Industrial Internet data processing method and system based on big data

Abstract

The present application relates to the field of data encryption technology, and specifically to an industrial Internet data processing method and system based on big data, the method comprising: obtaining each data sequence of each industrial area in an industrial park; for any two data sequences, respectively recorded as the first and second data sequences, randomly extracting U sequences of the same length corresponding to the same time series interval from the first and second data sequences; obtaining predicted correlations according to the sequences at each random extraction number in the first and second data sequences; obtaining a set of frequent sequences between the first and second data sequences according to the distribution of predicted correlations between the data sequences; constructing a binary tree of the first data sequence according to the set of frequent sequences between the data sequences; generating an initial plaintext state matrix according to the characteristics of the binary trees of each data sequence, and improving the AES encryption algorithm to achieve optimized security protection for big data encryption. The present application improves the security of industrial data encryption.

Description

Technical Field

[0001] The present application relates to the field of data encryption technology, and specifically to an industrial Internet data processing method and system based on big data. Background Art

[0002] In industrial parks, a large number of sensors are usually deployed to collect industrial production data for monitoring and controlling the production process. These sensor data contain important confidential information of industrial production, such as parameter adjustment of industrial production. Once data leakage occurs, it may cause significant losses to industrial production. In addition, in industrial production, sensor data is often used to monitor the operation of production. Once the data is tampered with, it may cause industrial production equipment to malfunction and even cause security risks. Therefore, it is necessary to protect the security of industrial Internet big data.

[0003] The traditional AES encryption algorithm directly selects 128-bit plaintext groups from historical data. There is a problem that the plaintext words in the initial plaintext matrix obtained by the traditional AES encryption algorithm have a large correlation with each other, which may cause attackers to crack the algorithm based on this feature, causing a great impact on the security of industrial Internet big data. Summary of the invention

[0004] In order to solve the above technical problems, this application provides an industrial Internet data processing method and system based on big data to solve existing problems.

[0005] In a first aspect, an embodiment of the present application provides an industrial Internet data processing method based on big data, the method comprising the following steps:

[0006] Obtain data series for each industrial area in the industrial park;

[0007] For any two data sequences, respectively denoted as the first data sequence and the second data sequence, randomly extract U sequences of the same length corresponding to the same time series interval from the first and second data sequences; obtain the first and second prediction value sequences according to the first randomly extracted sequences from the first and second data sequences; obtain the prediction correlation between the first and second data sequences according to the first and second data sequences and the first and second prediction value sequences at each random extraction number;

[0008] According to the distribution of the predicted correlation between the first data sequence and the remaining data sequences except the first data sequence, a high correlation sequence of the first data sequence is obtained; the high correlation sequences of all data sequences are input into the PrefixSpan algorithm to obtain each frequent high correlation sequence and its support; the frequent high correlation sequences having both the first and second data sequences are formed into a frequent sequence set between the first and second data sequences; according to each frequent high correlation sequence in the frequent sequence set and its support, a frequent correlation index between the first and second data sequences is obtained; according to the frequent correlation index between the first data sequence and the remaining data sequences except the first data sequence, a binary tree of the first data sequence is constructed;

[0009] For each data sequence, the same acquisition method as the first data sequence is used to obtain the corresponding binary trees, and the binary trees of each data sequence are clustered. Data of a preset length are randomly extracted from different clustering clusters to generate an initial plaintext state matrix. The AES encryption algorithm is combined with the initial plaintext state matrix to encrypt the industrial Internet big data, thereby realizing big data encryption optimization and security protection.

[0010] Preferably, obtaining the first and second prediction value sequences according to the sequences randomly extracted for the first time from the first and second data sequences comprises:

[0011] The ARIMA model is used to predict the first randomly selected sequence in the first data sequence to obtain a first predicted value sequence;

[0012] The first randomly selected sequence in the first data sequence is used as the endogenous variable sequence of the ARIMAX model, and the first randomly selected sequence in the second data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain a second predicted value sequence.

