Electronic scale weighing tamper-proofing method and system based on active learning

Abstract

The application discloses a kind of electronic scale weighing tamper-proofing method and system based on active learning, including S1, original data set is generated and preprocessed by multidimensional sensor, form standardized data set;S2, based on active learning algorithm, establish dynamic anomaly detection model, calculate the confidence score of data;S3, the standardized data set is encrypted using multi-level encryption algorithm, and a chain data storage structure is constructed based on encrypted associated data set;S4, construct cloud end traceability management platform and full life cycle management model;S5, configure modular intelligent hardware;S6, through real-time multi-node collaborative monitoring technology, establish linkage verification mechanism, and comprehensively analyze the abnormal behavior of multiple nodes through intelligent discrimination algorithm;S7, combined with mobile terminal application and near field communication technology, provide real-time weighing data and traceability information for users.The application has the advantages of strong tamper-proofing ability, high detection accuracy, high supervision efficiency and convenient user supervision.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information tamper-proofing, and particularly relates to an electronic scale weighing tamper-proofing method and system based on active learning. BACKGROUND

[0002] With the rapid development of weighing technology and intelligent devices, electronic scales are increasingly widely used in commercial, industrial, logistics and other fields. However, the authenticity and credibility of electronic scale weighing data have gradually attracted attention. Especially in the transaction scenario, cheating behaviors such as tampering with weighing data by hardware modification or software backdoor are rampant, which seriously affects fair transactions and regulatory efficiency. Electronic scale cheating behaviors are usually achieved through methods such as falsifying weighing signals, tampering with sensor data, circuit modification, or unauthorized disassembly of devices, which brings many challenges to both transaction parties and regulatory authorities.

[0003] In the prior art, traditional electronic scale tamper-proofing solutions are usually based on static detection rules or hardware encryption mechanisms. These methods are insufficient when faced with complex and varied cheating methods, which is manifested in the following aspects:

[0004] 1. Lack of dynamic detection capability: Traditional methods mainly rely on fixed rules or threshold settings, and have poor adaptability to new cheating methods. They cannot update the detection mechanism in real time to identify complex and diversified cheating behaviors.

[0005] 2. Insufficient data security protection: The data encryption mechanism in the prior art is mostly single-layer encryption or simple signature verification, which is easy to be cracked or tampered with, and it is difficult to ensure the integrity and tamper-proofing of the weighing data.

[0006] 3. Complex and lagging regulatory process: The tracing and verification of electronic scale device data mostly rely on manual intervention, and cannot realize dynamic tracking and rapid verification of device lifecycle data, resulting in low regulatory efficiency.

[0007] 4. Lack of multi-node collaborative detection capability: Traditional electronic scale tamper-proofing technology usually focuses on a single device, and cannot build a cross-device and cross-scenario collaborative detection mechanism, making it difficult to identify large-scale collaborative cheating behaviors.

[0008] 5. Consumers lack supervision means: In the prior art, consumers cannot directly participate in the supervision of weighing data, and lack tools for real-time acquisition of weighing data and traceability information, resulting in incomplete supervision closed loop.

[0009] Therefore, how to provide an electronic scale weighing tamper-proofing method and system based on active learning is a problem that those skilled in the art need to solve. SUMMARY

[0010] One purpose of the present application is to propose an electronic scale weighing anti-tampering method and system based on active learning, which introduces active learning algorithm, multi-level data encryption technology, multi-node cooperative monitoring technology and consumer supervision module, and describes in detail the technical solutions for dynamically detecting weighing data anomalies, ensuring data integrity, preventing device tampering and realizing real-time supervision, which has the advantages of strong anti-tampering ability, high detection accuracy, high supervision efficiency and convenient user supervision.

[0011] An electronic scale weighing anti-tampering method based on active learning according to an embodiment of the present application comprises the following steps:

[0012] S1, obtaining real-time weighing data, device running state data and external environment parameters of the electronic scale, generating an original data set, and pre-processing to form a standardized data set;

[0013] S2, establishing a dynamic anomaly detection model, inputting the standardized data set into the dynamic anomaly detection model, calculating the confidence score of the data in the standardized data set, selecting high-confidence data for online updating of the dynamic anomaly detection model, and manually labeling low-confidence data, and updating the dynamic anomaly detection model again after labeling;

[0014] S3, encrypting the standardized data set, and constructing a unique identifier set, associating and storing the encrypted data and the unique identifier set to generate an encrypted associated data set, constructing a chain data storage structure based on the encrypted associated data set, and embedding traceability information in the chain data storage structure to generate a chain traceability data set;

[0015] S4, constructing a cloud-based traceability management platform and a full life cycle management model, recording data of each link of the electronic scale device through the full life cycle management model, and dynamically tracking and real-time verifying each link data through time stamp and version control;

[0016] S5, integrating modular intelligent hardware, triggering the hardware locking function when detecting sensor signal anomalies and unauthorized changes to the circuit, recording the event and generating an alarm information at the same time;

[0017] S6, establishing a linkage verification mechanism among multiple electronic scale nodes through real-time multi-node cooperative monitoring technology, identifying and positioning cheating behaviors in complex scenarios;

[0018] S7, combining mobile application and near field communication technology to provide real-time weighing data and traceability information for users.

[0019] Optionally, the S1 specifically comprises:

[0020] S11. Acquire real-time weighing data M(t) of the electronic scale through multi-dimensional sensors, and simultaneously collect equipment operating status data and external environmental parameters. The equipment operating status data includes sensor signal drift V. drift (t), Equipment operating frequency F sys (t), where the external environmental parameters include the ambient temperature T. env Humidity H env Atmospheric pressure P env and vibration disturbance intensity A vib Generate the original dataset D containing multi-source data. raw ={M(t),V drift (t),F sys (t),T env H env ,P env A vib};

[0021] S12, regarding the original dataset D raw Dynamic noise suppression is performed by calculating the noise characteristic vector N in real time. char (t) Adjust the noise threshold function ∈(t) to filter out random high-frequency noise introduced by vibration, environmental changes and equipment interference, and generate the noise-suppressed dataset D. filtered ;

[0022] S13. To address data gaps caused by sensor malfunctions and data acquisition interruptions in the dataset, perform adaptive data completion based on a time series model. Analyze the noise-suppressed dataset D. filtered By analyzing the temporal correlations and cross-constraints of the data across each dimension, a complete dataset D is generated. completed ;

[0023] S14. For the completed dataset D... completed Normalization and formatting are performed, and the real-time weighing data M(t) and sensor signal drift V are normalized using a dimensional normalization function. drift (t) and equipment operating frequency F sys (t) is dimensionless based on the unified timestamp T. std Realign all data dimensions to generate a standardized dataset D. standard ={T std ,M'(t),V' drift (t),F' sys (t),T' env ,H' env ,P' env ,A' vib}

[0024] Optionally, S2 specifically includes:

