Industrial equipment predictive maintenance method, device, storage medium, and program product

Abstract

The disclosed embodiments disclose a method, device, storage medium, and program product for predictive maintenance of industrial equipment. The method includes: each node trains a prediction model based on local data, generates local model parameters, and encrypts them using a public key to generate encrypted model parameters; the nodes are divided into core nodes and edge nodes, and the core nodes are divided into primary nodes and secondary nodes; the edge nodes send the encrypted model parameters to the corresponding core nodes for local aggregation, generate local aggregation parameters, and then send them to the primary node, which performs global aggregation to generate a global encrypted sum; the central server uses a private key to decrypt the verified global encrypted sum, obtains the global model parameters, and sends them to each node for updating the prediction model, and uses the updated prediction model to perform predictive maintenance on industrial equipment. This method can achieve distributed and efficient prediction while enhancing data security, optimize maintenance processes, reduce costs, and improve equipment operation efficiency.

Description

Technical Field

[0001] The present disclosure relates to the field of industrial Internet technology, and in particular to a method, device, storage medium, and program product for predictive maintenance of industrial equipment. Background Art

[0002] With the development of industry, intelligent manufacturing systems are increasing in number, and predictive maintenance of equipment has become a key technology for improving production efficiency and reducing costs. However, due to the need to protect data privacy and trade secrets, direct data sharing poses risks. Existing technical solutions, such as traditional centralized data processing and simple federated learning, cannot fully address the needs of data privacy protection and multi-party collaboration. Summary of the Invention

[0003] In view of this, the embodiments of the present disclosure provide a method, device, storage medium, and program product for predictive maintenance of industrial equipment, which can achieve distributed and efficient prediction, optimize maintenance processes, reduce costs, and improve equipment operating efficiency while enhancing data security.

[0004] In a first aspect, the present disclosure provides a method for predictive maintenance of industrial equipment, which employs the following technical solutions:

[0005] Generate a public key and a private key, send the public key to each node, and store the private key in a central server; each node represents an industrial device;

[0006] Each node trains a prediction model based on local data, generates local model parameters, and encrypts them using the public key to generate encrypted model parameters;

[0007] Divide multiple nodes into multiple core nodes and multiple edge nodes, and divide multiple core nodes into one master node and several secondary nodes; one core node is associated with several edge nodes, and one master node is associated with several secondary nodes;

[0008] The edge node sends the encryption model parameters to the corresponding core node, and the core node locally aggregates the local encryption model parameters and the received encryption model parameters to generate local aggregated parameters;

[0009] The secondary node sends the local aggregation parameter to the primary node, and the primary node globally aggregates the local local aggregation parameter and the received local aggregation parameter to generate a global encrypted sum;

[0010] After the global encryption and verification, the master node encrypts and sends the global encryption to the central server;

[0011] The central server uses a private key to decrypt the global encryption sum, obtains the global model parameters and sends them to each node for updating the prediction model until the global aggregation times reach a preset threshold or the performance of the prediction model reaches the convergence standard, and uses the prediction model to perform predictive maintenance on industrial equipment.

[0012] Optionally, the method includes training a prediction model based on local data and generating local model parameters, specifically including:

[0013] Preprocessing the local data;

[0014] Perform floating point formatting on pre-processed local data;

[0015] Use non-associative addition and non-associative multiplication to protect the privacy of local data after floating-point formatting;

[0016] The prediction model is trained based on the SGD function and the privacy-protected local data.

[0017] Generate local model parameters.

[0018] Optionally, the expression of the non-associative addition is:

[0019] a⊕b=(a+b)+R(a,b); and / or,

[0020] The expression of the non-associative multiplication is:

[0021] a⊙b=(a×b)+R(a×b);

[0022] Where R(·) is a random perturbation function; a and b represent the two data items involved in the operation.

[0023] Optionally, the expression of the non-associative addition is:

[0024] a⊕b=(a+b)+R(a,b); and / or,

[0025] The expression of the non-associative multiplication is:

[0026] a⊙b=(a×b)+R(a×b);

[0027] Where R(·) is a random perturbation function; a and b represent the two data items involved in the operation.

[0028] Optionally, the expression of the non-associative addition is:

[0029] a⊕b=(a+b)+R(a,b); and / or,

[0030] The expression of the non-associative multiplication is:

[0031] a⊙b=(a×b)+R(a×b);

[0032] Where R(·) is a random perturbation function; a and b represent the two data items involved in the operation.

[0033] Optionally, dividing the multiple core nodes into one primary node and several secondary nodes includes:

[0034] The primary node is selected from the plurality of core nodes using a rotation mechanism or a random mechanism, and the remaining core nodes are recorded as the secondary nodes.

