A method for securely storing confidential data

Abstract

The present invention relates to the field of data processing, and more specifically, to a method for secure storage of confidential item data, the method comprising: obtaining a number of data points of confidential items; iteratively encrypting all data points to achieve secure storage; wherein, in one iterative encryption process, the method comprises: obtaining the deflection angle of the principal component; calculating the extraction probability according to the deviation angle of the principal component; screening the data points to be processed according to the extraction probability, adjusting the description vector and the modulus of the data points to be processed according to a preset encryption key; calculating the distribution difference, and if the distribution difference is greater than a preset threshold, the iteration ends. Through effective data encryption, the correlation of data points in different dimensions is destroyed, thereby improving the security of data.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and more specifically, to a method for securely storing confidential item data. Background Art

[0002] Some confidential item data contains some core information of confidential items, such as the design parameter data of a new product that has not yet been launched. Once this core information is leaked, it is easy to cause the confidential items to lose their competitiveness, thereby making the confidential items lose their necessary value. In order to ensure the security of confidential item data, confidential item data needs to be encrypted.

[0003] The data points of design parameters of confidential items have obvious distribution characteristics. These distribution characteristics are caused by the strong correlation between data points in different dimensions. This correlation can easily be used as a breakthrough for decryption, thereby reducing the security of encryption. Traditional encryption methods often use scrambling or changing numerical values ​​for encryption. Although these encryption methods can change the spatial correlation or statistical correlation between data to a certain extent, they cannot effectively destroy the correlation between different dimensions. Therefore, how to effectively destroy the correlation between different dimensions through encryption becomes the research focus of this invention.

[0004] The patent application document with application number CN105760768A shows a method for secure data storage. The method in the patent application document uses a more conventional method of changing numerical values ​​for encryption. Although this encryption method can destroy statistical correlation to a certain extent, it cannot effectively destroy the correlation between data of different dimensions. Therefore, the method in the patent application document is not suitable for solving the technical problems in the present invention. Summary of the invention

[0005] In order to solve the problem of how to effectively destroy the correlation between different dimensions through encryption, the present invention proposes a method for securely storing confidential item data, which includes the following steps:

[0006] Obtain several data points of confidential items;

[0007] Iteratively encrypt all data points for secure storage;

[0008] Among them, in one iterative encryption process, it includes:

[0009] Performing principal component analysis on all data points to obtain several principal component directions and corresponding eigenvalues; obtaining the outermost data point on the straight line located between the principal component direction and the geometric center of all data points, recording it as a reference data point, obtaining a description vector composed of the data point and the reference data point, and obtaining a principal component skewness angle composed of the description vector and the corresponding principal component direction; taking the eigenvalue as the weight, performing a weighted summation on the principal component skewness angle and then taking the inverse to obtain the extraction probability;

[0010] Randomly extract a preset number of data points from all data points as data points to be processed according to the extraction probability, and adjust the description vector and modulus of the data points to be processed according to the preset encryption key;

[0011] Calculate the distribution difference , , perform principal component analysis on all adjusted data points to obtain the principal component directions and eigenvalues, Indicates the adjustment before and after The absolute value of the difference of the eigenvalues, Indicates the adjustment before and after The angle between the principal component directions, represents the number of principal component directions, represents the angle between the jth principal component direction before and after adjustment, represents the hyperbolic tangent function;

[0012] If the distribution difference is greater than the preset threshold, the iteration ends.

[0013] The present invention determines the extraction probability of each data point according to the necessity of encryption of each data point, so that data points with higher encryption necessity can have more opportunities to be selected, thereby effectively meeting the encryption requirements and faster hiding the correlation between dimensions; further, after each encryption process, the encryption effect is evaluated, and the encryption cutoff control is performed according to the evaluation results, so as to prevent insufficient encryption or over-encryption; further, when calculating the extraction probability, the necessity of encryption of each data point is accurately reflected by introducing indicators such as the principal component deviation angle, thereby providing a basis for accurately selecting data points for encryption; further, when evaluating the encryption effect, the encryption effect of the data point is accurately reflected by introducing the angle between the principal component directions before and after encryption and the difference in the change in the principal component directions before and after encryption, thereby providing a basis for accurate encryption cutoff control.

[0014] Preferably, the principal component analysis of all data points is performed to obtain a number of principal component directions and corresponding eigenvalues, including:

[0015] The PCA algorithm is used to perform principal component analysis on all data points to obtain several principal component directions and the eigenvalues ​​corresponding to each principal component direction.

