Intelligent factory model data set security privacy protection method and device, equipment and medium

Abstract

The invention relates to an intelligent factory model data set security privacy protection method and device, equipment and a medium, and the method comprises the steps: employing a preset information entropy formula to calculate and determine an information entropy score corresponding to each Hash bit in a first Hash code, sorting the information entropy scores corresponding to all Hash bits in the first Hash code from large to small, performing dimension reduction on the Hash bits corresponding to the column index set to determine a column index set corresponding to the Hash bits with the preset Hash code length, and performing dimension reduction on the Hash bits corresponding to the column index set to determine a second Hash code corresponding to the model data set; and encrypting the model data set by adopting a preset Paillier homomorphic encryption algorithm to determine a homomorphic ciphertext corresponding to the model data set, and uploading the second hash code corresponding to the model data set and the homomorphic ciphertext corresponding to the model data set to a central aggregation server in the intelligent factory. According to the invention, the accuracy and efficiency of production scheduling are improved, and meanwhile, privacy and data security are ensured.

Description

Technical Field

[0001] The present application relates to the field of data security, and particularly to a method for protecting the security and privacy of an intelligent factory model data set, a corresponding device, an electronic device, and a computer-readable storage medium. Background Art

[0002] In the scenario of an intelligent factory, data security and privacy protection are very important and are also a hot research field at present. With the development of Industry 4.0 and the Internet of Things, the intelligent factory generates a large amount of data. The intelligent factory can use technologies such as cloud computing and artificial intelligence to achieve intelligent scheduling through the analysis of these data, coordinate different devices and systems, so as to optimize the production process and support decision-making. Some data in the factory is highly sensitive and its original data cannot be leaked. Therefore, we need secure privacy computing technology to ensure that the original data of each participating party (such as production workshops, production lines) is not leaked, and the intelligent scheduling of the intelligent factory is realized through data aggregation and analysis by a high-computing-power central server or computing cluster.

[0003] In the workshop environment with high privacy requirements, collaborative training has become a key problem that needs to be solved urgently, and its importance is self-evident. With the rapid development of intelligent factories, various production data in the workshops is becoming increasingly rich, and these data contain huge value. However, these data often involve the company's trade secrets, core knowledge of production processes, and employees' personal privacy. In traditional technologies, there are problems such as excessive computing resource occupation, low efficiency, and poor privacy for various production data in the workshops, which has become a technical problem that needs to be overcome urgently.

[0004] In summary, in view of the problems in the prior art that various production data in the workshops have excessive computing resource occupation, low efficiency, and poor privacy, the applicant of the present application has made corresponding explorations in consideration of solving this problem. Summary of the Invention

[0005] The purpose of the present application is to solve the above problems and provide a method for protecting the security and privacy of an intelligent factory model data set, a corresponding device, an electronic device, and a computer-readable storage medium.

[0006] To achieve the various purposes of the present application, the present application adopts the following technical solutions:

[0007] A method for protecting the security and privacy of an intelligent factory model data set proposed for one of the purposes of the present application includes:

[0008] Obtaining the model data sets after local training of each production workshop in the intelligent factory, where the model data set represents the production environment data of a certain production workshop in the intelligent factory at a specific moment;

[0009] Invoke the preset locality - sensitive hashing algorithm based on information entropy, project each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component;

[0010] Use the preset information entropy formula to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the set of column indices corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the set of column indices to determine the second hash code corresponding to the model dataset;

[0011] Use the preset Paillier homomorphic encryption algorithm to encrypt the model dataset to determine the homomorphic ciphertext corresponding to the model dataset, and upload the second hash code corresponding to the model dataset and the homomorphic ciphertext corresponding to the model dataset to the central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model dataset.

[0012] Optionally, the step of invoking the preset locality - sensitive hashing algorithm based on information entropy, projecting each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and constructing the first hash code corresponding to the model dataset includes:

[0013] Obtain the model dataset X = [x1, x2, …, x n ∈ R n×d , where xi i represents the i - th data component in the model dataset X, d represents the number of model parameters, and the locality - sensitive hashing algorithm based on information entropy uses a family of hash functions F = {f1, f2, …, f L} ∈ R d×L , each of its hash functions is randomly generated, and the number of hash functions is L. Then the hash bit of the data component x is defined as:

[0014]

[0015] where h(x) represents the result of the data component x under the hash function, and each data component x generates a binary vector H(x) = {h1(x), h2(x), …, h L (x)} ∈ {0, 1} L ;

[0016] Project each data component \(x\) in the model dataset \(X\) onto each hash function in the hash function family \(F\) to determine the corresponding hash bits for each data component, and combine the hash bits corresponding to each data component to form the first hash code corresponding to the model dataset. The first hash code corresponding to the model dataset \(X\) is a binary matrix \(B\), which is expressed as:

[0017]

[0018] wherein, the binary matrix \(B\) describes the hash bits of the model dataset \(X\), and \(h\) i \(\in \{0, 1\}\) n×1 represents the value of all data components in the model dataset \(X\) at the \(i\)-th hash bit of the binary matrix \(B\).

[0019] Optionally, the steps of calculating and determining the information entropy scores corresponding to each hash bit in the first hash code, sorting all the information entropy scores corresponding to the hash bits in the first hash code from largest to smallest to determine the set of column indices corresponding to the hash bits of the preset hash code length, and performing dimensionality reduction on the hash bits corresponding to the set of column indices to determine the second hash code corresponding to the model dataset include:

[0020] The expression of the information entropy formula is:

[0021]

[0022] wherein, \(p(h\) i ) represents the probability distribution of the value of the hash bit \(h\) i . In the hash bit \(h\) i , \(m = 2\), which means that the value of the element in the hash bit \(h\) i is 0 or 1. By calculating the information entropy score corresponding to each hash bit, it is denoted as \(score\in R\) L ;

[0023] Sort all the information entropy scores corresponding to the hash bits in the first hash code from largest to smallest, and select the top \(k\) hash bits with the highest scores from largest to smallest to construct the set of column indices \(D\), where \(k\) represents the preset hash code length, and the dimensionality reduction means determining a low-dimensional second hash code by removing the hash bits in the set of column indices \(D\).

