CN112507373B - A remote access method for industrial field data in the industrial Internet - Google Patents

Abstract

The invention discloses a remote access method for industrial field data in the industrial Internet, belongs to the technical field of data security, and is used for solving the problem of low security of existing remote data transmission encryption. The method includes: the client initiates an access request to access data of the data server to the intermediate server; the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server; in response to the access request, The data server performs a first transformation on the industrial field data requested to be accessed according to the first deep neural network, and obtains encrypted industrial field data and sends it back to the client through the intermediate server; the client encrypts the encrypted industrial field data according to the second deep neural network. The second transformation is performed on the obtained industrial field data to obtain decrypted industrial field data; wherein, the first transformation process and the second transformation process are mutually inverse. The present invention has higher encryption complexity and higher security.

Description

A remote access method for industrial field data in the industrial Internet

technical field

The invention relates to the technical field of data security, in particular to a remote access method for industrial field data in the industrial Internet.

Background technique

The essence and core of the Industrial Internet is to closely connect and integrate equipment, production lines, factories, suppliers, products and customers through the Industrial Internet platform. With the development of the Industrial Internet, remote access to industrial field data enables access to current networked field equipment data through remote clients, which saves manpower and time cost of running the field, and facilitates unified and centralized management of field data.

Remote Database Access (RDA) is a key technology to support the interconnection and interoperability of databases in the network environment. It is becoming mature in the current network technology, and the database has realized the application situation of the client/server architecture. RDA has become an important link in the construction of any information system. Common RDA services can be divided into five categories: dialog management services, transaction management services, control services, resource processing services, and database language services. The first two types of services are mainly related to database management, and the last three types of services are mainly related to database access. The Dialog Management Service provides facilities for managing RDA conversations, the Transaction Management Service supports the management of transactions, the Control Service is used to determine the status of pending RDA operations and cancel those operations, the Resource Handling Service is used to manage database resources, and the Database Language (DBL) Services involve defining and undoing DBL operations, calling previously defined operations, performing DBL operations, etc.

In the existing remote database access method, during the remote access of the client to the data on the server, the data needs to be encrypted for back and forth transmission. However, the traditional encryption method is to perform a linear transformation on the original data or a nonlinear transformation with an explicit expression. , its security is not high enough, and it is easy to be detected by attacks. If the security of the encryption algorithm is improved, the encryption and decryption costs are too high, and it is not suitable for a large amount of short-term data interaction.

SUMMARY OF THE INVENTION

The invention provides a remote access method for industrial field data in the industrial Internet, which is used to solve the problems of low data transmission encryption security or high encryption and decryption cost in the existing remote database access. The method for remotely accessing industrial field data in the industrial Internet provided by the present invention adopts a new encryption method. This encryption method is a non-display nonlinear transformation, and has higher complexity compared to the prior art when ensuring the same amount of calculation. and higher security.

The present invention provides a remote access method for industrial field data in the industrial Internet, comprising:

The client initiates an access request to the intermediate server to access the data of the data server;

The intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server;

In response to the access request, the data server performs a first transformation on the industrial field data requested to be accessed according to the first deep neural network, to obtain encrypted industrial field data and send it to the intermediate server;

The intermediate server transmits the encrypted industrial field data back to the client;

The client performs a second transformation on the encrypted industrial field data according to the second deep neural network to obtain decrypted industrial field data; wherein the first transformation process and the second transformation process are mutually inverse.