[0013] Preferably, obtaining the prediction correlation between the first and second data sequences according to the first and second data sequences and the first and second prediction value sequences at each random extraction number includes:

[0014] Obtaining a prediction difference coefficient of the first randomly selected sequence according to the difference between the corresponding elements of the first and second predicted value sequences and the corresponding actual value sequence; the actual value sequence is a sequence consisting of elements after the first data sequence and having the same length as the first data sequence;

[0015] For each random draw number sequence, the predicted difference coefficient is obtained by the same method as the predicted difference coefficient of the first random draw sequence;

[0016] Obtaining the prediction feature bonding weight between the first and second data sequences according to the industrial regions and data types to which the first and second data sequences belong;

[0017] Calculate the mean of the predicted difference coefficients for all random draws, and take the ratio of the predicted feature bonding weight to the predicted difference coefficient mean as the first ratio; take the ratio of the randomly drawn sequence length to the total length of the first data sequence as the second ratio; and take the product of the first ratio and the second ratio as the predicted correlation between the first and second data sequences.

[0018] Preferably, obtaining the prediction difference coefficient of the first randomly selected sequence according to the difference between corresponding elements of the first and second predicted value sequences and the corresponding actual value sequence comprises:

[0019] For each element, the difference between the first predicted value sequence and the corresponding element in the corresponding actual value sequence is calculated as the first difference, and the difference between the second predicted value sequence and the corresponding element in the corresponding actual value sequence is calculated as the second difference;

[0020] The absolute value of the ratio of the second difference to the first difference is calculated, and the average of the absolute values ​​of all elements is used as the predicted difference coefficient of the first randomly selected sequence.

[0021] Preferably, obtaining the prediction feature bonding weight between the first and second data sequences according to the industrial regions and data types to which the first and second data sequences belong includes:

[0022] Obtain the Euclidean distance between the industrial regions to which the first and second data series belong;

[0023] When the first and second data sequences belong to the same type of data sequences, the ratio of the value 2 to the Euclidean distance is used as the prediction feature bonding weight between the first and second data sequences;

[0024] When the first and second data sequences do not belong to the same type of data sequences, the ratio of the value 1 to the Euclidean distance is used as the prediction feature bonding weight between the first and second data sequences.

[0025] Preferably, obtaining a high correlation sequence of the first data sequence according to the distribution of predicted correlations between the first data sequence and each remaining data sequence except the first data sequence comprises:

[0026] The prediction correlations between the first data sequence and the remaining data sequences except the first data sequence are arranged in ascending order to form a prediction correlation sequence;

[0027] A clustering algorithm is used to cluster the prediction correlation sequence to obtain two cluster clusters. The cluster cluster with the largest prediction correlation mean is recorded as a high-feature bonding cluster, and the prediction correlations in the high-feature bonding cluster are arranged in order from small to large to form a high correlation sequence of the first data sequence.

[0028] Preferably, obtaining the frequent correlation index between the first and second data sequences according to each frequent high correlation sequence in the frequent sequence set and its support degree includes:

[0029] The frequent correlation coefficient is obtained according to the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support;

[0030] The maximum frequent correlation coefficient between any two data sequences is obtained, and the ratio of the frequent correlation coefficient to the maximum frequent correlation coefficient is used as the frequent correlation index between the first and second data sequences.

[0031] Preferably, obtaining the frequent correlation coefficient according to the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support includes:

[0032] The minimum support is a preset parameter; the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support is calculated, and the sum of the differences of all frequent high correlation sequences in the frequent sequence set is calculated;

[0033] The first data sequence is used as the endogenous variable sequence of the ARIMAX model, and the second data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain the prediction correlation corresponding to the first data sequence;

[0034] The second data sequence is used as the endogenous variable sequence of the ARIMAX model, and the first data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain the prediction correlation corresponding to the second data sequence;

[0035] Calculate the mean of the two predicted correlations;

[0036] The product of the mean and the sum is taken as the frequent correlation coefficient.

