A big data BI analysis system

Abstract

The present invention discloses a big data BI analysis system, which specifically relates to the technical field of big data analysis, including determining the probability distribution of production feature data through the production feature data of production components in historical operation, analyzing the probability density of each production feature of the production component and the objective function of each feature, obtaining historical feature information of the production component, collecting real-time production feature data of similar production components during operation, indirectly measuring the fluctuation amplitude through peak and valley values ​​extracted after continuous wavelet transform, obtaining stability information of similar production components, and obtaining difference information of similar production components by comparing the similarity with historical feature data, comprehensively analyzing the stability information and difference information of similar production components, and selectively replacing and adjusting similar production components. The present invention helps to accurately evaluate the performance and stability of components and optimize decision-making in the production process.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and more specifically, to a big data BI analysis system. Background Art

[0002] With the rapid development of information technology and data processing technology, especially the rise of big data technology, enterprises and organizations have generated a large amount of structured and unstructured data in production, operation, management and other aspects;

[0003] In the prior art, the invention patent CN109299082B, a big data analysis method, can freely select the required analysis components, solve various data analysis problems, and adapt to various application scenarios. The invention patent CN110990384B, a big data platform BI analysis method, improves the data quality of data cleaning, thereby enabling the BI analysis system to provide more accurate analysis results.

[0004] However, when applied to the production process, existing technologies only rely on the analysis of historical data or real-time data, and lack methods to combine the two. This makes it difficult to accurately and comprehensively evaluate production components, unable to timely identify changes in component status during the production process, and difficult to dynamically adjust according to the actual status of production components.

[0005] In order to solve the above-mentioned defects, a technical solution is now provided. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a big data BI analysis system to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A big data BI analysis system includes a generation component extraction module, a historical data analysis module, a scoring module, a real-time data analysis module, an evaluation module, and a threshold comparison module, with signal connections between the modules;

[0009] Generate a component extraction module to identify the same production components or production components with similar functions in the production process, mark them as similar production components, and obtain production feature data of similar production components;

[0010] A historical data analysis module is used to use the production characteristic data of the production component in the historical operation to smooth the distribution of the production characteristic data using a kernel function using a kernel density estimation method to determine the probability distribution of the production characteristic data;

[0011] The scoring module is used to obtain the historical feature information of the production component by obtaining the probability density of each production feature of the production component and the objective function of each feature, and generate an attention scoring coefficient;

[0012] The real-time data analysis module is used to collect real-time production characteristic data of similar production components during operation. The peak and valley values ​​extracted after continuous wavelet transform are used to indirectly measure the fluctuation amplitude, thereby obtaining stability information of similar production components. The module also obtains difference information of similar production components by comparing the similarity with historical characteristic data.

[0013] The evaluation module is used to comprehensively analyze the stability information and difference information of similar production components to obtain the real-time evaluation coefficient of the production components;

[0014] The threshold comparison module is used to compare the attention score coefficient and real-time evaluation coefficient of the production component with the pre-set threshold, and selectively replace and adjust similar production components.

[0015] In a preferred embodiment, obtaining historical characteristic information of a production component includes:

[0016] The historical characteristic information of the production components is represented by the attention score coefficient;

[0017] The logic for obtaining the attention score coefficient is as follows: determining the production characteristics of similar production components, obtaining a sample data set of the production characteristics of each similar production component based on the historical data of each similar production component, and marking the sample data set of the production characteristics of each similar production component as: ,in, , n=, 1, 2, 3, ..., N, N is a positive integer, n is the number of the same type of production components, is the sample data of the production characteristics of the n-th similar production component, i is the number of sample data of the production characteristics of the n-th similar production component;

[0018] The kernel density estimation method is used to smooth the distribution of sample data of production characteristics using the kernel function. The kernel density estimation formula of production characteristics is: ,in, is the probability density estimate when the sample data of the j-th production feature in the n-th similar production component is x, j = 1, 2, 3, ..., J, J is a positive integer, j is the number of the production feature in the same production component, is the number of the sample data of the jth production feature in the nth similar production component, m=1, 2, 3, ..., i, i is a positive integer, h is the bandwidth parameter, K is the Gaussian kernel function, ;