[0016] Preferably, obtaining the outermost data point on the straight line located between the principal component direction and the geometric center of all data points as the reference data point comprises:

[0017] The geometric center of all data points is obtained, and a straight line is fitted using the geometric center of all data points and the direction of the principal component. Any outermost data point on the fitted straight line is obtained and recorded as a reference data point.

[0018] Preferably, the step of obtaining a description vector formed by the data point and the reference data point comprises:

[0019] The vector obtained by subtracting the coordinates of the data point from those of the reference data point is recorded as the description vector of the data point.

[0020] Preferably, randomly selecting a preset number of data points from all data points as data points to be processed according to the extraction probability includes:

[0021] The normalized value of the extraction probability of the data point is used as the probability of the data point being selected, and a preset number of data points are randomly selected from all the data points as the data points to be processed.

[0022] The present invention increases the random characteristics of data encryption by randomly extracting data points for encryption, thereby increasing the difficulty of decryption.

[0023] Preferably, the adjusting the description vector and the modulus length of the data point to be processed according to a preset encryption key includes:

[0024] Obtaining a number of preset encryption keys for each data point to be processed;

[0025] Taking each dimension of the data point as a coordinate axis, a coordinate system is constructed; obtaining a reference data point on the principal component direction corresponding to the maximum eigenvalue of the data point to be processed and recording it as a target data point;

[0026] Recording a description vector formed by a target data point and the data point to be processed as a target description vector; obtaining an angle between the target description vector and each coordinate axis in the coordinate system as a dimension angle; adding a preset encryption key to the dimension angle to obtain an adjusted dimension angle;

[0027] Add the modulus length of the target description vector to the preset encryption key to obtain the adjusted modulus length;

[0028] Obtain an adjusted target description vector according to the adjusted dimension angle and the adjusted module length;

[0029] An adjusted data point is obtained according to the adjusted target description vector and the target data point.

[0030] The present invention adjusts the modulus and direction of the target vector to make the data points deviate more from the direction of the principal component, thereby making the distribution direction characteristics of the data points weaker and effectively hiding the correlation between dimensions.

[0031] Preferably, after adjusting the description vector and the modulus length of the data point to be processed according to the preset encryption key, the method further includes:

[0032] The value of each dimension of the target data point is added to the encryption key to obtain the adjusted value of each dimension, and the adjusted target data point is located using the adjusted values ​​of all dimensions of the target data point.

[0033] The present invention changes the coordinates of the target data point to prevent the target data point from being used as a fulcrum to decrypt various data points during decryption, thereby improving data security.

[0034] Preferably, the method of taking the eigenvalue as the weight, performing weighted summation on the principal component skewness angles and then taking the inverse thereof to obtain the extraction probability comprises:

[0035] Taking the eigenvalue as the weight, weighted summation is performed on all principal component skewness angles of the data point to obtain the distribution deviation of the data point;

[0036] The inverse of the distribution deviation of the data point is taken as the extraction probability of the data point.

[0037] The present invention reflects the explicitness of the distribution characteristics of the data points by analyzing the consistency of each data point with the majority distribution direction, and then sets the extraction probability of the data points according to the explicitness, thereby providing a basis for hiding the distribution characteristics faster and destroying the correlation between data of different dimensions.

[0038] The present invention has the following beneficial effects:

[0039] The present invention determines the extraction probability of each data point according to the encryption necessity of each data point, so that data points with higher encryption necessity can have more chances to be selected, thereby effectively meeting the encryption requirements and faster hiding the correlation between dimensions;

[0040] Furthermore, after each encryption process, the encryption effect is evaluated, and encryption cutoff control is performed according to the evaluation results, so as to prevent insufficient encryption or over-encryption.

[0041] Furthermore, when calculating the extraction probability, the necessity of encrypting each data point is accurately reflected by introducing indicators such as the principal component deviation angle, thereby providing a basis for accurately selecting data points for encryption;

[0042] Furthermore, when evaluating the encryption effect, the encryption effect of the data point can be accurately reflected by introducing the angle between the principal component directions before and after encryption and the difference in the change in the principal component directions before and after encryption, providing a basis for accurate encryption cutoff control. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0044] Figure 1 This is a flowchart of the steps of a method for securely storing confidential item data according to an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of data point distribution before adjustment provided by an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of the distribution of data points after adjustment provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0048] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0049] See also Figure 1 , which shows a flowchart of a method for securely storing confidential item data provided by an embodiment of the present invention, the method comprising the following steps:

[0050] S1: Obtain several data points of confidential items.

[0051] Specifically, data of each dimension of each confidential item is collected, and each dimension of the confidential item is used as a coordinate axis to construct a coordinate system; the data of all dimensions of each confidential item are projected into the coordinate system to obtain the data point corresponding to each confidential item, and if there are multiple confidential items, several data points are obtained.

