CN116502237B - Digital twin platform security collaboration method and system - Google Patents
Digital twin platform security collaboration method and system Download PDFInfo
- Publication number
- CN116502237B CN116502237B CN202310748377.4A CN202310748377A CN116502237B CN 116502237 B CN116502237 B CN 116502237B CN 202310748377 A CN202310748377 A CN 202310748377A CN 116502237 B CN116502237 B CN 116502237B
- Authority
- CN
- China
- Prior art keywords
- model
- gradient
- digital twin
- message array
- twin platform
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000005520 cutting process Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 35
- 238000012549 training Methods 0.000 claims description 17
- 238000003860 storage Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000005540 biological transmission Effects 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 13
- 238000004590 computer program Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 9
- 230000007246 mechanism Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000013461 design Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000002776 aggregation Effects 0.000 description 3
- 238000004220 aggregation Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/50—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems
- G06F21/57—Certifying or maintaining trusted computer platforms, e.g. secure boots or power-downs, version controls, system software checks, secure updates or assessing vulnerabilities
- G06F21/577—Assessing vulnerabilities and evaluating computer system security
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Bioethics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Mathematical Physics (AREA)
- Storage Device Security (AREA)
- Alarm Systems (AREA)
Abstract
The invention relates to the technical field of digital twinning, and provides a digital twinning platform safety cooperation method and system. The digital twin platform safety cooperation method is applied to a system comprising a digital twin platform and a plurality of side terminals, and comprises the following steps: obtaining model gradient data of a local model of an edge terminal and a trust identification of the edge terminal; determining whether to perform gradient clipping on the model gradient data according to the trust identification; randomly disturbing and randomly scrambling the model gradient values or the model gradient values after gradient cutting to obtain a shuffling result; and updating model parameters according to the shuffling result. The embodiment of the invention improves the safety and privacy of data transmission.
Description
Technical Field
The invention relates to the technical field of digital twinning, in particular to a digital twinning platform safety cooperation method, a digital twinning platform safety cooperation system, electronic equipment and a corresponding storage medium.
Background
With the advancement of national energy strategy, the power grid scale is rapidly expanding, and the following demands are rapidly growing: new energy, energy storage, lean management of power grid mass equipment, full consumption of distributed power, efficient access of novel loads of electric vehicles and the like, power grid and user-friendly interaction and the like. Along with the collection of massive user side data, at present, a fusion terminal, an energy controller, a concentrator, an edge internet of things agent, an intelligent gateway and other multi-type side terminals interact with devices such as an intelligent switch, a photovoltaic grid-connected controller, an electric vehicle charging pile controller, a user ammeter and the like on the same user side, and the side terminals control and schedule information communication with a digital twin cloud platform. Although the side terminals collect information of different devices on the same user side, the side terminals usually belong to different management departments, data barriers exist between the side terminals for isolation, and each side terminal locally reserves original user data collected by the corresponding device. Information islands exist among different data owners, and in order to obtain a model with better performance, the data must be integrated together, but the problems of information security and data privacy protection cannot be ignored.
Federal learning requires that all party data must be stored locally, model aggregation is performed through interactive model parameters, and federal learning design can solve the problem of partial data island, but the federal learning design has the defects of a plurality of vulnerabilities, weaknesses and the like, and the existing security holes of the federal learning design can be utilized by internal participants and external attackers to destroy the security of the federal learning system.
Disclosure of Invention
The embodiment of the invention aims to provide a digital twin platform security collaboration method and system, designs a trust mechanism framework based on mixed differential privacy, and optimally designs a novel differential privacy algorithm for a local hash coding technology and the like so as to at least solve part of problems in the background technology.
In order to achieve the above object, a digital twin platform security collaboration method is applied to a system including a digital twin platform and a plurality of side terminals, and the method includes: obtaining model gradient data of a local model of an edge terminal and a trust identification of the edge terminal; determining whether to perform gradient clipping on the model gradient data according to the trust identification; randomly disturbing and randomly scrambling the model gradient values or the model gradient values after gradient cutting to obtain a shuffling result; and updating model parameters according to the shuffling result.