[0019] Determine the objective function of the production feature and label the objective function of the production feature as: , where j = 1, 2, 3, ..., J, J is a positive integer, and j is the number of the production feature of the same type of production component;

[0020] Calculate the feature attention weight coefficient. The calculation formula of the feature attention weight coefficient is: ,in, is the feature attention weight coefficient of the j-th production feature of the n-th similar production component, is the maximum value in the sample data of the jth production feature of the nth similar production component, is the minimum value in the sample data of the jth production feature of the nth similar production component, is the probability density estimate of the j-th production feature of the n-th similar production component, is the objective function of the j-th production feature of the n-th similar production component;

[0021] Calculate the attention score coefficient, the calculation formula is: ;in, is the attention score coefficient of the nth similar production component, It is the specific weight coefficient of different production characteristics in the same production component.

[0022] In a preferred embodiment, obtaining stability information of similar production components includes:

[0023] The stability information of similar production components is expressed by the stability difference coefficient;

[0024] The logic for obtaining the stable difference coefficient is: obtaining the real-time production feature data of the production components within the monitoring interval, and marking the real-time production feature data of the production components within the monitoring interval as: , where q = 1, 2, 3, ..., Q, Q is a positive integer, and q is the number of the production feature data in the monitoring interval of the j-th production feature;

[0025] By performing discrete continuous wavelet transform on discrete real-time production characteristic data, the changes in the time series of production characteristic data in the monitoring interval are determined, and the peak coefficient and valley coefficient of the production characteristic data in the monitoring interval are identified;

[0026] The expression of the peak coefficient of real-time production characteristic data is: Where, is the peak coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, is the local peak value of the translation position;

[0027] The expression of the valley coefficient of real-time production characteristic data is: Where, is the valley coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, is the local valley value of the translation position;

[0028] Calculate the peak-to-valley difference coefficient. The calculation formula for the peak-to-valley difference coefficient is: in, is the peak-valley difference coefficient of the j-th production feature, g = 1, 2, 3, ..., G, G is a positive integer, and g is the number of the peak and valley values ​​captured in the monitoring interval;

[0029] Calculate the stable difference coefficient, the calculation formula is: in, is the coefficient of stable variation.

[0030] In a preferred embodiment, obtaining difference information of similar production components includes:

[0031] The difference information of similar production components is expressed by the portrait similarity coefficient;

[0032] The logic for obtaining the portrait similarity coefficient is as follows: based on the real-time production feature data and historical production feature data of the production component within the monitoring interval, the historical production feature data of the production component is marked as: Determine the portrait similarity coefficient using similarity calculation method;

[0033] Calculate the portrait similarity coefficient using the following formula: in, is the image similarity coefficient.

[0034] In a preferred embodiment, obtaining a real-time evaluation coefficient of a production component includes:

[0035] The stable difference coefficient and the portrait similarity coefficient are used to construct a real-time evaluation model to generate a real-time evaluation coefficient. The expression of the real-time evaluation coefficient is: ;in, is the real-time evaluation coefficient, is the proportional coefficient of the stable difference coefficient, is the proportional coefficient of the image similarity coefficient, and Both are greater than 0.

[0036] In a preferred embodiment, comparing the attention score coefficient and the real-time evaluation coefficient of the production component with a preset threshold value includes:

[0037] Set the real-time evaluation coefficient threshold and the attention score coefficient threshold, compare the real-time evaluation coefficient and the attention score coefficient of the production component with the real-time evaluation coefficient threshold and the attention score coefficient threshold, and generate the following situation:

[0038] If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is greater than the attention score coefficient threshold, no warning signal is generated;

[0039] If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is less than the attention score coefficient threshold, an adjustment signal is generated;

[0040] If the real-time evaluation coefficient is greater than the real-time evaluation coefficient threshold, an early warning signal is generated.