[0052] The confidential items may be components of newly developed products that have not yet been launched on the market, and the data of each dimension of the confidential items may be design parameter data.

[0053] S2: Iteratively encrypt all data points for secure storage.

[0054] It should be noted that, in order to achieve better encryption, the data points need to be encrypted multiple times to prevent a single encryption from failing to achieve a good effect.

[0055] Among them, in one iterative encryption process, steps S20 to S23 are included.

[0056] S20: Perform principal component analysis on all data points to obtain several principal component directions and corresponding eigenvalues; obtain the outermost data point on the straight line located between the principal component direction and the geometric center of all data points, record it as the reference data point, obtain the description vector composed of the data point and the reference data point, and obtain the principal component skewness angle composed of the description vector and the corresponding principal component direction; use the eigenvalue as the weight, perform weighted summation on the principal component skewness angle, and then take the inverse to obtain the extraction probability.

[0057] It should be noted that there is a correlation between different dimensional data of confidential items. Assuming that the dimensional data of confidential items are design parameter data, one of the design parameter data is rated power, and the other is maximum speed. There is a correlation between maximum speed and rated power. The traditional encryption method cannot hide the correlation between different dimensional data well, so an encryption method is needed to destroy this correlation.

[0058] It should be further explained that among all confidential items, some confidential items have strong correlations in different dimensions of data, while some confidential items have poor correlations in different dimensions of data. The confidential items with strong correlations should be analyzed with emphasis on encryption. Therefore, it is necessary to analyze the correlation of data of different dimensions of each data point to determine whether each data point needs to be encrypted.

[0059] Preferably, as an example, all data points are subjected to principal component analysis to obtain a number of principal component directions and corresponding eigenvalues; the outermost data point is obtained on a straight line located between the principal component direction and the geometric center of all data points, and a description vector formed by the data point and the reference data point is obtained; a principal component deflection angle formed by the description vector and the corresponding principal component direction is obtained; the eigenvalue is used as a weight, and the principal component deflection angle is weighted and summed, and then the inverse is taken to obtain the extraction probability, including:

[0060] The principal component analysis algorithm (PCA algorithm for short) is used to perform principal component analysis on all data points to obtain several principal component directions and the eigenvalues ​​corresponding to each principal component direction.

[0061] The geometric center of all data points is obtained, and a straight line is fitted using the geometric center of all data points and the direction of the principal component. Any outermost data point on the fitted straight line is obtained and recorded as a reference data point.

[0062] The vector obtained by subtracting the coordinates of the data point from those of the reference data point is recorded as the description vector of the data point, and the angle formed by the description vector and the corresponding principal component direction is recorded as the principal component skewness angle of the data point.

[0063] Taking the eigenvalue as the weight, weighted summation is performed on all principal component skewness angles of the data point to obtain the distribution deviation of the data point; and the inverse of the distribution deviation of the data point is taken as the extraction probability of the data point.

[0064] It can be understood that the principal component direction reflects the distribution direction of all data points, where the larger the eigenvalue, the more obvious the distribution direction of all data points in the corresponding principal component direction. If the distribution deviation of the data point in the principal component direction where the distribution direction is particularly obvious is smaller, it means that the dimensional correlation of the data point is stronger, the encryption necessity of the data point is greater, and the dimensional correlation characteristics of the data point need to be destroyed. This implementation uses the extraction probability to reflect the probability of a data point being extracted to be de-encrypted.

[0065] S21: randomly extracting a preset number of data points from all data points as data points to be processed according to the extraction probability, and adjusting the description vector and the module length of the data points to be processed according to a preset encryption key.

[0066] It should be noted that in order to destroy the correlation between data of different dimensions of data points, the coordinate positions of the data points can be adjusted to destroy the positional relationship between the data points, effectively destroy the distribution correlation between the data points, and then destroy the dimensional correlation of the data points.

[0067] Preferably, as an example, randomly selecting a preset number of data points from all data points as data points to be processed according to the extraction probability, and adjusting the description vector and the modulus length of the data points to be processed according to the preset encryption key, including:

[0068] Get the number of dimensions of the data points to be processed . Randomly select within the preset first value range A data is randomly selected as the angle key of each data point to be processed, and a data is randomly selected within the preset second value range as the modulus key of each data point to be processed. The angle key and the modulus key are both recorded as the preset encryption key. , the second value range is This is described as an example, and other embodiments may take other values, and this embodiment does not make any specific limitation.

