Industrial data priority calculation method based on correlation relationship and sequence processing

Abstract

The present invention discloses an industrial data priority calculation method based on correlation relationship and sequence processing, which relates to the field of industrial big data technology, including: performing sequence processing on industrial time series data to obtain first industrial time series data; calculating a time series correlation matrix based on the first industrial time series data, and evaluating the correlation strength between the first industrial time series data in combination with equipment association and process flow dependency; calculating the first priority score of each first industrial time series data, and obtaining the first priority of all first industrial time series data based on the first priority score; obtaining the second priority based on the correlation strength and the first priority score in combination with historical abnormal event information. The present invention improves the accuracy and adaptability of industrial time series data priority calculation by comprehensively considering data correlation, time series characteristics and dynamic factors, and provides reliable data support for decision optimization and abnormality detection of intelligent manufacturing systems.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial big data processing, and in particular to an industrial data priority calculation method based on correlation relationship and sequence processing. Background Art

[0002] With the advent of the Industrial 4.0 era and the rapid development of intelligent manufacturing, industrial systems have generated massive amounts of time series data. These data contain key information such as equipment operating status, production process parameters, and environmental monitoring indicators. However, faced with such a huge amount of data, how to effectively process and analyze these data has become an important challenge. Traditional data processing methods often use simple statistical analysis or fixed threshold judgment, which is difficult to adapt to the complex and changing industrial environment. In recent years, some researchers have begun to try to apply machine learning and deep learning technologies to industrial data analysis, such as long short-term memory networks (LSTM) for time series prediction and convolutional neural networks (CNN) for anomaly detection. These methods have achieved good results in certain specific scenarios, but there are still problems such as insufficient generalization ability and high computational complexity. On the other hand, methods based on correlation analysis, such as Granger causality test and mutual information analysis, have also been introduced into the field of industrial data processing to discover the intrinsic connection between data. However, these methods often only focus on the static relationship between data and ignore the dynamic characteristics of industrial systems.

[0003] The existing industrial time series data processing technologies have the following main deficiencies: First, most methods fail to fully consider the combined influence of the correlation relationship between data and the time series characteristics, resulting in the inability to effectively identify and utilize the dynamic associations between data. Second, traditional data processing algorithms often show problems of inefficiency and insufficient precision when faced with long-sequence, high-dimensional industrial time series data. In addition, the existing methods lack a dynamic evaluation mechanism for the importance of data, making it difficult to flexibly adjust the priority of data processing according to actual production conditions. Finally, when faced with emergencies or abnormal situations, the existing methods are not adaptable enough and are prone to missing key information, affecting the timeliness and accuracy of decision-making. These problems seriously restrict the effectiveness and efficiency of industrial big data analysis and affect the overall performance of intelligent manufacturing systems. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to provide an industrial data priority calculation method based on correlation relationship and sequence processing to solve the problems in the prior art that data correlation and timing characteristics are not fully considered, processing efficiency is low, and priority assessment is inaccurate.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: perform serial processing on industrial time series data to obtain first industrial time series data; based on the first industrial time series data, calculate the time series correlation matrix, and at the same time combine the equipment association and process flow dependency to evaluate the correlation strength between the first industrial time series data; calculate the first priority score of each of the first industrial time series data, and obtain the first priority of all the first industrial time series data based on the first priority score; based on the correlation strength and the first priority score, combined with historical abnormal event information, obtain the second priority, and complete the industrial data priority calculation based on correlation relationship and serial processing.

[0007] As a preferred scheme of the industrial data priority calculation method based on correlation relationship and sequence processing described in the present invention, wherein: the industrial time series data includes equipment operating parameters, production process data, quality inspection data, environmental monitoring data, and energy consumption data; the sequence processing includes time period division, segment-by-segment aggregation averaging and feature extraction.

[0008] As a preferred solution of the industrial data priority calculation method based on correlation relationship and sequence processing described in the present invention, wherein: based on the first industrial time series data, calculating the time series correlation matrix includes the following steps: based on the sequence processing result, constructing a feature vector set of the first industrial time series data ; Calculate the correlation coefficient between any two feature vectors of the first industrial time series data ;in, Indicates i The first industrial time series data and the j The correlation coefficient between the feature vectors of the first industrial time series data, i and j The total number of industrial time series data from 1 to the first n Any integer between i ≠ j Based on the correlation coefficient Constructing a time series dependency matrix .

[0009] Combining the equipment correlation and process dependency, evaluating the correlation strength between the first industrial time series data includes the following steps: e and process dependency R p , determine the initial relevance assessment According to the timing correlation matrix and the initial correlation assessment , evaluate the correlation strength between the first industrial time series data.

[0010] As a preferred solution of the industrial data priority calculation method based on correlation relationship and sequence processing described in the present invention, wherein: the time series correlation matrix It is expressed by the following formula:

[0011] ;

[0012] in, is the total number of first industrial time series data; in the time series correlation matrix middle, is the correlation coefficient A specific example of i =1, j =2.

[0013] The initial relevance assessment It is expressed by the following formula:

[0014] ;

[0015] ;

[0016] ;

[0017] in, is the device association R e and process dependency R p The weight coefficient of ≤ 1; P is the physical connectivity, 0 ≤P≤ 1; I is the information flow, 0 ≤I≤ 1; E is the energy flow, 0 ≤E≤ 1; Q is the direct connectivity in the process flow diagram, 0 ≤Q≤ 1; M is the similarity of material flow, 0 ≤M≤ 1; T' is the degree of temporal dependence, 0≤T'≤ 1.

[0018] The correlation strength is expressed by the following formula:

[0019] ;

[0020] Among them, S ij It is i and j The strength of the correlation between time series data.

[0021] As a preferred solution of the industrial data priority calculation method based on correlation relationship and sequence processing described in the present invention, wherein: calculating the first priority score of each of the first industrial time series data includes the following steps: using the RNN model to perform a preliminary priority evaluation on the first industrial time series data to obtain a preliminary priority evaluation value ; Based on the preliminary priority assessment value , combined with the impact of historical data, determine the first correction priority value The historical data impact refers to the impact of the first industrial time series data in the historical period on the t The function of calculating the first modified priority value of the first industrial time series data in a time period; based on the first modified priority value , combined with the impact of dynamic factors, determine the first priority score The dynamic factors include the change rate of the first industrial time series data, the quality of the first industrial time series data, the industrial system fluctuation, and the industrial system status.

[0022] The obtaining of the first priority of all the first industrial time series data based on the first priority score includes sorting all the first industrial time series data according to the first priority score to obtain the first priority.

[0023] As a preferred solution of the industrial data priority calculation method based on correlation relationship and sequence processing described in the present invention, wherein: the preliminary priority evaluation value It is expressed by the following formula:

[0024] ;

[0025] in, For the t The preliminary priority assessment value of the first industrial time series data in each time period; For the recurrent neural network RNN ​​in the t Hidden state output for each time period; For RNN t Reset gate output for a time period; , is the weight matrix; is the bias term; is the hyperbolic tangent activation function.

[0026] The first modified priority value It is expressed by the following formula:

[0027] ;

[0028] in, For the i The first industrial time series data is t The feature vector of each time period; For the i The first industrial time series data in the historical period The eigenvector of is the time decay weight; uThe number of historical time periods that need to be considered when calculating the similarity between first industry time series data in different time periods.

[0029] The first priority score It is expressed by the following formula:

[0030] ;

[0031] in, For the t The change rate of the first industrial time series data in a time period; For the t The first industrial time series data quality factor for each time period; For the t Industrial system fluctuation factors for a period of time; For the t Industrial system status in a time period; For the t -Industrial system status for 1 time period.

[0032] As a preferred solution of the method for calculating the priority of industrial data based on correlation relationship and sequence processing described in the present invention, the process of calculating the second priority includes the following steps: based on the preliminary correlation evaluation Correlation with historical abnormal events of the target industrial system to obtain correlation evaluation after considering abnormal events ; Based on the correlation evaluation after considering abnormal events and the strength of correlation , and obtain the final correlation evaluation value ; Based on the final relevance evaluation value and the first priority score , get the second priority.

[0033] To further solve the above technical problems, the present invention provides the following technical solutions: an industrial data priority calculation system based on correlation relationship and sequence processing, comprising: a data processing module, which performs sequence processing on industrial time series data to obtain first industrial time series data; a correlation calculation module, which is used to calculate the time series correlation matrix based on the first industrial time series data, and at the same time, combines the equipment correlation and process flow dependency to evaluate the correlation strength between the first industrial time series data; a first priority calculation module, which is used to calculate the first priority score of each of the first industrial time series data, and obtain the first priority of all the first industrial time series data based on the first priority score; a second priority calculation module, which is used to obtain the second priority based on the correlation strength and the first priority score, combined with historical abnormal event information, to complete the industrial data priority calculation based on correlation relationship and sequence processing.

[0034] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the industrial data priority calculation method based on correlation relationship and sequence processing as described above are implemented.

[0035] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the industrial data priority calculation method based on correlation relationship and sequence processing as described above are implemented.

[0036] Beneficial effects of the present invention: The present invention realizes a comprehensive and accurate priority assessment of industrial time series data by combining the RNN model, historical data influence and dynamic factors. The present invention introduces the correlation strength and historical abnormal event information for dynamic adjustment, which greatly improves the accuracy and adaptability of priority calculation. The improved correlation coefficient and adaptive weight mechanism adopted by the present invention effectively capture the complex dynamic correlation between data, and overcome the limitation of traditional methods that ignore time series characteristics. Through multi-stage priority calculation and adjustment, it can better cope with emergencies and abnormal situations in industrial environments, and provide strong support for the timely discovery of key information. Overall, the present invention significantly improves the efficiency and accuracy of industrial big data analysis, lays a solid foundation for decision optimization and anomaly detection of intelligent manufacturing systems, and has important theoretical significance and practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0038] Figure 1 An overall flow chart of an industrial data priority calculation method based on correlation relationship and sequence processing provided by one embodiment of the present invention.

[0039] Figure 2 A computer device diagram of the industrial data priority calculation method based on correlation relationship and sequence processing of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0041] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0042] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides an industrial data priority calculation method based on correlation relationship and sequence processing.

[0043] The existing industrial time series data processing methods have the following main problems: First, most methods fail to fully consider the correlation relationship between data and the combined influence of time series characteristics, resulting in the inability to effectively identify and utilize the dynamic associations between data. Secondly, traditional data processing algorithms often show problems of inefficiency and insufficient precision when facing long-sequence, high-dimensional industrial time series data. In addition, the existing methods lack a dynamic evaluation mechanism for the importance of data, making it difficult to flexibly adjust the priority of data processing according to actual production conditions. Finally, when faced with emergencies or abnormal situations, the existing methods are not adaptable enough and are prone to missing key information, affecting the timeliness and accuracy of decision-making. These problems seriously restrict the effectiveness and efficiency of industrial big data analysis and affect the overall performance of intelligent manufacturing systems.

[0044] The present application provides an effective solution to the above-mentioned problems. Next, multiple embodiments will be combined to explain in detail how to implement the industrial data priority calculation method based on correlation relationship and sequence processing.

[0045] Figure 1 The overall flow chart of the industrial data priority calculation method based on correlation relationship and sequence processing is shown, including:

[0046] S1: Perform sequence processing on the industrial time series data to capture the time series characteristics and dynamic changes of the industrial time series data to obtain the first industrial time series data.

[0047] Specifically, sequence processing includes time period division, segment-by-segment aggregation averaging, and feature extraction. In industrial data analysis, it is common practice to divide according to production cycle, shift or other relevant time scales. Segment-by-segment aggregation averaging is a data dimension reduction and smoothing technique used to reduce the impact of noise and highlight the overall trend of the data. Feature extraction includes statistics (such as mean, variance, kurtosis, etc.), frequency domain features, time domain features, etc. In this embodiment, sequence processing adopts common technical means in this field, so it is not repeated here.

[0048] It should be noted that before performing sequence processing on industrial time series data, preprocessing operations need to be performed first. Since there are some data quality problems in the collected original industrial time series data, in the data preprocessing part, it is necessary to perform preparatory operations such as missing value filling on the original industrial time series data to obtain the preprocessed industrial time series data.

[0049] Furthermore, industrial time series data include equipment operating parameters (such as temperature, pressure, vibration, etc.), production process data (such as output, raw material consumption, production speed, etc.), quality inspection data (such as product size, weight, defect rate, etc.), environmental monitoring data (such as air quality, noise level, etc.), and energy consumption data (such as electricity, water, gas usage, etc.). These five types of data are selected as the industrial time series data to be studied in the present invention because: first, the above-mentioned types of industrial time series data comprehensively cover all key aspects of the industrial production process, from equipment operation to production process, from quality control to environmental monitoring, and then to energy management. This comprehensive data coverage can provide a panoramic view of the operation of industrial systems. Secondly, the above-mentioned types of industrial time series data all have obvious time dimensions, and are usually collected continuously at fixed or variable time intervals to form a time series. For example, equipment operating parameters may be recorded once per second or minute, quality inspection data may change with production batches, and environmental monitoring data may be recorded every hour or every day. Thirdly, there are often complex relationships between these different types of industrial time series data. For example, changes in equipment operating parameters may affect production process data, which in turn affects quality inspection results; changes in environmental conditions may affect equipment performance and energy consumption. Then, these data types are core data in industrial production, and are of great significance for production efficiency, product quality, equipment maintenance, energy management, and environmental compliance. Finally, these diverse data types provide a rich source of information for advanced analysis. By analyzing these data, predictive maintenance, optimization of production processes, improvement of quality control, and improvement of energy efficiency can be carried out.

[0050] S2: Based on the first industrial time series data, the time series correlation matrix is ​​calculated, and the correlation strength between the first industrial time series data is evaluated by combining the equipment association and process flow dependency.

[0051] S2.1: Calculating the time series correlation matrix of the first industrial time series data, including the following steps:

[0052] First, based on the sequence processing results in S1, the feature vector set of the first industrial time series data is constructed , for the First industrial time series data:

[0053] ;

[0054] in, For the The first industrial time series data is The feature vector of each time period; For the The first industrial time series data is The feature vector of each time period; For the The first industrial time series data is The feature vector of each time period; the feature vector set of the first industrial time series data In, through Represents an element, The first industrial time series data is The feature vector of the time period. It is expressed by the following formula:

[0055] ;

[0056] in, For the The first industrial time series data is The average value of the time period; For the The first industrial time series data is The variance of the time period; For the The first industrial time series data is Skewness of the time period; For the The first industrial time series data is Kurtosis of the time period; For the The first industrial time series data is time period frequency domain features; For the The first industrial time series data is time period time domain features.

[0057] Secondly, calculate the correlation coefficient between any two feature vectors of the first industrial time series data For any two first industrial time series data and , calculate the correlation between their eigenvectors, specifically using the improved Pearson correlation coefficient:

[0058] ;

[0059] in, is the total number of time periods; and They are and The average eigenvector of Indicates The first industrial time series data and the The correlation coefficient between the first industrial time series data is a measure of the similarity or correlation between the feature vector sequences of the two first industrial time series data.

[0060] Preferably, in this embodiment, the Pearson correlation coefficient is improved, the basic structure of the Pearson correlation coefficient is retained, but it is extended to the case of multi-dimensional vectors to adapt to the feature vectors that need to be processed. This improvement can calculate the correlation between two feature vector sequences, not just the correlation between single variables.

[0061] Again, based on the correlation coefficient Construct the time series correlation matrix C.

[0062] Assume that n The first industrial time series data is constructed n × n The correlation matrix C is:

[0063] ;

[0064] Among them, 1 is the time series correlation matrix C The elements on the diagonal represent the correlation between each first industrial time series data and itself (always 1); is the total number of first industry time series data; time series correlation matrix C is a symmetric matrix, where , indicating the and The correlation between the first industrial time series data and the and The correlation between the first industrial time series data is the same. In this matrix, for A specific example ofi =1, j =2.

[0065] Preferably, in an actual industrial system, a change in one variable may not affect another variable immediately, but only after a period of time. By calculating the correlation coefficients at different lag times, these time dependencies that may be overlooked can be discovered. Therefore, after constructing the time series correlation matrix C, the present invention can calculate the correlation coefficients at different lag times to capture the potential time lag effect. Define the maximum lag time as L, then:

[0066] ;

[0067] where k is the lag time, -L ≤ k ≤ L.

[0068] Based on this, the strongest association between the two time series is determined by selecting the correlation coefficient with the largest absolute value, regardless of whether the association is immediate or lagged. For each pair of first industrial time series data, the correlation coefficient with the largest absolute value is selected as the final correlation measure:

[0069] ;

[0070] in, For the i and j The final selected correlation coefficient between the first industrial time series data represents the strongest correlation between the two first industrial time series data after considering all possible time lags.

[0071] S2.2: According to the device association R e and process dependency R p Determine initial relevance assessment :

[0072] ;

[0073] ;

[0074] ;

[0075] in, is the device association R e and process dependency R p The weight coefficient of ≤ 1; P is the physical connectivity, 0 ≤ P ≤ 1; I is the degree of information circulation, 0 ≤ I ≤ 1; Eis the energy flow rate, 0 ≤ E ≤ 1; Q is the direct connection degree in the process flow diagram, 0 ≤ Q ≤ 1; M is the similarity of material flow, 0 ≤ M ≤ 1; T’ is the timing dependence, 0≤ T’ ≤ 1.

[0076] Among them, physical connectivity P The expression is:

[0077] ;

[0078] Information Currency I The expression is:

[0079] ;

[0080] Energy Fluidity E The expression is:

[0081] ;

[0082] Direct connection Q The expression is:

[0083] ;

[0084] Logistics Flow Similarity M The expression is:

[0085] ;

[0086] Timing Dependency T’ The expression is:

[0087] .

[0088] S2.3: Based on the time series correlation matrix and preliminary correlation assessment , evaluate the strength of correlation between the first industrial time series data.

[0089] The expression for the strength of correlation is:

[0090] ;

[0091] Among them, S ij It is i and j The strength of the correlation between time series data.

[0092] Preferably, a new correlation strength matrix can be constructed through the correlation strength formula: S :

[0093] ;

[0094] This new correlation strength matrix combines time series correlation and preliminary correlation assessment to provide a more comprehensive and accurate correlation measurement; through standardized and consistent representation, the correlation strength between different first industrial time series data pairs can be directly compared, which is easy to interpret and use; it condenses multiple indicators into a single strength indicator, simplifies the analysis process and retains adjustment flexibility; it better adapts to complex industrial scenarios and combines statistical characteristics with actual process relationships; it provides a basis for anomaly detection, data dimensionality reduction, and cluster analysis; it supports visual display and helps discover potential patterns. This approach not only improves system understanding and anomaly detection, but also lays the foundation for subsequent analysis, decision-making, and system optimization.

[0095] S3: Calculate the first priority score of each piece of first industrial time series data, and obtain the first priority of all first industrial time series data based on the first priority score.

[0096] S3.1: Use the RNN model to perform preliminary priority assessment on the first industrial time series data and obtain preliminary priority assessment values :

[0097] ;

[0098] in, For the t The preliminary priority assessment value of the first industrial time series data in each time period; is the hidden state output of the recurrent neural network RNN ​​in the tth time period; For RNN t Reset gate output for a time period; , is the weight matrix; is the bias term; is the hyperbolic tangent activation function.

[0099] It should be noted that the first industrial time series data is t The time period is input into the system, then converted into a format acceptable to RNN, and then input into the RNN model, which processes and outputs the hidden state. and reset gate output (Since this is a conventional technique, it will not be described here.) Then use the RNN output to calculate the preliminary priority evaluation value using the above preliminary priority evaluation formula .

[0100] S3.2: Based on preliminary priority assessment , combined with the impact of historical data, determine the first correction priority value :

[0101] ;

[0102] in, For the i The first industrial time series data is t The feature vector of a time period (which can be the current time period); For the i The first industrial time series data in the historical period The eigenvector of is the time decay weight, which is used to adjust the importance of the first industrial time series data in different historical time periods. It increases with the time interval (i.e. t - t 'becomes larger) and decreases; the time interval is t - t ', indicating the historical time period t 'Distance to current time period t interval; u The number of historical time periods that need to be considered when calculating the similarity between first industry time series data in different time periods.

[0103] It should be noted that the impact of historical data refers to the impact of the first industrial time series data in the historical period on the second industrial time series data. t The role of calculating the first corrected priority value of the first industrial time series data in a time period (or the current time period). The specific explanation is as follows: The impact of historical data can consider the historical trends and patterns of the data, rather than just relying on the data at the current moment. By calculating the similarity between the current data and the historical data (using cosine similarity ), we can determine whether the current state conforms to the historical pattern. We use time decay weights to adjust the importance of different historical time periods to ensure that more recent data has a greater influence. In the formula, the influence of historical data is mainly reflected in the summation term This item calculates the difference between the current first industry time series data and the past u The weighted similarity of the first industrial time series data in each time period is then converted into an adjustment factor for the preliminary priority assessment value through an exponential function and a fractional form.

[0104] S3.3: Based on the first revised priority value , combined with the impact of dynamic factors, determine the first priority score :

[0105] ;

[0106] in, For the t The change rate of the first industrial time series data in a time period; For the t The first industrial time series data quality factor for each time period; For the t Industrial system fluctuation factors for a period of time; For the t Industrial system status in a time period; For the t -Industrial system status for 1 time period.

[0107] It should be noted that the impact of dynamic factors includes the change rate of the first industrial time series data, the quality of the first industrial time series data, the fluctuation of the industrial system, and the status of the industrial system.

[0108] S3.4: Score based on first priority , sort all the first industrial time series data and get the first priority.

[0109] S4: Based on the strength of correlation and the first priority score , combined with historical abnormal event information, the second priority is obtained, and the industrial data priority calculation based on correlation relationship and sequence processing is completed.

[0110] S4.1: Based on preliminary relevance assessment Correlation with historical abnormal events of the target industrial system to obtain correlation evaluation after considering abnormal events :

[0111] ;

[0112] in, is the impact factor of historical abnormal events; is the number of historical abnormal events related to the current First Industry time series data; is the total number of historical abnormal events.

[0113] S4.2: Relevance assessment based on consideration of unusual events and the strength of correlation , and obtain the final correlation evaluation value :

[0114] ;

[0115] in, is the weight coefficient.

[0116] Based on the final relevance evaluation value and the first priority score , get the second priority :

[0117] ;

[0118] in, is the dynamic weight factor, 0 ≤ ≤ 1; is the adjustment factor.

[0119] In summary, the present invention realizes a comprehensive and accurate priority assessment of industrial time series data by combining the RNN model, historical data influence and dynamic factors. The present invention introduces the correlation strength and historical abnormal event information for dynamic adjustment, which greatly improves the accuracy and adaptability of priority calculation. The improved correlation coefficient and adaptive weight mechanism adopted by the present invention effectively capture the complex dynamic correlation between data, and overcome the limitation of traditional methods that ignore the time series characteristics. Through multi-stage priority calculation and adjustment, it can better cope with emergencies and abnormal situations in industrial environments, and provide strong support for the timely discovery of key information. Overall, the present invention significantly improves the efficiency and accuracy of industrial big data analysis, lays a solid foundation for decision optimization and anomaly detection of intelligent manufacturing systems, and has important theoretical significance and practical application value.

[0120] Example 2

[0121] According to an embodiment of the present invention, an industrial data priority calculation system based on correlation relationship and sequence processing is provided, including a data processing module, a correlation calculation module, a first priority calculation module, and a second priority calculation module.

[0122] The data processing module is used to perform sequence processing on the industrial time series data, capture the time series characteristics and dynamic changes of the industrial time series data, and obtain the first industrial time series data. The sequence processing includes time period division, segment-by-segment aggregation and averaging, and feature extraction.

[0123] The correlation calculation module calculates the time series correlation matrix based on the first industrial time series data, and evaluates the correlation strength between the first industrial time series data by combining the equipment association degree and the process flow dependency.

[0124] The first priority calculation module is used to calculate the first priority score of each first industrial time series data, and obtain the first priority of all first industrial time series data based on the first priority score.

[0125] The second priority calculation module obtains the second priority based on the correlation strength and the first priority score, combined with the historical abnormal event information, and completes the industrial data priority calculation based on correlation relationship and sequence processing.

[0126] Example 2, reference Figure 2 , is an embodiment of the present invention, which is different from the previous embodiment in that:

[0127] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0128] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0129] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0130] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0131] Example 4

[0132] As an embodiment of the present invention, a method for calculating the priority of industrial data based on correlation relationship and sequence processing is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0133] One month's production data of a large manufacturing enterprise was selected as the experimental data set. The data includes the operating parameters of 100 devices, production process data, quality inspection data, environmental monitoring data and energy consumption data, totaling about 10 million records. The present invention is compared with three existing methods: traditional time series analysis (TSA), single LSTM network method (LSTM) and simple correlation analysis (SCA).

[0134] Evaluation metrics include:

[0135] Priority accuracy (PAR): the proportion of high-priority data correctly identified;

[0136] Anomaly Detection Rate (ADR): The proportion of anomaly events that are successfully detected;

[0137] Processing time (PT): the average time (in seconds) required to process 1 million pieces of data;

[0138] Resource Utilization Rate (RUR): Average CPU and memory usage during processing;

[0139] Scalability Score (SS): Performance as data volume increases (1-10 points).

[0140] Table 1: Performance of each method on different evaluation indicators

[0141] ,

[0142] Among them, priority accuracy PAR and anomaly detection rate ADR: As can be seen from Table 1, the present invention is significantly superior to other methods in both PAR and ADR indicators. This shows that the present invention can more accurately identify high-priority data and abnormal events. Especially in ADR, the present invention is 92.7%, which is 7.5 percentage points higher than the second best LSTM method of 85.2%, which is crucial for timely detection and handling of abnormal situations. From the PT indicator in Table 1, it can be seen that the data processing speed of the present invention is the fastest, which is twice as fast as the slowest TSA method. Table 3 further proves that as the data scale increases, the advantages of the present invention become more obvious. When processing 50 million data, the present invention is more than twice as fast as TSA and 33% faster than LSTM.

[0143] Table 2: Priority accuracy (PAR) (%) of each method on different data types

[0144] ,

[0145] Table 2 shows that the present invention performs well in processing various types of data, and PAR is always maintained above 93%. In contrast, other methods perform poorly on certain data types, such as TSA and SCA, whose accuracy is significantly reduced when processing quality inspection data. Therefore, the present invention has better versatility and adaptability.

[0146] Table 3: Processing time (PT) of each method on data sets of different sizes (seconds)

[0147] ,

[0148] Although the resource utilization rate of the present invention is 65%, which is higher than TSA and SCA, it is lower than 80% of LSTM. Considering the high accuracy and processing speed of the present method, this resource utilization level is acceptable.

[0149] The scalability score of the present invention is 9.2, which is much higher than that of other methods, indicating that it can maintain good performance when processing larger-scale data.

[0150] In summary, through the above theoretical experiments and data analysis, it can be seen that the present invention is significantly superior to the existing technical solutions in many aspects such as priority accuracy, anomaly detection capability, processing efficiency, data type adaptability and scalability. In particular, when processing large-scale, multi-type industrial time series data, the present invention shows obvious superiority. These advantages enable the present invention to better meet the high requirements for data processing in modern industrial production, help improve production efficiency, reduce failure risks, and provide more reliable data support for decision-making.