[0024] Optionally, the steps of encrypting the model dataset using a preset Paillier homomorphic encryption algorithm to determine the corresponding homomorphic ciphertext of the model dataset include:

[0025] Obtain the model dataset \(X = [x1, x2, \ldots, x\) after local training in each production workshop in the intelligent factoryn ∈ R n×d , where x i represents the i-th data component in the model data set X, and d represents the number of model parameters;

[0026] Call the preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ);

[0027] Based on the Paillier encryption formula, for each data component x i ∈ X, randomly generate an r i , and use the public key (n, g) to calculate the ciphertext c i , to determine the encrypted ciphertext vector c = [c1, c2,..., c d .

[0028] Optionally, the step of calling the preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ), includes:

[0029] Select a first prime number p and a second prime number q, calculate and determine n = p × q and λ = lcm(p - 1, q - 1), where λ = lcm(p - 1, q - 1) represents the least common multiple of p - 1 and q - 1;

[0030] Select a base satisfying gcd(L(g λ mod n 2 ), n) = 1, where

[0031] Calculate the private key parameter μ = (L(g λ mod n 2 )) -1 mod n, to determine the public key (n, g) and the private key (λ, μ).

[0032] Optionally, the expression of the Paillier encryption formula is:

[0033]

[0034] where x i is the i-th data component in the model data set X, r i is the random number for each encryption, and it satisfies

[0035] Optionally, the production environment data includes sensor data, device status operation data, and production demand data.

[0036] An intelligent factory model dataset security and privacy protection device provided to meet another object of the present application, comprising:

[0037] A production data acquisition module, configured to acquire the model datasets of each production workshop in the intelligent factory after local training, wherein the model datasets represent the production environment data of a certain production workshop in the intelligent factory at a specific moment;

[0038] A first hash code determination module, configured to call a preset locality-sensitive hashing algorithm based on information entropy, project each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component;

[0039] A second hash code determination module, configured to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code by using a preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from large to small to determine the column index set corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the column index set to determine the second hash code corresponding to the model dataset;

[0040] A security and privacy protection module, configured to encrypt the model dataset by using a preset Paillier homomorphic encryption algorithm to determine the homomorphic ciphertext corresponding to the model dataset, and upload the second hash code corresponding to the model dataset and the homomorphic ciphertext corresponding to the model dataset to a central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model dataset.

[0041] An electronic device provided to meet another object of the present application, comprising a central processing unit and a memory, wherein the central processing unit is configured to call and run a computer program stored in the memory to execute the steps of the intelligent factory model dataset security and privacy protection method of the present application.

[0042] A computer-readable storage medium provided to meet another object of the present application, which stores a computer program implemented according to the intelligent factory model dataset security and privacy protection method in the form of computer-readable instructions. When the computer program is called and run by a computer, it executes the steps included in the corresponding method.

[0043] Compared with the prior art, in view of the problems of excessive computing resource occupation, low efficiency, and poor privacy of various production data in the workshop in the prior art, the present application includes but is not limited to the following beneficial effects:

[0044] First, this application first uses the local sensitive hashing algorithm based on information entropy to remove redundant and invalid hash bits in the hash code, and then uses the homomorphic encryption Paillier algorithm to encrypt the uploaded data of each participant (data provider) within the smart factory, and uses the low-dimensional LSH algorithm to remove abnormal data that is too high or too low, and finally performs homomorphic aggregation calculations on the encrypted data. The Paillier algorithm supports computing operations on ciphertext data, thereby ensuring that the data remains encrypted throughout the processing process to prevent the leakage of sensitive information.

[0045] Second, due to the obfuscation of encryption, the server that performs data aggregation cannot remove abnormal data uploaded by participants. This application uses local sensitive hashing (LSH) to remove abnormal data without leaking privacy. When processing production data, it can efficiently identify and remove abnormal data caused by factors such as equipment failure, sensor error, and system failure, providing accurate and privacy-protected data support for subsequent intelligent scheduling decisions;

[0046] Third, the local sensitive hashing algorithm based on information entropy of the present application selects effective hash bits by information entropy, reduces the impact of invalid or redundant data, optimizes hash mapping, makes calculations more efficient, and removes redundant hash bits to reduce storage requirements and computing burdens, thereby increasing the speed of data processing. At the same time, it also significantly improves accuracy and retains hash bits with high information content, ensuring that data mapping can better maintain the aggregation of similar data points, thereby improving the accuracy of data processing. The interference of hash bits is reduced, making the final hash code more effective and reducing calculation errors. The computational complexity is greatly reduced. Due to the selection of low-dimensional hash bits, the dimensions required for subsequent data calculations are smaller, and the overall computational complexity is significantly reduced. This is particularly important for large-scale data sets, especially in smart factory environments that require rapid response.

[0047] Fourthly, the Paillier homomorphic encryption algorithm of this application ensures the privacy and security of data of all parties during data upload, transmission, calculation and aggregation. Even if the data is in an encrypted state, encrypted calculations can still be performed to protect the confidentiality of the data and avoid privacy risks caused by data leakage. In smart factories, data exchange between multiple parties is a common requirement. Through the Paillier encryption algorithm, even if the data is intercepted during transmission, the attacker cannot obtain useful information from the encrypted data, thereby effectively preventing the risk of data leakage or tampering. With the increasingly stringent laws on data privacy protection, the use of homomorphic encryption algorithms can ensure that the data processing process complies with the requirements of laws and regulations and reduce compliance risks.