Preferably, acquiring model gradient data of a local model of the side terminal includes: the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters; all the side terminals perform local model training to obtain a trained local model; and determining model gradient data according to the trained local model.
Preferably, determining whether to perform gradient clipping on the model gradient data according to the trust identification comprises: when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and when the trust is identified as trusted, not performing gradient clipping on the model gradient data.
Preferably, the random perturbation is performed by the side terminal and the random scrambling is performed by the digital twin platform.
Preferably, the random perturbation is performed by the side terminal, comprising: randomly selecting a hash algorithm from a preset hash algorithm set; calculating the model gradient value or the model gradient value after gradient clipping by adopting the selected hash algorithm to obtain a hash value; and calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value according to the disturbance probability to obtain a disturbance result.
Preferably, the random scrambling is performed by the digital twin platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; and taking the message arrangement result after the circulation is finished as the calculation result of the random scrambling.
The invention also provides a digital twin platform safety cooperation system, which comprises a digital twin platform and a plurality of side terminals, and further comprises: the data acquisition module is used for acquiring model gradient data of a local model of the side terminal and trust identification of the side terminal; the selecting and cutting module is used for determining whether to perform gradient cutting on the model gradient data according to the trust identification; the privacy algorithm module is used for randomly disturbing and randomly scrambling the model gradient value or the model gradient value after gradient cutting to obtain a shuffling result; and a model updating module for updating model parameters according to the shuffling result.
Preferably, acquiring model gradient data of a local model of the side terminal includes: the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters; all the side terminals perform local model training to obtain a trained local model; and determining model gradient data according to the trained local model.
Preferably, determining whether to perform gradient clipping on the model gradient data according to the trust identification comprises: when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and when the trust is identified as trusted, not performing gradient clipping on the model gradient data.
Preferably, the random perturbation is performed by the side terminal and the random scrambling is performed by the digital twin platform.
Preferably, the random perturbation is performed by the side terminal, comprising: randomly selecting a hash algorithm from a preset hash algorithm set; calculating the model gradient value or the model gradient value after gradient clipping by adopting the selected hash algorithm to obtain a hash value; and calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value according to the disturbance probability to obtain a disturbance result.
Preferably, the random scrambling is performed by the digital twin platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; and taking the message arrangement result after the circulation is finished as the calculation result of the random scrambling.
The invention also provides an electronic device, comprising: at least one processor; a memory coupled to the at least one processor; the memory stores instructions executable by the at least one processor, and the at least one processor implements the steps of the digital twin platform security collaboration method by executing the instructions stored by the memory.
There is also provided in the present invention a machine-readable storage medium having stored thereon instructions that when executed by a processor cause the processor to be configured to perform the steps of implementing the aforementioned digital twin platform security collaboration method.
There is also provided in the present invention a computer program product comprising a computer program which, when executed by a processor, implements the steps of the digital twin platform security collaboration method described previously.
The technical scheme has the following beneficial effects:
(1) The trust mechanism framework based on the mixed differential privacy can perform gradient calculation according to the user type, and complete the common training of multiple data sources on the joint model on the premise of guaranteeing the data information safety of each side terminal.
(2) The novel differential privacy algorithm is optimally designed for the local hash coding technology and the like, so that the safety of data transmission is improved, and the problem that the user privacy and analysis precision are difficult to be simultaneously considered in the existing method is solved.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain, without limitation, the embodiments of the invention. In the drawings:
FIG. 1 schematically illustrates a step schematic of a digital twin platform security collaboration method according to an embodiment of the present invention;
FIG. 2 schematically illustrates an architectural diagram of a system including a digital twinning platform and a number of edge terminals in accordance with an embodiment of the present invention;
FIG. 3 schematically illustrates an implementation of a digital twin platform security collaboration method in accordance with an embodiment of the present invention;
FIG. 4 schematically illustrates a flow diagram of a trust mechanism for hybrid differential privacy in accordance with an embodiment of the present invention;
fig. 5 schematically illustrates a structural diagram of a digital twin platform security collaboration system according to an embodiment of the present invention.
