A real-time production process traceability monitoring system based on the Internet of Things

Abstract

The present application relates to the technical field of production process traceability monitoring, and discloses a real-time production process traceability monitoring system based on an Internet of Things, which comprises a monitoring module configured to encapsulate business data and an environment matching degree index generated by comparing real-time environment resonance and a baseline fingerprint into an active data packet; and a central processing module configured to authenticate data credibility and update the baseline fingerprint in a traceable manner in response to a batch qualification management event, wherein the present application enables each production record to carry internal evidence of its context environment when generated, realizes a change from static evidence to dynamic self-evidence, solves the vacuum dilemma of data credibility, and simultaneously, the system has a dynamic self-healing capability through a mechanism of reverse calibration of a physical monitoring model by commercial achievements, and improves the robustness of long-term operation in a complex environment.

Description

Technical Field

[0001] This invention relates to a real-time production process traceability and monitoring system based on the Internet of Things, belonging to the field of production process traceability and monitoring technology. Background Technology

[0002] In current industrial production management and supervision practices, deploying IoT monitoring systems to collect business data from key processes and adding timestamps for archiving is a common technical means to ensure product quality traceability. This method provides basic data credentials for production records. However, when this technical means is applied to production activities with strict requirements for the integrity and auditability of data records, the inherent limitations of its information structure begin to affect the effectiveness of management and supervision. Specifically, a production record usually only contains business data values ​​and their corresponding time information. The record itself does not contain objective information to explain the stability of the local physical environment at the time of its generation. When conducting quality audits or fault tracing, it is impossible to rule out the possibility of data distortion caused by instantaneous abnormal sensor states or sudden interference in the local environment based solely on the record itself. There is a lack of direct technical connection between the record and its generation environment.

[0003] To compensate for this deficiency, a common technical approach is to add an independent equipment status monitoring unit to the monitoring system. However, this essentially establishes a parallel status data acquisition system. In management practice, complex correlation analysis and logical judgment are still required afterward for business data and status data from different sources, without changing the incompleteness of the original business data records in the information structure. In addition to the inherent defects at the data structure level, existing monitoring technologies also have limitations at the logical level of information acquisition and authentication, making it difficult to handle complex management inquiries. For example, Chinese utility model patent CN218734718U discloses an IoT-based device data acquisition light, device monitoring system, and device. The core of its technical solution is to collect data from the device's built-in three-color indicator light through an external sampling circuit. The system transmits the signals, which are then used by the Internet of Things (IoT) module to remotely transmit these light signals that indicate the device's operating, waiting, or abnormal status. While this solution addresses the issue of remote visibility of device status to some extent, its core technology remains a passive collection and forwarding of the device's own diagnostic results. It cannot independently and objectively assess the device's operating environment, nor can it answer a crucial auditing question: when a device's indicator light is green (in normal operation), is the resulting business data necessarily reliable? This method inherits the information limitations of the original indicator light system and fails to address the fundamental problem of the lack of intrinsic connection between data records and their generating environment.

[0004] Specifically, the technology suffers from the following shortcomings: 1. There is a lack of structured intrinsic correlation between business data and the local physical environment at the time of its generation. The data record itself lacks the ability to self-verify its context, and its credibility relies on external or offline corroboration; 2. The baseline model used to determine whether the environment is normal is usually static. When production equipment undergoes normal wear and tear or a slow, benign evolution in operating conditions, this static baseline cannot adapt synchronously, which may lead to misjudging new normal operating conditions as abnormal and generating erroneous management alerts. Therefore, how to establish a new data organization and authentication mechanism that enables each business data record to contain internal verification of its generated environment state, and enables the environmental baseline model in this mechanism to dynamically self-correct based on the final results of production activities, becomes the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a real-time production process traceability and monitoring system based on the Internet of Things. Its main purpose is to solve the problems that the data records in the existing production traceability system do not have the ability to prove their context, and that their static environmental baseline model cannot adapt to the long-term benign evolution of production conditions.

[0006] To achieve the above objectives, the present invention provides a real-time production process traceability and monitoring system based on the Internet of Things, which includes a monitoring module and a central processing module:

[0007] The monitoring module is configured to: store an environmental resonance baseline fingerprint, which characterizes the local physical environment of the monitoring module in a healthy operating state; where changes in the local physical environment caused by benign aging of equipment or evolution of operating conditions may cause the environmental resonance baseline fingerprint to become invalid, thereby triggering a false alarm problem of misjudging a new healthy operating state as a continuous abnormality; when reporting business data, the current environmental resonance signal is collected in real time, and the correlation between the current environmental resonance signal and the environmental resonance baseline fingerprint is calculated to generate an environmental matching degree index. Then, the business data and the environmental matching degree index are encapsulated into an active data packet and reported.