[0048] Fifthly, this application optimizes scheduling and improves data quality. By performing anomaly detection on encrypted data through the locality-sensitive hashing algorithm based on information entropy, it can quickly identify and isolate data that does not conform to the normal pattern, avoiding the negative impact of such abnormal data on subsequent analysis and decision-making. Anomaly detection can discover abnormal devices, process deviations, or production problems in the production process, and take timely measures to reduce the occurrence of faults.

[0049] In summary, through anomaly detection and data cleaning based on the locality-sensitive hashing algorithm of information entropy, and by using the Paillier homomorphic encryption algorithm for data encryption, the system demonstrates multiple advantages in aspects such as improving data processing efficiency, ensuring data privacy and security, enhancing data quality, and optimizing scheduling. This not only improves the accuracy and efficiency of production scheduling, but also ensures privacy and data security while enhancing the operability and robustness of the smart factory in a large-scale production environment. Brief Description of the Drawings

[0050] The above and / or additional aspects and advantages of this application will become apparent and easy to understand from the following description of the embodiments in conjunction with the drawings, where:

[0051] Figure 1 is a schematic flowchart of the method for protecting the security and privacy of the smart factory model dataset of this application;

[0052] Figure 2 is an exemplary network architecture adopted by the security and privacy computing system in the embodiment of this application;

[0053] Figure 3 is a principle block diagram of the device for protecting the security and privacy of the smart factory model dataset in the embodiment of this application;

[0054] Figure 4 is a schematic structural diagram of the computer device in the embodiment of this application. Detailed Embodiments

[0055] The embodiments of this application will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain this application and should not be construed as limiting this application.

[0056] Those skilled in the art can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application means the presence of the stated features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more of the associated listed items.

[0057] Those skilled in the art can understand that, unless otherwise defined, all terms used herein (including technical terms and scientific terms) have the same meaning as the general understanding of those of ordinary skill in the art to which the present application belongs. It should also be understood that terms such as those defined in a general dictionary should be understood as having a meaning consistent with the meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0058] Those skilled in the art can understand that the "client", "terminal", and "terminal device" used herein include both devices with a wireless signal receiver that only has the ability to receive and no ability to transmit, and devices with both receiving and transmitting hardware that can conduct two-way communication on a two-way communication link. Such devices can include: cellular or other communication devices such as personal computers, tablets, etc., which have a single-line display or a multi-line display or a cellular or other communication device without a multi-line display; PCS (Personal Communications Service), which can combine voice, data processing, fax, and / or data communication capabilities; PDA (Personal Digital Assistant), which can include a radio frequency receiver, a pager, Internet / intranet access, a web browser, a notepad, a calendar, and / or a GPS (Global Positioning System) receiver; conventional laptop and / or palm computers or other devices, which are conventional laptop and / or palm computers or other devices with and / or including a radio frequency receiver. The "client", "terminal", and "terminal device" used herein can be portable, transportable, installed in a vehicle (air, sea, and / or land), or suitable for and / or configured to run locally, and / or run in a distributed form at any other location on the earth and / or in space. The "client", "terminal", and "terminal device" used herein can also be a communication terminal, an Internet access terminal, a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback function, or can also be a smart TV, a set-top box, etc.

[0059] The hardware referred to by names such as "server", "client", and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer, and is a hardware device with the necessary components disclosed by the von Neumann principle, including a central processing unit (including an arithmetic unit and a controller), a memory, an input device, and an output device. The computer program is stored in its memory, and the central processing unit loads the program stored in the external memory into the memory for execution, executes the instructions in the program, and interacts with the input / output devices to complete specific functions.

[0060] It should be noted that the concept of "server" in this application can similarly be extended to apply to server clusters. According to the network deployment principles understood by those skilled in the art, the various servers should be logically divided. Physically, these servers can either be independent of each other but can be invoked through interfaces, or be integrated into a single physical computer or a set of computer clusters. Those skilled in the art should understand this flexibility and should not be restricted by this when implementing the network deployment method of this application.

[0061] One or several technical features of this application, unless expressly specified, can either be deployed on the server and accessed by the client remotely invoking the online service interface provided by the server, or be directly deployed and run on the client for access.

[0062] The neural network models cited or possibly cited in this application, unless expressly specified, can either be deployed on a remote server and remotely invoked by the client, or be deployed on a client capable of handling the device and directly invoked. In some embodiments, when it runs on the client, its corresponding intelligence can be obtained through transfer learning to reduce the requirements for the client's hardware operating resources and avoid excessive occupation of the client's hardware operating resources.

[0063] All kinds of data involved in this application, unless expressly specified, can either be remotely stored on the server or stored on the local terminal device, as long as it is suitable for being invoked by the technical solution of this application.

[0064] Those skilled in the art should be aware that although the various methods of this application are described based on the same concept and thus show commonality with each other, unless otherwise specified, these methods can be executed independently. Similarly, for the various embodiments disclosed in this application, they are all proposed based on the same inventive concept. Therefore, for concepts with the same expression, as well as concepts that are only appropriately transformed for convenience although the concept expressions are different, they should be equally understood.

[0065] For the various embodiments to be disclosed in this application, unless expressly pointed out that there is a mutually exclusive relationship between them, the relevant technical features involved in each embodiment can be cross-combined to flexibly construct new embodiments, as long as this combination does not deviate from the creative spirit of this application and can meet the requirements in the prior art or solve certain deficiencies in the prior art. Those skilled in the art should be aware of this flexibility.