Detailed Description
The following describes the detailed implementation of the embodiments of the present invention with reference to the drawings. It should be understood that the detailed description and specific examples, while indicating and illustrating the invention, are not intended to limit the invention.
Fig. 1 schematically shows a step schematic of a digital twin platform security collaboration method according to an embodiment of the present invention. As shown in fig. 1, a digital twin platform security collaboration method is applied to a system including a digital twin platform and a plurality of side terminals, and the method includes:
s01, obtaining model gradient data of a local model of an edge terminal and a trust identification of the edge terminal;
s02, determining whether to perform gradient clipping on the model gradient data according to the trust identification;
s03, randomly disturbing and randomly scrambling the model gradient value or the model gradient value after gradient cutting to obtain a shuffling result; in this step, the model gradient value is selected or the model gradient value after gradient clipping is determined according to step S02. And if gradient clipping is not performed according to the trust mark, the processing object is a model gradient value. And if gradient clipping is carried out according to the trust mark, the processing object is a model gradient value after gradient clipping.
S04, updating model parameters according to the shuffling result. In federal learning, the updated model parameters here are gradient parameters of the global model in the updated data twinning platform.
Through the embodiment, the side terminals are distinguished through the trust identification, so that the gradient influence of the non-trust side terminals on the aggregation model is reduced. And privacy enhancement of user data, especially gradient data, in the system transmission process is realized through random disturbance and random scrambling.
On the basis of the federal learning framework, the embodiment realizes the privacy protection of federal learning by applying a differential privacy mechanism, and a novel differential privacy algorithm is designed to reduce the overall privacy loss and the accuracy loss of the global model. Fig. 2 schematically shows an architecture diagram of a system comprising a digital twin platform and several edge terminals in an embodiment according to the invention. As shown in fig. 2, different devices belonging to the same user have the same user index ID, but different devices have different attribute information. The attribute information owned between any two devices is different. Specifically, for m side terminals installed in the same area, the side terminal 1 collects a pieces of attribute information of a-station devices of n users in the area, the side terminal 2 collects b pieces of attribute information of b-station devices of n users in the area, and the rest side terminals are analogically available. M terminals at the side are used as data owners) The model is intended to be co-trained based on n users for final use in the area distribution service model.
The differential privacy mechanism comprising a random disturbance algorithm and a random ordering algorithm in the embodiment of the invention mainly comprises the following parts:
and (3) local calculation: the side terminal i is based on a local databaseGlobal model broadcasted by digital twin cloud platformw t G As local training parameters, i.e.w t i =w t G Performing local model training to obtainw t+1 i 。（tRepresenting the current communication round
Local disturbance: the side terminal adopts the random disturbance algorithm designed by the invention to disturb the side terminal data model.
Arranging models: and a shuffler module is designed in the digital twin platform, so that random replacement of data uploaded by the side terminal is completed, a dual privacy amplification effect is realized, and a shuffled result is output.
Model aggregation: the digital twin platform weight aggregate updates the global model.
In some optional embodiments of the present invention, obtaining model gradient data of a local model of an edge terminal includes: the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters; all the side terminals perform local model training to obtain a trained local model; and determining model gradient data according to the trained local model. Federal learning is a distributed machine learning model, essentially by co-training a global model on behalf of all user devices with multiple user devices. And after receiving the global model parameters, the side terminal obtains a local model through local training and obtains model gradient data corresponding to the local model. And carrying out corresponding treatment on the model gradient in the subsequent step.