[0035] Optionally, the global encryption and verification are performed, specifically including:

[0036] The master node encrypts the global data and sends it to each secondary node;

[0037] YY+242226P

[0038] Each secondary node encrypts and audits the global data respectively;

[0039] If the review is passed, the secondary node sends a first signal to the primary node;

[0040] If the review fails, the secondary node sends a second signal to the primary node;

[0041] The master node counts the proportion of the first signal based on the number of received first signals and the number of received second signals;

[0042] If the proportion is greater than the preset proportion threshold, it means that the global encryption has passed the verification;

[0043] If the proportion is not greater than the preset proportion threshold, it means that the global encryption has failed verification.

[0044] In a second aspect, the embodiments of the present disclosure further provide an industrial equipment predictive maintenance system, which employs the following technical solutions:

[0045] A key generation and distribution module, configured to generate public and private keys, send the public key to each node, and store the private key in a central server; wherein each node represents one industrial device;

[0046] A prediction model training module is used for each node to train a prediction model based on local data, generate local model parameters and encrypt them using the public key to generate encrypted model parameters;

[0047] A node division module is used to divide multiple nodes into multiple core nodes and multiple edge nodes, and to divide multiple core nodes into one main node and several secondary nodes; wherein one core node is associated with several edge nodes, and a main node is associated with several secondary nodes;

[0048] A local aggregation module, configured for the edge node to send the encryption model parameters to the corresponding core node, and the core node to locally aggregate the local encryption model parameters and the received encryption model parameters to generate local aggregated parameters;

[0049] A global aggregation module, configured for the secondary node to send the local aggregation parameter to the primary node, and the primary node to globally aggregate the local local aggregation parameter and the received local aggregation parameter to generate a global encrypted sum;

[0050] A global encryption and sending module, configured for, after the global encryption and verification are completed, the master node encrypts and sends the global data to the central server;

[0051] A global encryption and decryption module is used for the central server to decrypt the global encryption and decryption using a private key, obtain global model parameters and send them to each node, and update the prediction model until the global aggregation number reaches a preset number threshold or the performance of the prediction model reaches the convergence standard, and the prediction model is used to perform predictive maintenance on industrial equipment.

[0052] In a third aspect, the embodiments of the present disclosure further provide a computer device that adopts the following technical solution:

[0053] The computer device comprises:

[0054] at least one processor; and,

[0055] a memory communicatively connected to the at least one processor; wherein,

[0056] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform any of the above-mentioned methods for predictive maintenance of industrial equipment.

[0057] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute any of the above-mentioned industrial equipment predictive maintenance methods.

[0058] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, comprising a computer program / instruction, which implements the steps of any of the above methods when executed by a processor.

[0059] The predictive maintenance method for industrial equipment provided by the disclosed embodiments uses a public and private key encryption mechanism to ensure the security of local model parameters generated by each node during transmission, preventing the leakage of sensitive data and enhancing the system's privacy protection capabilities. Nodes are divided into core nodes and edge nodes, and core nodes are further divided into primary nodes and secondary nodes, enabling distributed processing. This structure effectively utilizes the computing resources of each node in the network and reduces the burden on the central server. Through local and global aggregation, model parameters can be effectively integrated while ensuring data security. This method can improve the model's generalization ability and prediction accuracy. By distributing computing tasks to each node, computing resources in the network can be more effectively utilized, reducing the computing pressure on the central server and improving the overall system's operating efficiency. By continuously updating and optimizing the predictive model, the maintenance needs of industrial equipment can be more accurately predicted, thereby achieving predictive maintenance, reducing unplanned downtime, and improving the operating efficiency and lifespan of equipment. Furthermore, it can reduce production interruptions and repair costs caused by equipment failures, thereby reducing overall maintenance costs, ensuring the continuous operation of the production line, and thus improving production efficiency and the company's economic benefits. By setting preset thresholds or performance convergence criteria, the performance of the prediction model can be monitored, ensuring that maintenance is performed when the model reaches optimal performance, thus avoiding unnecessary maintenance operations. The entire process is highly automated, reducing manual intervention. Through intelligent model updates and maintenance, the efficiency and accuracy of maintenance work are improved. This method also supports the dynamic addition and removal of nodes, offering good scalability and adaptability to industrial equipment networks of varying sizes.

[0060] The above description is only an overview of the technical solution of the present disclosure. In order to more clearly understand the technical means of the present disclosure, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present disclosure more obvious and easy to understand, the following is a preferred embodiment.

[0061] The detailed description is as follows with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0063] Figure 1 A schematic diagram of a process flow of a method for predictive maintenance of industrial equipment provided by an embodiment of the present disclosure;

[0064] Figure 2A flowchart of a local training method for a prediction model provided in an embodiment of the present disclosure;

[0065] Figure 3 A block diagram of the principle of the industrial equipment predictive maintenance system provided by the embodiment of the present disclosure;

[0066] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0067] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0068] It should be clear that the following embodiments of the present disclosure are described through specific concrete examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.