[0008] The central processing module is configured to: receive active data packets and authenticate the credibility of business data based on environmental matching indicators; to solve the false alarm problem, execute a traceable baseline recalibration rule: the traceable baseline recalibration rule is triggered by a batch qualification management event associated with a preset production batch and originating from the enterprise management system; once triggered, perform statistical cluster analysis on the environmental resonance information in multiple historical active data packets archived during the preset production batch, and when a new stable statistical cluster is confirmed to be formed, generate an updated environmental resonance baseline fingerprint based on the new stable statistical cluster and transmit it to the monitoring module to replace its original environmental resonance baseline fingerprint.

[0009] Preferably, the process of generating the environmental resonance baseline fingerprint includes: during the system debugging phase, under standard production conditions, collecting the background resonance signal of the local environment where the monitoring module is located, and processing the background resonance signal to obtain the initial environmental resonance baseline fingerprint.

[0010] Preferably, both the environmental resonance baseline fingerprint and the current environmental resonance signal are spectral features. The spectral features are generated by performing a fast Fourier transform on the background vibration signal or background electromagnetic noise signal collected by the monitoring module to extract the main frequency components and corresponding amplitude information in the signal spectrum.

[0011] Preferably, the monitoring module is further configured to: respond to an event where the environmental matching index is lower than a preset authentication threshold; perform morphological analysis on the current environmental resonance signal to determine a management attribution type characterizing the abnormal nature of the current environmental resonance signal; further include the management attribution type in the reported active data packet; the central processing module is further configured to: trigger a differentiated management response workflow based on the management attribution type.

[0012] Preferably, the monitoring module is further configured to: generate a residual signal representing the difference between the current environmental resonance signal and the environmental resonance baseline fingerprint during the comparison process; determine one or more degradation characteristic indicators by quantifying the energy of the residual signal at a specific frequency defined in the warning frequency characteristic library based on a preset warning frequency feature library related to equipment degradation; further include one or more degradation characteristic indicators in the active data packet; and the central processing module is further configured to: analyze the changing trend of one or more time-seriesd degradation characteristic indicators to generate predictive maintenance alarms.

[0013] Preferably, the rule for the central processing module to analyze the changing trend is as follows: perform linear regression analysis on the time-series degradation characteristic indicators to calculate a trend slope that characterizes the growth rate of the degradation characteristic indicators. Among them, when the trend slope satisfy When this happens, a predictive maintenance alert is generated, in which... This is a preset slope threshold.

[0014] Preferably, the central processing module is further configured to: respond to a management event related to a planned change in the production environment; at the scheduled effective time of the planned change, send an instruction to the monitoring module to cause it to actively suspend the authentication function based on the environmental resonance baseline fingerprint and enter the baseline learning mode; in the baseline learning mode, construct a candidate baseline fingerprint by performing cluster analysis on the newly collected environmental resonance information; and respond to an instruction confirming that the new production state has stabilized, set the candidate baseline fingerprint as the new environmental resonance baseline fingerprint of the monitoring module.

[0015] Preferably, the instruction confirming that the new production status has stabilized is generated by production management personnel through the human-machine interface after the first batch of products under the new production configuration passes the first process inspection.

[0016] Preferably, the morphological analysis includes: calculating the energy standard deviation of the current environmental resonant signal in the time domain; when the energy standard deviation is higher than a first preset morphological threshold, determining the administrative attribution type as an impulsive anomaly; calculating the spectral entropy of the current environmental resonant signal in the frequency domain; when the spectral entropy is lower than a second preset morphological threshold, determining the administrative attribution type as a periodic anomaly; when the spectral entropy is higher than the second preset morphological threshold, determining the administrative attribution type as a broadband random anomaly.

[0017] Preferably, the monitoring module is further configured to encapsulate the data obtained by compressing the spectral characteristics of the current environmental resonance signal in the active data packet; the central processing module is further configured to associate the received active data packet with the corresponding production batch number and archive it as the data basis for executing the traceable baseline recalibration rule.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. This invention encapsulates business data with real-time collected environmental matching indicators to form an active data package and reports it. This structure enables each production record to carry internal verification of its own generation environment at the time of generation. In management and supervision activities, the certification center processing module can directly verify the credibility of business data based on the environmental matching indicators embedded in the data package. This method transforms the traditional data credibility model that relies on external offline auditing into a new model in which the data record itself provides real-time, online context self-verification, solving the long-standing problem of record context vacuum in production traceability systems.

[0020] 2. Upon receiving a batch qualification management event from the enterprise management system, a traceable baseline recalibration rule is triggered. This rule retrieves and analyzes all historical active data packets archived within the production cycle of a specific qualified batch. When it is confirmed that the environmental resonance information in these data has formed a new stable statistical cluster, the system generates an updated environmental resonance baseline fingerprint based on this cluster and transmits it to the monitoring module. This process constructs an information feedback loop that reverse-calibrates the physical monitoring model based on business management results, enabling the monitoring system to have dynamic self-healing capabilities. This avoids the problem of false alarms caused by baseline failure due to benign aging of equipment or evolution of operating conditions, which could lead to misjudging new healthy operating conditions as continuous abnormalities.