[0066] In a workshop environment with high privacy requirements, collaborative training has become a key issue that urgently needs to be solved, and its importance is self-evident. With the rapid development of smart factories, various types of production data in the workshop are becoming increasingly rich. These data contain huge value and play a crucial role in improving production efficiency, optimizing resource allocation, and enhancing the accuracy of decision-making. However, these data often involve the company's trade secrets, core knowledge of production processes, and employees' personal privacy. Therefore, how to achieve effective collaborative training among different systems, different devices, and even different enterprises in the workshop while protecting data privacy has become a technical problem that urgently needs to be overcome.

[0067] In a workshop environment with high privacy requirements, collaborative training not only faces the challenge of data privacy protection but also, due to the complexity and uncertainty of the workshop environment, often cannot guarantee the accuracy and integrity of the uploaded data. This uncertainty may stem from various factors such as equipment failures, sensor errors, human operation mistakes, or malicious attacks, all of which may cause abnormalities or errors in the data uploaded by the workshop. The existence of abnormal data not only affects the effect of collaborative training but may also lead to model misjudgment, decision-making mistakes, and even production safety accidents. Therefore, in the collaborative training system of smart factories, an auxiliary means is needed that can retain the plaintext similarity while ensuring privacy to promptly identify and eliminate these potential threats. By detecting abnormal data, not only can the accuracy and reliability of collaborative training in smart factories be improved, but also the production risks and cost losses caused by data errors can be effectively reduced. It is not only an important part of data security and privacy protection in smart factories but also a key guarantee for the efficient and stable operation of smart factories.

[0068] Against the backdrop of coexisting high privacy requirements and workshop environment uncertainty, collaborative training in smart factories not only faces the challenges of data privacy protection and abnormal data detection but also often needs to operate under limited performance resources. Lower performance requirements mean that the central server and each workshop device need to minimize computational complexity and resource consumption while ensuring functional integrity when processing data, conducting collaborative training, and performing anomaly detection. This requires completing privacy computing with as small a computational overhead as possible while ensuring the above two points.

[0069] Please refer to Figure 1 and Figure 2 , in one embodiment of the smart factory model dataset security and privacy protection method of the present application, it includes:

[0070] Step S10: Obtain the model datasets after local training of each production workshop in the smart factory, where the model datasets represent the production environment data of a certain production workshop in the smart factory at a specific moment;

[0071] The security and privacy computing system in the terminal device can respond to security and privacy protection events of the intelligent factory model dataset, and obtain the model datasets after local training in each production workshop in the intelligent factory. Among them, the model dataset represents the production environment data of a certain production workshop in the intelligent factory at a specific moment; among them, the production environment data includes sensor data, equipment status operation data, and production demand data.

[0072] Specifically, obtain the model datasets X = [x1, x2, …, x n ∈ R n×d , where x i represents the i-th data component in the model dataset X, and the data component is a d-dimensional vector.

[0073] Step S20: Invoke the preset information entropy-based locality-sensitive hashing algorithm, project each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component;

[0074] After obtaining the model datasets after local training in each production workshop in the intelligent factory, invoke the preset information entropy-based locality-sensitive hashing algorithm, project each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component;

[0075] In a specific embodiment, the step of invoking the preset information entropy-based locality-sensitive hashing algorithm, projecting each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and constructing the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component includes:

[0076] The security and privacy computing system in the terminal device obtains the model dataset X = [x1, x2, …, x n ∈ R n×d , where x i represents the i-th data component in the model dataset X, d represents the number of model parameters, and the information entropy-based locality-sensitive hashing algorithm uses a hash function family F = {f1, f2, …, f L} ∈ R d×L , each of its hash functions is randomly generated, and the number of hash functions is L, then the hash bit of the data component x is defined as:

[0077]

[0078] Among them, h(x) represents the result of the data component x under the hash function. Each data component x generates a binary vector H(x) = {h1(x), h2(x), …, h L (x)} ∈ {0, 1} L ;

[0079] Project each data component x in the model data set X onto each hash function in the hash function family F to determine the corresponding hash bits of each data component, and combine the corresponding hash bits of each data component into the first hash code corresponding to the model data set. The first hash code corresponding to the model data set X is a binary matrix B, which is expressed as:

[0080]

[0081] Among them, the binary matrix B describes the hash bits of the model data set X. Among them, h i ∈ {0, 1} n×1 represents the value of all data components in the model data set X at the i-th hash bit of the binary matrix B.

[0082] Step S30: Calculate and determine the information entropy scores corresponding to each hash bit in the first hash code using a preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the set of column indices corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the set of column indices to determine the second hash code corresponding to the model data set;

[0083] Call the preset low-dimensional locality-sensitive hashing bit algorithm based on information entropy, project each data component in the model data set onto each hash function to determine the corresponding hash bits of each data component, after constructing the first hash code corresponding to the model data set according to the corresponding hash bits of each data component, calculate and determine the information entropy scores corresponding to each hash bit in the first hash code using a preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the set of column indices corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the set of column indices to determine the second hash code corresponding to the model data set;

[0084] In some embodiments, the Locality-Sensitive Hashing (LSH) algorithm is a technique for approximate similarity search and is widely used in the processing of high-dimensional data. Its core idea is: by mapping high-dimensional data to a low-dimensional space, points that are similar in the low-dimensional space have a relatively high probability of falling into the same hash bucket, while dissimilar points are assigned to different hash buckets. The advantage of the LSH algorithm is that it can greatly reduce the computational complexity of high-dimensional data processing and can effectively identify the similarity between data points. However, locality-sensitive hashing often requires generating relatively long hash codes to meet the retrieval requirements. However, long hash codes consume a large amount of storage space and carry a large number of redundant hash bits. We use information entropy selection to remove redundant and invalid hash bits to improve data processing efficiency. The low-dimensional LSH algorithm can efficiently identify and eliminate abnormal data caused by factors such as equipment failures, sensor errors, and system failures when processing production data.