[0069] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on this disclosure, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0070] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The illustrations only show components related to the present disclosure and are not drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0071] Additionally, in the following description, specific details are provided to provide a thorough understanding of the examples. However, one skilled in the art will appreciate that the aspects described can be practiced without these specific details.

[0072] Reference Figure 1 The present disclosure provides a method for predictive maintenance of industrial equipment, comprising the following steps:

[0073] S1: Generate public and private keys, send the public key to each node, and store the private key in the central server.

[0074] Each node represents an industrial device.

[0075] S2: Each node trains a prediction model based on local data, generates local model parameters, and encrypts them using the public key to generate encrypted model parameters.

[0076] S3: Divide multiple nodes into multiple core nodes and multiple edge nodes, and divide multiple core nodes into one primary node and several secondary nodes.

[0077] Among them, one core node is associated with several edge nodes, and the main node is associated with several secondary nodes.

[0078] S4: The edge node sends the encryption model parameters to the corresponding core node. The core node locally aggregates the local encryption model parameters and the received encryption model parameters to generate local aggregated parameters.

[0079] S5: The secondary node sends the local aggregation parameters to the primary node, and the primary node globally aggregates the local local aggregation parameters and the received local aggregation parameters to generate a global encrypted sum.

[0080] S6: After global encryption and verification, the master node will encrypt and send the global data to the central server.

[0081] S7: The central server uses the private key to decrypt the global encryption and obtain the global model parameters and sends them to each node for updating the prediction model until the global aggregation times reach the preset threshold or the performance of the prediction model reaches the convergence standard, and the prediction model is used to perform predictive maintenance on industrial equipment.

[0082] The disclosed predictive maintenance method for industrial equipment uses a public and private key encryption mechanism to ensure the security of local model parameters generated by each node during transmission, preventing the leakage of sensitive data and enhancing the system's privacy protection capabilities. Nodes are divided into core nodes and edge nodes, and core nodes are further divided into primary nodes and secondary nodes, enabling distributed processing. This structure effectively utilizes the computing resources of each node in the network and reduces the burden on the central server. Through local and global aggregation, model parameters can be effectively integrated while ensuring data security. This method can improve the model's generalization ability and prediction accuracy. By distributing computing tasks to each node, computing resources in the network can be more effectively utilized, reducing the computing pressure on the central server and improving the overall system's operating efficiency. By continuously updating and optimizing the predictive model, the maintenance needs of industrial equipment can be more accurately predicted, thereby achieving predictive maintenance, reducing unplanned downtime, and improving the operating efficiency and lifespan of equipment. Furthermore, it can reduce production interruptions and repair costs caused by equipment failures, thereby reducing overall maintenance costs, ensuring the continuous operation of the production line, and thus improving production efficiency and the company's economic benefits. By setting preset thresholds or performance convergence criteria, the performance of the prediction model can be monitored, ensuring that maintenance is performed when the model reaches optimal performance, thus avoiding unnecessary maintenance operations. The entire process is highly automated, reducing manual intervention. Through intelligent model updates and maintenance, the efficiency and accuracy of maintenance work are improved. This method also supports the dynamic addition and removal of nodes, offering good scalability and adaptability to industrial equipment networks of varying sizes.

[0083] In summary, this system can enhance data security and protect data privacy while achieving distributed and efficient prediction, optimizing maintenance processes, reducing costs, and improving equipment operation efficiency.

[0084] In S1, a pair of keys, including a public key (pk) and a private key (sk), is generated using a suitable homomorphic encryption algorithm (e.g., RSA, ECC, BGV, or FV). The public key is distributed to each node in the system in a secure manner, such as by sending the public key over the network, and the security of the distribution process is ensured to prevent the key from being intercepted or tampered with during transmission. The private key is securely stored on a central server. The private key is the key in the decryption process, so it must be kept secure to prevent unauthorized access. The public key and private key can be generated by a designated management agency or by a central server. If the management agency is responsible for generating them, the management agency is responsible for distributing the public key and private key. If the central server is responsible for generating them, the central server distributes the public key to each node and saves the private key to ensure that only the central server can decrypt the subsequently obtained global encryption sum, thereby maintaining the security of the entire system and the privacy of the data.

[0085] During key distribution, verify node identity through digital signatures or certificate authentication to ensure that each node is trustworthy. Configure each node in the system to ensure that they use the same encryption parameters and protocol versions to ensure consistency in the encryption process. Implement a key management policy, including key update, rotation, and deprecation processes, to address potential security risks.