[0025] S21. Construct a dynamic anomaly detection model f based on an active learning algorithm. init (x; Θ0), where Θ0 is the initial parameter set of the model, and x is the input data vector, and the standardized dataset D is used. standard Input dynamic anomaly detection model f init (x; Θ0) are classified, and the corresponding confidence score set C(D) is output. standard )={C(x1),C(x2),…,C(x n )}, where C(x) i ) represents data x in a standardized dataset. i Confidence level belonging to the normal category;

[0026] S22, Based on the confidence score set C(D) standard A dynamic threshold δ(t) is set, which is adjusted in real time according to the confidence distribution characteristics, and the standardized dataset D is then processed. standard Divided into high-confidence dataset D high and low-confidence dataset D low :

[0027] D high ={x∈D standard |C(x)≥δ(t)};

[0028] D low ={x∈D standard |C(x)<δ(t)};

[0029] S23, The high-confidence data set D high Input dynamic anomaly detection model f init (x; Θ0), the initial parameter set Θ0 is adaptively optimized using an incremental learning algorithm, and the updated dynamic anomaly detection model is represented as f opt (x;Θ new ), where Θ new The optimized parameter set;

[0030] S24. For the low-confidence data set D low Manual annotation is performed to generate an annotated dataset. This annotated dataset is then used as new training samples and added to the dynamic anomaly detection model f. opt (x;Θ new In the training set, the dynamic anomaly detection model f is trained using an active learning framework. opt (x;Θ new A second optimization is performed to complete the deep update of the dynamic anomaly detection model parameters, and the final optimized dynamic anomaly detection model f is output. final (x;Θfinal );

[0031] S25. Utilize the final optimized dynamic anomaly detection model f final (x;Θ final The system performs classification and anomaly detection on the real-time input dataset, outputs abnormal data to record and track during device operation, and integrates it into an anomaly-tagged dataset D. anomaly .

[0032] Optionally, S3 specifically includes:

[0033] S31. For the standardized dataset D standard Encryption processing is performed on the standardized dataset D using multi-level encryption algorithms. standard Each data record in the dataset is processed in layers, and a symmetric encryption algorithm is used to process the standardized dataset D. standard The data in the middle is subjected to basic encryption to generate the first layer of encrypted data set D. enc1 ={E sym (x i ;K sym )|i=1,2,...,n}, where x i To standardize the data items in the dataset, K sym Based on the encryption key, E sym Represents a symmetric encryption function;

[0034] S32, Regarding the first-layer encrypted data set D enc1 The unique identifier set H(D) for each piece of data is calculated using a hash algorithm. enc1 )={h i |h i =Hash(E sym (x i ;K sym ),i=1,2,...,n}, where h i This is an irreversible identifier corresponding to the encrypted data record;

[0035] S33, The first layer of encrypted data set D enc1 With the set of unique identifiers H(D) enc1 Perform associated storage to generate an encrypted associated data set D. assoc ={(h i E sym (x i ;K sym ))∣i=1,2,...,n};

[0036] S34. Construct a linked data storage structure L(D) assoc ):

[0037] L(Dassoc )={(h i-1 ,h i ,h i+1 )|i=2,3,...,n-1};

[0038] In this chained data storage structure, each node contains a unique identifier h for the current record. i Pre-record identifier h i-1 and subsequent record identifier h i+1 ;

[0039] S35. Embed traceability information on the basis of the chain-like data storage structure. The traceability information includes the electronic scale device identifier ID. device Operation timestamp T op and operation log R op Generate a complete chain-source data set D for recording and tracking encrypted data. trace ={(L(D)} assoc ), ID device ,T op ,R op )}.

[0040] Optionally, S4 specifically includes:

[0041] S41. The chain-based traceability data set D trace Uploaded to the cloud-based traceability management platform, the chain traceability data set D is managed based on a full lifecycle management model. trace Initialization is performed to obtain the initialized source data set T. init ={D trace Meta device}, where Meta device Including electronic scale device identifier ID device Production date T prod Calibration parameter C calib Maintenance Records R maint and usage scenario information S usage ;

[0042] S42. Based on the timestamp generation mechanism, for the traceability data set T init Each data record in the database is appended with a unique timestamp T. stamp (i) The timestamp calculation rule is as follows:

[0043] T stamp (i)=T upload +Δt(i);

[0044] Among them, T upload This represents the baseline time for data upload, where Δt(i) is the relative time offset for each record upload.

[0045] S43. Based on the version control mechanism, trace the source data set T. init Perform historical status management and generate multi-version traceability data sets. in This indicates the state of the source dataset in version z;

[0046] S44. Combining the full lifecycle management model, the multi-version traceability data set T versioned The equipment's production, calibration, maintenance, and usage information are linked and stored in multiple dimensions to generate a full lifecycle linked data set T. lifecycle ={(T init Meta device ,T versioned )};

[0047] S45. Utilize a cloud-based traceability management platform to manage the entire lifecycle-related data set T. lifecycle Dynamic tracking and real-time verification are performed. After verification is successful, a set of verification records is generated. When verification fails, an exception handling mechanism is triggered to store the exception records as an exception data set.

[0048] Optionally, S5 specifically includes:

[0049] S51. Configure a sensor signal locking module to monitor the sensor output signal S in real time. signal (t) and calculate the signal deviation ΔS(t):

[0050] ΔS(t)=|S signal (t)-S ref |;

[0051] Among them, S ref For sensor calibration reference values, when ΔS(t) > δ S At that time, the sensor signal is locked and a sensor abnormality signal S is generated. anomaly (t)=S signal (t), where δ S The preset deviation threshold;

[0052] S52. Configure a circuit integrity monitoring module to dynamically detect the circuit's operating status, collect real-time current and voltage, and calculate the circuit's characteristic state C. integrity (t), when When this occurs, a circuit malfunction is detected, triggering the hardware lockout function and recording the circuit malfunction status, where R... normal Indicates the normal operating range of the circuit;

[0053] S53. Configure an anti-tamper trigger module to monitor the integrity of the device's hardware casing and the anti-tamper point signal D. secure(t) Determine whether the device has been disassembled without authorization, when D secure When (t) = 0, a disassembly alarm signal A is generated. tamper (t) triggers the anti-disassembly locking mechanism and records disassembly anomaly information;

[0054] S54, Integrated sensor abnormal signal S anomaly (t), Circuit characteristic state C integrity (t) and disassembly alarm signal A tamper (t), generating an exception record set R anomaly (t)={S anomaly (t),C integrity (t),A tamper (t)}, and R anomaly (t) Stored in a cloud-based traceability management platform;

[0055] S55, Regarding the abnormal record set R anomaly (t) Generate alarm information, upload the alarm information to the cloud traceability management platform and bind it to the full life cycle associated data set for tracking device status and supporting anomaly handling.

[0056] Optionally, S6 specifically includes:

[0057] S61. Based on real-time multi-node collaborative monitoring technology, configure a multi-node collaborative monitoring module to construct an electronic scale node set N = {n1, n2, ..., n}. k}, each node n i This indicates the collected local real-time weighing data. and local real-time weighing data Uploaded to the central monitoring system to form a global data set. Where k is the total number of nodes in the set N of the electronic scale nodes;

[0058] S62, Regarding the global data set D global (t) Perform multidimensional consistency verification by calculating the difference matrix ΔD(t) between node data to detect deviations between node data. When any element ΔD in the difference matrix ΔD(t) is... ij (t)>δ D At that time, the marked node n i and n j Abnormal node:

[0059]

[0060] Where, δ D N represents the set of electronic scale nodes, which is a preset threshold.

