An internet of things dynamic weighing and data tracing method and system

Abstract

The application discloses a kind of Internet of Things dynamic weighing and data tracing method and system, it is related to industrial internet data processing technical field, comprising: obtaining the dynamic weighing data of weighing event and at least one group of auxiliary data synchronized therewith;Based on the auxiliary data, through the preset credibility model, the credibility weight value is generated for the dynamic weighing data;Based on the weight value, one or more dynamic weighing data are statistically analyzed by weighting, and weighted statistical analysis result is obtained;And the change trend of the weighted statistical analysis result and credibility weight value is fused, two-dimensional fusion diagnosis is carried out, and the diagnosis result that can accurately decouple problem source is obtained;According to the diagnosis result, high credibility trace record or control instruction is generated.The application realizes the intelligent distinction of production process anomaly and measurement system anomaly by real-time quantification and management of data credibility, and improves the reliability of dynamic weighing system and the value of trace data.

Description

Technical Field

[0001] This invention relates to the field of industrial Internet of Things (IoT) data processing technology, and in particular to an IoT dynamic weighing and data traceability method and system. Background Technology

[0002] In modern automated production lines, especially in the manufacturing processes of high-value products such as pharmaceuticals and precision electronic components, online dynamic weighing is a crucial quality control step. However, existing technologies generally suffer from a fundamental "diagnostic blind spot": when weighing data is abnormal, the system cannot effectively distinguish whether the anomaly stems from "real fluctuations in the production process" or "a problem with the measurement system itself." This ambiguity leads to significant economic losses. On some high-value production lines, the inability to effectively differentiate between vibration interference and insufficient filling volume results in monthly losses from the incorrect rejection of qualified products and triggers several unnecessary production line shutdowns for troubleshooting, severely impacting production efficiency and quality control levels. Summary of the Invention

[0003] This invention aims to provide a method and system for dynamic weighing and data traceability in the Internet of Things, enabling real-time quantification and management of the reliability of measurement data, and based on this, intelligent root cause diagnosis, in order to solve the problems of data quality not being able to be assessed and fault diagnosis having a single dimension in the existing technology.

[0004] In view of the above problems, in a first aspect, the present invention provides an Internet of Things (IoT) dynamic weighing and data traceability method, comprising:

[0005] Acquire dynamic weighing data of the weighing event and at least one set of auxiliary data synchronized with it;

[0006] Based on the at least one set of auxiliary data, a confidence weight value is generated for the dynamic weighing data through a preset confidence model;

[0007] Based on the aforementioned credibility weight value, a weighted statistical analysis is performed on one or more dynamic weighing data to obtain the weighted statistical analysis results.

[0008] By combining the weighted statistical analysis results and the confidence weight values, a two-dimensional fusion diagnosis is performed to obtain the diagnostic results.

[0009] Based on the diagnostic results, a traceability record or control instruction is generated.

[0010] Secondly, the present invention also provides an Internet of Things (IoT) dynamic weighing and data traceability system, comprising:

[0011] The data acquisition module is used to acquire dynamic weighing data of weighing events and at least one set of auxiliary data synchronized with them;

[0012] The weight generation module is used to generate a confidence weight value for the dynamic weighing data based on the at least one set of auxiliary data and through a preset confidence model.

[0013] The analysis module is used to perform weighted statistical analysis on one or more dynamic weighing data based on the credibility weight value, and obtain the weighted statistical analysis results.

[0014] The diagnostic module is used to fuse the weighted statistical analysis results and the confidence weight values ​​to perform a two-dimensional fusion diagnosis and obtain a diagnostic result.

[0015] The decision management module is used to generate traceability records or control instructions based on the diagnostic results.

[0016] The technical solution provided in this application has at least the following technical effects or advantages:

[0017] This invention assigns a quantified confidence weight to each piece of weight data, enabling objective assessment of data quality throughout the traceability chain. During quality audits, it provides a clear understanding of the data's generation environment and confidence level, offering a more robust data basis for decision-making.