[0069] Obtain a reference data point on the principal component direction corresponding to the maximum eigenvalue of the data point to be processed and record it as the target data point; record the description vector formed by the target data point and the data point to be processed as the target description vector; obtain the angle between the target description vector and each coordinate axis in the coordinate system and record it as the dimension angle; add the angle key in the preset encryption key to the dimension angle to obtain the adjusted dimension angle; add the modulus length of the target description vector to the modulus length key in the preset encryption key to obtain the adjusted modulus length;

[0070] Obtain an adjusted target description vector according to the adjusted dimension angle and the adjusted module length;

[0071] An adjusted data point is obtained according to the adjusted target description vector and the target data point.

[0072] In particular, if the cumulative sum of the angle key and the dimension angle exceeds 2π, the direction of the adjusted dimension angle is the same as the cumulative sum of the angle key and the dimension angle minus 2π. For example, if the cumulative sum of the angle key and the dimension angle key is , The direction and position of The direction positions are the same, that is, they are all vertically upward.

[0073] It can be understood that the encryption key is randomly generated. By adding the dimension angle and the encryption key, the deviation of the data points to be processed from the original distribution direction can be made more random. Therefore, this encryption method can effectively destroy the distribution direction of the data points, and thus effectively destroy the correlation between different dimensions of the data points to be processed.

[0074] It should be noted that all data points to be processed are encrypted based on the target data points. Therefore, in order to prevent the target data point information from being exposed and causing the encrypted data to be cracked, the target data points need to be encrypted.

[0075] Specifically, the value of each dimension of the target data point is added to the encryption key to obtain the adjusted value of each dimension, and the adjusted target data point is located using the adjusted values ​​of all dimensions of the target data point.

[0076] It should be noted that the encryption key of the target data point is Random data in between.

[0077] S22: Calculate the distribution difference.

[0078] It should be noted that in order to prevent insufficient encryption from being easily cracked, or excessive encryption from causing low encryption efficiency, the encryption situation needs to be evaluated.

[0079] Preferably, as an example, calculating the distribution difference includes:

[0080] , where principal component analysis is performed on all adjusted data points to obtain the principal component directions and eigenvalues, Indicates the adjustment before The eigenvalues ​​and the adjusted The absolute value of the difference of the eigenvalues, Indicates the adjustment before After adjustment of the principal component direction The angle between the principal component directions, represents the number of principal component directions, represents the angle between the direction of the jth principal component before adjustment and the direction of the jth principal component after adjustment. represents the hyperbolic tangent function.

[0081] It is understandable that through It reflects the change in the distribution direction of all data points before and after adjustment. The larger the value, the greater the change in the distribution direction of the data points, which in turn indicates that the dimensional correlation of the data points has changed significantly. It reflects the adjustment difference of different distribution directions. The smaller the value, the smaller the adjustment difference of different distribution directions. The smaller the adjustment difference, the more it means that all data points are rotated without destroying the distribution correlation of data points. Therefore, further adjustment is needed to destroy this distribution relationship. Figure 2 Schematic diagram of data point distribution before adjustment and Figure 3 As shown in the schematic diagram of the data point distribution after adjustment, the distribution of the data points before and after the adjustment only undergoes a rotational change, and their relative position relationship does not change. Therefore, the data points need to be further adjusted to hide this distribution relationship.

[0082] S23: If the distribution difference is greater than a preset threshold, the iteration ends.

[0083] Preferably, as an example, the distribution difference is compared with a preset threshold, and if the distribution difference is greater than the preset threshold, the iteration ends.

[0084] It can be understood that the greater the distribution difference, the greater the degree of damage to the distribution characteristics of the data points. When the distribution difference reaches a certain level, further encryption can be stopped to prevent over-encryption.

[0085] After the iteration is completed, the encrypted data points of the confidential items are obtained and stored.

[0086] S3: Decrypt the encrypted data point.

[0087] Preferably, as an example, decrypting the encrypted data point includes:

[0088] The encrypted data points are numbered from left to right and from top to bottom. The numbers corresponding to the data points to be processed in each iterative encryption process are used as the numbering keys, and the numbers of the target data points are used as the reference keys. The reference keys, the encoding keys of the data points to be processed in all iterative processes, and the preset encryption keys of the data points to be processed are stored in the permission space, which is only open to specific personnel. For example, the encrypted data points are numbered 1, 2, 3, 4, and 5. In the third iterative encryption process, the data points numbered 2 and 4 are extracted as the data points to be processed, and 2 and 4 are used as the numbering keys.

[0089] A specific person obtains the base key in the permission space, uses the base key to locate the encrypted data point of the target data point, and subtracts the preset encryption key from the encrypted data point of the target data point to obtain the target data point.

