Device predictive maintenance system and method based on block chain network

Abstract

The invention discloses an equipment predictive maintenance system and method based on a block chain network, and belongs to the technical field of production equipment maintenance, and the method comprises the steps: collecting equipment operation data in real time through a sensor deployed on production equipment, and transmitting the data to a block chain data storage library composed of distributed nodes after preprocessing; the storage library comprises a data acquisition node, a model management node and a maintenance execution node, and equipment operation parameters and maintenance information are isolated and stored through an independent channel; based on historical data, constructing an equipment fault model, predicting a potential fault type and probability of the equipment, and transmitting the model and a prediction result to a block chain data storage library; generating a corresponding maintenance plan according to the prediction result, and automatically distributing the maintenance plan to a maintenance execution node; and after the maintenance is completed, the log is recorded to the block chain data storage library and is bound with the original plan, so that the traceability of the whole process is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production equipment maintenance, and in particular to an equipment predictive maintenance system and method based on a blockchain network. BACKGROUND

[0002] With the continuous progress of intelligent manufacturing technology, equipment safety has become an important guarantee for enterprise production. Industrial artificial intelligence, as a key technology for optimizing industrial production methods and decision-making modes, is helping enterprises improve product quality and service levels by deeply integrating artificial intelligence and manufacturing technology. However, as the core of industrial production, machine equipment inevitably experiences performance degradation and health state decline during long-term operation. Coupled with the increasing number of equipment components and the increasing complexity of operating environments, maintenance work has become increasingly difficult.

[0003] In the single-order design production mode, there are various types of equipment, and the production line needs to be adjusted frequently according to orders. Such adjustments not only exacerbate the complexity of equipment combinations, but also lead to diversification of processed products. For single-order design production enterprises, to match customer demand, production line equipment needs to be dynamically adjusted. To cope with the multi-project driven production mode, enterprises widely use milling machines, punch machines, bending machines, cutting machines and other equipment, which further increases the difficulty of maintenance work. In this context, the predictive maintenance solution has become a key way to solve complex maintenance problems, improve production efficiency and ensure safe production, as it can predict equipment failures in advance and optimize maintenance plans.

[0004] However, the data-driven predictive maintenance solution highly depends on Internet of Things technology to realize real-time collection and transmission of equipment data. However, the application of Internet of Things technology exposes two major problems: first, the lack of decentralization capability leads to low data interaction efficiency, making it difficult to support large-scale and high-concurrency equipment monitoring needs; second, the risk of privacy leakage seriously threatens enterprise data security and customer privacy. These problems not only limit the application effect of Internet of Things technology in predictive maintenance, but also restrict the overall improvement of intelligent manufacturing efficiency. SUMMARY

[0005] The present application aims to provide an equipment predictive maintenance system and method based on a blockchain network to solve the problems raised in the background.

[0006] To solve the above technical problems, the present application provides the following technical solution: an equipment predictive maintenance method based on a blockchain network, comprising: Step S100: Real-time collection of pre-set type of equipment operation data through physical sensors deployed on single-order design production equipment, and pre-processing of collected raw data to generate standardized equipment data sets; Step S200: Construct a blockchain data repository composed of multiple distributed nodes, including data collection nodes, model management nodes, and maintenance execution nodes; write the preprocessed standardized equipment data set to the first distributed storage unit through the data collection node; the first distributed storage unit is isolated through an independent communication channel and is used to store equipment operating parameters and historical data sets; Step S300: The model management node reads the historical equipment data set from the first distributed storage unit, constructs and trains the equipment fault prediction model; the trained fault prediction model is uploaded to the first distributed storage unit through the model management node, and the model calling permission management is triggered through the smart contract; after inputting the real-time equipment data into the fault prediction model, the equipment fault prediction result is outputted; Step S400: Based on the fault prediction result, a maintenance plan is generated and written to the second distributed storage unit through the model management node; the second distributed storage unit is isolated through another independent communication channel and only allows the maintenance execution node to access, and is used to store maintenance plans and maintenance logs; the maintenance execution node real-time monitors the second distributed storage unit, executes maintenance operations after obtaining the newly published maintenance plan; after the maintenance is completed, the maintenance log is recorded to the second distributed storage unit, and is associated with the original maintenance plan through hash binding to form a closed-loop traceability chain.