[0018] The dual-dimensional diagnostic logic of this invention can effectively distinguish whether the root cause of a problem stems from fluctuations in the production process or from a fault or interference in the measurement system itself. This avoids the ambiguity of traditional diagnostics, guides more precise maintenance and process adjustments, and reduces unnecessary downtime for troubleshooting.

[0019] By employing weighted statistical analysis, the system can automatically reduce the influence of low-reliability data in statistical calculations. This enables process control analysis to effectively suppress interference from environmental vibrations and other noises, resulting in analysis results that more accurately reflect the intrinsic state of the production process and reduce false alarms.

[0020] By monitoring the long-term trend of reliability weight values, this invention can provide early warnings before the performance of the measurement system significantly deteriorates or malfunctions. This allows equipment maintenance to be upgraded from reactive response to proactive prevention, helping to ensure the continuity and stability of production. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating an IoT dynamic weighing and data traceability method according to the present invention.

[0022] Figure 2 This is a schematic diagram of the structure of an Internet of Things (IoT) dynamic weighing and data traceability system. Detailed Implementation

[0023] This invention relates to an IoT dynamic weighing and data traceability method and system, which enables real-time quantification and management of the reliability of measurement data, and performs intelligent root cause diagnosis based on this, in order to solve the problems of data quality not being able to be assessed and fault diagnosis having a single dimension in the prior art.

[0024] The above technical solutions will now be described in detail with reference to the accompanying drawings and specific embodiments to provide a better understanding of them. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments used only to explain the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. Furthermore, it should be noted that, for ease of description, only the parts related to the present invention are shown in the drawings, not all of them.

[0025] Example 1

[0026] This embodiment uses a high-speed biopharmaceutical vial filling and checkweighing production line as a specific application scenario to illustrate the complete process, engineering details and application scenarios of the method of the present invention in detail.

[0027] Figure 1 This is a flowchart illustrating an IoT dynamic weighing and data traceability method according to the present invention.

[0028] A dynamic weighing and data traceability method for the Internet of Things includes the following steps:

[0029] Collect the raw multidimensional data stream of weighing events, perform time alignment and segmentation, and construct weighing event data packets;

[0030] The weighing event data packet is parsed to obtain the final weight value, and the auxiliary data in the weighing event data packet is feature extracted to obtain an auxiliary feature vector. The preset credibility model is calculated using the auxiliary feature vector to obtain a credibility weight value. The final weight value is associated with the credibility weight value to obtain a credibility weighing data pair.

[0031] Weighted statistics are calculated for one or more pairs of confidence-weighted data to obtain a set of weighted statistics. Based on the set of weighted statistics, at least one weighted and adapted outlier criterion is applied to obtain weighted statistical analysis results.

[0032] By combining the weighted statistical analysis results with the credibility weight values ​​in the credibility weighing data pairs, a diagnostic rule base match is performed to obtain a diagnostic conclusion report;

[0033] Based on the diagnostic conclusion report, traceability records are formatted to obtain high-reliability traceability records, which are then stored in the traceability database. Based on the diagnostic conclusion report, instructions are generated and issued to obtain control instructions.

[0034] It should be further explained that, in the specific implementation process, the detailed process of collecting the raw multidimensional data stream of weighing events, performing time alignment and segmentation, and constructing the weighing event data packet is as follows:

[0035] The process of acquiring the raw multidimensional data stream of a weighing event includes: synchronously and at high speed sampling a multimodal sensor group deployed on a dynamic weighing unit to obtain the raw multidimensional data stream. The multimodal sensor group includes a weighing sensor and at least one set of auxiliary sensors, preferably including a triaxial accelerometer and an acoustic sensor. The raw multidimensional data stream is a data set that includes a weighing data stream, a vibration data stream, and an acoustic data stream, all of which are continuous digital signal time series with precise hardware timestamps.