[0086] Implement batch processing technology to improve the efficiency of homomorphic operations. This means that when performing homomorphic encryption operations, multiple data points can be processed simultaneously, reducing the time and computing resources required for encryption and decryption. Design a layered network topology to improve system scalability. The core layer handles more complex tasks, while the edge layer is responsible for simpler tasks closer to the data source. This ensures that the system maintains efficient operation as nodes are added. Allow nodes to dynamically join or exit the network based on system needs, adapting to changes in the number of nodes, increasing system flexibility and adapting to changing workloads. Conduct thorough testing and verification before the system is officially launched to ensure the security and reliability of the key generation, distribution, and storage processes.

[0087] In S2, local data is the running data of each node itself. Each node maintains the version of local data to ensure data version synchronization.

[0088] Reference Figure 2 The flowchart of the local training method for the prediction model is shown. "Training the prediction model based on local data and generating local model parameters" includes the following steps:

[0089] S21: preprocessing local data;

[0090] S22: Formatting floating-point numbers on the pre-processed local data;

[0091] S23: Use non-associative addition and non-associative multiplication to protect the privacy of local data after floating-point formatting;

[0092] S24: Train the prediction model based on the SGD function and the privacy-protected local data to generate local model parameters.

[0093] In S21, each node preprocesses local data, which includes cleaning, standardization, and normalization. Preprocessing ensures data quality and provides a consistent input format for model training.

[0094] In S22, the IEEE 754 standard is used to uniformly format local data into floating-point numbers, allowing each node to process and store data in a unified format, ensuring data consistency and accuracy, and preparing for subsequent mathematical operations.

[0095] In S23, non-associative operations are introduced to enhance data privacy by modifying addition and multiplication operations to introduce random perturbations. The non-associative addition is: Non-associative multiplication is a⊙b=(a×b)+R(a×b). Here, R(·) is a random perturbation function that depends on a and b, representing the two data items involved in the operation. Through non-associative addition and non-associative multiplication, even if the calculation result can be observed, the original data cannot be directly inferred. This allows necessary calculations to be performed on local data while protecting data privacy.

[0096] In one implementation of S24, the expression of the SGD function is:

[0097]

[0098] Among them, θ t+1 is the prediction model parameter for the next iteration; θ t is the prediction model parameter of the current iteration; θ represents the prediction model parameter; η is the learning rate, which is used to control the update step size; is the loss gradient under the current prediction model parameters; ξ t The random perturbation term introduced in each iteration is a random number drawn from a Gaussian distribution. Its function is to add Gaussian noise to the gradient update to protect privacy, similar to the Laplace mechanism or Gaussian mechanism in differential privacy. t is the current number of training iterations.

[0099] The SGD function combines the SGD algorithm with non-associative operations. When calculating the gradients of model parameters, a local gradient clipping mechanism is designed to limit the size of the gradient. This protects data privacy while training the prediction model, ensuring that data privacy is not leaked even in a distributed training environment.

[0100] In another implementation of S24, the expression of the SGD function is:

[0101]

[0102] Before training begins, each node is assigned an initial privacy budget ε, which is adjusted according to the sensitivity of the local data to ensure that sensitive data is more protected. t The ε value used at iteration t is dynamically adjusted in each iteration based on the progress of the prediction model training and the current privacy requirements. This dynamic adjustment is achieved through a model performance feedback mechanism. Specifically, in each iteration, if the prediction model performance does not meet the preset target, ε is reduced to increase privacy protection. Conversely, if the prediction model performance meets the preset target, ε is increased to reduce noise and improve the prediction model accuracy.

[0103] This method of dynamically adjusting the ε value can find a balance between protecting privacy and optimizing model performance. The SGD function not only includes the update of model parameters, but also integrates privacy protection measures, allowing each node to effectively protect data privacy while training the model locally.

[0104] Furthermore, a privacy budget warning mechanism is implemented to monitor the usage of ε and the remaining privacy budget. If the privacy budget is close to being exhausted, the warning system will trigger an adjustment strategy, such as reducing the learning rate or increasing the noise level.

[0105] We can also design secure data sharing mechanisms using a secure data sharing protocol based on homomorphic encryption. We can also design a feature selection mechanism based on oblivious transfer (OT) to ensure data privacy during the feature selection process. We can also implement a secure data quality assessment protocol without exposing the original data.

[0106] After the prediction model training is complete, the prediction model parameters at this point are recorded as local model parameters. Model performance can also be evaluated using secure multi-party computation techniques, such as cross-validation with secure multi-party computation to ensure that node data is not leaked, and the area under the receiver operating characteristic (ROC) curve (AUC) is calculated to assess the model's predictive power. During the evaluation process, the data of each node remains encrypted, preventing the disclosure of any sensitive information. Adam optimizers or other optimization algorithms based on non-associative operations can also be implemented to accommodate non-associative addition and multiplication.

[0107] The local model parameters obtained by public key encryption are generated to generate encrypted model parameters and are ready to be shared with other nodes. The formula is: c i =Encrypt(pk,§ i ), where i is the sequence number of the node; c i is the encryption model parameter of the i-th node, § i is the local model parameter of the i-th node, and Encrypt is the encryption function.