[0007] Further, in step S100, the physical sensors deployed on the single-design production equipment are used to real-time collect and monitor the preset type of equipment operating data, the collected raw data is preprocessed, the median filtering algorithm is used to remove transient noise, the linear interpolation method is used to supplement the missing data segment caused by temporary sensor failure, all data is normalized and mapped to the [0, 1] interval; a timestamp, a unique equipment identifier, and a data source label are added to each data to generate a standardized equipment data set.

[0008] Further, step S200 includes: Step S201: Deploy three types of functional nodes including data collection nodes, model management nodes, and maintenance execution nodes in the blockchain network to form a blockchain data repository; wherein the data collection node is deployed on the workshop department server and is used to real-time collect and transmit equipment operating data; the model management node is deployed on the management department server and is used to construct and train the equipment fault prediction model and generate the maintenance plan; the maintenance execution node is deployed on the maintenance department terminal device and is used to receive and execute the maintenance plan and record the maintenance log; each node is configured with a unique digital identity certificate, and the certificate issuance and permission allocation are performed through the blockchain network; The permission assignment setting includes that the data collection node only has the permission to write the equipment operation data to the first distributed storage unit; the model management node has the permission to read the data from the first distributed storage unit, write the fault prediction model, and generate the maintenance plan; and the maintenance execution node only has the permission to read the maintenance plan from the second distributed storage unit and write the maintenance log. In step S203, three smart contracts are preset in the blockchain network, including a data uploading smart contract, a data query smart contract, and a model calling smart contract. The rules and processes of data interaction between nodes are defined through the smart contracts. The data uploading smart contract is used to process the process of uploading the equipment operation data by the data collection node to the blockchain network. The data query smart contract is used to process the query request for the stored data on the blockchain between nodes. The model calling smart contract is used to process the process of calling the fault prediction model by the model management node to perform equipment fault prediction. The data isolation between nodes is realized through independent communication channels. The data collection node and the model management node are connected through a first communication channel, and the model management node and the maintenance execution node are connected through a second communication channel.

[0009] In step S204, after the data collection node completes the preprocessing of the equipment operation data, the preset data uploading smart contract is automatically called. The smart contract analyzes the digital identity certificate of the data collection node, extracts the certificate signature, validity period, and permission range. The certificate signature is verified through the blockchain network. If the certificate is expired or the signature is invalid, the uploading process is terminated and an error code is returned. The data uploading smart contract compares the node identifier with the preset permission list, and only allows the data collection node to initiate the uploading request, and prohibits other nodes from directly writing data. In step S205, the SHA-256 algorithm is used to generate a unique hash value for the preprocessed standardized data set. The smart contract queries the hash index table stored in the blockchain data storage. If the hash value to be uploaded already exists, it is determined as duplicate data and the storage is rejected, and a duplicate event log is recorded. If the hash value is unique, the data is encrypted using an asymmetric encryption algorithm, and a data collection timestamp, a device unique identifier, and a sensor type label are added to form a standardized data block. The encrypted data block is written to the first distributed storage unit through the first communication channel, and the consensus mechanism of the blockchain network is triggered to propagate the data block to all participating nodes. Each node independently verifies the consistency of the data block hash value and checks whether the signature of the data collection node is valid, and then updates the blockchain data storage. The first distributed storage unit limits the operation of unauthorized nodes through an access control list, and only allows the model management node to call the historical data for model training in the subsequent process.

[0010] Further, step S300 includes: Step S301: The model management node calls a preset data query smart contract to extract a historical data set of the target device from a first distributed storage unit, the historical data set including device operating parameters and recorded historical device failure types; feature extraction is performed from the historical data set, the correlation between features is calculated, and correlation threshold R1 and correlation threshold R2 are set; the Pearson correlation coefficient algorithm is used to preferentially locate a feature group with a correlation lower than threshold R1 to construct an initial low-correlation feature set; the correlation threshold R1 is calculated by calculating the Pearson correlation coefficient of all feature pairs in the historical data set, and the M quantile of the absolute value distribution is taken as the value of R1; for feature pairs in the initial low-correlation feature set, a distance correlation coefficient is used to calculate a nonlinear correlation degree, and a dynamic threshold R2 is set; the correlation threshold R2 is based on the distance correlation coefficient distribution of all feature pairs in the historical data set, and the N quantile is selected as R2; wherein M and N are quantile parameters of the Pearson correlation coefficient and the distance correlation coefficient, respectively; feature pairs with a correlation lower than R2 are screened out to construct a low-correlation feature set, low-correlation feature groups in the low-correlation feature set are combined using an automatic feature generation algorithm, a new feature vector is generated, and the new feature vector is updated in the data set; based on the updated data set, in combination with the device failure types recorded in the historical data set, a random forest algorithm is used to train a device failure prediction model, and the model performance is optimized through cross-validation; Step S302: The trained failure prediction model is stored in the first distributed storage unit after being encrypted, and a model version identifier and training metadata are bound; a preset model calling smart contract is triggered, and model calling rules are defined, including: only the model management node is authorized to initiate a model calling request after authentication and to perform integrity checking on input data, including data source verification and hash value matching; real-time device data is input into the failure prediction model after preprocessing, and a device failure type and a failure probability are output; the model output result is attached with a blockchain network signature, and the prediction result is written into the second distributed storage unit through the model management node.