[0036] The original multidimensional data stream is time-aligned and segmented to construct a weighing event data package. Specifically, this process includes: segmenting the original multidimensional data stream into event data streams based on an external trigger signal to obtain an event data stream set. The external trigger signal is generated by a photoelectric sensor deployed at the entrance of the weighing unit. When the item to be inspected enters the weighing area, the photoelectric sensor level changes, generating a start trigger signal; when the item to be inspected completely leaves the weighing area, an end trigger signal is generated. Event segmentation means extracting data within a time window from the continuous original multidimensional data stream based on the timestamps of the start and end trigger signals. The data within this window constitutes an event data stream set. A globally unique item ID, obtained from an upstream workstation and bound to the physical entity of the item to be inspected, is assigned to the event data stream set. The item ID, the event data stream set, and the environmental and equipment status parameters collected within the time window are encapsulated into a structured data object to obtain the weighing event data package. The environmental and equipment status parameters include slowly varying parameters such as ambient temperature, ambient humidity, and the number of operating cycles since the last calibration of the equipment.

[0037] It should be further explained that, in the specific implementation process, the weighing event data packet is parsed to obtain the final weight value, and auxiliary data in the weighing event data packet is feature extracted to obtain an auxiliary feature vector. The preset credibility model is calculated using the auxiliary feature vector to obtain a credibility weight value. The final weight value is then associated with the credibility weight value to obtain a credibility weighing data pair. The detailed implementation process is as follows:

[0038] The weighing event data packet is parsed to obtain the final weight value. Specifically, the process includes: extracting the weighing data stream from the weighing event data packet; applying a preset digital low-pass filter to the weighing data stream to filter out known high-frequency electrical noise, obtaining a smooth weighing data stream; acquiring the smooth weighing data stream and setting a sliding window of size W (e.g., 5 data points) and a first-order derivative threshold T; starting from the first point of the data stream, calculating the linear regression slope of the data points within the sliding window as an estimate of the first-order derivative at the center point of the window; sliding the window forward by one data point and repeating the calculation until the entire data stream is traversed, obtaining a first-order derivative sequence; in the first-order derivative sequence, finding the first continuous subsequence with a length greater than L (e.g., 50 data points) where the absolute value of all points is less than the threshold T, the start and end points of this subsequence define the boundary of the stable period; extracting data points from all smooth weighing data streams within the boundary of the stable period to form a stable weight value set. The final weight value is obtained by summing all data points in the stable weight value set and then dividing by the number of data points in the set.

[0039] Feature extraction is performed on the auxiliary data in the weighing event data packet to obtain an auxiliary feature vector. Specifically, this process includes: extracting the vibration data stream and the acoustic data stream from the weighing event data packet; calculating the root mean square energy value, peak-to-peak value, kurtosis factor, margin factor, and spectral entropy of the vibration data stream along the three orthogonal axes to obtain a set of vibration feature values; calculating the maximum short-time energy value, zero-crossing rate, and the first twelve Mel-frequency cepstral coefficients of the acoustic data stream to obtain a set of acoustic feature values; and arranging the vibration feature value set, the acoustic feature value set, and the environmental and equipment state parameters extracted from the weighing event data packet in a preset order to construct a one-dimensional vector containing N values, which is the auxiliary feature vector.

[0040] The confidence weight value is obtained by calculating a preset confidence model using the auxiliary feature vector. The specific process includes: acquiring the auxiliary feature vector and initializing a base weight value of W_base (e.g., 100) and a final confidence weight value of W_final, setting the initial value of W_final equal to W_base. A preset attenuation rule base containing M rules is traversed. For the first rule in the rule base, the rule includes a condition part and an operation part. The condition part is, for example, "the root mean square value of vibration is greater than 0.8g," and the operation part is, for example, "multiplied by an attenuation factor of 0.5." It is determined whether the corresponding feature value in the auxiliary feature vector satisfies the condition part of the rule. If it does, the operation part is executed, updating W_final to the result of multiplying W_final by the attenuation factor of the rule. The process continues to traverse the second, third, and so on up to the Mth rule, updating W_final sequentially. After the traversal is complete, the final value of W_final is the confidence weight value.