[0037] Preferably, constructing a binary tree of the first data sequence according to frequent correlation indexes between the first data sequence and each remaining data sequence except the first data sequence comprises:

[0038] Perform threshold segmentation on the frequent correlation indexes between the first data sequence and each remaining data sequence except the first data sequence, obtain a segmentation threshold, and record the data sequence whose frequent correlation index is greater than the segmentation threshold as a highly correlated data sequence of the first data sequence;

[0039] The frequent correlation indexes in the highly correlated data sequence are sequentially arranged from small to large, from top to bottom and from left to right to construct a binary tree of the first data sequence.

[0040] In the second aspect, an embodiment of the present application also provides an industrial Internet data processing system based on big data, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, it implements the steps of any one of the above-mentioned industrial Internet data processing methods based on big data.

[0041] This application has at least the following beneficial effects:

[0042] This application is based on the degree of fit between data sequences of any two types of industrial production data in the collected data, and adopts ARIMA model and ARIMAX model to analyze the prediction feature adhesion relationship between the two sequences, and then combines K-means clustering algorithm and PrefixSpan algorithm to construct a frequent correlation index to comprehensively reflect the correlation fit degree between industrial production process data sequences. The beneficial effect is that considering the industrial scenario, there is a strong degree of correlation between data sequences of multiple types of industrial production data, avoiding the problem of large error in prediction correlation based only on the two data sequences; at the same time, a binary tree is constructed based on this, and the Chameleon clustering algorithm is used to obtain the initial plaintext state matrix. The beneficial effect is that the correlation between adjacent plaintext bytes is reduced, and the traditional AES encryption algorithm is avoided. The problem of large correlation between bytes in the initial plaintext byte matrix when directly selecting continuous 128 bits of plaintext from industrial production process data exists, thereby improving the security of industrial data encryption. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 A flowchart of the industrial Internet data processing method based on big data provided for this application;

[0045] Figure 2 An improved flowchart of the AES encryption algorithm provided for this application. DETAILED DESCRIPTION

[0046] In order to further explain the technical means and effects adopted by this application to achieve the predetermined invention purpose, the following is a detailed description of the industrial Internet data processing method and system based on big data proposed in this application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0048] The specific scheme of the industrial Internet data processing method and system based on big data provided by this application is described in detail below with reference to the accompanying drawings.

[0049] An embodiment of the present application provides an industrial Internet data processing method and system based on big data.

[0050] Specifically, the following industrial Internet data processing methods based on big data are provided, please refer to Figure 1 , the method comprises the following steps:

[0051] Step S001, collecting data sequences of each industrial area in the industrial park.

[0052] Different industrial areas in an industrial park may have different industrial production. Therefore, this embodiment divides the entire industrial park into N industrial areas according to location, places the same sensor in each industrial area, and collects production process data of each industrial area. For example, temperature and humidity sensors are used to obtain temperature and humidity data of the production process of the industrial area, sound sensors are used to obtain sound data of the production process of the industrial area, and air detectors are used to obtain dust concentration in the production process of the industrial area. The specific type of sensor used is adjusted by the implementer according to the industrial production situation.

[0053] In this embodiment, m data types are obtained from the industrial park. For the data sequence obtained for each data type, the interval between two adjacent data collections in the data sequence is t, and the number of data collection moments is T. In this embodiment, t is 1 second and T is 600.

[0054] Due to the complexity of the industrial production environment, the collected data may be missing, and the dimensions of different data are inconsistent. Therefore, this embodiment uses a linear interpolation method to first fill the data, and then performs Z-Score standardization on the filled data to remove the dimension effect. Among them, the linear interpolation method and Z-Score are both well-known technologies and will not be described in detail in this embodiment.

[0055] Based on the above collected data, the data series in each industrial area in the industrial park can be obtained.

[0056] Step S002, based on the correlation between any two industrial production data sequences in the collected data, determine the initial plaintext state matrix of the AES encryption algorithm.

[0057] There may often be a certain correlation between different industrial production data series. For example, in the material crushing area in the industrial park, when the hardness of the material is high, it often produces a large load on the crusher. In order to provide greater crushing performance, the current of the crusher may increase, and the sound of the crushing machine also increases.