[0011] Further, step S400 includes: Step S401: The model management node reads real-time device failure prediction results from the second distributed storage unit, dynamically generates a structured maintenance plan in combination with a preset maintenance strategy knowledge base, the plan including a failure type, a maintenance action, a spare parts list, and a safety risk level, and writes the plan into the second distributed storage unit through defined access permissions, allowing only authenticated maintenance execution nodes to read the maintenance plan, while binding the maintenance plan hash value with the original prediction result hash value to form a bidirectional traceability chain; after the maintenance plan is generated, a real-time alarm is triggered and pushed to the maintenance execution node; Step S402: the maintenance execution node obtains the authorized maintenance plan and performs the maintenance operation, including spare parts delivery, on-site maintenance and work time recording, generates a structured log after the maintenance is completed, the log contains actual replacement spare parts information, operation steps and equipment parameter changes; the log is written into the second distributed storage unit through the maintenance execution node, and the log hash value is bound with the corresponding maintenance plan hash value to build a closed-loop traceability chain, if the log is detected to be inconsistent with the maintenance plan, an audit event is automatically generated and pushed to the management department, a new maintenance plan is formulated by the management personnel and stored in the second distributed storage unit, and the new maintenance plan is synchronized to all network nodes through the block chain consensus mechanism to form an abnormal handling closed loop.

[0012] A device predictive maintenance system based on a block chain network, comprising a data acquisition module, a block chain storage module, a device fault prediction module and a maintenance scheduling module. The data acquisition module collects real-time device operation data of a preset type through physical sensors deployed on the target single-design production equipment, and pre-processes the collected raw data to generate a standardized device data set. The block chain storage module constructs a block chain data storage library composed of a plurality of distributed nodes, including a data acquisition node, a model management node and a maintenance execution node; the pre-processed standardized device data set is written into a first distributed storage unit through the data acquisition node; the first distributed storage unit is isolated through an independent communication channel and is used to store device operation parameters and historical data sets. The device fault prediction module reads the historical device data set from the first distributed storage unit through the model management node, constructs and trains a device fault prediction model; the trained fault prediction model is uploaded to the first distributed storage unit through the model management node, and triggers the intelligent contract to realize the model calling permission management; after the real-time device data is input into the fault prediction model, the device fault prediction result is output. The maintenance scheduling module generates a maintenance plan based on the fault prediction result and writes it into a second distributed storage unit through the model management node; the second distributed storage unit is isolated through another independent communication channel and only allows the maintenance execution node to access, and is used to store the maintenance plan and the maintenance log; the maintenance execution node real-time monitors the second distributed storage unit, obtains the newly published maintenance plan and performs the maintenance operation; after the maintenance is completed, the maintenance log is recorded to the second distributed storage unit, and the original maintenance plan is associated through hash binding to form a closed-loop traceability chain.

[0013] Compared with the prior art, the beneficial effects achieved by the present application are: The application adopts physical sensors to collect equipment operation data in real time, and generates a standardized equipment data set through preprocessing to ensure the integrity and consistency of the data; a multi-node distributed storage library is established through blockchain technology to realize the safe storage and circulation of data; meanwhile, the equipment fault prediction model is managed through a blockchain smart contract to ensure the efficiency and transparency of data interaction, and the data access and operation of each node are strictly controlled, greatly improving the reliability of data processing; The application realizes accurate prediction of equipment failure by combining sensor data with historical failure data to build and train an equipment failure prediction model, and generates a maintenance plan based on the prediction result, greatly improving the maintenance efficiency and accuracy and reducing equipment downtime and production loss; The application can perform maintenance operations after obtaining the authorized maintenance plan by setting the maintenance execution node, and records detailed maintenance logs, which are bound to the original maintenance plan through blockchain technology to realize full traceability and ensure the transparency and auditability of the maintenance process; if an abnormality occurs, an audit event is automatically generated and a processing mechanism is triggered to ensure the efficiency and compliance of equipment maintenance, thereby improving the operation reliability and production efficiency of the equipment. BRIEF DESCRIPTION OF DRAWINGS

