CN-121280189-B - Technician training and project dynamic authorization engine method based on blockchain

CN121280189BCN 121280189 BCN121280189 BCN 121280189BCN-121280189-B

Abstract

The invention provides a technician training and project dynamic authorization engine method based on a blockchain, which comprises the steps of obtaining an original training event sequence of a technician, constructing a decision vector according to the training data chain, calculating authorization adaptation score through an intelligent contract to obtain an authority ID and recording a change log chain, monitoring the authority change log in real time, evaluating risks through an authority freezing decision model, calculating freezing adaptation score, triggering the intelligent contract freezing authority when the freezing adaptation score exceeds a threshold value, generating an early warning notice and a uplink record, constructing a feedback vector based on the authority freezing record, calculating an optimization adaptation score through a certificate optimization decision model, issuing a certificate unique mark as an NFT (network File) certificate through the intelligent contract after reaching standards, and synchronously adjusting the task role of the technician and uplink certificate. The invention improves the flexibility, efficiency and transparency of the system and can meet the requirements of modern enterprises on high-efficiency, accurate and traceable management.

Inventors

YANG XI
ZHANG YANBIN
WEN HUAN
LIN CHEN
LIU SANJUN
LIANG JIAYOU
YANG LI
ZHONG MING
WEI TIANCAI

Assignees

广州分析测试中心科力技术开发公司
广州海关技术中心
广东省科学院测试分析研究所(中国广州分析测试中心)
广州双合软件有限公司
苏交科集团广东检测认证有限公司

Dates

Publication Date: 20260512
Application Date: 20250912

Claims (2)

1. A blockchain-based technician training and project dynamic authorization engine method, the method comprising the steps of: S1, acquiring an original training event sequence of a technician, processing dynamic interference with an LSTM model through an attention mechanism, generating a refining vector, uploading the refining vector to a blockchain storage certificate after the associated technician is identified, forming a training data chain, and outputting an assessment result; S2, constructing a decision vector according to the training data chain, calculating an authorization adaptation score through an intelligent contract, generating an authorization token when the authorization adaptation score reaches a threshold value, updating access rights and uploading a chain certificate to obtain a rights ID and recording a change log chain; S3, monitoring a permission change log in real time, evaluating risks through a permission freezing decision model, calculating freezing adaptation scores, triggering intelligent contract freezing permission when the freezing adaptation scores exceed a threshold value, generating early warning notification and a uplink record, and obtaining a risk ID for identifying a freezing instance; S4, constructing a feedback vector based on the authority freezing record, calculating an optimization adaptation score through a certificate optimization decision model, and issuing a certificate unique identifier as an NFT (network File) certificate after reaching standards through an intelligent contract, and synchronously adjusting the task role of a technician and uploading the certificate, wherein the feedback vector is a risk ID and an early warning notice; the original training event sequence comprises a unique identification of a technician, a course module interaction intensity vector sequence and an assessment response record; The dynamic interference is processed through an attention mechanism and an LSTM model to generate a refining vector, and the refining vector is uploaded to a blockchain storage certificate after being identified by a correlation technician to form a training data chain, which is specifically as follows: integrating the original training event sequence into a unified input vector; Inputting the input vector into an attention mechanism and an LSTM model to generate a refining vector, wherein the attention mechanism and the LSTM model are specifically as follows: In the multi-head attention layer, three attention heads jointly calculate the attention score of the node to obtain the attention score of the corresponding node, and the attention weighted sequence is input into an interference adaptation LSTM unit to update the state to obtain the hidden state of the current time step; When the norm threshold of the hidden state of the current time step is smaller than a preset threshold, the hidden state of the current time step is used as a refining vector to be associated with the unique identification of the technician to form a structured record, and the structured record is uploaded to a blockchain storage certificate to form a training data chain; The S1 further includes: Inputting the assessment response record into an LSTM model, extracting a time sequence feature vector through cell state updating, combining the assessment response record and the time sequence feature vector to perform product fusion, and revising through interference deviation to obtain a revised feature vector; The S2 further includes: integrating the training data chain and the assessment result into a decision vector; Inputting the decision vector into an intelligent contract decision model to obtain an authorized decision vector; Calculating an authorization adaptation score based on the authorization decision vector and a training data chain and an assessment result, wherein the authorization adaptation score is used for evaluating linkage suitability of a current decision; When the authorization adaptation score reaches a preset threshold value, an intelligent contract function is activated, an authorization token is generated by taking the authorization adaptation score as an input vector parameter, and the authorization token is linked with a project management system in real time through an API (application program interface), so that corresponding access items are updated; the intelligent contract decision model comprises a threshold activating layer and an adaptive weight feedforward unit, wherein, The threshold activating layer processes the achievement qualification judgment by using a sigmoid function and converts the achievement qualification judgment into a condition that the achievement qualification judgment is close to 1 and a condition that the achievement qualification judgment is close to 0; The adaptive weight feedforward unit includes: the first layer is a linear layer, and the weight matrix is initialized to be uniformly distributed; the second layer is a ReLU activation layer for nonlinear conversion; the third layer is an output layer, and a sigmoid function is adopted to compress the value to a [0,1] interval to form a final authorization decision vector; the step S3 further includes: Real-time monitoring authority ID and recording change log chains are integrated into a monitoring vector; inputting the monitoring vector into a right freezing decision model, wherein the right freezing decision model comprises a time sequence embedding layer and a risk adaptation GRU model, and outputting a risk assessment vector; The freezing function takes the risk assessment vector as a parameter, generates a freezing token, updates the authority state on the blockchain to be frozen, and sends out an early warning notice through an event log to warn possible abuse risk; The time sequence embedding layer processes the log chain of record change through a position coding function, wherein the embedding dimension of each log node is matched with the changing dimension, the risk adaptation GRU model is a three-layer GRU network, and the structure of each layer is as follows: The first layer is a GRU layer, an input gate and a forget gate are processed through a door control mechanism by using a sigmoid function, and historical information is captured; The second layer is an attention enhancement layer, and a self-attention mechanism is applied to each node to highlight key changes; The third layer is an output layer, a final risk vector is obtained through linear transformation, and an output result is generated through ReLU activation; the step S4 specifically comprises the following steps: Acquiring a feedback vector, taking the risk ID as a risk reference point, extracting an event chain of early warning information, and forming multidimensional historical data; Inputting the feedback vector into a certificate optimization decision model and outputting a final decision vector, wherein the certificate optimization decision model comprises an event feedback embedding layer and an adaptive transducer coding unit; Calculating an optimization adaptation score based on the final decision vector, the risk ID and the early warning information, and activating an intelligent contract optimization function when the optimization adaptation score reaches a threshold value, wherein the intelligent contract optimization function takes the final decision vector as an input vector, generates an electronic certificate as NFT (network File transfer function) evidence, and updates task role allocation; all changes are fed back through the blockchain event log, so that synchronous consensus of certificate issuing and role adjustment is ensured, and the system is delay-free and cannot be tampered; The event feedback embedding layer processes event chains in early warning information through a position coding function, the embedding dimension of each event node is matched with the notification attribute, and the adaptive transducer coding unit comprises a three-layer structure: The first layer is a multi-head self-attention layer and is used for capturing the dependency relationship among events, and calculating the weighted influence of nodes through dot product scaling and position offset of query-key-value vectors; The second layer is a normalized layer to be used for preventing gradient explosion; The third layer is a feedforward network layer, which is used for nonlinear projection and optimizing a final decision vector.
2. The blockchain-based technician training and project dynamic authorization engine method of claim 1, wherein S1 further comprises: After the training data chain is formed, the attention mechanism, the attention score output by the LSTM model and the LSTM state updating result are combined, a hash chain is generated through SHA-256, and the hash chain is stored through an intelligent contract consensus mechanism, so that a tamper-proof on-chain path is formed.