[0085] In a specific embodiment, the steps of calculating and determining the information entropy scores corresponding to each hash bit in the first hash code using a preset information entropy formula, sorting the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine a set of column indices corresponding to the hash bits of a preset hash code length, and performing dimensionality reduction on the hash bits corresponding to the set of column indices to determine the second hash code corresponding to the model dataset include:

[0086] The expression of the information entropy formula is:

[0087]

[0088] where p(h i ) represents the probability distribution of the value of hash bit h i . In hash bit h i , m = 2, which means the values of the elements in hash bit h i are 0 or 1. By calculating the information entropy score corresponding to each hash bit, which is denoted as score ∈ R L ;

[0089] Sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest, and select the top k hash bits with the highest scores from largest to smallest to construct a set of column indices D, where k represents the preset hash code length. Here, the dimensionality reduction means determining a low-dimensional second hash code by removing the hash bits in the set of column indices D.

[0090] As can be seen from the above steps, in this application, through the locality-sensitive hashing algorithm based on information entropy (LSH bit selection algorithm), the hash codes generated by LSH are optimized to remove redundant and invalid hash bits. The traditional LSH algorithm may generate high-dimensional hash codes, resulting in an excessive computational burden and possibly containing useless hash bit information. Through the information entropy selection technique, the system can retain the effective information while eliminating the irrelevant parts, thus obtaining a low-dimensional LSH algorithm for data in intelligent factories. This low-dimensional LSH algorithm can efficiently process a large amount of data in workshop data and production scheduling, reduce the computational overhead, and improve the detection accuracy. The locality-sensitive hashing algorithm based on information entropy in the embodiments of this application reduces the dimension and simplifies the hash codes. By scoring the information entropy values of the hash bits and selecting the top k hash bits, redundant hash bits can be effectively reduced, thereby greatly reducing the dimension of the hash codes. The lower dimension of the hash codes can effectively reduce the occupancy of storage space, and then accelerate the subsequent similarity query and data retrieval processes.

[0091] Furthermore, the locality-sensitive hashing algorithm based on information entropy can map similar data components to the same hash bucket through the designed hash function. Therefore, by reasonably selecting a subset of hash bits (i.e., the column index set D), it can reduce the dimension while retaining sufficient locality-sensitive features to ensure that similar data components can be correctly mapped to the same bucket.

[0092] Furthermore, the locality-sensitive hashing algorithm based on information entropy can greatly reduce the storage space. The binary vectors generated by the original data set through hash transformation are more storage-space-saving than the original real-valued vectors. Screening out meaningful hash bits through information entropy further reduces the storage requirements. By removing redundant hash bits (a reduction of about 20%), you reduce the storage requirements for the data set and at the same time reduce the computational amount, thereby improving the processing efficiency.

[0093] Furthermore, the locality-sensitive hashing algorithm based on information entropy can significantly improve the computational efficiency and reduce the computational complexity. The operations of information entropy scoring and redundant hash bit removal can reduce the computational burden in subsequent processing. By reducing the number of hash bits that need to be calculated for each query, the computational amount is reduced and the query efficiency is improved. The locality-sensitive hashing algorithm based on information entropy can perform fast retrieval through hash buckets, which is much more efficient than directly calculating the similarity between data components (such as Euclidean distance or cosine similarity). By reducing the number of hash bits, the matching of data components during the query process becomes faster.

[0094] Furthermore, sorting the hash bits using information entropy can identify which hash bits contain more information, so more discriminative hash bits can be selected. This enables the algorithm to better adapt to the characteristics of different datasets in practical applications, enhancing the robustness and stability of locality-sensitive hashing. Selecting hash bits with larger amounts of information can make the final hash code more effectively represent the differences between data components, thereby improving the accuracy and robustness of the algorithm in practical applications.

[0095] Furthermore, balance between time efficiency and space efficiency: By evaluating the information entropy of the hash bits and removing redundant hash bits, you can reduce the storage overhead while maintaining a high query efficiency. In this way, your algorithm can find a better balance between memory usage and computational efficiency, especially suitable for processing large-scale datasets.

[0096] Furthermore, since the property of LSH is to map similar data components to the same hash bucket through a hash function, the optimized hash bit selection can enhance this effect. Selecting hash bits with larger amounts of information means that the similarity of data components can be better retained, thus improving the local sensitivity and accuracy of the algorithm.

[0097] Furthermore, supporting large-scale data processing, the optimized locality-sensitive hashing algorithm can efficiently process large-scale datasets, especially suitable for real-time systems with a large amount of data, such as production scheduling control model inference in smart factories, image retrieval, recommendation systems, and big data analysis.

[0098] In summary, through the selection of hash bits and information entropy scoring, it is possible to effectively reduce the computational and storage complexities while maintaining high local sensitivity and accuracy, thus achieving more efficient and accurate data processing and retrieval.

[0099] Step S40: Encrypt the model dataset using a preset Paillier homomorphic encryption algorithm to determine the corresponding homomorphic ciphertext of the model dataset, and upload the second hash code corresponding to the model dataset and the homomorphic ciphertext corresponding to the model dataset to the central aggregation server in the smart factory to complete the security and privacy protection of the smart factory model dataset.