[0021] 3. When the monitoring module determines that the environmental matching index is lower than the preset authentication threshold, it will perform morphological analysis on the current environmental resonance signal to determine a management attribution type that characterizes the abnormal nature of the signal, and encapsulate it in an active data packet. After receiving the attribution type, the central processing module can trigger a differentiated management response workflow accordingly. This mechanism utilizes the same environmental resonance signal collected for authentication credibility, and provides the management system with a preliminary classification of the cause of the anomaly without changing the hardware configuration, so that subsequent investigation and maintenance work can be scheduled more accurately.

[0022] 4. During the process of comparing the current environmental resonance signal with the environmental resonance baseline fingerprint in the monitoring module, a residual signal representing the difference between the two is generated. Based on a preset warning frequency feature library related to equipment degradation, one or more degradation characteristic indicators are determined by quantifying the energy of the residual signal at a specific frequency. The central processing module can generate predictive maintenance alarms by performing trend analysis on the time-series degradation characteristic indicators. This method extracts information that can characterize the long-term health status evolution of the equipment from the comparison process used for the immediate credibility of authentication data, enabling a single system to simultaneously monitor the credibility of current records and predict future equipment risks. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the core functions and data closed-loop process of the system of this invention;

[0024] Figure 2 This is a response curve of the environmental matching index of the present invention to external interference;

[0025] Figure 3 This is a diagram of the physical deployment and information interaction architecture of the system of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0027] This invention discloses a real-time production process traceability and monitoring system based on the Internet of Things (IoT). The system architecture consists of at least one monitoring module deployed at each workstation on the production floor, and a central processing module. The monitoring module's task is to collect business data and encapsulate it into a live data packet. The central processing module's task is to receive and parse this data packet to perform credibility authentication of production process records and dynamic maintenance of the monitoring model. The two modules interact via an industrial IoT communication protocol. In application scenarios with strict requirements for the integrity and auditability of production process records, a record containing only business data and a timestamp cannot respond to audit inquiries regarding the stability of the environment at the moment of data generation and the reliability of the sensors themselves. This vacuum in the record's context is a long-standing technical problem in the field of management and supervision. To address this problem, the system of this invention is configured to encapsulate a live data packet embedding environmental state self-certification information. The procedure is as follows: During the initial deployment and debugging phase of the system, under a verified standard production condition, the monitoring module utilizes its... An integrated accelerometer sensor acquires a 60-second local environmental background resonance signal. The module's built-in microcontroller performs a Fast Fourier Transform on this signal to extract its spectral characteristics, which are then stored in non-volatile memory as a reference characterizing the healthy operating state—this is the environmental resonance baseline fingerprint. After the system enters the real-time operation phase, the monitoring module executes a combined data acquisition and encapsulation process at predetermined intervals, e.g., once per second. Specifically, in the very short instant before acquiring business data, e.g., within 100 milliseconds, it first acquires a segment of the current environmental resonance signal and immediately calculates the correlation between the spectral characteristics of this real-time signal and the pre-stored environmental resonance baseline fingerprint, thus deriving a floating-point number between 0 and 1—this is the environmental matching index. Subsequently, the monitoring module encapsulates the business data, the environmental matching index, and the optional compressed spectral characteristics of the current environmental resonance signal into a structured active data packet and reports it. A numerical example is when the temperature sensor reading is... When the monitoring module calculates the correlation between the current environment and the baseline to be 0.98, the core fields of the generated active data packet are {Business Data: 25.1, Environment Matching Degree: 0.98}. After receiving the data packet, the central processing module's first action is to verify whether the environment matching degree index is higher than a preset authentication threshold, such as 0.95. A data record with a value higher than this threshold is confirmed by the system as having authenticated credibility, while one with a value lower than this threshold is marked as questionable. In this way, each production record carries internal verification of its own generation environment when it is generated, changing the judgment of data credibility from relying on external corroboration to a method in which the data record itself provides real-time contextual self-verification.

[0028] To acquire effective environmental resonance signals, the accelerometer in the monitoring module is fixed to the main load-bearing structure of the production equipment during deployment. Its sampling frequency is set to be more than 10 times the highest characteristic frequency of the key moving parts of the equipment, and its range covers 5 times the vibration amplitude during normal operation. After obtaining the spectral feature vectors of the current environmental resonance signal and the environmental resonance baseline fingerprint, the environmental matching degree index is calculated using a cosine similarity algorithm. This involves treating the two spectral feature vectors as vectors in a multi-dimensional space for calculation. The calculation result mainly reflects the similarity of the spectral shape and is not sensitive to the drift of the overall signal energy. The key thresholds required for system operation are determined through standardized calibration procedures executed during the initial deployment phase. The authentication threshold is determined by continuously collecting and calculating 1000 environmental matching degree indices under standard production conditions to form a health status index set. The average value of this dataset minus three times the standard deviation is taken as the authentication threshold. The trend slope threshold in the predictive maintenance function... Then, by applying a known, linearly-progressing simulated fault to the key component under controlled conditions, and performing linear regression analysis on the degradation characteristic index sequence collected during the process, 80% of the calculated slope is taken as the initial setting value of the threshold.