Description

Technician training and project dynamic authorization engine method based on blockchain Technical Field The invention belongs to the field of blockchains, and particularly relates to a blockchain-based technician training and project dynamic authorization engine method. Background With the rapid development of information technology and intelligence, technician training and project authorization management have become critical issues facing many businesses and institutions. The existing technician training system is generally based on a traditional offline or online training platform, adopts a fixed course system and task allocation mode, lacks flexibility and individuation, and is difficult to accurately learn path planning according to actual requirements and capabilities of technicians. The traditional project authorization management system is also usually static, and once the authority allocation is set, dynamic adjustment is difficult, so that a technician cannot acquire tasks and project resources matched with the capabilities of the technician in real time, the work efficiency of the technician is affected, and the overall execution effect of the project is reduced. In addition, the existing system has certain defects in the aspects of data security, authority management and task traceability, most of data records and task authorizations lack transparency and traceability, and are easy to be manually interfered and tampered, so that the trust and the fairness of the system are affected. Blockchain technology is gradually applied to the fields of data management and authority control due to the characteristics of decentralization, non-tampering and transparency traceability, but the practical application of the blockchain technology in technician training and project authorization management still has a plurality of challenges. The existing blockchain technology is applied to the data recording and transaction verification level mostly, the difficult problem between dynamic authorization and personalized training path design of technicians is not solved, and the existing intelligent contract and decentralization characteristics are not fully utilized to realize real-time and intelligent training and authorization decision. Therefore, how to combine the emerging technology, especially the combination of blockchain and artificial intelligence, to achieve the intelligence and individualization of technician training and dynamically adjust project authorization becomes a problem to be solved urgently. Disclosure of Invention The invention aims to provide a technician training and project dynamic authorization engine method based on a blockchain, which not only solves the defects of the traditional system in terms of authority management, training customization and dynamic adjustment, but also greatly improves the flexibility, efficiency and transparency of the system, and can meet the requirements of modern enterprises on high-efficiency, accurate and traceable management. To achieve the above object, in a first aspect of the present invention, there is provided a blockchain-based technician training and project dynamic authorization engine method, the method comprising the steps of: S1, acquiring an original training event sequence of a technician, processing dynamic interference with an LSTM model through an attention mechanism, generating a refining vector, uploading the refining vector to a blockchain storage certificate after the associated technician is identified, forming a training data chain, and outputting an assessment result; S2, constructing a decision vector according to the training data chain, calculating an authorization adaptation score through an intelligent contract, generating an authorization token when the authorization adaptation score reaches a threshold value, updating access rights and uploading a chain certificate to obtain a rights ID and recording a change log chain; S3, monitoring a permission change log in real time, evaluating risks through a permission freezing decision model, calculating freezing adaptation scores, triggering intelligent contract freezing permission when the freezing adaptation scores exceed a threshold value, generating early warning notification and a uplink record, and obtaining a risk ID for identifying a freezing instance; s4, constructing a feedback vector based on the authority freezing record, calculating an optimization adaptation score through a certificate optimization decision model, issuing a certificate unique mark as NFT evidence by an intelligent contract after reaching standards, and synchronously adjusting the task role of a technician and uploading the evidence, wherein the feedback vector is a risk ID and an early warning notice. Further, the original training event sequence comprises a unique identification of a technician, a course module interaction intensity vector sequence and an assessment response record; The dynamic interfere