[0090] Specific personnel obtain the numbered secret key and the preset encryption secret key in the permission space, locate the corresponding data point to be processed in the encrypted data point according to the numbered secret key, obtain the adjusted target description vector of the data point to be processed according to the target data point and the data point to be processed, and subtract the corresponding preset encryption secret key from the angle and modulus length between the adjusted target description vector and each dimension to obtain the target description vector. The data point is located according to the target description vector and the target data point.

[0091] At this point, this embodiment is completed.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for securely storing confidential item data, characterized in that: include: Obtain several data points of confidential items; Iteratively encrypt all data points for secure storage; Among them, in one iterative encryption process, it includes: Performing principal component analysis on all data points to obtain several principal component directions and corresponding eigenvalues; obtaining the outermost data point on the straight line located between the principal component direction and the geometric center of all data points, recording it as a reference data point, obtaining a description vector composed of the data point and the reference data point, and obtaining a principal component skewness angle composed of the description vector and the corresponding principal component direction; taking the eigenvalue as the weight, performing a weighted summation on the principal component skewness angle and then taking the inverse to obtain the extraction probability; Randomly extract a preset number of data points from all data points as data points to be processed according to the extraction probability, and adjust the description vector and modulus of the data points to be processed according to the preset encryption key; Calculate the distribution difference , , perform principal component analysis on all adjusted data points to obtain the principal component directions and eigenvalues, Indicates the adjustment before and after The absolute value of the difference of the eigenvalues, Indicates the adjustment before and after The angle between the principal component directions, represents the number of principal component directions, represents the angle between the jth principal component direction before and after adjustment, represents the hyperbolic tangent function; If the distribution difference is greater than the preset threshold, the iteration ends.

2. A method for securely storing confidential item data according to claim 1, characterized in that: The principal component analysis is performed on all data points to obtain several principal component directions and corresponding eigenvalues, including: The PCA algorithm is used to perform principal component analysis on all data points to obtain several principal component directions and the eigenvalues ​​corresponding to each principal component direction.

3. A method for securely storing confidential item data according to claim 1, characterized in that: The step of obtaining the outermost data point on the straight line located between the principal component direction and the geometric center of all data points as the reference data point comprises: The geometric center of all data points is obtained, and a straight line is fitted using the geometric center of all data points and the direction of the principal component. Any outermost data point on the fitted straight line is obtained and recorded as a reference data point.

4. A method for securely storing confidential item data according to claim 1, characterized in that: The obtaining of a description vector formed by the data point and the reference data point comprises: The vector obtained by subtracting the coordinates of the data point from those of the reference data point is recorded as the description vector of the data point.

5. A method for securely storing confidential item data according to claim 1, characterized in that: The step of randomly selecting a preset number of data points from all data points as data points to be processed according to the extraction probability includes: The normalized value of the extraction probability of the data point is used as the probability of the data point being selected, and a preset number of data points are randomly selected from all the data points as the data points to be processed.

6. A method for securely storing confidential item data according to claim 1, characterized in that: The step of adjusting the description vector and the modulus of the data point to be processed according to the preset encryption key includes: Obtaining a number of preset encryption keys for each data point to be processed; Taking each dimension of the data point as a coordinate axis, a coordinate system is constructed; obtaining a reference data point on the principal component direction corresponding to the maximum eigenvalue of the data point to be processed and recording it as a target data point; Recording a description vector formed by a target data point and the data point to be processed as a target description vector; obtaining an angle between the target description vector and each coordinate axis in the coordinate system as a dimension angle; adding a preset encryption key to the dimension angle to obtain an adjusted dimension angle; Add the modulus length of the target description vector to the preset encryption key to obtain the adjusted modulus length; Obtain an adjusted target description vector according to the adjusted dimension angle and the adjusted module length; An adjusted data point is obtained according to the adjusted target description vector and the target data point.

7. A method for securely storing confidential item data according to claim 6, characterized in that: After adjusting the description vector and the modulus length of the data point to be processed according to the preset encryption key, the method further includes: The value of each dimension of the target data point is added to the encryption key to obtain the adjusted value of each dimension, and the adjusted target data point is located using the adjusted values ​​of all dimensions of the target data point.

8. A method for securely storing confidential item data according to claim 1, characterized in that: The method of taking the eigenvalue as the weight, performing weighted summation on the principal component skewness angle, and then taking the inverse thereof to obtain the extraction probability includes: Taking the eigenvalue as the weight, weighted summation is performed on all principal component skewness angles of the data point to obtain the distribution deviation of the data point; The inverse of the distribution deviation of the data point is taken as the extraction probability of the data point.

Images