[0029] In long-term production practices, equipment may undergo slow, benign evolution due to normal wear and tear or changing operating conditions, causing the original environmental resonance baseline fingerprint to gradually become ineffective. Without adjustment, the system may misinterpret the new operating state as a continuous anomaly, generating false alarms. Therefore, this system employs a traceable baseline recalibration rule to dynamically correct the monitoring model. The execution flow of this rule is as follows: The central processing module subscribes to batch qualification management events in the enterprise management system through an information interface. This event serves as confirmation of the validity of all process records within a production cycle. When the central processing module receives a batch qualification event, for example, associated with production batch number BN20251009, it is triggered to execute a background recalibration task. First, it retrieves all historical activity records archived by all relevant monitoring modules during the production batch BN20251009. The system first generates a data packet; then, it performs a statistical cluster analysis on the massive amount of environmental resonance information within this time window. When the analysis finds that more than 80% of the data points have stably formed a new statistical cluster that is different from the old baseline in the latter half of the production cycle, the system determines that the centroid of the new cluster represents the latest healthy environmental resonance state of the current workstation. Finally, the central processing module calculates the updated environmental resonance baseline fingerprint based on the new stable statistical cluster and transmits it to the corresponding monitoring module via the network to replace its original environmental resonance baseline fingerprint. At the same time, the system records this model update event driven by management results in the audit log. Through such an information loop of reverse calibration of the physical monitoring model by management results, the monitoring system can adapt to the benign evolution of the production environment and avoid false alarms caused by baseline failure.

[0030] In production management, when the system issues an environmental anomaly alarm, a generic alarm cannot provide effective guidance for subsequent maintenance activities. Therefore, the system is also configured to provide a management attribution prediction when necessary. This mechanism is implemented by the monitoring module being triggered when the calculated environmental matching index is lower than a preset authentication threshold (e.g., 0.95). It then performs two parallel morphological analyses on the recently acquired environmental resonance signal that caused the alarm: first, it calculates the energy standard deviation of the signal in the time domain; a standard deviation higher than the first preset morphological threshold indicates drastic fluctuations in signal energy, thus determining the management attribution type as an impact-type anomaly; second, it calculates the spectral entropy of the signal in the frequency domain; a spectral entropy lower than the second preset morphological threshold means that energy is concentrated in a few frequencies. Based on the rate, the management attribution type is determined to be periodic anomaly, and vice versa for broadband random anomaly. The morphological threshold is determined during the system debugging phase by collecting samples of known types of interference events and performing statistical analysis. Finally, the monitoring module attaches the generated management attribution type, such as periodic, to the active data packet and reports it together. After receiving this information, the central processing module can trigger differentiated response workflows according to preset management rules. For example, alarms attributed to periodicity are automatically generated into work orders and dispatched to the equipment maintenance department, while broadband random alarms are given priority to be notified to electrical engineers. This mechanism uses the same environmental resonance signal collected for authentication credibility to provide the management system with a preliminary classification of the cause of the anomaly, enabling subsequent troubleshooting work to be scheduled more accurately.

[0031] To predict the evolution trend of equipment health status, the system also provides a predictive maintenance mechanism based on residual signal analysis. The degradation of critical components is a gradual process; the weak abnormal signals they generate, superimposed on the background resonance, may not be sufficient to trigger alarms for judging the reliability of immediate data. However, their continuous growth trend indicates a risk of failure. To capture this trend, the monitoring module generates a residual signal representing the difference between the current environmental resonance signal and the environmental resonance baseline fingerprint. Based on a specific equipment degradation-related early warning frequency feature library, the monitoring module determines one or more degradation characteristic indicators by quantifying the energy of the residual signal at specific frequencies defined in the early warning frequency feature library. These degradation characteristic indicators are encapsulated in active data packets and continuously received and archived by the central processing module. The central processing module performs trend analysis on the time-series degradation characteristic indicators. The rule is to perform linear regression analysis on the degradation characteristic indicators over a recent period (e.g., 1000 data points) to calculate a trend slope characterizing the growth rate of the degradation characteristic indicator. When the trend slope satisfy At that time, the system will issue a predictive maintenance alert, in which It is a preset slope threshold based on statistical analysis of the component's historical failure data; in this way, the system extracts information that can characterize the long-term health status evolution of the equipment from the comparison process of the real-time credibility of the authentication data, so that a system can simultaneously monitor the credibility of the current record and predict the future risks of the equipment.