[0058] However, the traditional AES encryption algorithm directly obtains plaintext groups from the collected data, and there is a problem of large correlation between the plaintext words in the initial plaintext state matrix, resulting in low security when encrypting industrial Internet big data.

[0059] Therefore, the purpose of this embodiment is to group data sequences with higher correlation into one category, so as to construct an initial plaintext state matrix for data sequences with lower similarity.

[0060] In this embodiment, there are N industrial areas, each of which has m types of data sequences. data sequences, each of which has a total of T data elements.

[0061] In the industrial production process, there may be more information transfer between two data sequences with strong correlation. When the data in one sequence changes, the data in the other sequence will often change accordingly. Therefore, when the two data sequences have a strong correlation, the comprehensive prediction is more likely to be close to the actual value.

[0062] Based on the above analysis, from the xth data sequence in the industrial park The length of random sampling is U The result of the first extraction is recorded as , using the ARIMA model as the forecasting model, the input is the data sequence , the output is The first prediction value sequence at time .

[0063] Furthermore, from the yth data sequence Get the data sequence Sequences with the same time interval are denoted as , using the ARIMAX model as the prediction model, the data series As the exogenous variable sequence in the ARIMAX model, the data sequence As the forecast data series of the ARIMAX model, the output is also The second prediction value sequence at time .

[0064] Get data series at the same time after The elements at each moment constitute the actual value sequence Based on the above analysis, the relationship between the predicted results and the actual values ​​of all randomly selected data sequences between the xth data sequence and the yth data sequence is used to construct the prediction correlation :

[0065]

[0066]

[0067]

[0068] in, is a data sequence , The predicted coefficient of variance for the first random draw, is the number of elements in the extracted sequence, is the first prediction value sequence The gth data value in is the second prediction value sequence The gth data value in is a sequence of actual values The g-th data value in, where is the first difference, is the second difference.

[0069] It should be noted that when the error of the combined prediction of the two sequences is greater than the error of the single sequence prediction, that is, The larger the value, the smaller the prediction correlation between the two sequences, and the larger the comprehensive prediction error. Similarly, the calculation method of the prediction difference coefficient of the remaining random extraction times is the same as the method of the prediction difference coefficient of the first random extraction time sequence, so that the prediction difference coefficient of each random extraction time sequence can be obtained by the above method.

[0070] in, is a data sequence , The predicted features and weights between is a data sequence , The Euclidean distance between industrial areas.

[0071] It should be noted that when the industrial areas where the two data series are located are closer, that is, The smaller it is, the more likely it is that the two data sequences have a greater spatial correlation, and the predicted value during comprehensive prediction is more likely to be close to the actual value. At the same time, when the two data sequences belong to the same type of data sequences, the greater the feature correlation and weight between the two data sequences, that is, the greater the prediction feature correlation and weight. The bigger.

[0072] in, is a data sequence , The predictive correlation between It is the data sequence The length of a randomly drawn sequence, is a data sequence The total length of It is the data sequence The number of random draws, is a data sequence , The u-th prediction difference coefficient between is an adjustment factor, which is used to prevent the denominator from being zero. Its value is 0.001, where is the first ratio, is the second ratio.

[0073] It should be noted that when the predicted feature stickiness and weight between two data sequences are larger, that is, The larger the value, the more likely the two data series are to be spatially correlated, and the better the comprehensive prediction can be made. At the same time, the smaller the comprehensive prediction error between the two series, that is, The smaller the value, the more likely the two data series are to have a stronger predictive correlation. The larger the length of the input sequence to be predicted, the larger the The larger the value is, the more samples are used for prediction and the more accurate the prediction results are. The larger it is, the more correlated the two data series are, and the more similar the changing trends are.

[0074] At this point, the prediction correlation between any two data series can be obtained. The larger the value, the better the comprehensive prediction effect of the two data series, and the more correlation and adhesion characteristics between the two data series. However, the above processing method only considers the characteristics between the two data series. In the actual industrial production process, there is often a strong degree of correlation between multiple data series. The prediction correlation between data series cannot reflect this feature, and there is a problem of large errors.