[0041] In a preferred embodiment, the threshold value (e.g., 0.8g) in the condition section is not arbitrarily set. This value can be determined using a standard engineering calibration method. For example, the weighing unit can be placed on a six-degree-of-freedom vibration table, and without placing the object to be inspected, a sinusoidal vibration excitation at a specific frequency (e.g., 50Hz) increasing in steps from 0.1g to 2.0g can be applied to the vibration table. Simultaneously, the noise level (e.g., standard deviation) of the static output signal of the weighing sensor is continuously monitored. When a non-linear, sharp increase in the noise level is observed at an inflection point, the corresponding vibration intensity can be determined as the preferred threshold value for this condition section.

[0042] The final weight value is associated with the credibility weight value to obtain a credibility weighing data pair. This association represents the creation of a new data structure containing at least four fields: a unique item ID, a timestamp, the final weight value, and the credibility weight value. This data structure constitutes the credibility weighing data pair.

[0043] It should be further explained that, in the specific implementation process, weighted statistics are calculated for one or more pairs of the aforementioned confidence-weighted data to obtain a set of weighted statistics. Based on the set of weighted statistics, at least one weighted and adapted outlier criterion is applied to obtain the weighted statistical analysis results. The detailed implementation process is as follows:

[0044] The process of calculating a weighted statistic for one or more of the aforementioned confidence weighted data pairs to obtain a weighted statistic set includes: storing a predetermined number (e.g., the most recent P) of the aforementioned confidence weighted data pairs in a first-in-first-out (FIFO) data buffer; calculating the weighted arithmetic mean for the P confidence weighted data pairs in the data buffer; the calculation of the weighted arithmetic mean is as follows: multiplying the final weight value of each data pair by its corresponding confidence weight value to obtain P product terms; summing the P product terms to obtain a total product term; summing the P confidence weight values ​​to obtain a total weight; and dividing the total product term by the total weight to obtain the weighted arithmetic mean. Similarly, the weighted standard deviation of the P confidence weighted data pairs is calculated. The calculated weighted arithmetic mean and weighted standard deviation, etc., are stored in a set, which is the weighted statistic set.

[0045] Based on the weighted statistical set, at least one weighted and adapted outlier criterion is applied to obtain a weighted statistical analysis result. The specific process includes: acquiring the latest confidence-based weighing data pair and comparing its final weight value with the upper and lower control limits calculated based on the weighted statistical set to obtain a single-point outlier result; backtracking and querying the historical K-1 consecutive confidence-based weighing data pairs in the data buffer, combining them with the latest data pair to form an analysis window containing K data points; and applying a preset pattern outlier rule library to the analysis window. Taking the rule "K consecutive points falling on the same side of the center line" as an example, the adaptation process is as follows: Initialize a same-side counter C to 0 and a weight accumulator S to 0; obtain the weighted average of all data points in the analysis window as the center line; starting from the first data point in the window, determine whether its final weight value is greater than the center line; if it is, increment the counter C by 1 and accumulate the confidence weight value corresponding to the data point into S; continue to determine the second to the Kth data point; after traversal, determine whether the counter C is equal to K; if C is equal to K, further determine whether the value of the weight accumulator S is greater than a preset pattern confidence threshold; if S is also greater than the threshold, then confirm that the pattern has been triggered. Logically combine the single-point anomaly detection result and the triggering status of all pattern anomaly detection rules to obtain a Boolean vector, which is the weighted statistical analysis result.