In an optional embodiment, the data server performs a first transformation on the industrial field data requested to be accessed according to the first deep neural network, and obtains encrypted industrial field data and sends it to the intermediate server, including:

Divide the industrial field data X requested to be accessed into two groups to obtain the first group of industrial field data X ₁ and the second group of industrial field data X ₂ ;

The first transformation is performed on the first group of industrial field data X ₁ and the second group of industrial field data X ₂ using the following formula:

和

作为加密后的工业现场数据发送给所述中间服务器；Output data obtained by first transforming the first group of industrial field data X ₁ and the second group of industrial field data X ₂

and

Send to the intermediate server as encrypted industrial field data;

为第一深度神经网络模型的中间输出；in,

is the intermediate output of the first deep neural network model;

X=(a _1,1 ,…,a _1,n ,a _2,1 ,…,a _2,n ,…, _am,1 ,…,am _,n ) ^T ∈R ^mn , (a _{j, 1} ,…,a _j,n ) is

The jth data in the industrial field data requested to be accessed, j=1,...,m; n is the number of parameters in each industrial field data, m is the total number of data copies in the industrial field data requested to be accessed ;

X ₁ =(a _1,1 ,…,a _1,n ,…,a _i,1 ,…,a _i,n ) ^T ∈R ⁱⁿ ,

X ₂ =(a _i+1,1 ,…,ai+1,n,…, _am,1 ,…,am _,n ) ^T ∈R ^(mi)n ,

Among them, the expression of the function m ⁽¹⁾ () is: m ⁽¹⁾ (x)=W ₂ ·σ(W ₁ x+b1)+b ₂ , the expression of the function m ⁽²⁾ () The formula is: m ⁽²⁾ (x)=W ₄ ·σ(W ₃ x+b ₃ )+b ₄ ,

The expression for this function σ() is:

W ₁ , W ₂ , W ₃ , W ₄ , b ₁ , b ₂ , b ₃ , and b ₄ are preset matrices, which are the intermediate parameters of the model;

Among them, W ₁ ∈R ^in×in indicates that W ₁ is an in×in-dimensional real matrix, W ₂ ∈R ^(mi)n×in indicates that W ₂ is a (mi)n×in-dimensional real matrix, and W ₃ ∈R ^{(mi )n×(mi)n} indicates that W ₃ is a (mi)n×(mi)n-dimensional real matrix, and W ₄ ∈R ^(mi)n×in indicates that W ₄ is a (mi)n×in-dimensional real matrix;

b ₁ , b ₄ ∈ R ⁱⁿ means b ₁ , b ₄ is an in-dimensional real vector; b ₂ , b ₃ ∈ R ^(mi)n means b ₂ , b ₃ is a (mi)n-dimensional real vector.

In an optional embodiment, the client performs a second transformation on the encrypted industrial field data according to the second deep neural network, and the obtained decrypted industrial field data includes:

和

进行第二变换：The client uses the following formula for the received encrypted industrial field data

and

Do the second transformation:

为第二深度神经网络模型的中间输出；in,

is the intermediate output of the second deep neural network model;

The results Y ₁ and Y ₂ output after the second transformation are combined to obtain the decrypted industrial field data.

In an alternative embodiment, i=[m/2], and [m/2] represents the largest integer not exceeding m/2.

In an optional embodiment, the method for remotely accessing industrial field data in the industrial Internet further includes:

Collect a certain amount of industrial field data in advance as sample data;

The first deep neural network is used to train the sample data to obtain the values of W ₁ , W ₂ , W ₃ , W ₄ and b ₁ , b ₂ , b ₃ , and b ₄ .

In an optional embodiment, the W ₁ , W ₂ , W ₃ , W ₄ , b ₁ , b ₂ , b ₃ , b ₄ obey a standard Gaussian distribution.

In an optional embodiment, the industrial field data includes at least sensory data and operational data.

In an optional embodiment, the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server, including:

When receiving the access request sent by the client, the intermediate server verifies the access authority of the logged-in user corresponding to the access request;

If the access authority verification of the login user corresponding to the access request passes, the intermediate server responds to the access request and establishes a communication connection relationship with the client and the data server respectively;

The intermediate server forwards the access request to the data server.

In an optional embodiment, the access rights of the logged-in user include, but are not limited to: compliance of the user account and access rights of the user account to different data.