[0108] In S3, nodes are divided into core nodes and edge nodes. In a distributed system, core nodes typically bear more computing and coordination responsibilities, aggregating data from edge nodes, performing more complex computations, and participating in global decision-making. Edge nodes, typically located near data sources, collect and initially process data, sending it to core nodes but not directly participating in global aggregation or decision-making. For example, consider three factories participating in federated learning of a predictive model. Each factory has production equipment for daily industrial manufacturing and control equipment that manages these production equipment. The production equipment is classified as edge nodes, and the control equipment is classified as core nodes. Each core node manages a group of edge nodes, forming a hierarchical network structure that helps reduce communication overhead and improve data processing efficiency. By configuring multiple core nodes, the system can achieve better fault tolerance and load balancing. If a core node fails, other core nodes can take over its tasks, ensuring stable system operation.

[0109] The core nodes are further divided into primary and secondary nodes. There is one primary node, responsible for final global aggregation, and several secondary nodes. A rotational or random mechanism is used to select the primary node from multiple core nodes, and the remaining core nodes are recorded as secondary nodes. This prevents a single node from concentrating power for a long time, reduces the risk of system attacks or failures, and increases system fairness and robustness. It also helps balance the load of each node, improving the efficiency and scalability of the entire system.

[0110] Furthermore, the system dynamically adjusts node roles and responsibilities based on the current workload and node health. When assigning node roles, security must be considered to ensure the security and reliability of core nodes (especially master nodes), as they play a critical role in the system. This role division enables distributed systems to more efficiently process large amounts of data while maintaining system stability and reliability. Each node performs tasks according to its role and responsibilities, working together to achieve the goals of the entire system.

[0111] In S4, edge nodes send their own encrypted model parameters to their corresponding core nodes. The core nodes then perform a local aggregation of their own and the received encrypted model parameters using a secure aggregation protocol based on homomorphic encryption. During the local aggregation process, core nodes can implement an anomaly detection mechanism to identify and address possible malicious updates. After completing the local aggregation, the core nodes generate encrypted local aggregate parameters.

[0112] In S5, the secondary node sends its local aggregate parameters, which have been securely processed using homomorphic encryption, to the primary node. The primary node then uses the homomorphic encryption secure aggregation protocol to globally aggregate its own local aggregate parameters with the received local aggregate parameters, generating a global encrypted sum, denoted as Cglobal, which is the encrypted aggregation result of all node encrypted model parameters.

[0113] The master node design is a decentralized aggregation mechanism based on secret sharing, which can increase the robustness and anti-attack capabilities of the system. During the global aggregation process, the master node continues to perform anomaly detection to ensure the accuracy and security of the aggregation results.

[0114] In S6, the master node sends the global encrypted sum to each secondary node (which can be sent by broadcast), and each secondary node audits the global encrypted sum. The audit includes checking the encryption integrity and correctness of the global encrypted sum. If the audit passes, the secondary node sends a first signal to the master node (the first signal is used to indicate a vote in favor). If the audit fails, the secondary node sends a second signal to the master node (the second signal is used to indicate a vote in disfavor). The master node counts the proportion of the first signal based on the number of first signals and the number of second signals received. If the proportion is greater than a preset proportion threshold (for example, 2 / 3), it means that the global encrypted sum has passed verification, and it is determined that all core nodes have reached a consensus. If the proportion is not greater than the preset proportion threshold, it means that the global encrypted sum has not passed verification, and it is determined that all core nodes have not reached a consensus. The location of the error is detected and corrected. If necessary, manual intervention or external arbitrator intervention is used. If necessary, the prediction model is retrained and the nodes are re-divided. This voting mechanism combines fully homomorphic encryption with a Byzantine fault-tolerant protocol, enabling reliable distributed data aggregation and analysis while protecting the privacy of individual data. It is able to resist some malicious nodes (for example, malicious nodes accounting for 1 / 3 of the total number of core nodes), protect the privacy of individual data and achieve reliable distributed data aggregation. It is suitable for industrial Internet of Things environments that require high security and privacy protection, such as smart manufacturing, supply chain management, energy management and other fields.

[0115] In S7, the central server uses the private key to decrypt Cglobal and obtain the global model parameters. The central server sends the global model parameters back to all nodes (can be sent via broadcast). Each node uses the global model parameters to update its local prediction model. Each node locally evaluates the performance of the updated model. The central server collects performance feedback and checks whether the model has met the convergence standard or other performance indicators. If abnormal data (such as extreme values) is detected during the model update or performance evaluation process, the system will trigger an alarm. Using zero-knowledge proof technology, the existence of an anomaly can be proved without revealing the specific device where the anomaly occurred, thus protecting data privacy.