[0014] The accompanying drawings are used to provide a further understanding of the application, and constitute a part of the specification, together with the embodiments of the application, to explain the application, and do not constitute a limitation on the application. In the drawings: Fig. 1 It is a technical architecture diagram of a device predictive maintenance method based on a blockchain network; Fig. 2 It is a method flowchart of a device predictive maintenance method based on a blockchain network. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0016] Please refer to Figs. 1-2 The application provides a technical solution: a device predictive maintenance method based on a blockchain network, comprising: Step S100: through physical sensors deployed on single-design production equipment, real-time collection of preset types of equipment operation data, and preprocessing of collected raw data to generate a standardized equipment data set; Step S200: Construct a blockchain data repository composed of multiple distributed nodes, including data collection nodes, model management nodes, and maintenance execution nodes; write the preprocessed standardized equipment data set to the first distributed storage unit through the data collection node; the first distributed storage unit is isolated through an independent communication channel and is used to store equipment operating parameters and historical data sets; Step S300: The model management node reads the historical equipment data set from the first distributed storage unit, constructs and trains the equipment fault prediction model; the trained fault prediction model is uploaded to the first distributed storage unit through the model management node, and the model calling permission management is triggered through the smart contract; after inputting the real-time equipment data into the fault prediction model, the equipment fault prediction result is outputted; Step S400: Based on the fault prediction result, a maintenance plan is generated and written to the second distributed storage unit through the model management node; the second distributed storage unit is isolated through another independent communication channel and only allows the maintenance execution node to access, and is used to store the maintenance plan and the maintenance log; the maintenance execution node real-time monitors the second distributed storage unit, executes the maintenance operation after obtaining the newly published maintenance plan; after the maintenance is completed, the maintenance log is recorded to the second distributed storage unit and is associated with the original maintenance plan through hash binding to form a closed-loop traceability chain.

[0017] In step S100, the preset type of equipment operating data is collected and monitored in real time through the physical sensor deployed on the single-design production equipment, the raw data collected is preprocessed, the median filtering algorithm is used to remove transient noise, the linear interpolation method is used to supplement the missing data segment caused by temporary failure of the sensor, all data is normalized and mapped to the [0, 1] interval; a timestamp, a unique equipment identifier and a data source label are added to each data to generate a standardized equipment data set.

[0018] Step S200 includes: Step S201: Deploy three types of functional nodes including data collection nodes, model management nodes and maintenance execution nodes in the blockchain network to form a blockchain data repository; wherein the data collection node is deployed on the workshop department server and is used to collect and transmit equipment operating data in real time; the model management node is deployed on the management department server and is used to construct and train the equipment fault prediction model and generate the maintenance plan; the maintenance execution node is deployed on the maintenance department terminal equipment and is used to receive and execute the maintenance plan and record the maintenance log; each node is configured with a unique digital identity certificate and the certificate issuance and permission allocation are performed through the blockchain network; The permission assignment set includes that the data collection node only has the permission to write the equipment operation data to the first distributed storage unit; the model management node has the permission to read the data from the first distributed storage unit, write the fault prediction model, and generate the maintenance plan; and the maintenance execution node only has the permission to read the maintenance plan from the second distributed storage unit and write the maintenance log. In the blockchain network, three kinds of smart contracts are preset, including a data uploading smart contract, a data query smart contract, and a model calling smart contract. The rules and processes of data interaction between nodes are defined through the smart contracts. The data uploading smart contract is used to process the process of uploading the equipment operation data by the data collection node to the blockchain network. The data query smart contract is used to process the query request for the stored data on the blockchain between nodes. The model calling smart contract is used to process the process of calling the fault prediction model by the model management node to perform equipment fault prediction. The data isolation between nodes is realized through independent communication channels. The data collection node and the model management node are connected through a first communication channel, and the model management node and the maintenance execution node are connected through a second communication channel.