[0100] Calculate and determine the information entropy scores corresponding to each hash bit in the first hash code using a preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine a set of column indexes corresponding to the hash bits of a preset hash code length, perform dimensional reduction on the hash bits corresponding to the set of column indexes to determine a second hash code corresponding to the model dataset, and then use a preset Paillier homomorphic encryption algorithm to encrypt the model dataset to determine a homomorphic ciphertext corresponding to the model dataset. Upload the second hash code corresponding to the model dataset and the homomorphic ciphertext corresponding to the model dataset to a central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model dataset.

[0101] In a specific embodiment, the step of encrypting the model dataset using a preset Paillier homomorphic encryption algorithm to determine the homomorphic ciphertext corresponding to the model dataset includes:

[0102] Step S401, obtain a model dataset X = [x1, x2, …, x n ∈ R n×d , where x i represents the i-th data component in the model dataset X, and d represents the number of model parameters;

[0103] Step S402, call a preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ);

[0104] Step S403, based on the Paillier encryption formula, for each data component x i ∈ X, randomly generate an r i , and use the public key (n, g) to calculate the ciphertext c i to determine the encrypted ciphertext vector c = [c1, c2, …, c d .

[0105] In some embodiments, the expression of the Paillier encryption formula is:

[0106]

[0107] where x i is the i-th data component in the model dataset X, r i is a random number for each encryption, and it satisfies

[0108] In a further embodiment, the step of calling a preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ), includes:

[0109] Step S4021: Select a first prime number p and a second prime number q, calculate and determine n = p × q and λ = lcm(p - 1, q - 1), where λ = lcm(p - 1, q - 1) represents the least common multiple of p - 1 and q - 1;

[0110] Step S4022: Select a base satisfying gcd(L(g λ mod n 2 ), n) = 1, where

[0111] Step S4023: Calculate the private key parameter μ = (L(g λ mod n 2 )) -1 mod n to determine the public key (n, g) and the private key (λ, μ).

[0112] More specifically, first, each workshop collects data in the production environment in real time, including sensor data, equipment operation status, production requirements and other information. These data are sent to the model training module for dynamic modeling to ensure the adaptability of the scheduling system to the workshop environment. Data modeling includes the analysis of workshop production progress, equipment status and material flow, and generates a prediction model for scheduling decisions. During the process of uploading model data, the system uses the Paillier homomorphic encryption algorithm to encrypt the production environment data uploaded by each production workshop in the intelligent factory. The production environment data includes sensor data, equipment status operation data, and model data of production requirements data uploaded, specifically including:

[0113] 1) Obtain the model data set X = [x1, x2,..., x n ∈ R n×d , where x i represents the i-th data component in the model data set X, which represents the eigenvalue of the model data set, and d represents the number of model parameters.

[0114] 2) Call a preset to generate Pallier public-private key pairs to generate public-private key pairs, which includes:

[0115] a. Select a first prime number p and a second prime number q, calculate and determine n = p × q and λ = lcm(p - 1, q - 1), where λ = lcm(p - 1, q - 1) represents the least common multiple of p - 1 and q - 1;

[0116] b. Select a base Satisfy gcd(L(g λ mod n 2 ), n) = 1, where

[0117] c. Calculate the private key parameter μ = (L(g λ mod n 2 )) -1 mod n to determine the public key (n, g) and the private key (λ, μ).

[0118] Based on the above steps, the Paillier encryption formula is expressed as:

[0119]

[0120] where x i is the i-th data component in the model data set X, r i is a random number for each encryption, which satisfies Its encryption process includes:

[0121] Based on the Paillier encryption formula, for each data component x i ∈ X, randomly generate an r i , and then use the public key (n, g) to calculate the ciphertext c i . Finally, the encrypted ciphertext vector c = [c1, c2,..., c d can be obtained.

[0122] In a further embodiment, the Paillier homomorphic encryption algorithm allows the correspondence of addition between ciphertext domains, which means that if there are two ciphertexts c1 and c2, then the product c * = c1 × c2 mod n 2 , and c * corresponds to x1 + x2 after decryption. In the encrypted state, data can be directly added without decrypting the data itself.

[0123] In a further embodiment, the second hash code corresponding to the model data set obtained in the above embodiment and the homomorphic ciphertext corresponding to the model data set are uploaded to the central aggregation server in the intelligent factory. The system processes the hash code of the model data set through a locality-sensitive hashing algorithm based on information entropy, removing abnormally high or low data points, which may be caused by sensor failures, equipment anomalies, production anomalies, etc. The low-dimensional locality-sensitive hashing bit selection algorithm can quickly identify data that does not conform to the normal pattern and remove its homomorphic ciphertext; after data cleaning, all valid data will be aggregated and calculated through homomorphic encryption. As described above, only the addition of plaintext parameters needs to be converted into the multiplication of ciphertext parameters, and then the decryption operation can be completed on the client to obtain the plaintext accumulation result of the parameters, and the updated model parameters can be obtained. The aggregated encrypted data can be further used for production scheduling decisions without the need for decryption operations, ensuring the privacy of data during the aggregation process.

[0124] As can be seen from the above embodiments, compared with the prior art, in view of the problems of excessive computing resource occupation, low efficiency, and poor privacy in various production data in the workshop in the prior art, the present application includes but is not limited to the following beneficial effects:

[0125] First, the present application uses a locality-sensitive hashing algorithm based on information entropy to remove redundant and invalid hash bits in the hash code, then encrypts the data uploaded by each participant (data provider) inside the intelligent factory using the homomorphic encryption Paillier algorithm, and uses the low-dimensional LSH algorithm to remove abnormally high or low data. Finally, the encrypted data is homomorphically aggregated and calculated. The Paillier algorithm supports computational operations on ciphertext data, thus ensuring that the data remains encrypted throughout the processing process and preventing the leakage of sensitive information.