[0116] The central server or master node can publish appropriate statistical information, such as average output, total energy consumption, etc., without disclosing the specific data of individual nodes to ensure data privacy and system transparency.

[0117] If the global aggregation times do not reach the preset times threshold, or the prediction model performance does not reach the convergence standard, repeat the above steps and continue to update the prediction model until the global aggregation times reach the preset times threshold or the prediction model performance reaches the convergence standard.

[0118] When publishing global model parameters, we use functional encryption technology to allow each node to attempt predictions using the updated prediction model in an encrypted domain, ensuring data security during the prediction process. We also implement a prediction API with differential privacy protection to further protect the prediction model and user privacy.

[0119] The resulting predictive model is used to predict industrial equipment failures, enabling routine maintenance and ensuring safe and efficient production. A secure incremental learning mechanism can be designed to enable the predictive model to be updated and optimized based on newly generated data. A secure model rollback mechanism can also be implemented to restore the predictive model to its previous stable state when performance degrades.

[0120] In the technical solution for building a predictive maintenance method for industrial equipment, the hardware architecture design for intelligent data connection and collection equipment is a fundamental and critical component. To this end, a hardware architecture was designed to ensure efficient and secure data collection and processing. First, a high-performance processor, such as the ARM Cortex-A series, was selected to provide powerful computing power to support complex data processing and analysis tasks. Furthermore, a dedicated encryption hardware accelerator was integrated to enhance data security during collection and transmission. The hardware architecture also includes a scalable storage system capable of handling large-scale data storage needs and providing sufficient data support for subsequent data analysis and model training. Regarding data collection, multi-interface data collection capabilities were implemented, supporting common industrial communication protocols such as Modbus and Profinet, ensuring seamless connection and data collection with various industrial equipment.

[0121] To improve the efficiency and flexibility of data collection, a configurable data acquisition interface was designed, and a real-time data collection and caching mechanism was implemented to facilitate rapid response and processing of data generated by industrial equipment. Regarding network connectivity, the device supports both wired and wireless network connections, including Ethernet, WiFi, and 5G, ensuring stable and rapid data transmission to core nodes or central servers.

[0122] Security is a top priority in the design, so a secure network communication protocol, such as TLS / SSL, is used to protect data transmission. Network fault detection and automatic recovery are implemented to ensure stable and reliable network connections. To fully leverage the advantages of edge computing, the hardware architecture integrates edge computing capabilities, enabling local data preprocessing and analysis, as well as a dynamically loadable edge application framework. This allows the device to flexibly deploy and update edge computing applications based on actual needs.

[0123] Through these carefully designed hardware features, intelligent data connection and acquisition devices not only efficiently collect and process industrial equipment data, but also provide strong hardware support for predictive maintenance systems for industrial equipment while ensuring data security and privacy. This lays a solid foundation for efficient and secure operations in areas such as smart manufacturing, supply chain management, and energy management.

[0124] Software system design is crucial for intelligent data connectivity and acquisition devices. To ensure high-performance operation and data processing, an efficient and secure software system was developed. First, an operating system and runtime environment suitable for real-time operations, such as FreeRTOS or RTThread, were selected to ensure system responsiveness and stability. Furthermore, secure boot and firmware update mechanisms were implemented to ensure continuous updates and secure maintenance throughout the device's lifecycle. Application isolation and resource management mechanisms were also employed to optimize system performance and prevent potential security threats.

[0125] In terms of data acquisition and preprocessing, we developed a configurable data acquisition engine that can adapt to different industrial equipment and data formats, implement data cleaning and standardization, ensure data quality, and lay a solid foundation for subsequent analysis and model training. We also use data compression and efficient storage to reduce storage space requirements and improve data transmission efficiency.

[0126] Secure communication and data transmission are key components of the system. This communication protocol, based on fully homomorphic encryption, allows computation to be performed directly on encrypted data without decryption, thereby protecting data privacy while maximizing its usefulness. Data fragmentation and resumable transmission, as well as data integrity verification and anti-tampering mechanisms, can also be used to ensure the security and reliability of data during transmission.

[0127] Device management and monitoring features, including remote device configuration and management interfaces, allow system administrators to easily perform device maintenance and configuration. This device health monitoring and fault diagnosis functionality helps identify and resolve potential device issues. Secure logging and auditing, which records key system operations and status changes, provides additional assurance for secure system operation.

[0128] Through the design and implementation of these software systems, intelligent data connection acquisition devices can not only collect and process data efficiently and securely, but also seamlessly integrate with other components in the industrial equipment predictive maintenance system to jointly build a robust and reliable industrial Internet of Things solution.

[0129] To improve system performance, we can analyze system bottlenecks and optimize computationally intensive operations, such as optimizing memory usage and management strategies to ensure the most efficient use of system resources. Intelligent energy management methods can be used to extend device battery life, which is particularly important in energy-constrained environments.