In an optional embodiment, the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server, including:

In an optional embodiment, the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server, including:

Step A1, the intermediate server calculates the signal-to-noise ratio SINR value of the client according to its own transmission frequency; the SINR value is expressed as the ratio of the signal to the interference signal plus noise:

Among them, σ _i1 represents the SINR value of the i1th client, P _m1 represents the transmission frequency of the intermediate server, g _i1 represents the corresponding path gain when the intermediate server transmits signals to the i1th client, and ω _i1 represents as The corresponding path gain coefficient when the intermediate server transmits a signal to the i1th client, the value is [0.5, 0.8], N represents the total number of clients that send access requests to the intermediate server, and g _j1 represents the intermediate server to the intermediate server. The path gain corresponding to the j1th client transmitting signal, ω is the error factor in the calculation process, the value is [0.05, 0.1]; p _m is the average transmission frequency of the intermediate server;

Step A2, obtain the user data of the client according to the SINR value of the client:

Wherein, L _i1 represents the user data of the i1th client, log is the logarithmic calculation symbol, det represents the pre-designed parameter, and the value range is [2, 10], Q _S represents the current value of the intermediate server Bandwidth, R _max represents the preset maximum number of client connections of the intermediate server;

Step A3, input the user data of the client into the preset user database for retrieval, and confirm whether there is matching target user data, if so, continue to perform step A4; otherwise, issue "unable to connect to the server" to the client hint;

Step A4, issue the double prompt of password verification and face verification to the described client;

Step A5, verify whether the target password provided by the client is correct, and compare the current face image provided by the client with the preset face image of the client pre-recorded by the intermediate server. When the face image is the same as the preset face image and the target password verification is passed, confirm that the client identity information passes the verification and establish a communication connection relationship between the intermediate server, the client and the data server;

Step A6: After establishing the connection relationship, the intermediate server forwards the access request to the data server.

The method for remotely accessing industrial field data in the industrial Internet provided by the present invention provides a new data transmission encryption method for remote access to industrial field data in the industrial Internet based on a deep neural network. This encryption method is a reversible non-display expression. Compared with the prior art, the linear transformation only has the same amount of computation, but has higher complexity and is more difficult to be broken, and can well overcome the problems existing in the prior art.

Other features and advantages of the present invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description, claims, and drawings.

The technical solutions of the present invention will be further described in detail below through the accompanying drawings and embodiments.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification, and are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the attached image:

1 is a flowchart of a method for remotely accessing industrial field data in an industrial Internet provided by the present invention;

2 is a flowchart of a method for encrypting industrial field data requested to be accessed through a first deep neural network provided by the present invention;

FIG. 3 is a flowchart of a method for decrypting encrypted industrial field data through a second deep neural network provided by the present invention.

Detailed ways

The method for remotely accessing industrial field data in the industrial Internet provided by the embodiment of the present invention is used to encrypt and decrypt the industrial field data in the industrial Internet through a deep neural network during the remote access process. The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

FIG. 1 is a flowchart of a method for remote access to industrial field data in an industrial Internet provided by the present invention. As shown in FIG. 1 , the method includes the following steps S1-S5:

S1: The client initiates an access request to the intermediate server to access the data of the data server.

In this embodiment, similar to the existing remote access server method, the client sends an access request for the data server on the industrial site to the intermediate server, and the access request carries the client identifier, the data server identifier, and the data requested to be accessed. identification and other information. The intermediate server may be a virtual server, or other device terminals that can realize remote transfer.

S2: The intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server.

In an optional embodiment, when the intermediate server receives the access request sent by the client, it first verifies the access authority of the logged-in user corresponding to the access request, and if the verification is passed, the intermediate server responds to the access request, respectively. A communication connection relationship with the client and the data server is established, and finally the intermediate server forwards the access request to the data server.

Among them, the access rights of the logged-in user include but are not limited to: the compliance of the user account and the access rights of the user account to different data. For example, data of multiple devices is stored in the data server, and the access rights of different registered users (users with remote access compliance) to data of different devices can be preset. For example, user account A can view the industrial data of device one. Account B can view the industrial data of all devices, which will not be repeated here.