[0019] After the data collection node completes the preprocessing of the equipment operation data, the preset data uploading smart contract is automatically called. The smart contract analyzes the digital identity certificate of the data collection node, extracts the certificate signature, validity period, and permission range. The certificate signature is verified through the blockchain network. If the certificate is expired or the signature is invalid, the uploading process is terminated and an error code is returned. The data uploading smart contract compares the node identifier with the preset permission list, and only allows the data collection node to initiate the uploading request, and prohibits other nodes from directly writing data. The SHA-256 algorithm is used to generate a unique hash value for the preprocessed standardized data set. The smart contract queries the hash index table stored in the blockchain data storage. If the hash value to be uploaded already exists, it is determined as duplicate data and storage is refused, and a duplicate event log is recorded. If the hash value is unique, the data is encrypted using an asymmetric encryption algorithm, and a data collection timestamp, a device unique identifier, and a sensor type label are added to form a standardized data block. The encrypted data block is written into the first distributed storage unit through the first communication channel, and the consensus mechanism of the blockchain network is triggered to propagate the data block to all participating nodes. Each node independently verifies the consistency of the data block hash value and checks whether the signature of the data collection node is valid, and then updates the blockchain data storage. The first distributed storage unit limits the operation of unauthorized nodes through an access control list, and only allows the model management node to call the historical data for model training in the subsequent process.

[0020] Step S300 includes: Step S301: The model management node calls a preset data query smart contract to extract a historical data set of the target device from a first distributed storage unit, the historical data set including device operating parameters and recorded historical device failure types; feature extraction is performed from the historical data set, the correlation between features is calculated, and correlation thresholds R1 and R2 are set; the Pearson correlation coefficient algorithm is used to preferentially locate feature groups with a correlation lower than the threshold R1, and an initial low-correlation feature set is constructed; the correlation threshold R1 is calculated by calculating the Pearson correlation coefficient of all feature pairs in the historical data set, and the M quantile of the absolute value distribution is taken as the value of R1; for feature pairs in the initial low-correlation feature set, a distance correlation coefficient is used to calculate a nonlinear correlation degree, and a dynamic threshold R2 is set; the correlation threshold R2 is based on the distance correlation coefficient distribution of all feature pairs in the historical data set, and the N quantile is selected as R2; wherein M and N are quantile parameters of the Pearson correlation coefficient and the distance correlation coefficient respectively; feature pairs with a correlation lower than R2 are screened out to construct a low-correlation feature set, low-correlation feature groups in the low-correlation feature set are combined using an automatic feature generation algorithm, a new feature vector is generated, and the new feature vector is updated in the data set; based on the updated data set, in combination with the device failure types recorded in the historical data set, a random forest algorithm is used to train a device failure prediction model, and the model performance is optimized through cross-validation; Step S302: The trained failure prediction model is stored in the first distributed storage unit after being encrypted, and a model version identifier and training metadata are bound; a preset model calling smart contract is triggered, and model calling rules are defined, including: only the model management node is authorized to initiate a model calling request after authentication and to perform integrity checking on input data, including data source verification and hash value matching; real-time device data is input into the failure prediction model after preprocessing, and device failure types and failure probabilities are output; the model output result is attached with a blockchain network signature, and the prediction result is written into the second distributed storage unit through the model management node.

[0021] Step S400 includes: Step S401: The model management node reads real-time device failure prediction results from the second distributed storage unit, dynamically generates a structured maintenance plan in combination with a preset maintenance strategy knowledge base, the plan including failure types, maintenance actions, spare parts lists, and safety risk levels, and writes the plan into the second distributed storage unit through defined access permissions, allowing only authenticated maintenance execution nodes to read the maintenance plan, while binding the maintenance plan hash value with the original prediction result hash value to form a two-way traceability chain; the maintenance plan triggers a real-time alarm and is pushed to the maintenance execution node after being generated; Step S402: the maintenance execution node obtains the authorized maintenance plan and then performs the maintenance operation, including spare parts delivery, on-site maintenance and work time recording, generates a structured log after the maintenance is completed, including actual replacement of spare parts information, operation steps and device parameter changes; the log is written into the second distributed storage unit by the maintenance execution node, and the log hash value is bound with the corresponding maintenance plan hash value to build a closed-loop traceability chain; if the log is detected to be inconsistent with the maintenance plan, an audit event is automatically generated and pushed to the management department, and a new maintenance plan is formulated by the management personnel and stored in the second distributed storage unit, and is synchronized to all network nodes through the block chain consensus mechanism to form an abnormal handling closed loop.