[0130] To enhance system reliability, we implement data backup and recovery mechanisms to prevent data loss. Through failover and fault tolerance, we ensure that the system can continue to operate even when some components fail. We also conduct long-term stability testing and optimization to ensure that the system maintains high performance after extended periods of operation. We conduct comprehensive security vulnerability scanning and remediation to ensure that the system is not vulnerable to known vulnerabilities. We implement intrusion detection and prevention to prevent unauthorized access and potential network attacks. We also conduct security emergency response to enable swift and effective responses to security incidents.

[0131] Through these comprehensive measures, intelligent data connection and collection equipment can not only meet the needs of predictive maintenance of industrial equipment, but also provide efficient and reliable services while protecting data privacy. This provides a solid technical foundation for smart manufacturing, supply chain management, energy management and other fields in the Industrial Internet of Things environment.

[0132] Reference Figure 3 The present disclosure provides an industrial equipment predictive maintenance system, comprising:

[0133] The key generation and distribution module 101 is used to generate public and private keys, send the public key to each node, and store the private key in a central server; each node represents an industrial device;

[0134] The prediction model training module 102 is used for each node to train the prediction model based on local data, generate local model parameters and encrypt them using the public key to generate encrypted model parameters;

[0135] The node division module 103 is used to divide multiple nodes into multiple core nodes and multiple edge nodes, and divide multiple core nodes into one main node and several secondary nodes; wherein one core node is associated with several edge nodes, and one main node is associated with several secondary nodes;

[0136] The local aggregation module 104 is used for the edge node to send the encryption model parameters to the corresponding core node, and the core node locally aggregates the local encryption model parameters and the received encryption model parameters to generate local aggregated parameters;

[0137] The global aggregation module 105 is used for the secondary node to send the local aggregation parameters to the primary node, and the primary node globally aggregates the local local aggregation parameters and the received local aggregation parameters to generate a global encrypted sum;

[0138] A global encryption and sending module 106 is configured to encrypt and send the global data to the central server by the master node after global encryption and verification;

[0139] The global encryption and decryption module 107 is used for the central server to decrypt the global encryption and obtain the global model parameters using the private key and send them to each node for updating the prediction model until the global aggregation times reach the preset times threshold or the performance of the prediction model reaches the convergence standard, and the prediction model is used to perform predictive maintenance on industrial equipment.

[0140] The various variations and specific examples of the industrial equipment predictive maintenance method provided above are also applicable to the industrial equipment predictive maintenance system provided in the present disclosure. Through the above detailed description of the industrial equipment predictive maintenance method, those skilled in the art can clearly understand the implementation method of the industrial equipment predictive maintenance system. For the sake of brevity of the specification, it will not be described in detail here.

[0141] The computer device according to an embodiment of the present disclosure includes a memory and a processor. The memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache). The non-volatile memory may include, for example, read-only memory (ROM), a hard disk, a flash memory, etc.

[0142] The processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the computer device to perform desired functions. In one embodiment of the present disclosure, the processor is configured to execute the computer-readable instructions stored in the memory, causing the computer device to execute all or part of the steps of the industrial equipment predictive maintenance method described in the various embodiments of the present disclosure.

[0143] Those skilled in the art should understand that in order to solve the technical problem of how to obtain a good user experience, this embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the scope of protection of this disclosure.

[0144] like Figure 4 The present invention provides a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Figure 4 The computer device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0145] like Figure 4 As shown, the computer device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device into a random access memory (RAM). Various programs and data required for the operation of the computer device are also stored in the RAM. The processor, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0146] Typically, the following devices can be connected to the I / O interface: input devices such as sensors or visual information acquisition devices; output devices such as display screens; storage devices such as tapes and hard disks; and communication devices. The communication device can allow the computer device to communicate with other devices (such as edge computing devices) wirelessly or by wire to exchange data. Figure 4 A computer device having various devices is shown, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0147] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processor, all or part of the steps of the industrial equipment predictive maintenance method of the embodiment of the present disclosure are performed.

[0148] For detailed description of this embodiment, please refer to the corresponding description of the above embodiments.

[0149] I will not go into details here.

[0150] According to an embodiment of the present disclosure, a computer-readable storage medium stores non-transitory computer-readable instructions, which, when executed by a processor, execute all or part of the steps of the industrial equipment predictive maintenance method described in the aforementioned embodiments of the present disclosure.

[0151] The above-mentioned computer-readable storage media include, but are not limited to, optical storage media (e.g., CD-ROMs and DVDs), magneto-optical storage media (e.g., MOs), magnetic storage media (e.g., magnetic tapes or mobile hard disks), media with built-in rewritable non-volatile memory (e.g., memory cards), and media with built-in ROM (e.g., ROM cartridges).