[0046] It should be further explained that, in the specific implementation process, the weighted statistical analysis results are combined with the credibility weight values ​​in the credibility weighing data pairs to perform diagnostic rule base matching, thereby obtaining a diagnostic conclusion report; based on the diagnostic conclusion report, traceability records are formatted to obtain high-credibility traceability records and store them in the traceability database; and based on the diagnostic conclusion report, instructions are generated and issued to obtain control instructions. The detailed implementation process is as follows:

[0047] The weighted statistical analysis results are combined with the credibility weight values ​​in the credibility weighing data pairs to perform a diagnostic rule base match, resulting in a diagnostic conclusion report. The specific process includes: using the weighted statistical analysis result vector as one query key and the credibility weight value in the credibility weighing data pair that triggered the analysis as another query key; using these two query keys to search and match a preset, tree-structured or matrix-structured diagnostic rule base; when a unique matching rule is found, its "conclusion" section is extracted and combined with the current event ID, timestamp, and other information to construct a structured data object containing fields such as diagnosis type, diagnosis details, and recommended measures, which is the diagnostic conclusion report.

[0048] The traceability record is formatted according to the diagnostic conclusion report to obtain a high-confidence traceability record, which is then stored in the traceability database. Instructions are generated and issued based on the diagnostic conclusion report to obtain control commands. The traceability record formatting process is as follows: a new data record structure is created; the final weight value and the confidence weight value are extracted from the confidence weighing data pair that triggered the diagnosis; the diagnosis type and diagnosis details text are extracted from the diagnostic conclusion report; the item's unique identifier, current timestamp, final weight value, confidence weight value, diagnosis type, and diagnosis details text are respectively filled into different fields of the new data record structure to obtain the high-confidence traceability record, which is then sent to the traceability database for a write operation. The instruction generation process is as follows: a suggested measures field is extracted from the diagnostic conclusion report; based on the content of this field, an instruction mapping table is queried to obtain the specific instruction code and target device address; the instruction code and target device address are formatted and encapsulated according to the communication protocol required by the target device to obtain the control command, which is then sent through the network interface.

[0049] Through the above steps, the present invention can realize real-time quantitative management of the reliability of dynamic weighing data in a specific engineering scenario, and based on this, perform accurate and intelligent fault root cause diagnosis and in-depth tracing.

[0050] Example 2

[0051] Figure 2 This is a schematic diagram of the structure of an IoT dynamic weighing and data traceability system according to the present invention. The system is used to execute the method of Embodiment 1, and preferably adopts an edge-cloud collaborative distributed system architecture to balance the needs of real-time performance, computing load, and data management.

[0052] An Internet of Things (IoT) dynamic weighing and data traceability system, comprising:

[0053] The data acquisition module is used to acquire dynamic weighing data of weighing events and at least one set of synchronous auxiliary data. Physically, this module corresponds to an integrated sensing unit deployed on the production site, which integrates weighing sensors and auxiliary sensor groups, along with its associated data acquisition hardware.

[0054] The weight generation module is used to generate confidence weight values ​​for the dynamic weighing data based on the at least one set of auxiliary data and a preset confidence model. The software entity of this module is preferably deployed on an edge computing node adjacent to the production line to ensure real-time computation.

[0055] The analysis module is used to perform weighted statistical analysis on one or more dynamic weighing data based on the credibility weight value, and obtain the weighted statistical analysis results.

[0056] The diagnostic module is used to integrate the weighted statistical analysis results and the confidence weight values ​​to perform a two-dimensional fusion diagnosis and obtain a diagnostic result.

[0057] The decision management module is used to generate traceability records or control instructions based on the diagnostic results. It is responsible for interacting with the upper-level information system and issuing control instructions.

[0058] In a preferred embodiment, the diagnostic module is configured to: determine the diagnostic result as a production process abnormality when the weighted statistical analysis result shows a process abnormality and the confidence weight value is in a preset high confidence range; and determine the diagnostic result as a measurement system abnormality when the weighted statistical analysis result shows a process abnormality and the confidence weight value is in a preset low confidence range.

[0059] In a preferred embodiment, the system employs an edge-cloud collaborative architecture. The data acquisition module and the weight generation module are deployed on edge computing nodes; some functions of the analysis module, the diagnostic module, and the decision management module are deployed on the edge computing nodes to perform real-time tasks, while the remaining functions are deployed on the cloud platform to perform non-real-time analysis and management tasks. The advantage of this architecture is that it places the tasks with the highest latency requirements at the edge, closer to the data source, while tasks requiring massive amounts of data and powerful computing power are processed in the cloud, achieving optimal resource allocation.