[0022] Taking a single-order design type numerical control machine tool as an example, each device is equipped with a vibration sensor, a temperature sensor and a current monitoring module, and key parameters such as spindle speed, bearing temperature and cutting force are collected in real time; the median filtering algorithm is applied to eliminate the instantaneous noise interference caused by the splashing of cutting fluid, and the median values of three consecutive sampling points are used to replace the abnormal values; for the 10-second data gap generated during the periodic calibration of the sensor, a transition value is generated based on the front and rear effective data segments by using the linear interpolation method; then the vibration amplitude, temperature value and other parameters are uniformly mapped to the [0, 1] interval through Min-Max standardization; each data packet is automatically attached with a time stamp, a device serial number and a sensor type code, and finally a standardized data set is formed; The model management node triggers the training process every week: extracts the historical data set of the last 6 months from the first storage unit, calculates the correlation between the features, and there is a low correlation of 0.12 between the spindle vibration signal and the X-axis current; an automatic feature generation algorithm is used to construct a "vibration energy-current change rate" composite feature; when training the model using the random forest algorithm, 200 decision trees and 5-fold cross-validation are set, and finally a prediction model with an F1 value of 0.92 is obtained; the trained model is stored in the first distributed storage unit, and the version number and training data hash are bound; the model calls the smart contract for verification; the on-site device data is input into the model after real-time preprocessing, and the output fault type is: spindle bearing wear and fault probability 85%; the prediction result is written into the second distributed storage unit by the model management node, triggering the generation and execution of the maintenance plan, and forming a complete closed-loop maintenance process.

[0023] It is obvious to a person skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be considered as limiting the claims involved.