[0061] Multi-node collaborative features C are extracted based on the difference matrix ΔD(t).collab (t):

[0062]

[0063] in, Represents node n i The average difference between node n and other nodes is used to measure the difference between node n and other nodes. i Consistency with overall data;

[0064] S63. Analyze the marked abnormal nodes using an intelligent discrimination algorithm, and construct a time series model F based on the time series trend of the node data. time (t), and combined with multi-node collaborative features C collab (t) Calculate the node anomaly score S node (i):

[0065]

[0066] Where g is the scoring calculation function, when the node score S node (i)>γ, where γ is the anomaly scoring threshold, when determining node n i This is a cheating node;

[0067] S64. Record the set N of nodes determined to be cheating. cheat ={n i |S node (i)>γ}, and record the associated abnormal data as Stored on a cloud-based traceability management platform;

[0068] S65. Generate a comprehensive anomaly report. The report includes the set of nodes identified as cheating and the associated set of abnormal data. Upload the comprehensive anomaly report to the cloud-based traceability management platform and bind it with the full lifecycle associated data set for device status tracking and abnormal behavior analysis.

[0069] Optionally, S7 specifically includes:

[0070] S71. Configure a near-field communication module to establish a real-time connection between the electronic scale device and the user's mobile terminal, and transmit the local real-time weighing data and traceability information collected by the electronic scale device to the user's mobile terminal through near-field communication technology.

[0071] S72. Build a mobile application to receive and parse local real-time weighing data and traceability information transmitted by the electronic scale device, generate data display in the mobile application, and graphically display the received local real-time weighing data and traceability information.

[0072] S73. Construct a consumer participation supervision module, design a user feedback interface, allow users to provide feedback on the accuracy of weighing data and equipment status, generate user feedback records, and upload user feedback records to the cloud traceability management platform, linking them to the full lifecycle related data set for improving equipment performance and optimizing traceability records;

[0073] S74. In the cloud-based traceability management platform, user feedback records are dynamically analyzed, and feedback analysis reports are generated by combining the full lifecycle related data set.

[0074] An active learning-based anti-tampering system for electronic scales includes:

[0075] The data acquisition module is used to collect real-time weighing data, equipment operating status data, and environmental parameters of the electronic scale, and generate a standardized dataset.

[0076] The dynamic anomaly detection module builds a dynamic anomaly detection model based on an active learning algorithm, performs dynamic analysis on a standardized dataset, and identifies data anomalies through confidence calculation and optimization.

[0077] The data encryption module uses multi-level encryption algorithms to perform layered encryption on standardized datasets, generates a chained data storage structure, and constructs a unique data identifier to achieve data integrity protection and tamper-proofing.

[0078] The cloud-based traceability management platform is used to store and manage the entire lifecycle data of electronic scales from production to use, and to dynamically track and verify the data in real time through timestamps and version control.

[0079] Modular smart hardware, including a sensor signal locking module, a circuit integrity monitoring module, and an anti-tampering trigger module, is used to monitor and handle abnormal equipment signals, circuit modifications, and unauthorized disassembly.

[0080] The multi-node collaborative monitoring module is used to establish a linkage verification mechanism between multiple electronic scale nodes. It uses intelligent discrimination algorithms to comprehensively analyze abnormal behavior and identify and locate cheating behavior in complex scenarios.

[0081] The consumer monitoring module includes a near-field communication module and a mobile application. Users can obtain weighing data and traceability information in real time through the mobile device, and a feedback interface is provided to generate and upload user feedback records.

[0082] The beneficial effects of this invention are:

[0083] (1) By combining active learning algorithms and dynamic anomaly detection models, this invention provides real-time analysis and dynamic optimization capabilities for weighing data, enabling the system to adapt to constantly changing cheating methods and effectively identify abnormal weighing behaviors in complex scenarios. In particular, it achieves high precision and high reliability in data analysis for detecting new tampering methods.

[0084] (2) This invention achieves integrity protection and anti-tampering of weighing data from collection to storage through multi-level encryption algorithms and chain-based data storage technology, ensuring that the data cannot be irreversibly altered during transmission and storage. Simultaneously, combined with a cloud-based traceability management platform, it enables dynamic tracking and real-time verification of the entire lifecycle of electronic scale equipment, significantly improving the efficiency of data management and supervision.

[0085] (3) This invention establishes a cross-device linkage verification mechanism through a multi-node collaborative monitoring module, and combines intelligent discrimination algorithm to conduct comprehensive analysis of abnormal behavior, effectively identifying and locating large-scale collaborative cheating behavior in complex scenarios, thereby enhancing the anti-cheating capability of the system.

[0086] (4) Through the consumer supervision module and near-field communication technology, the present invention enables users to obtain weighing data and traceability information in real time and provide feedback through mobile applications, thus constructing a closed-loop supervision system with user participation. This not only improves the transparency of the system, but also significantly enhances users' trust in and ability to supervise the weighing data. Attached Figure Description

[0087] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0088] Figure 1 This is a general framework diagram of an electronic scale weighing anti-tampering method and system based on active learning proposed in this invention. Detailed Implementation

[0089] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0090] refer to Figure 1 A method for preventing tampering of electronic scale weighing based on active learning includes the following steps:

[0091] S1. Real-time weighing data, equipment operating status data and external environmental parameters of the electronic scale are acquired through multi-dimensional sensors to generate a raw dataset, and the raw dataset is preprocessed to form a standardized dataset.

[0092] In this embodiment, S1 specifically includes:

[0093] S11. Acquire real-time weighing data M(t) of the electronic scale through multi-dimensional sensors, and simultaneously collect equipment operating status data and external environmental parameters. The equipment operating status data includes sensor signal drift V. drift (t), Equipment operating frequency F sys (t), where the external environmental parameters include the ambient temperature T. env Humidity H env Atmospheric pressure P env and vibration disturbance intensity A vib Generate the original dataset D containing multi-source data. raw ={M(t),V drift (t),F sys (t),T env H env ,P env A vib};

[0094] S12, regarding the original dataset D raw Dynamic noise suppression is performed by calculating the noise characteristic vector N in real time. char (t) Adjust the noise threshold function ∈(t) to filter out random high-frequency noise introduced by vibration, environmental changes and equipment interference, and generate the noise-suppressed dataset D. filtered ;

[0095] S13. To address data gaps caused by sensor malfunctions and data acquisition interruptions in the dataset, perform adaptive data completion based on a time series model. Analyze the noise-suppressed dataset D. filtered By analyzing the temporal correlations and cross-constraints of the data across each dimension, a complete dataset D is generated. completed ;

[0096] S14. For the completed dataset D... completed Normalization and formatting are performed, and the real-time weighing data M(t) and sensor signal drift V are normalized using a dimensional normalization function. drift (t) and equipment operating frequency F sys (t) is dimensionless based on the unified timestamp T. std Realign all data dimensions to generate a standardized dataset D. standard ={T std ,M'(t),V' drift (t),F' sys (t),T' env ,H' env ,P' env ,A' vib}

[0097] This implementation method collects data through multi-dimensional sensors and performs noise reduction, data completion, and formatting to generate a standardized dataset, providing high-quality input for subsequent dynamic anomaly detection and data tamper-proofing. Combining real-time noise suppression and adaptive data completion mechanisms not only improves data integrity and consistency but also significantly enhances the accuracy and robustness of the anomaly detection model, providing a solid foundation for the reliability of electronic scale weighing data.