[0060] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for dynamic weighing and data traceability in the Internet of Things, characterized in that, include: The system acquires dynamic weighing data of a weighing event and at least one set of auxiliary data synchronized with it. The at least one set of auxiliary data includes: a set of environmental state data for characterizing the state of the measurement environment; and a set of equipment state data for characterizing the state of the weighing equipment itself. Based on the at least one set of auxiliary data, an auxiliary feature vector is extracted. By traversing a preset attenuation rule library, it is determined whether the auxiliary feature vector meets the rule conditions and the corresponding attenuation operation is performed to generate a confidence weight value for the dynamic weighing data. Based on the aforementioned credibility weight value, a weighted statistical analysis is performed on one or more dynamic weighing data to obtain the weighted statistical analysis results. By integrating the weighted statistical analysis results and the confidence weight values, a two-dimensional fusion diagnosis is performed, which includes: When the weighted statistical analysis results show that the process is abnormal, and the confidence weight value is in the preset high confidence range, the diagnosis result is determined to be an abnormal production process. When the weighted statistical analysis results show that the process is abnormal, and the confidence weight value is in the preset low confidence range, the diagnosis result is determined to be an abnormality of the measurement system. Based on the diagnostic results, generate traceability records or control instructions.

2. The IoT dynamic weighing and data traceability method as described in claim 1, characterized in that, The environmental status data includes at least one of environmental vibration data, environmental acoustic data, temperature data, or humidity data; the equipment status data includes at least one of weighing equipment calibration status data, operating time data, or load history data.

3. The IoT dynamic weighing and data traceability method as described in claim 1, characterized in that, The weighted statistical analysis includes: Based on the aforementioned confidence weight value, calculate the weighted average or weighted standard deviation of one or more dynamic weighing data; A weighted statistical process control chart is generated based on the weighted average or the weighted standard deviation.

4. The IoT dynamic weighing and data traceability method as described in claim 3, characterized in that, The process of obtaining the weighted statistical analysis results further includes: based on the weighted statistical process control chart, applying at least one weighted and adapted outlier criterion to determine the process state; the weighted and adapted outlier criterion, in addition to applying the classic outlier criterion, also incorporates consideration of the confidence weight value corresponding to one or more data points.

5. The IoT dynamic weighing and data traceability method as described in claim 1, characterized in that, The dual-dimensional fusion diagnosis also includes: when the weighted statistical analysis results do not show process abnormalities, but the confidence weight value shows a downward trend or is in the preset low confidence range, the diagnosis result is determined to be a potential risk warning for the measurement system.

6. The IoT dynamic weighing and data traceability method as described in claim 1, characterized in that, The process of generating traceability records includes: associating the dynamic weighing data, the credibility weight value, and the diagnostic results, and storing them in the traceability database to form a high-credibility traceability data chain.

7. An Internet of Things (IoT) dynamic weighing and data traceability system, characterized in that, The system is used to execute the IoT dynamic weighing and data traceability method according to any one of claims 1-6, the system comprising: The data acquisition module is used to acquire dynamic weighing data of weighing events and at least one set of auxiliary data synchronized with them; The weight generation module is used to extract auxiliary feature vectors based on the at least one set of auxiliary data, determine whether the auxiliary feature vectors meet the rule conditions by traversing the preset attenuation rule library, and perform the corresponding attenuation operation to generate a confidence weight value for the dynamic weighing data. The analysis module is used to perform weighted statistical analysis on one or more dynamic weighing data based on the credibility weight value, and obtain the weighted statistical analysis results. The diagnostic module is used to fuse the weighted statistical analysis results and the confidence weight values ​​to perform a two-dimensional fusion diagnosis and obtain a diagnostic result. The decision management module is used to generate traceability records or control instructions based on the diagnostic results.

Images