S3: In response to the access request, the data server performs a first transformation on the industrial field data requested to be accessed according to the first deep neural network, and obtains encrypted industrial field data and sends it to the intermediate server.

Among them, the industrial field data is mainly divided into two parts: one part is operation data, which is used to record mouse movements, operation commands, etc.; the other part is sensor data, which is used to indicate the working state of the machine, such as wind speed, pressure, temperature, humidity etc. observation data.

In this embodiment, the first deep neural network is preset, and after receiving the data access request, the data server performs the first transformation on the industrial field data requested to be accessed according to the first deep neural network, and the transformed data is the original Encrypted data used for transmission during remote access in the embodiment.

S4: The intermediate server transmits the encrypted industrial field data back to the client.

S5: The client performs a second transformation on the encrypted industrial field data according to the second deep neural network to obtain decrypted industrial field data;

Wherein, the first transformation process and the second transformation process are mutually inverse.

In this embodiment, after receiving the industrial field data encrypted by the first deep neural network, the client performs a second transformation on the encrypted industrial field data by using the second deep neural network, and the second deep neural network performs a second transformation on the data. The transformation process of the first deep neural network and the transformation process of the data by the first deep neural network are inverse to each other. After the second change, the encrypted industrial field data is decrypted into the original data of the data server, and the user can use the client to view it normally.

The method for remotely accessing industrial field data in the industrial Internet provided by this embodiment encrypts and decrypts data based on a deep neural network. This encryption method is a reversible nonlinear transformation with no explicit expression. Compared with the existing encryption technology, it has the advantages of It has the same amount of computation, but has higher complexity and is more difficult to break, which can well overcome the problems existing in the existing technology.

In an optional embodiment, as shown in FIG. 2 , the method for encrypting the industrial field data requested to be accessed by using the first deep neural network includes the following steps S31-S33:

S31: Divide the industrial field data X requested to be accessed into two groups, and obtain the first group of industrial field data X ₁ and the second group of industrial field data X ₂ .

In this embodiment, it is assumed that the industrial field data requested to be accessed is in the form of a matrix A∈Rm ^×n , that is,

First, the data in matrix form is flattened as X, as the input data of the first deep neural network:

X=(a ₁₁ ,…,a _1n ,a ₂₁ ,…,a _2n ,…,a _m1 ,…,a _mn ) ^T ∈R ^mn

Among them, (a _i1 ,...,a _in ) is the ith piece of data in the industrial field data requested to be accessed, i=1,...,m; n is the number of parameters in each piece of industrial field data, m is the number of parameters in each piece of industrial field data The total number of data copies in the industrial field data requested to be accessed; R ^mn indicates that there are m copies of industrial field data in the one-dimensional column vector X, and each industrial field data has n parameters, that is, there are m×n data in X.

In this embodiment, the first i×n data in X are divided into the first group of industrial field data X ₁ , and the last (mi)×n data in X are taken as the second group of industrial field data X ₂ , namely: Take the 1st to i-th pieces of data in the industrial field data requested to be accessed as the first group of industrial field data X ₁ (corresponding to the 1st to i-th rows in matrix A), (mi)～m pieces of data are taken as the second group of industrial field data X ₂ (corresponding to the (mi)～mth row in matrix A), and the expressions of X ₁ and X ₂ are as follows:

X ₁ =(a _1,1 ,…,a _1,n ,…,a _i,1 ,…,a _i,n ) ^T ∈R ⁱⁿ ,

X ₂ =(a _i+1,1 ,…,a _i+1,n ,…, _am,1 ,…,am _,n ) ^T ∈R ^(mi)n ,

Obviously, X ₁ is a (i×n)×1 matrix, and X ₂ is a ((mi)×n)×1 matrix.

Preferably, i=[m/2], and [m/2] represents the largest integer not exceeding m/2. For example, if A is a 5×2 matrix, and i=2, the resulting X ₁ is a 4×1 matrix, and X ₂ is a 6×1 matrix.