[0075] Therefore, this embodiment uses the data sequence For example, calculate the data sequence The predicted correlation between any data sequence is sorted in order of predicted correlation from small to large, and recorded as the predicted correlation sequence .

[0076] Using K-means clustering algorithm, the input is the prediction correlation sequence , the cluster number of the K-means clustering algorithm is set to 2, and the output is the cluster cluster after the algorithm clustering. The cluster with a larger predicted correlation mean is recorded as a high correlation cluster, and all elements in the high correlation cluster are arranged in order from small to large predicted correlation, and the sequence composed of the sorted elements is recorded as a high correlation sequence.

[0077] Using the above method, a high correlation sequence is calculated for each data sequence, and the high correlation sequences corresponding to all data sequences are input into the PrefixSpan algorithm, and the minimum support of the PrefixSpan algorithm is set to 5%, and the output is each frequent high correlation sequence and its support. Among them, the PrefixSpan algorithm is a well-known technology and will not be described in detail in this embodiment.

[0078] Based on the above analysis, there will be data series , The frequent high correlation sequence of the data sequence , The frequent sequence set of data can be obtained , Frequent correlation index :

[0079]

[0080]

[0081] in, is a data sequence , The frequent correlation coefficient of is a data sequence , The comprehensive prediction correlation of , Yes The data sequence is used as the prediction sequence. The data series is used as an exogenous variable, and the predicted correlation is calculated. Yes The data sequence is used as the prediction sequence. The data sequence is used as an exogenous variable, and the predicted correlation is calculated. Z is the data sequence , The number of frequent highly correlated sequences in the frequent sequence set, is the support of the zth frequent high correlation sequence in the frequent sequence set, It is the minimum support set by the PrefixSpan algorithm. is a data sequence , The frequent correlation index of It is the largest frequent correlation index among all data sequences.

[0082] When the prediction correlation between the two sequences is greater, that is, The larger the value, the smaller the error between the two data series and the actual value when making a comprehensive prediction, and the more likely the two data series are to have correlation and viscosity. At the same time, the higher the support of the frequent high correlation sequence where the two data series are located, that is, The larger the value, the more times the two data sequences appear in the PrefixSpan algorithm, and the more frequent they are. , The more frequent the correlation viscosity, the higher the frequent correlation coefficient The larger the frequent correlation index is, The bigger.

[0083] So far, the data series is calculated The frequent correlation index between all other data sequences is used to obtain the segmentation threshold using the Otsu method. Data sequences with frequent correlation index greater than or equal to the segmentation threshold are recorded as high-correlation data sequences, and all high-correlation data sequences are recorded as data sequences. The corresponding binary tree node is specifically: take the largest frequent correlation index data sequence in the high correlation data sequence as the root node, and construct a binary tree from top to bottom and from left to right in the order of frequent correlation index from small to large, wherein if the frequent correlation indexes of two data sequences are the same, they are sorted from small to large according to the subscripts of the data sequences. Among them, the Otsu method is a well-known technology and will not be described in detail in this embodiment.

[0084] The binary tree corresponding to each data sequence is used as the initial subcluster and input into the Chameleon clustering algorithm to cluster the data sequences with related cohesion into one category. The output is the obtained Binary tree clusters. =4, based on the AES encryption algorithm, 32 bits of industrial sensor data are randomly extracted from each of the four clusters each time to form the initial plaintext state matrix of the AES encryption algorithm.

[0085] Step S003, improving the AES encryption algorithm according to the initial plaintext state matrix when encrypting the acquired industrial production process data, so as to realize encryption of the industrial production process data.

[0086] According to the above steps, the initial plaintext state matrix of the industrial production process data is obtained when it is encrypted, and it is used as the initial plaintext matrix in the traditional AES encryption algorithm. All industrial production process data are encrypted based on the improved AES encryption algorithm to obtain the corresponding ciphertext. Among them, the improved flow chart of the AES encryption algorithm is as follows Figure 2 As shown, the AES subsequent encryption process is a well-known technology, and the specific process will not be repeated here.