[0126] Second, due to the encryption obfuscation, the server performing data aggregation cannot remove the abnormal data uploaded by the participants. The present application uses locality-sensitive hashing (LSH) to remove abnormal data without revealing privacy. When processing production data, it can efficiently identify and remove abnormal data caused by factors such as equipment failures, sensor errors, and system failures, providing accurate and privacy-protected data support for subsequent intelligent scheduling decisions;

[0127] Thirdly, in the local sensitive hashing algorithm based on information entropy of this application, effective hash bits are selected through information entropy, reducing the influence of invalid or redundant data, optimizing hash mapping, making the calculation more efficient. The removal of redundant hash bits reduces storage requirements and computational burden, thereby improving the data processing speed. At the same time, it also significantly improves the accuracy and retains hash bits with high information content, ensuring that the mapping of data can better maintain the clustering of similar data points, thus improving the accuracy of data processing. Reducing the interference of hash bits makes the finally obtained hash code more effective and reduces computational errors. It greatly reduces the computational complexity. Since the selection of low-dimensional hash bits makes the dimensions to be processed in subsequent data calculations smaller, the overall computational complexity is significantly reduced, which is particularly important for large-scale data sets, especially in the intelligent factory environment that requires quick response.

[0128] Fourthly, the Paillier homomorphic encryption algorithm of this application ensures the privacy and security of each party's data during the processes of data uploading, transmission, calculation, and aggregation. Even when the data is in an encrypted state, encrypted calculations can still be performed, protecting the confidentiality of the data and avoiding privacy risks caused by data leakage. In an intelligent factory, data exchange among multiple parties is a common requirement. Through the Paillier encryption algorithm, even if the data is intercepted during transmission, attackers cannot obtain useful information from the encrypted data, thus effectively preventing the risks of data leakage or tampering. With the increasingly strict data privacy protection laws, adopting a homomorphic encryption algorithm can ensure that the data processing process complies with legal and regulatory requirements and reduces compliance risks.

[0129] Fifthly, this application optimizes scheduling and improves data quality. By performing anomaly detection on encrypted data through the local sensitive hashing algorithm based on information entropy, it can quickly identify and isolate data that does not conform to the normal pattern, avoiding the negative impact of such abnormal data on subsequent analysis and decision-making. Anomaly detection can discover abnormal devices, process deviations, or production problems in the production link and take timely measures to reduce the occurrence of faults. All in all, through anomaly detection and data cleaning using the local sensitive hashing algorithm based on information entropy, and data encryption using the Paillier homomorphic encryption algorithm, the system demonstrates multiple advantages in improving data processing efficiency, ensuring data privacy and security, improving data quality, and optimizing scheduling. This not only improves the accuracy and efficiency of production scheduling, but also ensures privacy and data security while enhancing the operability and robustness of the intelligent factory in a large-scale production environment.

[0130] Please refer to Figure 3, An intelligent factory model dataset security and privacy protection device provided to meet one of the purposes of this application, including a production data acquisition module 1100, a first hash code determination module 1200, a second hash code determination module 1300, and a security and privacy protection module 1400. Among them, the production data acquisition module 1100 is configured to acquire the model datasets after local training in each production workshop in the intelligent factory, where the model datasets represent the production environment data of a certain production workshop in the intelligent factory at a specific moment; the first hash code determination module 1200 is configured to call a preset locality-sensitive hashing algorithm based on information entropy, project each data component in the model dataset onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model dataset according to the hash bits corresponding to each data component; the second hash code determination module 1300 is configured to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code by using a preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the column index set corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the column index set to determine the second hash code corresponding to the model dataset; the security and privacy protection module 1400 is configured to encrypt the model dataset by using a preset Paillier homomorphic encryption algorithm to determine the homomorphic ciphertext corresponding to the model dataset, and upload the second hash code corresponding to the model dataset and the homomorphic ciphertext corresponding to the model dataset to the central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model dataset.

[0131] Based on any embodiment of this application, please refer to Figure 4 , Another embodiment of this application further provides an electronic device, which can be implemented by a computer device, as Figure 4 shown, the internal structure schematic diagram of the computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected through a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The control information sequence can be stored in the database. When the computer-readable instructions are executed by the processor, the processor can implement an intelligent factory model dataset security and privacy protection method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device can store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the intelligent factory model dataset security and privacy protection method of this application. The network interface of the computer device is used to connect and communicate with the terminal. Those skilled in the art can understand, Figure 4The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0132] In this embodiment, the processor is used to execute Figure 3 the specific functions of each module in. The memory stores the program codes and various types of data required to execute the above modules. The network interface is used for data transmission between the user terminal and the server. The memory in this embodiment stores the program codes and data required to execute all modules in the intelligent factory model dataset security and privacy protection device of this application, and the server can call the program codes and data of the server to execute the functions of all modules.

[0133] This application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to execute the steps of the intelligent factory model dataset security and privacy protection method according to any embodiment of this application.

[0134] This application also provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by one or more processors, the steps of the intelligent factory model dataset security and privacy protection method according to any embodiment of this application are implemented.

[0135] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments of this application can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0136] The above are only some embodiments of this application. It should be noted that for those of ordinary skill in the art, without departing from the principle of this application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of this application.

Claims

1. A method for protecting the security and privacy of an intelligent factory model dataset, characterized in that, Including: Obtain the model data sets after local training in each production workshop in the intelligent factory, where the model data sets represent the production environment data of a certain production workshop in the intelligent factory at a specific moment; Call the preset locality-sensitive hashing algorithm based on information entropy, project each data component in the model data set onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model data set according to the hash bits corresponding to each data component; Use the preset information entropy formula to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code, sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the column index set corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the column index set to determine the second hash code corresponding to the model data set; Use the preset Paillier homomorphic encryption algorithm to encrypt the model data set to determine the homomorphic ciphertext corresponding to the model data set, and upload the second hash code corresponding to the model data set and the homomorphic ciphertext corresponding to the model data set to the central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model data set.