In some optional embodiments of the invention, determining whether to gradient clip the model gradient data according to the trust flag comprises: when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and when the trust is identified as trusted, not performing gradient clipping on the model gradient data. In an actual application scene, a new side terminal installed for reasons of equipment debugging, test point verification and the like appears on site, when a digital twin platform cannot judge that the newly accessed side terminal is a new user or a third party, the side terminal is divided into a trusted user and an untrusted user and marked by adopting a trusted identifier, and the system carries out respective processing according to a trusted mechanism framework provided by the embodiment of the invention. The specific gradient clipping mode is implemented by referring to the prior art, and the invention does not relate to the improvement of the specific gradient clipping method. The embodiment provides gradient clipping for the non-trusted side terminal, so that gradient explosion in a system caused by malicious side terminal access is avoided.
In some alternative embodiments of the invention, the random perturbation is performed by the edge terminal and the random scrambling is performed by the digital twinning platform. Both random perturbation and random scrambling are ways to mask the true value. Random perturbation may be performed locally using existing OLH (Optimized Local Hashing) methods. Random scrambling mostly employs a shuffling algorithm such as Fei Xueye z conventional random permutation algorithm to shuffle messages. The embodiment realizes the privacy enhancement of the user data by separating the random disturbance from the random scrambling execution main body. Fig. 3 schematically shows an implementation diagram of a digital twin platform security collaboration method according to an embodiment of the present invention. As shown in fig. 3, the present embodiment adds a local perturbation module in the side terminal, for executing the foregoing random perturbation. A shuffler module is designed in the digital twinning platform for performing the random scrambling described previously.
In some optional embodiments of the invention, the random perturbation is performed by the edge terminal, comprising the steps of: gradient data terminating with sidev i ，i∈{1,2,…,dFor example, thev i Possibly after gradient clipping, randomly selecting a hash algorithm from a preset hash algorithm set, and collectingCalculating the gradient data by using the selected hash algorithm to obtain a hash value; comprising the following steps: at a given pointv i Under the Hash familyRandomly selecting a hash functionH i For a pair ofv i Encoding to obtain
；
Wherein the value range of the hash function is Ω, |Ω|=g。
Calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value by using the disturbance probability to obtain a disturbance result, wherein the method comprises the following steps: for datav i Corresponding hash addressPerforming local disturbance to obtain->：
Where w.p. denotes "probability",γis constant, finally obtain the firstiDisturbance results for individual users。
In some alternative embodiments of the invention, the random scrambling is performed by the digital twinning platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; the message arrangement result after the loop is finished is used as the calculation result of the random scrambling, and can be implemented by the following steps.
Step 1: splicing messages sent by all users to generate a message array, and supposing that the message array has r elements; the main execution body of this step is preferentially a shuffler module.
Step 2: setting the circulation times from 1, randomly taking the remainder of the last element of the array (the serial number of the last element is r in the first circulation), wherein the divisor is r+1, the dividend is the circulation times 3, and exchanging the element corresponding to the obtained serial number with the last element;
step 3: the last element is not regarded as the element in the message array any more, and the loop of the step 2 is continued with the new message array;
step 4: the loop is ended until only the first element is left, and a new message arrangement result is generated.
FIG. 4 schematically illustrates a flow diagram of a trust mechanism for hybrid differential privacy in accordance with an embodiment of the present invention. As shown in fig. 4, it mainly comprises the following steps:
step 1: initializing a model and parameters, including a server global model, privacy calculation parameters, and gradient calculation related parameters, wherein the gradient calculation related parameters include: gradient initial clipping value, convergence threshold, target loss function, etc.;
step 2: digital twin platform (server side) will global model parametersAnd an initial cut value +.>Sending the data to each side terminal;
step 3: all the side terminals train own models locally and prepare to upload new model gradient values;
step 4: for a trusted user, the trusted user transmits the model gradient value to the digital twin platform after adopting the random disturbance and random scrambling;
step 5-7: for an untrusted user, uploading gradient estimation information X of the round by the untrusted user, recalculating a gradient cut-off value by the digital twin platform, and transmitting the gradient cut-off value to the untrusted user, wherein the untrusted user transmits the model gradient value to the digital twin platform after adopting the random disturbance and random scrambling;
step 8: the digital twin platform will average the user gradients and update the model parameters.