[0152] For detailed description of this embodiment, please refer to the corresponding description in the aforementioned embodiments, which will not be repeated here.

[0153] The basic principles of the present disclosure have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this disclosure are merely illustrative and not restrictive, and should not be construed as necessarily possessed by each embodiment of the present disclosure. Furthermore, the specific details disclosed above are provided for illustrative purposes and to facilitate understanding, rather than as limitations. These details do not limit the present disclosure to necessarily being implemented using these specific details.

[0154] In the present disclosure, relational terms such as first and second, etc. are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The block diagrams of the devices, devices, equipment, and systems involved in the present disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "including," "comprising," "having," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0155] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0156] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0157] Various changes, substitutions, and modifications may be made to the technology described herein without departing from the teachings defined by the appended claims. Moreover, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of things, means, methods, and actions described above. Currently existing or later developed processes, machines, manufactures, compositions of things, means, methods, or actions that perform substantially the same function or achieve substantially the same results as the corresponding aspects described herein may be utilized. Accordingly, the appended claims include within their scope such processes, machines, manufactures, compositions of things, means, methods, or actions.

[0158] The above description of the disclosed aspects is provided to enable any person skilled in the art to make

[0159] Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0160] The above description has been provided for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A predictive maintenance method for industrial equipment, characterized in that: include: Generate a public key and a private key, send the public key to each node, and store the private key in a central server; each node represents an industrial device; Each node trains a prediction model based on local data, generates local model parameters, and encrypts them using the public key to generate encrypted model parameters; Divide multiple nodes into multiple core nodes and multiple edge nodes, and divide multiple core nodes into one master node and several secondary nodes; one core node is associated with several edge nodes, and one master node is associated with several secondary nodes; The edge node sends the encryption model parameters to the corresponding core node, and the core node locally aggregates the local encryption model parameters and the received encryption model parameters to generate local aggregated parameters; The secondary node sends the local aggregation parameter to the primary node, and the primary node globally aggregates the local local aggregation parameter and the received local aggregation parameter to generate a global encrypted sum; After the global encryption and verification, the master node encrypts and sends the global encryption to the central server; The global encryption and verification specifically include: The master node encrypts the global data and sends it to each secondary node; Each secondary node encrypts and audits the global data respectively; If the review is passed, the secondary node sends a first signal to the primary node; If the review fails, the secondary node sends a second signal to the primary node; The master node counts the proportion of the first signal based on the number of received first signals and the number of received second signals; If the proportion is greater than the preset proportion threshold, it means that the global encryption has passed the verification; If the proportion is not greater than the preset proportion threshold, it means that the global encryption has not passed the verification; The central server uses a private key to decrypt the global encryption sum, obtains the global model parameters and sends them to each node for updating the prediction model until the global aggregation times reach a preset threshold or the performance of the prediction model reaches the convergence standard, and uses the prediction model to perform predictive maintenance on industrial equipment.

2. The method for predictive maintenance of industrial equipment according to claim 1, characterized in that: Train the prediction model based on local data and generate local model parameters, including: Preprocessing the local data; Perform floating point formatting on pre-processed local data; Use non-associative addition and non-associative multiplication to protect the privacy of local data after floating-point formatting; The prediction model is trained based on the SGD function and the privacy-protected local data to generate local model parameters.

3. The method for predictive maintenance of industrial equipment according to claim 2, characterized in that: The expression of the non-associative addition is: a⊕b=(a+b)+R(a,b); and / or, The expression of the non-associative multiplication is: a⊙b=(a×b)+R(a×b); Where R(·) is a random perturbation function; a and b represent the two data items involved in the operation.

4. The method for predictive maintenance of industrial equipment according to claim 3, wherein: The expression of the SGD function is: ； in, is the prediction model parameter for the next iteration; is the prediction model parameter of the current iteration; is the learning rate; is the loss gradient under the current prediction model parameters; is the random perturbation term introduced in each iteration; is the current training iteration number.

5. The method for predictive maintenance of industrial equipment according to claim 3, characterized in that: The expression of the SGD function is: ； in, is the prediction model parameter for the next iteration; is the prediction model parameter of the current iteration; is the learning rate; is the loss gradient under the current prediction model parameters; is the random perturbation term introduced in each iteration; It is in t The ε value used in the iteration; ε is the privacy budget; is the current training iteration number.

6. The method for predictive maintenance of industrial equipment according to claim 1, wherein: The multiple core nodes are divided into one main node and several secondary nodes, including: The primary node is selected from the plurality of core nodes using a rotation mechanism or a random mechanism, and the remaining core nodes are recorded as the secondary nodes.

7. A computer device, characterized in that: The computer device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the industrial equipment predictive maintenance method described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, which are used to enable a computer to execute the industrial equipment predictive maintenance method according to any one of claims 1 to 6.

9. A computer program product comprising computer instructions, characterized in that When the computer instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images