2. The method for protecting the security and privacy of the intelligent factory model data set according to claim 1, characterized in that, The step of calling the preset locality-sensitive hashing algorithm based on information entropy, projecting each data component in the model data set onto each hash function to determine the hash bits corresponding to each data component, and constructing the first hash code corresponding to the model data set includes: Obtain the model dataset X = [x1, x2, …, x n ∈ R n×d , where xi i represents the i-th data component in the model dataset X, d represents the number of model parameters, and the information entropy-based locality-sensitive hashing algorithm uses a family of hash functions F = {f1, f2, …, f L} ∈ R d×L , each of whose hash functions is randomly generated, and the number of hash functions is L. Then the hash bit of the data component x is defined as: Among them, h(x) represents the result of the data component x under the hash function. Each data component x generates a binary vector H(x) = {h1(x), h2(x),..., h L (x)} ∈ {0, 1} L ; Project each data component \(x\) in the model data set \(X\) onto each hash function in the hash function family \(F\) to determine the hash bits corresponding to each data component, and combine the hash bits corresponding to each data component to form the first hash code corresponding to the model data set. The first hash code corresponding to the model data set \(X\) is a binary matrix \(B\), which is expressed as: Among them, the binary matrix B describes the hash bits of the model data set X, where h i ∈ {0, 1} n×1 represents the value of all data components in the model data set X at the i-th hash bit of the binary matrix B.

3. The method for protecting the security and privacy of the intelligent factory model data set according to claim 2, wherein, The step of using the preset information entropy formula to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code, sorting the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest to determine the column index set corresponding to the hash bits of the preset hash code length, and performing dimensionality reduction on the hash bits corresponding to the column index set to determine the second hash code corresponding to the model data set includes: The expression of the information entropy formula is: Among them, p(h i ) represents the probability distribution of the value of hash bit h i . In the hash bit h i , m = 2, which means that the value of the element in the hash bit h i is 0 or 1. By calculating the information entropy score corresponding to each hash bit, it is denoted as score ∈ R L ; Sort the information entropy scores corresponding to all hash bits in the first hash code from largest to smallest, and select the top \(k\) hash bits with the highest scores from largest to smallest to construct the column index set \(D\), where \(k\) represents the preset hash code length, and the dimensionality reduction means determining a low-dimensional second hash code by removing the hash bits in the column index set \(D\).

4. The method for protecting the security and privacy of the intelligent factory model data set according to claim 1, wherein, The step of using the preset Paillier homomorphic encryption algorithm to encrypt the model data set to determine the homomorphic ciphertext corresponding to the model data set includes: Obtain the model dataset X = [x1, x2, …, x n ∈ R n×d , where x i represents the i-th data component in the model dataset X, and d represents the number of model parameters; Call the preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ); Based on the Paillier encryption formula, for each data component x i ∈ X, a random number r i is randomly generated, and the ciphertext c i is calculated using the public key (n, g) to determine the encrypted ciphertext vector c = [c1, c2,..., c d .

5. The method for protecting the security and privacy of the intelligent factory model data set according to claim 4, wherein, The step of calling the preset Paillier homomorphic encryption algorithm to generate a public-private key pair, where the public-private key pair includes a public key (n, g) and a private key (λ, μ), includes: Select a first prime number p and a second prime number q, calculate and determine n = p × q and λ = lcm(p - 1, q - 1), where λ = lcm(p - 1, q - 1) represents the least common multiple of p - 1 and q - 1; Select a base Satisfy gcd(L(g λ mod n 2 ), n) = 1, where Calculate the private key parameter μ = (L(g λ mod n 2 )) -1 mod n to determine the public key (n, g) and the private key (λ, μ).

6. The method for protecting the security and privacy of the intelligent factory model data set according to claim 1, wherein The expression of the Paillier encryption formula is: where x i is the i-th data component in the model data set X, and r i is a random number for each encryption, which satisfies 7. The method for protecting the security and privacy of the intelligent factory model data set according to any one of claims 1 to 6, characterized in that, The production environment data includes sensor data, device status operation data, and production demand data.

8. An intelligent factory model dataset security and privacy protection device, characterized in that, Including: A production data acquisition module, configured to acquire the model data sets after local training in each production workshop in the intelligent factory, where the model data sets represent the production environment data of a certain production workshop in the intelligent factory at a specific moment; A first hash code determination module, configured to call the preset locality-sensitive hashing algorithm based on information entropy, project each data component in the model data set onto each hash function to determine the hash bits corresponding to each data component, and construct the first hash code corresponding to the model data set according to the hash bits corresponding to each data component; A second hash code determination module, configured to calculate and determine the information entropy scores corresponding to each hash bit in the first hash code by using the preset information entropy formula, sort the information entropy scores corresponding to all hash bits in the first hash code from large to small to determine the column index set corresponding to the hash bits of the preset hash code length, and perform dimensionality reduction on the hash bits corresponding to the column index set to determine the second hash code corresponding to the model data set; A security and privacy protection module, configured to encrypt the model data set by using the preset Paillier homomorphic encryption algorithm to determine the homomorphic ciphertext corresponding to the model data set, and upload the second hash code corresponding to the model data set and the homomorphic ciphertext corresponding to the model data set to the central aggregation server in the intelligent factory to complete the security and privacy protection of the intelligent factory model data set.

9. An electronic device, comprising a central processing unit and a memory, characterized in that, The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program implemented according to the method described in any one of claims 1 to 7 in the form of computer-readable instructions. When the computer program is called and run by the computer, it executes the steps included in the corresponding method.

Images