[0032] To address planned and acute production environment changes caused by production task switching in flexible manufacturing scenarios, the system is also configured to execute a baseline-managed and guided relearning procedure. This procedure relies on the central processing module subscribing to and parsing management events such as production line configuration change orders or new product model switching instructions from the Manufacturing Execution System (MES) or Enterprise Resource Planning (ERP) system through its standard information interface. When the central processing module receives a management event indicating a future production line change, it sets a predetermined state switching task for all affected monitoring modules. At the effective time specified in the change order, the central processing module automatically sends an instruction to all relevant monitoring modules to enter baseline learning mode. The monitoring modules receiving the instruction then actively... The authentication function based on the original environmental resonance baseline fingerprint is suspended, and the old baseline fingerprint is cleared. This proactive suspension operation can avoid false alarms that will inevitably occur due to baseline mismatch during production line changes. In baseline learning mode, the monitoring module continuously collects and reports new environmental resonance signals. The central processing module performs cluster analysis on these new data to initially construct a new candidate baseline fingerprint. After the first batch of new products rolls off the production line and passes the first process inspection, the production management personnel send a command to the central processing module through the human-machine interface to confirm that the new production status has stabilized. The central processing module responds to this command, officially sets the aforementioned candidate baseline fingerprint as the new environmental resonance baseline fingerprint for all relevant monitoring modules, and instructs them to restore their complete authentication function.

[0033] Example 1: This example illustrates a specific application of the technical solution in a particular industrial scenario. In the manufacturing of an aerospace composite component, a curing oven station heat-treats a batch of carbon fiber wing spars. This process requires accurate temperature recording. During the curing cycle, an auxiliary cooling device near the curing oven experiences intermittent abnormal vibrations due to bearing wear. The device is shut down after several minutes. During this period, the operational data recorded by the monitoring module deployed on the curing oven—namely, the curing temperature—shows that the values ​​are within the tolerance range required by the process requirements throughout. In a subsequent quality audit, auditors, based on independent equipment maintenance records, questioned the validity of the temperature records during this period. The problem lies in the inability to eliminate the potential risk of abnormal physical vibrations disturbing temperature sensor readings based solely on temperature data; that is, the limitations of traditional monitoring methods. The control records could not provide direct proof of the physical environment at the time of their generation. In this case, the records archived by the system of this invention were retrieved. Before the abnormal vibration occurred, the environmental matching index embedded in all active data packets reported by the monitoring module was stable above the authentication threshold of 0.95. During the abnormal vibration period shown in the maintenance records, the system log showed that although the business data, i.e. the temperature value, in the active data packets did not change, its environmental matching index had continuously dropped to below 0.7. Furthermore, when the environmental matching index was lower than the authentication threshold, the system triggered a management attribution prediction mechanism. The processor performed morphological analysis on the current environmental resonance signal and added a periodic abnormal management attribution type to these low-matching active data packets. This attribution type was consistent with the bearing wear fault recorded in the equipment maintenance log in terms of physical characteristics.

[0034] The system's environmental matching index, serving as the primary criterion for data credibility, identifies instability in the generating environment even when the business data itself remains unchanged, thereby adjusting the confidence level of the record in management. Signals deemed to be environmental anomalies directly serve as input, triggering management attribution type analysis and providing supervisors with a judgment on the nature of the disturbance source, transforming an environmental anomaly alarm into a management event with technical implications. Based on this set of active data packets containing environmental self-certification information and anomaly attribution types, the quality management department ultimately makes the management decision to isolate the batch of wing beams and perform further non-destructive testing. Auditors also confirm the integrity of the traceability monitoring system's management information under complex operating conditions. By binding the credibility certification of process data with the morphological analysis of the abnormal environment within the single information structure of the active data packet, the system solves the technical problem of the separation between process data records and equipment status information, which prevents the formation of a closed-loop evidence chain. This makes production records a management information carrier capable of dynamically representing the health of their contextual environment.

[0035] Example 2: To objectively verify the ability of the technical solution of the present invention to identify and characterize interference that does not affect business data but changes the local physical environment, the following experiment was conducted; the experiment was carried out in a simulated industrial environment, the core of which was a constant temperature test chamber, whose internal temperature was stably controlled at [temperature value missing]. The fluctuation range is no greater than The test platform is configured as follows: a monitoring module using the technical solution of this invention, serving as the test group, is installed on a fixed bracket inside the test chamber. This module integrates a temperature sensor and an acceleration sensor, wherein the sampling frequency of the acceleration sensor is set to 1kHz; a traditional monitoring node containing only a temperature sensor of the same specification serves as the control group, and is installed side by side on the same bracket; a programmable micro-amplitude vibration source is coupled to the fixed bracket to apply external environmental interference; a central processing module is used to synchronously record and process data from the test group and the control group; the test process is divided into three stages: the first stage is baseline establishment and stable operation, under constant temperature and no external vibration conditions, the monitoring module of the test group... First, a 60-second background vibration signal was collected, and its environmental resonance baseline fingerprint was processed and established according to the procedures in the specific implementation method. Then, the experimental group and the control group ran continuously for 10 minutes in this stable state. The second stage was a periodic interference test, in which a micro-amplitude vibration source was activated to apply a continuous vibration with a frequency of 80Hz to the support without producing a temperature rise effect, and maintained for 5 minutes. The third stage was an impact interference test, in which a series of instantaneous low-energy mechanical impacts were applied to the support at 10-second intervals using a pneumatic device, and maintained for 5 minutes. In all three stages, the experimental group reported an active data packet containing business data and environmental matching indicators, while the control group only reported business data, and the reporting frequency was 1Hz.