S32: Use the first predetermined formula (1) to perform a first transformation on the _first group of industrial field data X1 and the _second group of industrial field data X2:

为第一深度神经网络模型的中间输出；m⁽¹⁾()这一函数的表达式见公式(2)，m⁽²⁾()这一函数的表达式见公式(3)：in,

is the intermediate output of the first deep neural network model; the expression of m ⁽¹⁾ () is shown in formula (2), and the expression of m ⁽²⁾ () is shown in formula (3):

m ⁽¹⁾ (x)=W ₂ ·σ(W ₁ x+b ₁ )+b ₂ (2)

m ⁽²⁾ (x)=W ₄ ·σ(W ₃ x+b ₃ )+b ₄ (3)

In formulas (2) and (3), the expression of the function σ() is:

In equations (2) and (3), W ₁ ∈R ^in×in ,W ₂ ∈R ^(mi)n×in ,W ₃ ∈R ^(mi)n×(mi)n ,W ₄ ∈R ^{(mi) n×in} , b ₁ , b ₄ ∈ R ⁱⁿ , b ₂ , b ₃ ∈ R ^(mi)n . W ₁ ∈R ^in×in means that W ₁ is a real matrix of in×in dimension, W ₂ ∈ R ^(mi)n×in means that W ₂ is a real matrix of (mi)n×in dimension, for example, if m =5, n=2, i=2, then W ₁ is a 4×4-dimensional real matrix, W ₂ is a 6×4-dimensional real matrix, and W ₃ ∈R ^(mi)n×(mi)n represents W ₃ is a (mi)n×(mi)n-dimensional real matrix, W ₄ ∈R ^(mi)n×in means W ₄ is a (mi)n×in-dimensional real matrix; b ₁ ,b ₄ ∈R ⁱⁿ means b ₁ , b ₄ are in-dimensional real vectors; b ₂ , b ₃ ∈ R ^(mi)n represent b ₂ , b ₃ are (mi) n-dimensional real vectors.

Optionally, W ₁ , W ₂ , W ₃ , W ₄ , b ₁ , b ₂ , b ₃ , and b ₄ are preset matrices, which are used as encryption and decryption keys in the data transmission process provided by this embodiment of the present invention , W ₁ , W ₂ , W ₃ , W ₄ , the values of b ₁ , b ₂ , b ₃ , and b ₄ can be preset according to experience. Preferably, W ₁ , W ₂ , W ₃ , W ₄ , b ₁ , b ₂ , b ₃ , b ₄ obey a standard Gaussian distribution.

In an optional embodiment, before the method provided by the present invention is executed, a certain amount of industrial field data may be pre-collected as sample data, and then the first deep neural network is used to train the sample data to obtain the result. Describe the values of W ₁ , W ₂ , W ₃ , W ₄ and b ₁ , b ₂ , b ₃ , and b ₄ .

S33: Send the output data obtained by the first transformation of the first group of industrial field data X ₁ and the second group of industrial field data X ₂ to the intermediate server as encrypted industrial field data.

作为加密后的工业现场数据发送给所述中间服务器。至此，被请求访问的工业现场数据通过以公式(1)-(3)为代表的第一深度神经网络加密完成了。In this embodiment, the data server converts the output data obtained after the first transformation

It is sent to the intermediate server as encrypted industrial field data. So far, the industrial field data requested to be accessed is encrypted by the first deep neural network represented by formulas (1)-(3).

进行解密的方法包括如下步骤S51-S52：In an optional embodiment, as shown in FIG. 3, the encrypted industrial field data is processed by a second deep neural network.