[0098] S2. Establish a dynamic anomaly detection model based on the active learning algorithm. Input the standardized dataset into the dynamic anomaly detection model. Calculate the confidence score of the data in the standardized dataset, select high-confidence data for online updates of the dynamic anomaly detection model, manually label low-confidence data, and update the dynamic anomaly detection model again after labeling.

[0099] In this embodiment, S2 specifically includes:

[0100] S21. Construct a dynamic anomaly detection model f based on an active learning algorithm. init (x; Θ0), where Θ0 is the initial parameter set of the model, and x is the input data vector, and the standardized dataset D is used. standard Input dynamic anomaly detection model f init (x; Θ0) are classified, and the corresponding confidence score set C(D) is output. standard )={C(x1),C(x2),…,C(x n )}, where C(x) i ) represents data x in a standardized dataset. i Confidence level belonging to the normal category;

[0101] S22, Based on the confidence score set C(D) standard A dynamic threshold δ(t) is set, which is adjusted in real time according to the confidence distribution characteristics, and the standardized dataset D is then processed. standard Divided into high-confidence dataset D high and low-confidence dataset D low :

[0102] D high ={x∈D standard |C(x)≥δ(t)};

[0103] D low ={x∈D standard |C(x)<δ(t)};

[0104] S23, The high-confidence data set D high Input dynamic anomaly detection model f init(x; Θ0), the initial parameter set Θ0 is adaptively optimized using an incremental learning algorithm, and the updated dynamic anomaly detection model is represented as f opt (x;Θ new ), where Θ new The optimized parameter set;

[0105] S24. For the low-confidence data set D low Manual annotation is performed to generate an annotated dataset. This annotated dataset is then used as new training samples and added to the dynamic anomaly detection model f. opt (x;Θ new In the training set, the dynamic anomaly detection model f is trained using an active learning framework. opt (x;Θ new A second optimization is performed to complete the deep update of the dynamic anomaly detection model parameters, and the final optimized dynamic anomaly detection model f is output. final (x;Θ final );

[0106] S25. Utilize the final optimized dynamic anomaly detection model f final (x;Θ final The system performs classification and anomaly detection on the real-time input dataset, outputs abnormal data to record and track during device operation, and integrates it into an anomaly-tagged dataset D. anomaly .

[0107] This implementation method uses a dynamic anomaly detection model based on active learning, combined with confidence calculation and online optimization, to achieve dynamic anomaly analysis and updating of standardized data. This not only improves the ability to identify novel anomalies in complex data environments, but also constructs a manual annotation feedback mechanism for low-confidence data, thus forming a closed-loop optimization. This significantly improves the accuracy and adaptability of anomaly detection, providing technical assurance for the integrity and reliability of weighing data.

[0108] S3. Encrypt the standardized dataset using a multi-level encryption algorithm. Construct a set of unique identifiers and associate the encrypted data with the set of unique identifiers to generate an encrypted associated data set. Construct a chained data storage structure based on the encrypted associated data set. Embed traceability information consisting of the electronic scale's device identifier, operation timestamp, and operation record on the chained data storage structure to generate a complete chained traceability data set.

[0109] In this embodiment, S3 specifically includes:

[0110] S31. For the standardized dataset D standard Encryption processing is performed on the standardized dataset D using multi-level encryption algorithms. standardEach data record in the dataset is processed in layers, and a symmetric encryption algorithm is used to process the standardized dataset D. standard The data in the middle is subjected to basic encryption to generate the first layer of encrypted data set D. enc1 =(E sym (x i ;K sym )|i=1,2,...,n}, where x i To standardize the data items in the dataset, K sym Based on the encryption key, E sym Represents a symmetric encryption function;

[0111] S32, Regarding the first-layer encrypted data set D enc1 The unique identifier set H(D) for each piece of data is calculated using a hash algorithm. enc1 )={h i |h i =Hash(E sym (x i ;K sym ),i=1,2,...,n}, where h i This is an irreversible identifier corresponding to the encrypted data record;

[0112] S33, The first layer of encrypted data set D enc1 With the set of unique identifiers H(D) enc1 Perform associated storage to generate an encrypted associated data set D. assoc ={(h i E sym (x i ;K sym ))∣i=1,2,...,n};

[0113] S34. Construct a linked data storage structure L(D) assoc ):

[0114] L(D assoc )={(h i-1 ,h i ,h i+1 )|i=2,3,...,n-1};

[0115] In this chained data storage structure, each node contains a unique identifier h for the current record. i Pre-record identifier h i-1 and subsequent record identifier h i+1 ;

[0116] S35. Embed traceability information on the basis of the chain-like data storage structure. The traceability information includes the electronic scale device identifier ID. device Operation timestamp Top and operation log R op Generate a complete chain-source data set D for recording and tracking encrypted data. trace ={(L(D)} assoc ), ID device ,T op ,R op )}.

[0117] This implementation method combines multi-level encryption algorithms and a chain-store structure to ensure the integrity and immutability of weighing data. At the same time, by associating unique identifiers with traceability information, it enables dynamic tracking and real-time verification of weighing data throughout the entire process, improving data security and traceability management efficiency, and providing reliable technical protection for preventing cheating on electronic scales.

[0118] S4. Construct a cloud-based traceability management platform and a full lifecycle management model. Upload the encrypted chain traceability data set to the cloud-based traceability management platform. Record the data of the electronic scale equipment from production, calibration, maintenance to use through the full lifecycle management model. Dynamically track and verify the data of each stage through timestamps and version control.

[0119] In this embodiment, S4 specifically includes:

[0120] S41. The chain-based traceability data set D trace Uploaded to the cloud-based traceability management platform, the chain traceability data set D is managed based on a full lifecycle management model. trace Initialization is performed to obtain the initialized source data set T. init ={D trace Meta device}, where Meta device Including electronic scale device identifier ID device Production date T prod Calibration parameter C calib Maintenance Records R maint and usage scenario information S usage ;

[0121] S42. Based on the timestamp generation mechanism, for the traceability data set T init Each data record in the database is appended with a unique timestamp T. stamp (i) The timestamp calculation rule is as follows:

[0122] T stamp (i)=T upload +Δt(i);

[0123] Among them, T upload This represents the baseline time for data upload, where Δt(i) is the relative time offset for each record upload.