[0036] Throughout the experiment, the operational data, namely temperature readings, in both the control and experimental groups remained at [a certain level]. Within the specified range, no fluctuations caused by external interference were observed. The environmental matching index in the active data packets reported by the test group showed changes consistent with the environmental state, and the specific data is recorded in Table 1. During the stable operation in the first stage, the environmental matching index remained above 0.95. After entering the second stage, the index decreased and remained in the range of 0.70 to 0.75. This is because the 80Hz periodic vibration component was introduced into the current environmental resonance signal, which caused the result obtained after the correlation calculation of its spectral characteristics with the baseline fingerprint that does not contain this frequency component to decrease. In the third stage, the index changed periodically between a reference value close to 1.0 and a valley value below 0.60, reflecting the change in the environmental resonance state caused by each instantaneous impact.

[0037] Table 1: Comparison of key system indicators at different experimental stages.

[0038]

[0039] Experimental data shows that the monitoring system provided by the technical solution of this invention can distinguish different operating states of the local physical environment by changing the environmental matching degree index in the active data packet, while keeping its core business data stable. The experimental results verify that the system can provide contextual environmental state information at the time of production process records, thereby providing an objective technical basis for data credibility auditing in management and supervision activities.

[0040] Example 3: This example combines Figures 1 to 3 This describes a real-time production process traceability and monitoring system based on the Internet of Things (IoT). Figure 1 As shown, the process begins with the production process input. The data acquisition and encapsulation module generates an active data packet containing embedded environmental self-certification information by combining business data with environmental resonance signals. This data packet is sent to the credibility authentication module, which judges the validity of the data based on the comparison with the environmental resonance baseline fingerprint. The authentication results are divided into certified data for quality audit and questionable data that triggers differentiated management. On the other hand, the certified data is archived as a historical active data packet. When a batch qualification management event is received, the system triggers the traceability baseline recalibration module, which reversely calibrates the physical monitoring model based on business results by analyzing historical data and generates an updated baseline fingerprint to dynamically maintain the benchmark. At the same time, the residual signal generated during the data acquisition process is sent to the predictive maintenance analysis module to predict equipment deterioration trends and generate predictive maintenance alarms.

[0041] like Figure 2 The solid line represents the real-time changing environmental matching index, and the dashed line is the preset authentication threshold of 0.95. The curves intuitively show that during the initial stable operation phase from T+0 min to T+9 min, the index is consistently higher than the authentication threshold. During the periodic interference phase from approximately T+10 min to T+15 min, the index is consistently lower than the authentication threshold. After the subsequent impact interference phase from T+16 min onwards, the index fluctuates drastically around the authentication threshold.

[0042] like Figure 3 As shown, the architecture is as follows: the physical environmental resonance generated by production equipment such as curing ovens is collected by intelligent monitoring nodes containing acceleration and temperature sensors. The data is transmitted to the central processing block in the central application server through the Industrial Internet of Things. This server interacts with the database server that stores the production process traceability database. The database also interfaces with the enterprise management system ERP / MES through the management event interface. At the same time, the central application server is connected to the management terminal through the enterprise intranet, providing a human-machine interface for quality audit and production management personnel, thus forming a complete closed loop of data collection, processing, storage and application.

[0043] Example 4: In a continuously operating production environment, normal equipment wear or changes in operating conditions can cause the preset environmental resonance baseline fingerprint to drift, potentially misjudging new healthy operating conditions as abnormal. The system of this invention addresses this problem through a retrospective baseline recalibration rule triggered by batch qualification management events. Its core lies in statistical cluster analysis of historical data. The specific procedure for this analysis is as follows: When the central processing module receives a batch qualification management event associated with a specific production batch from the enterprise management system, it first retrieves all historical active data packets archived by the relevant monitoring module during the production period of that specific batch from the historical database. Subsequently, the system extracts the spectral features contained in these data packets and uses these features as a set of data points in a multidimensional space. A density-based spatial clustering algorithm, namely the DBSCAN algorithm, is used to identify whether a new data cluster center exists. The execution of this algorithm depends on two key parameters: neighborhood radius. With minimum sample size ;in, The value is determined based on the dispersion of the sample data collected during the initial calibration phase under healthy operating conditions. Specifically, it involves calculating the average Euclidean distance between all spectral feature vectors within the initial calibration sample set and setting 1.5 times this average distance as the mean. This value, this setting, matches the neighborhood radius to the inherent volatility of the environmental signals at that particular workstation; The value of is related to the total amount of data in the analysis window. It is set to 0.1% of the total number of data points in the batch. This setting associates the minimum number of samples with the total amount of data.