The above embodiment provides a trust mechanism framework, according to which the trusted user and the untrusted user of the side terminal respectively process, so as to improve the security of the system.
Based on the same inventive concept, the embodiment of the invention also provides a digital twin platform safety cooperation system. Fig. 5 schematically illustrates a structural diagram of a digital twin platform security collaboration system according to an embodiment of the present invention. As shown in fig. 5, the system includes a digital twin platform and a plurality of side terminals, and further includes: the data acquisition module is used for acquiring model gradient data of a local model of the side terminal and trust identification of the side terminal; the selecting and cutting module is used for determining whether to perform gradient cutting on the model gradient data according to the trust identification; the privacy algorithm module is used for randomly disturbing and randomly scrambling the model gradient value or the model gradient value after gradient cutting to obtain a shuffling result; and a model updating module for updating model parameters according to the shuffling result.
In some alternative embodiments, obtaining model gradient data for a local model of an edge termination includes: the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters; all the side terminals perform local model training to obtain a trained local model; and determining model gradient data according to the trained local model.
In some alternative embodiments, determining whether to gradient clip the model gradient data according to the trust identification comprises: when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and when the trust is identified as trusted, not performing gradient clipping on the model gradient data.
In some alternative embodiments, the random perturbation is performed by the edge side terminal and the random scrambling is performed by the digital twinning platform.
In some alternative embodiments, the random perturbation is performed by the edge terminal, comprising: randomly selecting a hash algorithm from a preset hash algorithm set; calculating the model gradient value or the model gradient value after gradient clipping by adopting the selected hash algorithm to obtain a hash value; and calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value according to the disturbance probability to obtain a disturbance result.
In some alternative embodiments, the random scrambling is performed by the digital twinning platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; and taking the message arrangement result after the circulation is finished as the calculation result of the random scrambling.
The specific limitation of each functional module in the digital twin platform security collaboration system can be referred to above for the limitation of the digital twin platform security collaboration method, and will not be repeated here. The various modules in the system described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In some embodiments of the present invention, there is also provided an electronic device including: at least one processor; a memory coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to perform the steps of the digital twin platform security collaboration method described above. The control module or processor herein has the functions of numerical computation and logical operation, and has at least a central processing unit CPU, a random access memory RAM, a read only memory ROM, various I/O ports, an interrupt system, and the like, which have data processing capabilities. The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel may be provided with one or more of the methods described above by adjusting the kernel parameters. The memory may include volatile memory, random Access Memory (RAM), and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip.
In one embodiment of the present invention, a machine-readable storage medium is provided having instructions stored thereon that, when executed by a processor, cause the processor to be configured to perform the steps of the digital twin platform security collaboration method described previously.
In one embodiment of the present invention, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the digital twin platform security collaboration method described above.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises an element.
The foregoing is merely exemplary of the present invention and is not intended to limit the present invention. Various modifications and variations of the present invention will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the invention are to be included in the scope of the claims of the present invention.
Claims (8)
1. The digital twin platform safety cooperation method is characterized by being applied to a system comprising a digital twin platform and a plurality of side terminals, and comprises the following steps:
obtaining model gradient data of a local model of an edge terminal and a trust identification of the edge terminal;
determining whether to perform gradient clipping on the model gradient data according to the trust identification;
randomly disturbing and randomly scrambling the model gradient values or the model gradient values after gradient cutting to obtain a shuffling result; the random scrambling is performed by the side terminal, and the random scrambling is performed by the digital twin platform;
the random perturbation is performed by the edge side terminal, comprising: randomly selecting a hash algorithm from a preset hash algorithm set; calculating the model gradient value or the model gradient value after gradient clipping by adopting the selected hash algorithm to obtain a hash value; calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value according to the disturbance probability to obtain a disturbance result;
the random scrambling is performed by the digital twin platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; taking the message arrangement result after the circulation is finished as the randomly scrambled calculation result;
and updating model parameters according to the shuffling result.