The method for decrypting includes the following steps S51-S52:

和

进行第二变换：S51: Use the second predetermined formula (5) to perform the encryption on the received industrial field data

and

Do the second transformation:

为第二深度神经网络模型的中间输出；函数m⁽¹⁾()和m⁽²⁾()仍旧采用第一变换中使用的上述公式(2)-(4)来计算，显然，通过数学推导可得知，计算出的

等于采用公式(1)计算过程中得到的

计算出的

等于采用公式(1)计算过程中得到的

计算出的

等于采用公式(1)计算过程中得到的

计算出的

等于采用公式(1)计算过程中得到的

计算出的

等于采用公式(1)计算过程中得到的

计算出的

等于采用公式(1)计算过程中得到的

in,

is the intermediate output of the second deep neural network model; the functions m ⁽¹⁾ () and m ⁽²⁾ () are still calculated by the above formulas (2)-(4) used in the first transformation, obviously, through mathematical derivation It can be seen that the calculated

is equal to that obtained in the calculation process using formula (1)

calculated

is equal to that obtained in the calculation process using formula (1)

calculated

is equal to that obtained in the calculation process using formula (1)

calculated

is equal to that obtained in the calculation process using formula (1)

calculated

is equal to that obtained in the calculation process using formula (1)

calculated

is equal to that obtained in the calculation process using formula (1)

因此计算出的

根据公式(1)中的

可以得到

而

因此Y₂＝X₂。可见，第二深度神经网络输出的Y₁和Y₂等于一开始输入第一深度神经网络的第一组工业现场数据X₁和第二组工业现场数据X₂，通过上述过程即可实现对第一深度神经网络输出数据的解密。Since formula (1) is used in the calculation process,

So calculated

According to formula (1)

can get

and

Therefore Y ₂ =X ₂ . It can be seen that Y ₁ and Y ₂ output by the second deep neural network are equal to the first group of industrial field data X ₁ and the second group of industrial field data X ₂ input to the first deep neural network at the beginning. Decryption of the output data of a deep neural network.

S52: Combine the output results Y ₁ and Y ₂ after the second transformation to obtain decrypted industrial field data.

In this embodiment, according to the method of dividing the industrial field data X requested to be accessed into X ₁ and X ₂ in step S31, the results Y ₁ and Y ₂ output after the second transformation are recombined into X, so that the present invention can be obtained. The raw data of the industrial field data requested for access.

In an optional embodiment, the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server, including:

In an optional embodiment, the intermediate server establishes a communication connection relationship with the client and the data server, and forwards the access request to the data server, including:

Step A1, the intermediate server calculates the signal-to-noise ratio SINR value of the client according to its own transmission frequency; the SINR value is expressed as the ratio of the signal to the interference signal plus noise:

Among them, σ _i1 represents the SINR value of the i1th client, P _m1 represents the transmission frequency of the intermediate server, g _i1 represents the corresponding path gain when the intermediate server transmits signals to the i1th client, and ω _i1 represents as The corresponding path gain coefficient when the intermediate server transmits a signal to the i1th client, the value is [0.5, 0.8], N represents the total number of clients that send access requests to the intermediate server, and g ^j1 represents the intermediate server to the intermediate server. The path gain corresponding to the j1th client transmitting signal, ω is the error factor in the calculation process, the value is [0.05, 0.1]; p _m is the average transmission frequency of the intermediate server;

Step A2, obtain the user data of the client according to the SINR value of the client:

Wherein, L _i1 represents the user data of the i1th client, log is the logarithmic calculation symbol, det represents the pre-designed parameter, and the value range is [2, 10], Q _S represents the current value of the intermediate server Bandwidth, R _max represents the preset maximum number of client connections of the intermediate server;

Step A3, input the user data of the client into the preset user database for retrieval, and confirm whether there is matching target user data, if so, continue to perform step A4; otherwise, issue "unable to connect to the server" to the client hint;

Step A4, issue the double prompt of password verification and face verification to the described client;

Step A5, verify whether the target password provided by the client is correct, and compare the current face image provided by the client with the preset face image of the client pre-recorded by the intermediate server. When the face image is the same as the preset face image and the target password verification is passed, confirm that the client identity information passes the verification and establish a communication connection relationship between the intermediate server, the client and the data server;

Step A6: After establishing the connection relationship, the intermediate server forwards the access request to the data server.