[0124] S43. Based on the version control mechanism, trace the source data set T. init Perform historical status management and generate multi-version traceability data sets. in This indicates the state of the source dataset in version z;

[0125] S44. Combining the full lifecycle management model, the multi-version traceability data set T versioned The equipment's production, calibration, maintenance, and usage information are linked and stored in multiple dimensions to generate a full lifecycle linked data set T. lifecycle ={(T init Meta device ,T versioned )};

[0126] S45. Utilize a cloud-based traceability management platform to manage the entire lifecycle-related data set T. lifecycle Dynamic tracking and real-time verification are performed. After verification is successful, a set of verification records is generated. When verification fails, an exception handling mechanism is triggered to store the exception records as an exception data set.

[0127] This implementation method uses a cloud-based traceability management platform combined with timestamps and version control mechanisms to achieve dynamic tracking and real-time verification of the entire lifecycle data of electronic scales, ensuring data integrity and immutability. It also supports the associated storage and updating of multiple versions of traceability data, improving the accuracy of anomaly detection and tracking, and providing comprehensive support for supervision and equipment management.

[0128] S5 integrates modular smart hardware, configured with a sensor signal locking module, a circuit integrity monitoring module, and an anti-tampering trigger module. When an abnormal sensor signal or unauthorized modification of the circuit is detected, the hardware locking function is triggered, and the event is recorded and an alarm message is generated.

[0129] In this embodiment, S5 specifically includes:

[0130] S51. Configure a sensor signal locking module to monitor the sensor output signal S in real time. signal (t) and calculate the signal deviation ΔS(t):

[0131] ΔS(t)=|S signal (t)-S ref |;

[0132] Among them, S ref For sensor calibration reference values, when ΔS(t) > δ S At that time, the sensor signal is locked and a sensor abnormality signal S is generated. anomaly (t)=S signal (t), where δS The preset deviation threshold;

[0133] S52. Configure a circuit integrity monitoring module to dynamically detect the circuit's operating status, collect real-time current and voltage, and calculate the circuit's characteristic state C. integrity (t), when When this occurs, a circuit malfunction is detected, triggering the hardware lockout function and recording the circuit malfunction status, where R... normal Indicates the normal operating range of the circuit;

[0134] S53. Configure an anti-tamper trigger module to monitor the integrity of the device's hardware casing and the anti-tamper point signal D. secure (t) Determine whether the device has been disassembled without authorization, when D secure When (t) = 0, a disassembly alarm signal A is generated. tamper (t) triggers the anti-disassembly locking mechanism and records disassembly anomaly information;

[0135] S54, Integrated sensor abnormal signal S anomaly (t), Circuit characteristic state C integrity (t) and disassembly alarm signal A tamper (t), generating an exception record set R anomaly (t)={S anomaly (t),C integrity (t),A tamper (t)}, and R anomaly (t) Stored in a cloud-based traceability management platform;

[0136] S55, Regarding the abnormal record set R anomaly (t) Generate alarm information, upload the alarm information to the cloud traceability management platform and bind it to the full life cycle associated data set for tracking device status and supporting anomaly handling.

[0137] This implementation method, through the combined action of sensor signal locking, circuit integrity monitoring, and anti-tampering triggering modules, can capture abnormal equipment signals, circuit modifications, and unauthorized disassembly in real time, generate complete anomaly records, and upload them to the cloud. This comprehensively improves the equipment's anti-tampering capabilities and the reliability of data traceability, providing all-round protection for the safe operation of electronic scales.

[0138] S6. Through real-time multi-node collaborative monitoring technology, a linkage verification mechanism between multiple electronic scale nodes is established. Through intelligent discrimination algorithm, the abnormal behavior of multiple nodes is comprehensively analyzed, and cheating behavior in complex scenarios is identified and located.

[0139] In this embodiment, S6 specifically includes:

[0140] S61. Based on real-time multi-node collaborative monitoring technology, configure a multi-node collaborative monitoring module to construct an electronic scale node set N = {n1, n2, ..., n}. k}, each node n i This indicates the collected local real-time weighing data. and local real-time weighing data Uploaded to the central monitoring system to form a global data set. Where k is the total number of nodes in the set N of the electronic scale nodes;

[0141] S62, Regarding the global data set D global (t) Perform multidimensional consistency verification by calculating the difference matrix ΔD(t) between node data to detect deviations between node data. When any element ΔD in the difference matrix ΔD(t) is... ij (t)>δ D At that time, the marked node n i and n j Abnormal node:

[0142]

[0143] Where, δ D N represents the set of electronic scale nodes, which is a preset threshold.

[0144] Multi-node collaborative features C are extracted based on the difference matrix ΔD(t). collab (t):

[0145]

[0146] in, Represents node n i The average difference between node n and other nodes is used to measure the difference between node n and other nodes. i Consistency with overall data;

[0147] S63. Analyze the marked abnormal nodes using an intelligent discrimination algorithm, and construct a time series model F based on the time series trend of the node data. time (t), and combined with multi-node collaborative features C collab (t) Calculate the node anomaly score S node (i):

[0148]

[0149] Where g is the scoring calculation function, when the node score S node (i)>γ, where γ is the anomaly scoring threshold, when determining node n i This is a cheating node;

[0150] S64. Record the set N of nodes determined to be cheating.cheat ={n i |S node (i)>γ}, and record the associated abnormal data as Stored on a cloud-based traceability management platform;

[0151] S65. Generate a comprehensive anomaly report. The report includes the set of nodes identified as cheating and the associated set of abnormal data. Upload the comprehensive anomaly report to the cloud-based traceability management platform and bind it with the full lifecycle associated data set for device status tracking and abnormal behavior analysis.

[0152] This implementation method combines difference matrix analysis with time series models through a multi-node collaborative monitoring module. By integrating data consistency and historical trends among nodes, it accurately identifies cheating behavior in complex scenarios. Intelligent discrimination algorithms are used to score and locate abnormal nodes, improving detection accuracy and efficiency, and providing a highly efficient and reliable solution for tamper-proofing electronic scales.

[0153] S7 combines mobile applications and near-field communication technology to provide users with real-time weighing data and traceability information.

[0154] In this embodiment, S7 specifically includes:

[0155] S71. Configure a near-field communication module to establish a real-time connection between the electronic scale device and the user's mobile terminal, and transmit the local real-time weighing data and traceability information collected by the electronic scale device to the user's mobile terminal through near-field communication technology.

[0156] S72. Build a mobile application to receive and parse local real-time weighing data and traceability information transmitted by the electronic scale device, generate data display in the mobile application, and graphically display the received local real-time weighing data and traceability information.

[0157] S73. Construct a consumer participation supervision module, design a user feedback interface, allow users to provide feedback on the accuracy of weighing data and equipment status, generate user feedback records, and upload user feedback records to the cloud traceability management platform, linking them to the full lifecycle related data set for improving equipment performance and optimizing traceability records;

[0158] S74. In the cloud-based traceability management platform, user feedback records are dynamically analyzed, and feedback analysis reports are generated by combining the full lifecycle related data set.