[0044] After the parameters are determined, the processor in the central processing module executes the DBSCAN algorithm and marks all data points as core points, boundary points, or noise points. After the algorithm is completed, the system performs a validity judgment on the clustering results. The judgment logic is as follows: if the number of data points contained in the largest identified cluster accounts for more than 70% of the total number of data points in the last third of the production cycle of that batch, then the system determines that the cluster represents a stable and healthy operating condition. Assuming a qualified batch contains 30,000 data points, then... When the value is set to 30, if the largest cluster contains 8,500 of the data points in the last 10,000 data points of the batch, the cluster is considered valid. After the cluster is confirmed to be valid, the system calculates the geometric center, i.e. the centroid, of all data points in the largest cluster and uses the spectral characteristics of the centroid as the updated environmental resonance baseline fingerprint. This fingerprint is then transmitted to the corresponding monitoring module via the network to replace the original environmental resonance baseline fingerprint.

[0045] Example 5: To establish an environmental resonance baseline fingerprint characterizing the healthy operating status of a production workstation when the system is initially deployed, the following pre-calibration procedure needs to be performed: First, production management personnel confirm that the production equipment and its auxiliary systems at the workstation are all under verified standard production conditions and eliminate all non-standard interference sources; then, the field engineer sends an instruction to the monitoring module to be calibrated to enter baseline learning mode; in this mode, the monitoring module continuously collects and reports the background resonance signal of its local environment, with a duration covering several complete standard production operation cycles; after receiving all resonance signal data during this period, the central processing module performs statistical cluster analysis on the dataset to identify and extract the most concentrated data subset, and calculates and sets the statistical central feature of this data subset as the initial environmental resonance baseline fingerprint of the monitoring module. This procedure provides a stable reference benchmark for subsequent calculation of environmental matching index.

[0046] Example 6: This example describes the method for establishing the data model required for predictive maintenance functions. To calibrate the predictive maintenance alarm function of the system, the warning frequency feature library in the central processing module needs to be populated with data. The procedure can be one of the following methods: First, technicians consult the technical manuals and specifications of key components, such as bearings and motors, of the monitored equipment to obtain their theoretical fault characteristic frequencies, and enter these frequency data and corresponding component and fault mode information into the warning frequency feature library. Second, during the controlled maintenance period, a known slight and reversible fault simulation signal is applied to the key components, such as adding a small counterweight to the shaft to simulate an unbalanced state. At the same time, the monitoring module collects the environmental resonance signal under this condition, and by performing spectrum analysis on the signal, the specific frequency response caused by the simulated fault is identified. This frequency, which has been verified by actual measurement, is recorded as a degradation feature and associated with the warning frequency feature library. After the warning frequency feature library is populated, the system can analyze the energy change trend of the residual signal at these specific frequencies during subsequent monitoring, i.e., calculate the trend slope. and with the preset slope threshold By making comparisons, early warnings can be issued for early signs of equipment degradation.

[0047] To further demonstrate the non-obviousness of the technical solution of the present invention, and the beneficial effects compared with conventional technical paths in the field, the following comparative examples are provided.

[0048] Comparative Example 1: To verify the limitations of conventional technical solutions in real industrial audit scenarios in the absence of the core technical features of the active data package and traceable baseline recalibration described in this invention, this comparative example experiment was conducted. This comparative example adopts a common enhanced monitoring deployment scheme in the prior art. This scheme has two independent monitoring nodes installed side by side on an aerospace composite material curing oven station that is exactly the same as in Example 1. These nodes are connected to the same data server via an industrial network: a high-precision temperature monitoring node for recording the curing temperature; and an independent equipment vibration monitoring node (integrated with an acceleration sensor) for recording the operating vibration status of the auxiliary cooling equipment of the curing oven. Both nodes report and archive their collected data independently with timestamps added at a frequency of 1Hz. The test process reproduced the real fault scenario described in Example 1: In the middle of the curing cycle, an auxiliary cooling device experienced intermittent abnormal vibration for 5 minutes due to bearing wear. The device was then manually shut down. The test aims to verify the effectiveness and probative value of the data archived by this conventional technical solution in the subsequent quality audit. The key data recorded and archived by the data server during the test are shown in the table below:

[0049] Table 2: Data Recording Table for Comparative Example 1.

[0050]

[0051] The experimental results show that the technical solution of independently archiving multi-source heterogeneous data presents the following technical challenges in quality auditing: temperature data records and vibration data records are two datasets that are completely separate in terms of information structure. Auditors can simultaneously observe acceptable temperature values ​​and abnormal vibration values, but they cannot establish a definite direct correlation between the two at any given time point based solely on the data itself; for temperature values ​​recorded at the three time points T0+20min, T0+22min, and T0+24min... and The possibility of abnormal physical vibrations causing instantaneous disturbances to temperature sensor readings cannot be ruled out. Since the temperature record itself does not carry any self-verifying information about the environmental state at the time of generation, its credibility relies entirely on offline inferences and judgments made by auditors based on external knowledge. The data record itself does not have self-verifying capabilities. The test results confirm that without the active data packet mechanism of this invention, which structures and encapsulates business data and environmental matching indicators at the time of acquisition, even if a separate equipment status monitoring unit is added to the conventional technical solution, the production record generated still has the inherent defect of a context vacuum. It cannot provide direct, objective, and online internal verification of the credibility of its own data in a strict quality audit, thus failing to solve the problems existing in the background technology.