[0087] The embodiment of the present application is based on the degree of fit between the data sequences of any two types of industrial production data in the collected data, and adopts the ARIMA model and the ARIMAX model to analyze the prediction feature adhesion relationship between the two sequences, and then combines the K-means clustering algorithm and the PrefixSpan algorithm to construct a frequent correlation index, which comprehensively reflects the degree of correlation fit between the industrial production process data sequences. The beneficial effect is that in the industrial scenario, the data sequences of multiple types of industrial production data have a strong degree of correlation, avoiding the problem of large error in prediction based only on the correlation between two data sequences;

[0088] At the same time, a binary tree is constructed based on this, and the Chameleon clustering algorithm is used to obtain the initial plaintext state matrix. The beneficial effect is that it reduces the correlation between adjacent plaintext bytes, avoids the problem of large correlation between bytes in the initial plaintext byte matrix when the traditional AES encryption algorithm directly selects continuous 128 bits of plaintext from industrial production process data, and improves the security of industrial data encryption.

[0089] Based on the same inventive concept as the above method, an embodiment of the present application also provides an industrial Internet data processing system based on big data, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the industrial Internet data processing methods based on big data described in the above method are implemented.

[0090] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0092] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. The industrial Internet data processing method based on big data is characterized by: The method comprises the following steps: Obtain data series for each industrial area in the industrial park; For any two data sequences, respectively denoted as the first data sequence and the second data sequence, randomly extract U sequences of the same length corresponding to the same time series interval from the first and second data sequences; obtain the first and second prediction value sequences according to the first randomly extracted sequences from the first and second data sequences; obtain the prediction correlation between the first and second data sequences according to the first and second data sequences and the first and second prediction value sequences at each random extraction number; According to the distribution of the predicted correlation between the first data sequence and the remaining data sequences except the first data sequence, a high correlation sequence of the first data sequence is obtained; the high correlation sequences of all data sequences are input into the PrefixSpan algorithm to obtain each frequent high correlation sequence and its support; the frequent high correlation sequences having both the first and second data sequences are formed into a frequent sequence set between the first and second data sequences; according to each frequent high correlation sequence in the frequent sequence set and its support, a frequent correlation index between the first and second data sequences is obtained; according to the frequent correlation index between the first data sequence and the remaining data sequences except the first data sequence, a binary tree of the first data sequence is constructed; For each data sequence, the same acquisition method as the first data sequence is used to obtain the corresponding binary trees, the binary trees of each data sequence are clustered, and data of a preset length are randomly extracted from different clusters to generate an initial plaintext state matrix. The AES encryption algorithm is used in combination with the initial plaintext state matrix to encrypt the industrial Internet big data, so as to achieve big data encryption optimization and security protection; The method for obtaining the frequent correlation index is: The frequent correlation coefficient is obtained according to the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support; The maximum frequent correlation coefficient between any two data sequences is obtained, and the ratio of the frequent correlation coefficient to the maximum frequent correlation coefficient is used as the frequent correlation index between the first and second data sequences.

2. The industrial Internet data processing method based on big data according to claim 1, characterized in that: The step of obtaining the first and second prediction value sequences according to the sequences randomly extracted for the first time from the first and second data sequences comprises: The ARIMA model is used to predict the first randomly selected sequence in the first data sequence to obtain a first predicted value sequence; The first randomly selected sequence in the first data sequence is used as the endogenous variable sequence of the ARIMAX model, and the first randomly selected sequence in the second data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain a second predicted value sequence.