2. The method of claim 1, wherein obtaining model gradient data for a local model of the edge termination comprises:
the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters;
all the side terminals perform local model training to obtain a trained local model;
and determining model gradient data according to the trained local model.
3. The method of claim 1, wherein determining whether to gradient clip the model gradient data based on the trust identification comprises:
when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and
and when the trust mark is trust, performing gradient clipping on the model gradient data.
4. The digital twin platform safety cooperation system is characterized by comprising a digital twin platform and a plurality of side terminals, and further comprising:
the data acquisition module is used for acquiring model gradient data of a local model of the side terminal and trust identification of the side terminal;
the selecting and cutting module is used for determining whether to perform gradient cutting on the model gradient data according to the trust identification;
the privacy algorithm module is used for randomly disturbing and randomly scrambling the model gradient value or the model gradient value after gradient cutting to obtain a shuffling result; the random scrambling is performed by the side terminal, and the random scrambling is performed by the digital twin platform;
the random perturbation is performed by the edge side terminal, comprising: randomly selecting a hash algorithm from a preset hash algorithm set; calculating the model gradient value or the model gradient value after gradient clipping by adopting the selected hash algorithm to obtain a hash value; calculating disturbance probability according to the value range of the hash value, and carrying out disturbance on the hash value according to the disturbance probability to obtain a disturbance result;
the random scrambling is performed by the digital twin platform, comprising: splicing the received messages into a message array; determining an element to be exchanged from the message array according to the number of the last element in the message array, the current cycle number and a random remainder algorithm, and exchanging with the last element; taking the elements except the last element in the message array as a new message array, and taking the new message array as a message array to be processed in the next cycle; repeating the steps of determining the elements to be exchanged, exchanging and determining a new message array, and ending the cycle when the new message array has only one element; taking the message arrangement result after the circulation is finished as the randomly scrambled calculation result; and
and the model updating module is used for updating model parameters according to the shuffling result.
5. The system of claim 4, wherein obtaining model gradient data for a local model of the edge termination comprises:
the method comprises the steps that a side terminal obtains global model parameters and initial cutoff values issued by a digital twin platform, and the global model parameters are used as local training parameters;
all the side terminals perform local model training to obtain a trained local model;
and determining model gradient data according to the trained local model.
6. The system of claim 4, wherein determining whether to gradient clip the model gradient data based on the trust identification comprises:
when the trust mark is not trusted, performing gradient clipping on the model gradient data by the digital twin platform; and
and when the trust mark is trust, performing gradient clipping on the model gradient data.
7. An electronic device, comprising: at least one processor;
a memory coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor, the at least one processor implementing the steps of the digital twin platform security collaboration method of any one of claims 1 to 3 by executing the instructions stored by the memory.