[0052] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A real-time production process traceability and monitoring system based on the Internet of Things, characterized in that, The system includes a monitoring module and a central processing module: The monitoring module is configured to: store an environmental resonance baseline fingerprint, which characterizes the local physical environment of the monitoring module in a healthy operating state; where changes in the local physical environment caused by benign aging of equipment or evolution of operating conditions may cause the environmental resonance baseline fingerprint to become invalid, thereby triggering a false alarm problem of misjudging a new healthy operating state as a continuous abnormality; when reporting business data, the current environmental resonance signal is collected in real time, and the correlation between the current environmental resonance signal and the environmental resonance baseline fingerprint is calculated to generate an environmental matching degree index. Then, the business data and the environmental matching degree index are encapsulated into an active data packet and reported. The central processing module is configured to: receive active data packets and authenticate the credibility of business data based on environmental matching indicators; to solve the false alarm problem, execute a traceable baseline recalibration rule: the traceable baseline recalibration rule is triggered by a batch qualification management event associated with a preset production batch and originating from the enterprise management system; once triggered, perform statistical cluster analysis on the environmental resonance information in multiple historical active data packets archived during the preset production batch, and when a new stable statistical cluster is confirmed to be formed, generate an updated environmental resonance baseline fingerprint based on the new stable statistical cluster and transmit it to the monitoring module to replace its original environmental resonance baseline fingerprint.

2. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 1, characterized in that, The process of generating the environmental resonance baseline fingerprint includes: during the system debugging phase, under standard production conditions, collecting the background resonance signal of the local environment where the monitoring module is located, and processing the background resonance signal to obtain the initial environmental resonance baseline fingerprint.

3. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 1, characterized in that, Both the environmental resonance baseline fingerprint and the current environmental resonance signal are spectral characteristics; The spectral features are generated by performing a fast Fourier transform on the background vibration signal or background electromagnetic noise signal collected by the monitoring module.

4. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 1, characterized in that, The monitoring module is also configured to respond to events where the environmental compatibility index falls below a preset authentication threshold; Morphological analysis is performed on the current environmental resonance signal to determine a managerial attribution type that characterizes the anomalous nature of the current environmental resonance signal; the managerial attribution type is further included in the reported active data packet; the central processing module is also configured to trigger a differentiated managerial response workflow based on the managerial attribution type.

5. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 1, characterized in that, The monitoring module is also configured to: generate a residual signal representing the difference between the current environmental resonance signal and the environmental resonance baseline fingerprint during the comparison process; determine one or more degradation characteristic indicators by quantifying the energy of the residual signal at a specific frequency defined in the warning frequency characteristic library based on a preset warning frequency feature library related to equipment degradation; further include one or more degradation characteristic indicators in the active data packet; and the central processing module is also configured to: analyze the changing trend of one or more time-seriesd degradation characteristic indicators to generate predictive maintenance alarms.

6. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 5, characterized in that, The central processing module analyzes the trend of change by performing linear regression analysis on the time-series degradation characteristic indicators to calculate the trend slope that characterizes the growth rate of the degradation characteristic indicators. Among them, when the trend slope satisfy When this happens, a predictive maintenance alert is generated, in which... This is a preset slope threshold.

7. The real-time production process traceability and monitoring system based on the Internet of Things according to claim 1, characterized in that, The central processing module is also configured to: respond to a management event related to a planned change in the production environment; and at the scheduled effective time of the planned change, send an instruction to the monitoring module to cause it to actively suspend the authentication function based on the environmental resonance baseline fingerprint and enter the baseline learning mode. In baseline learning mode, a candidate baseline fingerprint is constructed by performing cluster analysis on newly acquired environmental resonance information; In response to an instruction confirming that the new production state has stabilized, the candidate baseline fingerprint is set as the new environmental resonance baseline fingerprint for the monitoring module.

8. A real-time production process traceability and monitoring system based on the Internet of Things according to claim 7, characterized in that, The instruction confirming that the new production status has stabilized is generated by production management personnel through the human-machine interface after the first batch of products under the new production configuration passes the first process inspection.

9. A real-time production process traceability and monitoring system based on the Internet of Things according to claim 4, characterized in that, Morphological analysis includes: calculating the energy standard deviation of the current environmental resonant signal in the time domain; when the energy standard deviation is higher than the first preset morphological threshold, the administrative attribution type is determined to be an impulse anomaly; calculating the spectral entropy of the current environmental resonant signal in the frequency domain; when the spectral entropy is lower than the second preset morphological threshold, the administrative attribution type is determined to be a periodic anomaly; when the spectral entropy is higher than the second preset morphological threshold, the administrative attribution type is determined to be a broadband random anomaly.

10. A real-time production process traceability and monitoring system based on the Internet of Things according to claim 3, characterized in that, The monitoring module is also configured to further encapsulate the data obtained by compressing the spectral characteristics of the current environmental resonance signal in the active data packet; the central processing module is also configured to associate the received active data packet with the corresponding production batch number and archive it.

Images