The beneficial effects of the above technical solutions are: by obtaining the transmission frequency of the intermediate server and calculating the SINR value of the current client i1 to accurately determine the interference signal and noise in the transmission process, and then using the SINR value of the current client i1 to calculate the current The user data of the client i1 can avoid the disturbance of the interference signal, so that the obtained user data of the current client i1 is more accurate and true. The target user data to accurately preliminarily confirm whether the current client i1 has the authority to connect to the intermediate server, after confirming the authority, by further using the current face image and target password provided by the current client i1 to double-verify the identity, verify Only after the connection between the intermediate server and the data server can be realized, the security and reliability of the connection are ensured, and the security of the data in the data server is further ensured.

As will be appreciated by one skilled in the art, 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 having computer-usable program code embodied therein, including but not limited to disk storage, optical storage, and the like.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention. Thus, provided that these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims (8)

1. A remote access method for industrial field data in industrial Internet is characterized by comprising the following steps:

the client initiates an access request for accessing the data of the data server to the intermediate server;

the intermediate server establishes a communication connection relation with the client and the data server and forwards the access request to the data server;

responding to the access request, the data server performs first transformation on the industrial field data requested to be accessed according to a first deep neural network to obtain encrypted industrial field data, and the encrypted industrial field data is sent to the intermediate server;

the intermediate server transmits the encrypted industrial field data back to the client;

the client side carries out second transformation on the encrypted industrial field data according to a second deep neural network to obtain decrypted industrial field data; wherein the first transformation process and the second transformation process are reciprocal;

the method for the data server to perform the first transformation on the industrial field data requested to be accessed according to the first deep neural network to obtain the encrypted industrial field data and send the encrypted industrial field data to the intermediate server includes the following steps:

dividing the industrial field data X requested to be accessed into two groups to obtain a first group of industrial field data X₁And a second set of industrial field data X₂；

The following formula is adopted for the first group of industrial field data X₁And a second set of industrial field data X₂Performing a first transformation:

the first group of industrial field data X₁And a second set of industrial field data X₂Output data obtained by the first transformation

And

sending the encrypted industrial field data to the intermediate server;

wherein,

an intermediate output of the first deep neural network model;

X＝(a_1,1,…,a_1,n,a_2,1,…,a_2,n,…,a_m,1,…,a_m,n)^T∈R^mn，(a_j,1,…,a_j,n) J is 1, …, m in j-th data in the industrial field data requested to be accessed; n is the number of parameters in each piece of industrial field data, and m is the total number of data in the industrial field data requested to be accessed;

X₁＝(a_1,1,…,a_1,n,…,a_i,1,…,a_i,n)^T∈Rⁱⁿ，

X₂＝(a_i+1,1,…,a_i+1,n,…,a_m,1,…,a_m,n)^T∈R^(m-i)n，

wherein m is⁽¹⁾() The expression for this function is: m is⁽¹⁾(x)＝W₂·σ(W₁x+b₁)+b₂,m⁽²⁾() The expression for this function is: m is⁽²⁾(x)＝W₄·σ(W₃x+b₃)+b₄，

The expression of the function σ () is:

W₁,W₂,W₃,W₄，b₁,b₂,b₃,b₄the matrix is a preset matrix and is an intermediate parameter of the model;

wherein, W₁∈R^in×inRepresents W₁Is an in x in dimensional real matrix, W₂∈R^(m-i)n×inRepresents W₂Is (m-i) n × in dimension real matrix, W₃∈R^{(m-i)n×(m-i)n}Represents W₃Is (m-i) n x (m-i) n dimensional real matrix, W₄∈R^(m-i)n×inRepresents W₄Is (m-i) n x in dimension real matrix;

b₁,b₄∈R^ibdenotes b₁，b₄Is an in-dimensional real vector; b₂,b₃∈R^(m-i)nDenotes b₂，b₃Is (m-i) an n-dimensional real vector;

the client performs second transformation on the encrypted industrial field data according to a second deep neural network, and obtaining decrypted industrial field data includes:

the client side adopts the following formula to encrypt the received industrial field data

And

and carrying out a second transformation:

wherein,

is an intermediate output of the second deep neural network model;

the result Y output after the second transformation₁And Y₂And merging to obtain the decrypted industrial field data.