[0151] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An industrial data priority calculation method based on correlation relationship and sequence processing, characterized in that: include: Performing sequence processing on the industrial time series data to obtain first industrial time series data; Based on the first industrial time series data, a time series correlation matrix is ​​calculated, and the correlation strength between the first industrial time series data is evaluated by combining the equipment association degree and the process flow dependency relationship; Calculate a first priority score for each piece of the first industrial time series data, and obtain a first priority for all pieces of the first industrial time series data based on the first priority score; Based on the correlation strength and the first priority score, combined with historical abnormal event information, a second priority is obtained, and the industrial data priority calculation based on correlation relationship and sequence processing is completed; The industrial time series data includes equipment operating parameters, production process data, quality inspection data, environmental monitoring data, and energy consumption data; the sequence processing includes time period division, segment-by-segment aggregation and averaging, and feature extraction; Based on the first industrial time series data, calculating the time series correlation matrix includes the following steps: Based on the sequence processing results, the feature vector set D of the first industrial time series data is constructed i ; Calculate the correlation coefficient r between any two eigenvectors of the first industrial time series data ij ; Among them, r ij represents the correlation coefficient between the feature vectors of the i-th first industrial time series data and the j-th first industrial time series data, i and j are any integers from 1 to the total number n of first industrial time series data, i≠j; Correlation coefficient r ij Calculated by the following formula: Where T is the total number of time periods; and The first industrial time series data D i and D j The average eigenvector of is the feature vector of the i-th first industrial time series data in the t-th time period; is the feature vector of the jth first industrial time series data in the tth time period; Based on the correlation coefficient r ij Constructing a time series correlation matrix C; Combining the equipment association degree and the process dependency, evaluating the correlation strength between the first industrial time series data includes the following steps: According to the device association R e and process dependency R p , determine the preliminary relevance assessment C1; According to the time series correlation matrix C and the preliminary correlation evaluation C1, evaluating the correlation strength between the first industrial time series data; The timing correlation matrix C is expressed by the following formula: Where n is the total number of first industry time series data; in the time series correlation matrix C, r 12 is the correlation coefficient r ij A specific example, where i=1, j=2; The preliminary correlation evaluation C1 is expressed by the following formula: R p =max(Q,M,T’); Among them, α is the device association R e and process dependency R p The weight coefficient is 0≤α≤1; P is the physical connection, 0≤P≤1; I is the information flow, 0≤I≤1; E is the energy flow, 0≤E≤1; Q is the direct connection in the process flow chart, 0≤Q≤1; M is the similarity of material flow, 0≤M≤1; T' is the degree of temporal dependence, 0≤T'≤1; Among them, the expression of physical connectivity P is: The expression of information circulation I is: The expression of energy flow rate E is: The expression of direct connection degree Q is: The expression of logistics flow similarity M is: The expression of timing dependence T' is: The correlation strength is expressed by the following formula: Among them, S ij is the correlation strength between the i-th and j-th time series data; Calculating the first priority score of each piece of the first industrial time series data includes the following steps: Use the RNN model to perform a preliminary priority assessment on the first industrial time series data to obtain a preliminary priority assessment value Based on the preliminary priority assessment value Combined with the impact of historical data, determine the first revision priority value The historical data impact refers to the effect of the first industrial time series data in the historical time period on the calculation of the first revised priority value of the first industrial time series data in the t-th time period; Based on the first revised priority value Combined with the impact of dynamic factors, determine the first priority score B t ; The dynamic factors include the change rate of the first industrial time series data, the quality of the first industrial time series data, the industrial system fluctuation, and the industrial system status; The obtaining the first priority of all the first industrial time series data based on the first priority score comprises sorting all the first industrial time series data according to the first priority score to obtain the first priority; The initial priority assessment value It is expressed by the following formula: in, is the preliminary priority evaluation value of the first industrial time series data in the tth time period; h t is the hidden state output of the recurrent neural network RNN ​​in the tth time period; r t is the reset gate output of the RNN at the tth time period; W h , W r is the weight matrix; b is the bias term; tanh is the hyperbolic tangent activation function; The first modified priority value It is expressed by the following formula: in, is the feature vector of the i-th first industrial time series data in the t-th time period; is the characteristic vector of the i-th first industrial time series data in the historical time period t'; α t-t′ is the time decay weight; u is the number of historical time periods that need to be considered when calculating the similarity between the first industrial time series data in different time periods; The first priority score B t It is expressed by the following formula: Among them, A t is the change rate of the first industrial time series data in the tth time period; X t is the first industrial time series data quality factor in the tth time period; V t is the industrial system fluctuation factor in the tth time period; G t is the industrial system status in the tth time period; G t-1 is the industrial system status in the t-1th time period; The process of calculating the second priority level includes the following steps: Based on the preliminary correlation evaluation C1 and the historical abnormal event correlation of the target industrial system, a correlation evaluation C2 after considering the abnormal event is obtained; Evaluate C2 and correlation strength S based on correlation after considering abnormal events ij , and obtain the final correlation evaluation value C3; Based on the final relevance evaluation value C3 and the first priority score B t , get the second priority; The correlation evaluation C2 after considering abnormal events is expressed by the following formula: Among them, β is the impact factor of historical abnormal events; N a is the number of historical abnormal events related to the current first industry time series data; N total is the total number of historical abnormal events; The final correlation evaluation value C3 is expressed by the following formula: C3=C2 (1+λ S ij ); Among them, λ is the weight coefficient; Second priority B final It is expressed by the following formula: Among them, γ is the dynamic weight factor, 0≤γ≤1; K is the adjustment factor.

2. A system using the industrial data priority calculation method based on correlation relationship and sequence processing as claimed in claim 1, characterized in that: include: A data processing module performs sequence processing on the industrial time series data to obtain first industrial time series data; A correlation calculation module, used to calculate a time series correlation matrix based on the first industrial time series data, and evaluate the correlation strength between the first industrial time series data in combination with the equipment association degree and the process flow dependency; A first priority calculation module, used to calculate a first priority score for each piece of the first industrial time series data, and obtain a first priority for all of the first industrial time series data based on the first priority score; The second priority calculation module obtains the second priority based on the correlation strength and the first priority score in combination with the historical abnormal event information, and completes the industrial data priority calculation based on correlation relationship and sequence processing.

3. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method for calculating the priority of industrial data based on correlation relationship and sequence processing described in claim 1 are implemented.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for calculating the priority of industrial data based on correlation relationship and sequence processing described in claim 1 are implemented.

Images