[0159] This implementation combines near-field communication and mobile applications to transmit weighing data and traceability information in real time. It also introduces a user feedback mechanism to incorporate user participation into the weighing data supervision process, which not only improves data transparency but also achieves dynamic tracking and feedback loop improvement, providing comprehensive support for the reliability and tamper-proof of electronic scale data.

[0160] An active learning-based anti-tampering system for electronic scales includes:

[0161] The data acquisition module is used to collect real-time weighing data, equipment operating status data, and environmental parameters of the electronic scale, and generate a standardized dataset.

[0162] The dynamic anomaly detection module builds a dynamic anomaly detection model based on an active learning algorithm, performs dynamic analysis on a standardized dataset, and identifies data anomalies through confidence calculation and optimization.

[0163] The data encryption module uses multi-level encryption algorithms to perform layered encryption on standardized datasets, generates a chained data storage structure, and constructs a unique data identifier to achieve data integrity protection and tamper-proofing.

[0164] The cloud-based traceability management platform is used to store and manage the entire lifecycle data of electronic scales from production to use, and to dynamically track and verify the data in real time through timestamps and version control.

[0165] Modular smart hardware, including a sensor signal locking module, a circuit integrity monitoring module, and an anti-tampering trigger module, is used to monitor and handle abnormal equipment signals, circuit modifications, and unauthorized disassembly.

[0166] The multi-node collaborative monitoring module is used to establish a linkage verification mechanism between multiple electronic scale nodes. It uses intelligent discrimination algorithms to comprehensively analyze abnormal behavior and identify and locate cheating behavior in complex scenarios.

[0167] The consumer monitoring module includes a near-field communication module and a mobile application. Users can obtain weighing data and traceability information in real time through the mobile device, and a feedback interface is provided to generate and upload user feedback records.

[0168] This implementation method integrates data acquisition, dynamic anomaly detection, multi-level encryption, cloud-based traceability management, modular smart hardware, multi-node collaborative monitoring, and consumer supervision modules to achieve multi-dimensional anti-tampering management of the entire lifecycle of electronic scales. It can detect and locate abnormal behavior in real time, improve the accuracy of data integrity protection and cheating behavior identification, and build an efficient and secure electronic scale management system.

[0169] Example:

[0170] The following is a detailed description of an embodiment of the present invention. Taking a large logistics center's application of the anti-tampering technology during actual weighing as an example, this embodiment details how a fixed verification weight is concealed before weighing on the electronic scale, and how sensors monitor the status of the verification weight and compare it with the weighing data to effectively protect against data tampering, and the resulting beneficial effects. This embodiment not only fully solves the problem of electronic scales being tampered with or having their data tampered with during the weighing process, but also intuitively demonstrates the technical advantages and implementation effects of the present invention in practical applications.

[0171] In the daily weighing process of a logistics center, due to the high-frequency use of electronic scales, some operators or external criminals may cheat by modifying sensors or interfering with the data acquisition system, resulting in distorted weighing data and discrepancies between actual delivery and settlement, causing serious economic losses to the company. Traditional anti-tampering solutions typically rely only on digital detection and data encryption measures, which often fail to detect abnormal data in a timely manner in the face of complex on-site environments and malicious tampering. To completely solve this problem, this embodiment innovatively incorporates a fixed calibration weight hidden inside the electronic scale before weighing, and a dedicated sensor monitors the status of this weight in real time as a physical verification basis. When the electronic scale performs normal weighing, the system simultaneously collects the main weighing data and the calibration weight status data. By comparing the actual monitored data of the calibration weight with the preset benchmark value, it can determine whether data tampering has occurred. If the calibration weight status data deviates abnormally from the expected value, an alarm mechanism is immediately triggered, marking the weighing data as abnormal and automatically uploading it to the cloud traceability platform for subsequent tracing and processing. This solution effectively avoids the vulnerabilities that may exist if dynamic anomaly detection algorithms are relied upon alone, while providing an auxiliary detection method based on physical characteristics verification, ensuring that the overall system can still respond in a timely manner and take protective measures when faced with malicious tampering.

[0172] In its implementation, the logistics center chose to deploy the system in a large logistics warehouse in Hangzhou, Zhejiang Province. This warehouse handles a wide variety and large quantity of goods daily, requiring extremely high weighing accuracy. To ensure the anti-tampering system could withstand the high-load operating environment, a concealed, fixed calibration weight was added to the original electronic scale structure. This fixed weight, weighing 10 kg, was installed in a key part of the electronic scale using a special tamper-proof fixing device, ensuring it could not be easily removed or altered during any operation. Simultaneously, the built-in sensor module of the electronic scale was redesigned to collect real-time monitoring data of the calibration weight while simultaneously acquiring regular weighing data. During the system's debugging phase, extensive testing data determined the standard monitoring value of the calibration weight and the allowable error range (generally controlled within ±0.2 kg), which was used as a crucial basis for subsequent judgments regarding whether the weighing data had been tampered with. The entire system also incorporates a dynamic anomaly detection model built using active learning algorithms. It performs multi-level processing on all collected data and compares it with preset physical verification results. If there is a significant inconsistency between the two, the system will immediately determine that there is a risk of data tampering and will automatically issue an alarm within a short period of time.

[0173] In this embodiment, the logistics center underwent a three-month trial operation from January to March 2024, with the system monitoring all weighing data 24 / 7. A comparison with traditional electronic scale systems clearly demonstrates the significant effectiveness of this system in preventing data tampering, improving anomaly detection accuracy, and shortening anomaly response time. To more intuitively illustrate the beneficial effects of this embodiment, the table below, "Statistical Table of Anti-Tampering Implementation Effect of Electronic Scales in the Logistics Center," lists some key data from the trial operation period. This statistical table records several key indicators, including the number of anomaly detections, anomaly event response time, data integrity verification success rate, and user feedback adoption rate. All data are real data automatically recorded by the system during the trial operation. The data shown in the table reflects a significant difference in performance before and after the system's official launch, proving the superiority of this invention in preventing tampering.

[0174] Table 1: Statistical Table of Anti-Tampering Implementation Effect of Electronic Scales in Logistics Center

[0175] Indicator name Conventional system value Present system value Abnormal event detection times Average 25 times per month Average 3 times per month Response time (minutes) About 40 minutes About 5 minutes Data integrity verification rate About 90.5% Over 99.2% User feedback adoption rate About 18% Over 80% Weighing error fluctuation range ±1.5 kg Limited to within ±0.3 kg

[0176] In practical applications, logistics center managers reported that traditional systems often required lengthy manual checks when encountering anomalies, and struggled to detect concealed tampering in a timely manner, leading to inaccurate weighing data for a large number of goods and impacting subsequent loading, unloading, and logistics scheduling. However, with the implementation of this system, thanks to the dual protection of concealed fixed calibration weights and real-time sensor monitoring, the system can instantly detect data anomalies during the weighing process. The number of anomaly detections has been significantly reduced from 25 per month to 3, and the average system response time has been shortened from approximately 40 minutes in the traditional system to less than 5 minutes, greatly reducing logistics delays and economic losses caused by data anomalies. The data integrity verification rate has increased from 90.5% in the traditional system to over 99.2%, ensuring the authenticity and reliability of every weighing data point. Furthermore, the user feedback acceptance rate has increased from 18% to over 80%, indicating a significant improvement in the trust and satisfaction of operators and customers with the system. Meanwhile, in terms of weighing error, traditional systems often experience large data fluctuations due to tampering, with errors reaching ±1.5 kg. However, this system, through a combination of physical verification and digital detection, stabilizes the weighing error within ±0.3 kg, greatly improving weighing accuracy.

[0177] During the trial operation, the system collected over 150,000 data entries. Through an active learning algorithm, the data was compared and verified in real time. Upon detecting abnormal data, the system automatically generated an anomaly report and notified on-site technicians for verification within minutes. The warehouse manager stated in an interview that the system not only improved the accuracy of weighing data but also significantly reduced disputes and economic losses caused by data tampering, making the entire logistics scheduling process smoother and more efficient. Furthermore, the system established a comprehensive data traceability mechanism on the cloud-based traceability platform, leaving detailed records of every abnormal event. This facilitates subsequent problem analysis and equipment maintenance, ensuring the transparency and traceability of the entire system's operation.

[0178] After three months of continuous trial operation, the logistics center underwent a comprehensive evaluation at the end of March 2024. The evaluation results showed that the introduction of this system transformed the logistics center's weighing data anomalies from a high-incidence event to an extremely rare occurrence. On-site anomaly response time was significantly shortened, and the accuracy of data verification and traceability was greatly improved. During the trial operation, a total of 9 anomalies were detected, while the traditional system averaged 75 anomalies during the same period. Notably, after detecting anomalies, this system could complete alarm and initial handling within an average of 5 minutes, while the traditional system often required more than 40 minutes for manual verification in similar situations. The data demonstrates that the system has achieved significant results in ensuring data integrity, reducing operational risks, and improving logistics efficiency.

[0179] Throughout the implementation process, the logistics center not only achieved strict monitoring of every weighing data point, but also ensured timely response and handling of any anomalies detected through automatic system recording and real-time comparison. On-site operators generally reported that the system was simple and intuitive to operate, running stably even during peak periods, significantly reducing the risks associated with human error or tampering. After a period of trial operation, the system underwent continuous parameter adjustments and optimization based on actual on-site conditions, further improving detection accuracy and response speed, thus establishing an effective anti-tampering management process.

[0180] This embodiment fully demonstrates, through specific scenarios and detailed data, the electronic scale weighing anti-tampering technology solution proposed in this invention, which combines active learning with physical anti-tampering measures. It not only effectively solves the data tampering problem inherent in traditional electronic scales but also achieves significant economic and social benefits in practical applications. Trial operation data and user feedback both indicate that this technology solution can achieve real-time monitoring and efficient protection of weighing data, ensuring the authenticity, accuracy, and security of every weighing operation. Therefore, it provides a new technological option and implementation path for the logistics industry and other fields requiring high-precision weighing data support.

[0181] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for preventing tampering of electronic scale weighing based on active learning, characterized in that, Includes the following steps: S1. Acquire real-time weighing data, equipment operating status data, and external environmental parameters of the electronic scale, generate a raw dataset, and preprocess it to form a standardized dataset; S2. Establish a dynamic anomaly detection model. Input the standardized dataset into the dynamic anomaly detection model. Calculate the confidence score of the data in the standardized dataset, select high-confidence data for online updates of the dynamic anomaly detection model, manually label low-confidence data, and update the dynamic anomaly detection model again after labeling. S3. Encrypt the standardized dataset and construct a set of unique identifiers. Link the encrypted data and the set of unique identifiers for storage to generate an encrypted linked data set. Construct a chained data storage structure based on the encrypted linked data set and embed traceability information on the chained data storage structure to generate a chained traceability data set. S4. Construct a cloud-based traceability management platform and a full lifecycle management model. Record data of each stage of the electronic scale equipment through the full lifecycle management model, and achieve dynamic tracking and real-time verification of data at each stage through timestamps and version control. S5 integrates modular smart hardware. When abnormal sensor signals or unauthorized circuit modifications are detected, the hardware lock function is triggered, and the event is recorded and alarm information is generated. S6. By using real-time multi-node collaborative monitoring technology, a linkage verification mechanism is established between multiple electronic scale nodes to identify and locate cheating behavior in complex scenarios. S7 combines mobile applications and near-field communication technology to provide users with real-time weighing data and traceability information; S4 specifically includes: S41. Set up chain-based traceability data Uploaded to the cloud-based traceability management platform, the chain traceability data set is managed based on a full lifecycle management model. Initialization is performed to obtain the initialized source data set. ,in Including electronic scale device identifiers Production date Calibration parameters Maintenance records and usage scenario information ; S42. Based on the timestamp generation mechanism, for the traceability data set Each data record in the database is appended with a unique timestamp. The timestamp calculation rule is as follows: ； in, Indicates the base time for data upload. This refers to the relative time offset when each record is uploaded. S43. Based on the version control mechanism, trace the source data set. Perform historical status management and generate multi-version traceability data sets. ,in Indicates the source data set in the th... The status of each version; S44. Combining the full lifecycle management model, the multi-version traceability data set This data is stored in a multi-dimensional association with the equipment's production, calibration, maintenance, and usage information to generate a full lifecycle data set. ; S45. Utilize a cloud-based traceability management platform to manage the entire lifecycle-related data set. Dynamic tracking and real-time verification are performed. After verification is successful, a set of verification records is generated. When verification fails, an exception handling mechanism is triggered to store the exception records as an exception data set.

2. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, S1 specifically includes: S11. Obtain real-time weighing data of the electronic scale through multi-dimensional sensors. The system synchronously collects equipment operating status data and external environmental parameters, including sensor signal drift. Equipment operating frequency External environmental parameters include ambient temperature. ,humidity Atmospheric pressure and vibration disturbance intensity Generate the original dataset containing data from multiple sources. ; S12, For the original dataset Dynamic noise suppression is performed by calculating the noise characteristic vector in real time. Adjusting the noise threshold function The random high-frequency noise introduced by vibration, environmental changes, and equipment interference is filtered out to generate a noise-suppressed dataset. ; S13. To address data gaps in the dataset caused by sensor malfunctions and data acquisition interruptions, perform adaptive data completion based on a time series model, and analyze the dataset after noise suppression. By analyzing the temporal correlations and cross-constraints of the data across various dimensions, a complete dataset is generated. ; S14. Complete the dataset after completion. Normalization and formatting are performed on the real-time weighing data using a dimensional normalization function. Sensor signal drift and equipment operating frequency Dimensionless transformation is performed based on a unified timestamp. Realign all data dimensions to generate a standardized dataset. .

3. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, S2 specifically includes: S21. Construct a dynamic anomaly detection model based on active learning algorithm. ,in For the initial parameter set of the model, For the input data vector, the standardized dataset will be used. Input dynamic anomaly detection model Perform classification processing and output the corresponding confidence score set. ,in Represents data in a standardized dataset Confidence level belonging to the normal category; S22, Based on confidence score set Set dynamic threshold The dynamic threshold The threshold is adjusted in real time based on the confidence distribution characteristics, and the standardized dataset is... Divided into high-confidence datasets and low confidence datasets : ； ； S23, Set up high-confidence data Input dynamic anomaly detection model An incremental learning algorithm is used for the initial parameter set. After adaptive optimization, the updated dynamic anomaly detection model is represented as follows: ,in The optimized parameter set; S24. For low-confidence data sets Manual annotation is performed to generate an annotated dataset, which is then used as new training samples and added to the dynamic anomaly detection model. In the training set, the dynamic anomaly detection model is trained using an active learning framework. A second optimization is performed to complete the deep update of the dynamic anomaly detection model parameters, and the final optimized dynamic anomaly detection model is output. ; S25. Utilize the final optimized dynamic anomaly detection model The system performs classification and anomaly detection on the real-time input dataset, outputs abnormal data for recording and tracking during device operation, and integrates them into an anomaly-tagged dataset. .

4. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, S3 specifically includes: S31. Standardized dataset Encryption processing is performed on the standardized dataset using multi-layered encryption algorithms. Each data record in the dataset is processed in layers, and a symmetric encryption algorithm is used to process the standardized dataset. The data in the middle is subjected to basic encryption to generate the first layer of encrypted data set. ,in To standardize the data items in the dataset, Based on the encryption key, Represents a symmetric encryption function; S32, For the first layer of encrypted data set The set of unique identifiers for each piece of data is calculated using a hash algorithm. ,in This is an irreversible identifier corresponding to the encrypted data record; S33, The first layer of encrypted data set With unique identifier set Perform associated storage to generate an encrypted associated data set. ; S34. Construct a chained data storage structure : ； In this chained data storage structure, each node contains a unique identifier for the current record. Pre-record identifier and subsequent record identifiers ; S35. Embed traceability information on the basis of the chain-like data storage structure. The traceability information includes the electronic scale device identifier. Operation timestamp and operation records Generate a complete chain of traceability data for recording and tracking encrypted data. .

5. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, S5 specifically includes: S51. Configure a sensor signal lock-in module to monitor the sensor output signal in real time. And calculate the signal deviation. : ； in, To calibrate the sensor reference value, when At that time, the sensor signal is locked and a sensor abnormality signal is generated. ,in The preset deviation threshold; S52. Configure a circuit integrity monitoring module to dynamically detect the circuit's operating status, collect real-time current and voltage, and calculate the circuit's characteristic states. ,when When this occurs, a circuit malfunction is detected, triggering the hardware lockout function and recording the circuit malfunction status. Indicates the normal operating range of the circuit; S53. Configure an anti-tampering trigger module to monitor the integrity of the device's hardware casing and anti-tampering point signals. To determine whether a device has been disassembled without authorization, when At that time, a disassembly alarm signal is generated. This triggers the anti-disassembly locking mechanism and records disassembly anomaly information; S54, Integrating sensor abnormal signals Circuit characteristic states and disassembly alarm signal Generate an exception record set and will Stored on a cloud-based traceability management platform; S55, Regarding the collection of abnormal records Generate alarm information, upload the alarm information to the cloud traceability management platform, and bind it to the full lifecycle associated data set to track device status and support anomaly handling.

6. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, S6 specifically includes: S61. Based on real-time multi-node collaborative monitoring technology, configure a multi-node collaborative monitoring module to construct an electronic scale node set. Each node This indicates the collected local real-time weighing data. and local real-time weighing data Uploaded to the central monitoring system to form a global data set. ,in, It is a set of electronic scale nodes. The total number of nodes in the middle; S62, Regarding the global data set Perform multidimensional consistency verification by calculating the difference matrix of data between nodes. Detect the deviation between node data, when the difference matrix any element At that time, mark the node and Abnormal node: ； in, For the preset threshold, Represents the set of nodes of an electronic scale; Based on the difference matrix Extracting multi-node collaborative features : ； in, Represents a node The average difference between a node and other nodes is used to measure the node's performance. Consistency with overall data; S63. Analyze the marked abnormal nodes using intelligent discrimination algorithms, and construct a time series model based on the time series trend of the node data. And combined with multi-node collaborative features Computation node anomaly scoring : ； in, For the scoring calculation function, when the node scores... ,in When the abnormal scoring threshold is reached, the decision node is determined. This is a cheating node; S64. Record the set of nodes determined to be cheating. And record the associated abnormal data as Stored in the cloud-based traceability management platform; S65. Generate a comprehensive anomaly report. The report includes the set of nodes identified as cheating and the associated set of abnormal data. Upload the comprehensive anomaly report to the cloud-based traceability management platform and bind it with the full lifecycle associated data set for device status tracking and abnormal behavior analysis.

7. The method for preventing tampering of electronic scale weighing based on active learning according to claim 1, characterized in that, Specifically, S7 includes: S71. Configure a near-field communication module to establish a real-time connection between the electronic scale device and the user's mobile terminal, and transmit the local real-time weighing data and traceability information collected by the electronic scale device to the user's mobile terminal through near-field communication technology. S72. Build a mobile application to receive and parse local real-time weighing data and traceability information transmitted by the electronic scale device, generate data display in the mobile application, and graphically display the received local real-time weighing data and traceability information. S73. Construct a consumer participation supervision module, design a user feedback interface, allow users to provide feedback on the accuracy of weighing data and equipment status, generate user feedback records, and upload user feedback records to the cloud traceability management platform, linking them to the full lifecycle related data set for improving equipment performance and optimizing traceability records; S74. In the cloud-based traceability management platform, user feedback records are dynamically analyzed, and feedback analysis reports are generated by combining the full lifecycle related data set.

8. An active learning-based electronic scale weighing anti-tampering system, applied to the active learning-based electronic scale weighing anti-tampering method according to any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect real-time weighing data, equipment operating status data, and environmental parameters of the electronic scale, and generate a standardized dataset. The dynamic anomaly detection module builds a dynamic anomaly detection model based on an active learning algorithm, performs dynamic analysis on a standardized dataset, and identifies data anomalies through confidence calculation and optimization. The data encryption module uses multi-level encryption algorithms to perform layered encryption on standardized datasets, generates a chained data storage structure, and constructs a unique data identifier to achieve data integrity protection and tamper-proofing. The cloud-based traceability management platform is used to store and manage the entire lifecycle data of electronic scales from production to use, and to dynamically track and verify the data in real time through timestamps and version control. Modular smart hardware, including a sensor signal locking module, a circuit integrity monitoring module, and an anti-tampering trigger module, is used to monitor and handle abnormal equipment signals, circuit modifications, and unauthorized disassembly. The multi-node collaborative monitoring module is used to establish a linkage verification mechanism between multiple electronic scale nodes. It uses intelligent discrimination algorithms to comprehensively analyze abnormal behavior and identify and locate cheating behavior in complex scenarios. The consumer monitoring module includes a near-field communication module and a mobile application. Users can obtain weighing data and traceability information in real time through the mobile device, and a feedback interface is provided to generate and upload user feedback records.

Images