2. The method as claimed in claim 1, wherein i ═ m/2, [ m/2] denotes a maximum integer not exceeding m/2.

3. The method for remote access to industrial field data in an industrial internet as claimed in claim 1, further comprising:

acquiring a certain amount of industrial field data as sample data in advance;

training the sample data by adopting the first deep neural network to obtain the W₁,W₂,W₃,W₄And b₁,b₂,b₃,b₄The value of (c).

4. The method of claim 1, wherein W is a value of a threshold value of the remote access to the industrial field data in the industrial internet₁,W₂,W₃,W₄，b₁,b₂,b₃,b₄Obeying a standard gaussian distribution.

5. The method for remote access to industrial field data in an industrial internet as claimed in claim 1, wherein the industrial field data includes at least sensing data and operating data.

6. The method for remotely accessing industrial field data in the industrial internet as claimed in claim 1, wherein the intermediate server establishes a communication connection relationship with the client and the data server and forwards the access request to the data server, comprising:

when the intermediate server receives an access request sent by the client, verifying the access authority of a login user corresponding to the access request;

if the access authority of the login user corresponding to the access request passes the verification, the intermediate server responds to the access request and respectively establishes communication connection relations with the client and the data server;

the intermediate server forwards the access request to the data server.

7. The method for remotely accessing industrial field data in industrial internet as claimed in claim 6, wherein the access right of the login user includes but is not limited to: compliance of the user account and access rights of the user account to different data.

8. The method for remotely accessing industrial field data in the industrial internet as claimed in claim 1, wherein the intermediate server establishes a communication connection relationship with the client and the data server and forwards the access request to the data server, comprising:

step A1, the intermediate server calculates the signal-to-noise ratio (SINR) value of the client according to the self transmitting frequency; the SINR value is expressed as a ratio of signal to interference signal plus noise:

wherein σ_i1Expressed as SINR value, P, of the i1 th client_m1Expressed as the transmission frequency of the intermediate server, g_i1Expressed as the corresponding path gain, ω, when the intermediate server transmits a signal to said i1 th client_i1The path gain coefficient is expressed as a path gain coefficient corresponding to the signal transmitted by the intermediate server to the i1 th client, and the value is [0.5, 0.8 ]]N denotes the total number of clients sending access requests to the intermediate server, g_j1The path gain is expressed as the corresponding path gain when the intermediate server transmits a signal to the j1 th client, and ω is expressed as an error factor in the calculation process and takes the value of [0.05, 0.1%]；p_mIs the average transmit frequency of the intermediate server;

step a2, obtaining the user data of the client according to the SINR value of the client:

wherein L is_i1Representing the user data of the i1 th client, wherein log is a logarithm calculation symbol, det is a preset calculation parameter, and the value range is [2,10 ]]，Q_SExpressed as the current bandwidth, R, of the intermediate server_maxThe middle expressed as presetThe maximum number of client connections of the server;

step A3, inputting the user data of the client into a preset user database for retrieval, confirming whether matched target user data exists, and if yes, continuing to execute step A4; otherwise, sending a prompt of 'unable to connect to the server' to the client;

step A4, sending out double prompts of password authentication and face authentication to the client;

step A5, verifying whether the target password provided by the client is correct, comparing the current face image provided by the client with a preset face image of the client pre-recorded by an intermediate server, and when the current face image is confirmed to be the same as the preset face image and the target password is verified to be passed, confirming that the client identity information is verified to be passed and establishing the communication connection relation between the intermediate server and the client as well as between the intermediate server and a data server;

step A6, after the connection relationship is established, the intermediate server forwards the access request to the data server.

Images