Claims

1.A blockchain network-based device predictive maintenance method, characterized by comprising the steps of: The method comprises: Step S100: collecting preset type of equipment operation data in real time through physical sensors deployed on single design type production equipment, and pre-processing the collected raw data to generate a standardized equipment data set; Step S200: constructing a blockchain data repository composed of multiple distributed nodes, including a data collection node, a model management node, and a maintenance execution node; writing the pre-processed standardized equipment data set to a first distributed storage unit through the data collection node; the first distributed storage unit is isolated through an independent communication channel and is used to store equipment operation parameters and historical data sets; Step S300: the model management node reads the historical equipment data set from the first distributed storage unit, constructs and trains an equipment fault prediction model; the trained fault prediction model is uploaded to the first distributed storage unit through the model management node, and triggers the smart contract to realize model calling permission management; after inputting real-time equipment data into the fault prediction model, the equipment fault prediction result is outputted; Step S400: generating a maintenance plan based on the fault prediction result and writing it to a second distributed storage unit through the model management node; the second distributed storage unit is isolated through another independent communication channel and only allows the maintenance execution node to access, and is used to store maintenance plans and maintenance logs; the maintenance execution node monitors the second distributed storage unit in real time, executes maintenance operations after obtaining the newly published maintenance plan; after the maintenance is completed, the maintenance log is recorded to the second distributed storage unit and is associated with the original maintenance plan through hash binding to form a closed-loop traceability chain. 2.The device predictive maintenance method based on a blockchain network according to claim 1, wherein: The step S100 collects and monitors preset type of equipment operation data in real time through physical sensors deployed on single design type production equipment, pre-processes the collected raw data, removes transient noise by using a median filter algorithm, supplements missing data segments caused by temporary sensor failure by using a linear interpolation method, normalizes all data, and respectively maps them to the [0, 1] interval; a timestamp, a unique equipment identifier, and a data source label are added to each piece of data to generate a standardized equipment data set. 3.The device predictive maintenance method based on a blockchain network according to claim 1, characterized in that: The step S200 comprises: Step S201: deploying three types of functional nodes including a data collection node, a model management node, and a maintenance execution node in a blockchain network to form a blockchain data repository; the data collection node is deployed on a workshop department server and is used to collect and transmit equipment operation data in real time; the model management node is deployed on a management department server and is used to construct and train an equipment fault prediction model and generate a maintenance plan; the maintenance execution node is deployed on a maintenance department terminal device and is used to receive and execute the maintenance plan and record the maintenance log; each node is configured with a unique digital identity certificate, and the certificate issuance and permission allocation are performed through the blockchain network; The permission assignment setting includes that the data collection node only has the permission to write the equipment operation data to the first distributed storage unit; the model management node has the permission to read the data from the first distributed storage unit, write the fault prediction model, and generate the maintenance plan; and the maintenance execution node only has the permission to read the maintenance plan from the second distributed storage unit and write the maintenance log. The step S203 comprises: presetting three smart contracts in the blockchain network, including a data uploading smart contract, a data query smart contract, and a model calling smart contract; defining the rules and processes of data interaction between nodes through the smart contracts; the data uploading smart contract is used for processing the process of uploading the equipment operation data by the data collection node to the blockchain network; the data query smart contract is used for processing the query request for the stored data on the blockchain between nodes; and the model calling smart contract is used for processing the process of calling the fault prediction model by the model management node to perform equipment fault prediction; the nodes are connected through independent communication channels to realize data isolation, wherein the data collection node and the model management node are connected through a first communication channel, and the model management node and the maintenance execution node are connected through a second communication channel. 4.The device predictive maintenance method based on a blockchain network according to claim 3, characterized in that: The step S200 further comprises: The step S204 comprises: after the data collection node completes the preprocessing of the equipment operation data, automatically calling the preset data uploading smart contract, and the smart contract analyzing the digital identity certificate of the data collection node, extracting the certificate signature, validity period, and permission range; verifying whether the certificate signature is valid through the blockchain network, and if the certificate is expired or the signature is invalid, terminating the uploading process and returning an error code; the data uploading smart contract comparing the node identifier with the preset permission list, and only allowing the data collection node to initiate the uploading request and prohibiting other nodes from directly writing data; The step S205 comprises: generating a unique hash value for the preprocessed standardized data set by using the SHA-256 algorithm, querying the hash index table stored in the blockchain data storage by the smart contract, if the to-be-uploaded hash value already exists, determining that it is duplicate data and refusing to store, and recording a duplicate event log; if the hash value is unique, using an asymmetric encryption algorithm to encrypt the data, adding a data collection timestamp, a device unique identifier, and a sensor type label, and forming a standardized data block; writing the encrypted data block to the first distributed storage unit through the first communication channel, triggering the consensus mechanism of the blockchain network, and propagating the data block to all participating nodes; independently verifying the hash value consistency of the data block by each node and checking whether the signature of the data collection node is valid, and then updating the blockchain data storage; and the first distributed storage unit limiting the operation of unauthorized nodes through an access control list, and only allowing the model management node to call the historical data for model training in the subsequent process. 5.The device predictive maintenance method based on a blockchain network according to claim 1, characterized in that: The step S300 comprises: Step S301: The model management node calls a preset data query smart contract to extract a historical data set of the target device from a first distributed storage unit, the historical data set including device operating parameters and recorded historical device failure types; feature extraction is performed from the historical data set, the correlation between the features is calculated, and a correlation threshold R1 and a correlation threshold R2 are set; the Pearson correlation coefficient algorithm is used to preferentially locate a feature group with a correlation lower than the threshold R1, and an initial low-correlation feature set is constructed; the correlation threshold R1 is calculated by calculating the Pearson correlation coefficient of all feature pairs in the historical data set, and the M quantile of the absolute value distribution is taken as the value of R1; for the feature pairs in the initial low-correlation feature set, a distance correlation coefficient is used to calculate a nonlinear correlation degree, and a dynamic threshold R2 is set; the correlation threshold R2 is based on the distance correlation coefficient distribution of all feature pairs in the historical data set, and the N quantile is selected as R2; wherein M and N are quantile parameters of the Pearson correlation coefficient and the distance correlation coefficient, respectively; feature pairs with a correlation lower than R2 are selected to construct a low-correlation feature set, and an automatic feature generation algorithm is used to perform a preset logical combination on the low-correlation feature groups in the low-correlation feature set to generate a new feature vector and update it to the data set; based on the updated data set, in combination with the device failure types recorded in the historical data set, a random forest algorithm is used to train a device failure prediction model, and the model performance is optimized through cross-validation; Step S302: The trained failure prediction model is stored in the first distributed storage unit after being encrypted, and a model version identifier and training metadata are bound; a preset model calling smart contract is triggered, and model calling rules are defined, including: only the model management node is authorized to initiate a model calling request after authentication and to perform integrity checking on input data, including data source verification and hash value matching; real-time device data is input into the failure prediction model after preprocessing, and the device failure type and failure probability are output; the model output result is attached with a blockchain network signature, and the prediction result is written into the second distributed storage unit through the model management node. 6.The device predictive maintenance method based on a blockchain network according to claim 1, wherein: The step S400 includes: Step S401: The model management node reads the real-time device failure prediction result from the second distributed storage unit, dynamically generates a structured maintenance plan in combination with a preset maintenance strategy knowledge base, the plan including failure type, maintenance action, spare parts list, and safety risk level, and writes the plan into the second distributed storage unit through defined access permissions, allowing only authenticated maintenance execution nodes to read the maintenance plan, while binding the maintenance plan hash value with the original prediction result hash value to form a two-way traceability chain, and triggering a real-time alarm and pushing the alarm to the maintenance execution node after the maintenance plan is generated; Step S402: The maintenance execution node obtains the authorized maintenance plan and performs the maintenance operation, including spare parts delivery, on-site repair, and work time record. After the maintenance is completed, a structured log is generated, including actual replacement of spare parts information, operation steps, and device parameter changes. The log is written into the second distributed storage unit by the maintenance execution node, and the log hash value is bound with the corresponding maintenance plan hash value to construct a closed-loop traceability chain. If the log is detected to be inconsistent with the maintenance plan, an audit event is automatically generated and pushed to the management department, and a new maintenance plan is formulated by the management personnel and stored in the second distributed storage unit, which is synchronized to all network nodes through the blockchain consensus mechanism to form an abnormal handling closed loop. 7.A blockchain network based device predictive maintenance system, characterized in that: The system comprises a data acquisition module, a blockchain storage module, a device fault prediction module, and a maintenance scheduling module; The data acquisition module acquires real-time device operation data of a preset type through physical sensors deployed on the target single-design production equipment, and pre-processes the acquired raw data to generate a standardized device data set; The blockchain storage module constructs a blockchain data storage library composed of multiple distributed nodes, including a data acquisition node, a model management node, and a maintenance execution node; The pre-processed standardized device data set is written into the first distributed storage unit through the data acquisition node; The first distributed storage unit is isolated through an independent communication channel and is used to store device operation parameters and historical data sets; The device fault prediction module reads the historical device data set from the first distributed storage unit through the model management node, constructs and trains a device fault prediction model; The trained fault prediction model is uploaded to the first distributed storage unit through the model management node, and the model calling permission management is realized through the smart contract; after the real-time device data is input into the fault prediction model, the device fault prediction result is output; The maintenance scheduling module generates a maintenance plan based on the fault prediction result and writes it into the second distributed storage unit through the model management node; The second distributed storage unit is isolated through another independent communication channel and only allows the maintenance execution node to access, and is used to store the maintenance plan and the maintenance log; the maintenance execution node real-time monitors the second distributed storage unit and executes the maintenance operation after obtaining the newly released maintenance plan; After the maintenance is completed, the maintenance log is recorded in the second distributed storage unit, and the original maintenance plan is associated through hash binding to form a closed-loop traceability chain. 8.The device predictive maintenance method based on a blockchain network according to claim 7, characterized in that: The blockchain storage module comprises a node management unit and a smart contract unit; The node management unit is used to deploy the blockchain network nodes and define the operation permissions of the data acquisition node, the model management node, and the maintenance execution node; Data isolation and access control between nodes are realized through smart contracts; The smart contract unit integrates data upload, data query, and model calling smart contracts, defines data interaction rules, and isolates the first distributed storage unit and the second distributed storage unit through independent communication channels. 9.The device predictive maintenance method based on a blockchain network according to claim 7, characterized in that: The device fault prediction module comprises a feature extraction unit and a model construction unit; The feature extraction unit extracts a historical data set from a first distributed storage unit, screens a low-correlation feature group through feature heat map analysis, generates a derived feature, and updates the data set; The model construction unit trains a device fault prediction model based on a random forest algorithm, optimizes the model performance through cross-validation, and stores the encrypted model binding version identifier to the first distributed storage unit. 10.The device predictive maintenance method based on a blockchain network according to claim 7, wherein: The maintenance scheduling module includes a maintenance plan generation unit and a maintenance execution unit; The maintenance plan generation unit dynamically generates a structured maintenance plan by combining the prediction results with the preset maintenance strategy, writes into the second distributed storage unit by defining access permissions, and binds the prediction result hash value; The maintenance execution unit automatically checks whether the log and the maintenance plan are consistent by recording the maintenance operation log and associating the maintenance plan hash value, and triggers an audit event and generates a new plan when inconsistency is detected.

Images