3. The industrial Internet data processing method based on big data as claimed in claim 2, characterized in that: The step of obtaining the prediction correlation between the first and second data sequences according to the first and second data sequences and the first and second prediction value sequences at each random extraction number includes: Obtaining a prediction difference coefficient of the first randomly selected sequence according to the difference between the corresponding elements of the first and second predicted value sequences and the corresponding actual value sequence; the actual value sequence is a sequence consisting of elements after the first data sequence and having the same length as the first data sequence; For each random draw number sequence, the predicted difference coefficient is obtained by the same method as the predicted difference coefficient of the first random draw sequence; Obtaining the prediction feature bonding weight between the first and second data sequences according to the industrial regions and data types to which the first and second data sequences belong; Calculate the mean of the predicted difference coefficients of all random draw times, and use the ratio of the predicted feature bonding weight to the predicted difference coefficient mean as the first ratio; use the ratio of the length of the randomly drawn sequence to the total length of the first data sequence as the second ratio; and use the product of the first ratio and the second ratio as the predicted correlation between the first and second data sequences; The step of obtaining the prediction feature bonding weight between the first and second data sequences according to the industrial regions and data types to which the first and second data sequences belong includes: Obtain the Euclidean distance between the industrial regions to which the first and second data series belong; When the first and second data sequences belong to the same type of data sequences, the ratio of the value 2 to the Euclidean distance is used as the prediction feature bonding weight between the first and second data sequences; When the first and second data sequences do not belong to the same type of data sequences, the ratio of the value 1 to the Euclidean distance is used as the prediction feature bonding weight between the first and second data sequences.

4. The industrial Internet data processing method based on big data as claimed in claim 3 is characterized in that: The step of obtaining the prediction difference coefficient of the first randomly selected sequence according to the difference between the corresponding elements of the first and second predicted value sequences and the corresponding actual value sequence includes: For each element, the difference between the first predicted value sequence and the corresponding element in the corresponding actual value sequence is calculated as the first difference, and the difference between the second predicted value sequence and the corresponding element in the corresponding actual value sequence is calculated as the second difference; The absolute value of the ratio of the second difference to the first difference is calculated, and the average of the absolute values ​​of all elements is used as the predicted difference coefficient of the first randomly selected sequence.

5. The industrial Internet data processing method based on big data as claimed in claim 3, characterized in that: The step of obtaining a high correlation sequence of the first data sequence according to the distribution of predicted correlations between the first data sequence and each remaining data sequence except the first data sequence comprises: The prediction correlations between the first data sequence and the remaining data sequences except the first data sequence are arranged in ascending order to form a prediction correlation sequence; A clustering algorithm is used to cluster the prediction correlation sequence to obtain two cluster clusters. The cluster cluster with the largest prediction correlation mean is recorded as a high-feature bonding cluster, and the prediction correlations in the high-feature bonding cluster are arranged in order from small to large to form a high correlation sequence of the first data sequence.

6. The industrial Internet data processing method based on big data according to claim 1, characterized in that: The frequent correlation coefficient is obtained according to the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support, including: The minimum support is a preset parameter; the difference between the support of each frequent high correlation sequence in the frequent sequence set and the minimum support is calculated, and the sum of the differences of all frequent high correlation sequences in the frequent sequence set is calculated; The first data sequence is used as the endogenous variable sequence of the ARIMAX model, and the second data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain the prediction correlation corresponding to the first data sequence; The second data sequence is used as the endogenous variable sequence of the ARIMAX model, and the first data sequence is used as the exogenous variable sequence of the ARIMAX model to obtain the prediction correlation corresponding to the second data sequence; Calculate the mean of the two predicted correlations; The product of the mean and the sum is taken as the frequent correlation coefficient.

7. The industrial Internet data processing method based on big data according to claim 1, characterized in that: The step of constructing a binary tree of the first data sequence according to frequent correlation indexes between the first data sequence and each remaining data sequence except the first data sequence includes: Perform threshold segmentation on the frequent correlation indexes between the first data sequence and each remaining data sequence except the first data sequence, obtain a segmentation threshold, and record the data sequence whose frequent correlation index is greater than the segmentation threshold as a highly correlated data sequence of the first data sequence; The frequent correlation indexes in the highly correlated data sequence are sequentially arranged from small to large, from top to bottom and from left to right to construct a binary tree of the first data sequence.

8. An industrial Internet data processing system based on big data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the industrial Internet data processing method based on big data as described in any one of claims 1 to 7 are implemented.

Images