8. A machine-readable storage medium having instructions stored thereon that when executed by a processor cause the processor to be configured to implement the digital twin platform security collaboration method of any of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310748377.4A CN116502237B (en) | 2023-06-25 | 2023-06-25 | Digital twin platform security collaboration method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310748377.4A CN116502237B (en) | 2023-06-25 | 2023-06-25 | Digital twin platform security collaboration method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116502237A CN116502237A (en) | 2023-07-28 |
| CN116502237B true CN116502237B (en) | 2023-11-17 |
Family
ID=87323390
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310748377.4A Active CN116502237B (en) | 2023-06-25 | 2023-06-25 | Digital twin platform security collaboration method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116502237B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112367109A (en) * | 2020-09-28 | 2021-02-12 | 西北工业大学 | Incentive method for digital twin-driven federal learning in air-ground network |
| CN114125785A (en) * | 2021-11-18 | 2022-03-01 | 清华大学 | Low-delay high-reliability transmission method, device, equipment and medium for digital twin network |
| CN114945044A (en) * | 2022-07-25 | 2022-08-26 | 北京智芯微电子科技有限公司 | Method, device and equipment for constructing digital twin platform based on federal learning |
| CN115310360A (en) * | 2022-08-16 | 2022-11-08 | 天津大学 | Digital twin auxiliary industrial Internet of things reliability optimization method based on federal learning |
| CN116167084A (en) * | 2023-02-24 | 2023-05-26 | 北京工业大学 | Federal learning model training privacy protection method and system based on hybrid strategy |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3809220B1 (en) * | 2019-10-14 | 2023-01-18 | Honda Research Institute Europe GmbH | Method and system for semi-supervised deep anomaly detection for large-scale industrial monitoring systems based on time-series data utilizing digital twin simulation data |
-
2023
- 2023-06-25 CN CN202310748377.4A patent/CN116502237B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112367109A (en) * | 2020-09-28 | 2021-02-12 | 西北工业大学 | Incentive method for digital twin-driven federal learning in air-ground network |
| CN114125785A (en) * | 2021-11-18 | 2022-03-01 | 清华大学 | Low-delay high-reliability transmission method, device, equipment and medium for digital twin network |
| CN114945044A (en) * | 2022-07-25 | 2022-08-26 | 北京智芯微电子科技有限公司 | Method, device and equipment for constructing digital twin platform based on federal learning |
| CN115310360A (en) * | 2022-08-16 | 2022-11-08 | 天津大学 | Digital twin auxiliary industrial Internet of things reliability optimization method based on federal learning |
| CN116167084A (en) * | 2023-02-24 | 2023-05-26 | 北京工业大学 | Federal learning model training privacy protection method and system based on hybrid strategy |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116502237A (en) | 2023-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110942154B (en) | Data processing method, device, equipment and storage medium based on federal learning | |
| CN111080288B (en) | Block chain consensus achieving method and device based on directed acyclic graph | |
| CN109246176B (en) | Multi-controller synchronization method and device based on block chain in software defined network | |
| CN110198213A (en) | A kind of system based on privacy sharing random number common recognition algorithm | |
| Yin et al. | A blockchain-based incremental update supported data storage system for intelligent vehicles | |
| CN114945044B (en) | Method, device and equipment for constructing digital twin platform based on federal learning | |
| CN105187218B (en) | A kind of digitized record signature, the verification method of multi-core infrastructure | |
| US10700934B2 (en) | Communication control device, communication control method, and computer program product | |
| CN113726525A (en) | Energy industry cloud network data tracing method and device | |
| CN112000993A (en) | Block chain-based data storage verification method, equipment and storage medium | |
| CN116502237B (en) | Digital twin platform security collaboration method and system | |
| CN112417478B (en) | Data processing method, device, equipment and storage medium | |
| CN111371560B (en) | Certificateless fault-tolerant aggregation signature method and system applied to Internet of vehicles | |
| CN117390448A (en) | Client model aggregation method and related system for inter-cloud federal learning | |
| CN114329127B (en) | Feature binning method, device and storage medium | |
| CN108683599B (en) | Preprocessing-based method and system for determining maximum flow of flow network | |
| CN114095246B (en) | Network access identity authentication method of power distribution terminal | |
| CN112332994B (en) | False information identification method, false information identification system, computer equipment and storage medium | |
| CN111291464A (en) | Dynamic equivalence method and device for power system | |
| CN114844719B (en) | Cross-network terminal identity authentication method, device and system of communication network | |
| CN116633704B (en) | Graph calculation method and device | |
| CN116401071B (en) | Resource allocation method and system for edge calculation | |
| CN114938349B (en) | Internet of things data processing method and device, computer equipment and storage medium | |
| US20240275621A1 (en) | Blockchain generation method and apparatus | |
| CN107368378A (en) | A kind of maps processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |