is further configured to: define the residual signal as a model error signal in response to the occurrence confidence being lower than the preset threshold; adaptively correct one or more model parameters in the digital twin model using the model error signal.

6. The remote monitoring system for an automated packaging apparatus of claim 5, wherein, The adaptive correction of one or more model parameters in the digital twin model comprises: parsing the process beat parameter into an actuator action event, and determining one or more target branches associated with the actuator action event by using the pre-stored pneumatic network topology matrix; extracting a segmented error signal associated with the target branch in time from the model error signal; updating only the dynamic physical parameters related to the target branch in the digital twin model by using the segmented error signal.

7. The remote monitoring system for an automated packaging apparatus of claim 6, wherein, The extracting a segmented error signal associated with the target branch in time from the model error signal comprises: performing multi-scale wavelet transform on the model error signal to obtain wavelet coefficients distributed in time-frequency domain; identifying one or more adjacent branches adjacent to the target branch by using the pneumatic network topology matrix, and predicting time-frequency features of crosstalk signals induced in the adjacent branches by the actuator action event based on the topological relationship between the adjacent branches and the target branch; constructing a wavelet domain mask based on the time-frequency features; filtering the wavelet coefficients by using the wavelet domain mask, and then performing inverse wavelet transform to reconstruct and obtain the purified segmented error signal.

8. The remote monitoring system for an automated packaging apparatus of claim 7, wherein, The data processing device also stores a packaging process feature library, which stores reference time-frequency features corresponding to a plurality of preset packaging action types of the automatic packaging equipment; The predicting time-frequency features of crosstalk signals induced in the adjacent branches by the actuator action event comprises: identifying the packaging action type corresponding to the actuator action event from the actuator action event; extracting the reference time-frequency features corresponding to the identified packaging action type from the packaging process feature library; calculating the time-frequency features of the crosstalk signals based on the reference time-frequency features and the topological relationship between the adjacent branches and the target branch.

9. The remote monitoring system for an automated packaging apparatus of claim 1, wherein, The obtaining time delay information comprises: determining a sensing unit that receives the residual error signal with the strongest energy or the earliest time from the plurality of sensing units as a reference sensing unit; performing Fourier transform on the residual error signal of the reference sensing unit and the residual error signal of at least one other sensing unit except the reference sensing unit to obtain their respective frequency spectrum signals; calculating cross-power spectrum based on the obtained frequency spectrum signals, and performing phase transformation weighting on the cross-power spectrum to obtain a whitened spectrum; performing inverse Fourier transform on the whitened spectrum to obtain a cross-correlation function; identifying the peak position of the cross-correlation function to determine the time delay between the reference sensing unit and the other sensing unit; combining one or more time delays thus determined into the time delay information.

10. The remote monitoring system for an automated packaging apparatus of claim 9, wherein, The combining the pre-stored pneumatic network topology matrix to perform inversion calculation to determine the branch identifier of the micro-leakage event comprises: discretizing the target pneumatic network into a plurality of candidate leakage points based on the pneumatic network topology matrix; For each of the candidate leak points, a theoretical time delay when a leak occurs at the candidate leak point is calculated by using the pneumatic network topology matrix; The calculated theoretical time delay is compared with the time delay information to calculate a leak location likelihood value corresponding to each of the candidate leak points; A candidate leak point with a largest leak location likelihood value is determined, and a branch in which the candidate leak point is located is determined as the branch identification of the micro-leak event.

Images