[0041] Technical effects and advantages of the present invention:

[0042] The present invention analyzes the distribution of historical production characteristic data of similar production components, quantifies production components with better performance among similar production components, and collects real-time production characteristic data of production components. By comparing with the historical production characteristic data and analyzing the stability of the real-time production characteristic data, the historical status of the current production component is evaluated. The present invention helps to accurately evaluate the performance and stability of the component and optimize decision-making in the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0044] Figure 1 This is a structural diagram of a big data BI analysis system of the present invention. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Example

[0046] like Figure 1 A structural diagram of a big data BI analysis system is given, including a generation component extraction module, a historical data analysis module, a scoring module, a real-time data analysis module, an evaluation module, and a threshold comparison module, with signal connections between the modules.

[0047] Generate a component extraction module to identify the same production components or production components with similar functions in the production process, mark them as similar production components, and obtain production feature data of similar production components;

[0048] A historical data analysis module is used to use the production characteristic data of the production component in the historical operation to smooth the distribution of the production characteristic data using a kernel function using a kernel density estimation method to determine the probability distribution of the production characteristic data;

[0049] The scoring module is used to obtain the historical feature information of the production component by obtaining the probability density of each production feature of the production component and the objective function of each feature, and generate an attention scoring coefficient;

[0050] The real-time data analysis module is used to collect real-time production characteristic data of similar production components during operation. The peak and valley values ​​extracted after continuous wavelet transform are used to indirectly measure the fluctuation amplitude, thereby obtaining stability information of similar production components. The module also obtains difference information of similar production components by comparing the similarity with historical characteristic data.

[0051] The evaluation module is used to comprehensively analyze the stability information and difference information of similar production components to obtain the real-time evaluation coefficient of the production components;

[0052] The threshold comparison module is used to compare the attention score coefficient and real-time evaluation coefficient of the production component with the pre-set threshold, and selectively replace and adjust similar production components.

[0053] In the factory, the characteristic data of each production component is recorded and stored in real time through sensors, edge devices or embedded controllers. The characteristic data of different production components are different, while the characteristic data of the same production components or production components that can achieve the same production effect are the same.

[0054] Storing characteristic data of the same production components or production components that can achieve the same production effect in the cloud or in the same database makes the analysis of the same type of components faster when analyzing the same production components.

[0055] Among them, the characteristic data of the production components may include the operation data, performance data, maintenance data and environmental data of the production components. Each type of data serves as a characteristic of the production component. The same production components or production components with similar functions in the production process are marked as similar production components. Similar production components have the same type and quantity of production characteristics.

[0056] A production component in the production process is analyzed, and the probability distribution of the production feature data is determined through the production feature data of the production component in historical operation. Because the probability distribution of production feature data is not easy to obtain, the kernel density estimation method is adopted to smooth the distribution of production feature data using the kernel function to obtain the probability density of each production feature of the production component, and the relationship between each production feature and production demand, production process and production requirements is combined to determine the objective function of each feature to obtain the historical feature information of the production component. After collection, the historical feature information of the production component is represented by the attention score coefficient, and the performance of the same production feature of the same production component is quantified according to the attention score coefficient.

[0057] Components produced in the same category share the same production characteristics. By analyzing sample data of these production characteristics, the following information can be obtained:

[0058] By analyzing sample data, we can understand the performance of similar components under different production characteristics, including the impact of characteristics on components, the stability of components under specific conditions, and the performance advantages and disadvantages.

[0059] Based on the sample data of production characteristics, methods such as kernel density estimation can be used to obtain the probability density function of each production characteristic. The probability density function describes the distribution trend of a production characteristic within a specific range. If the probability density of a certain characteristic value is high, it means that the component performs relatively stably and frequently near the characteristic value. If the probability density of the characteristic value is low, it may indicate that the component performs poorly or unstable under this condition.

[0060] Based on the probability density and performance of similar production components under different production characteristics, component selection can be optimized. By evaluating the stability of components under key characteristics, the best performing components can be selected for production.

[0061] The logic for obtaining the attention score coefficient is as follows: determining the production characteristics of similar production components, obtaining a sample data set of the production characteristics of each similar production component based on the historical data of each similar production component, and marking the sample data set of the production characteristics of each similar production component as: ,in, , n=, 1, 2, 3, ..., N, N is a positive integer, n is the number of the same type of production components, is the sample data of the production characteristics of the n-th similar production component, i is the number of sample data of the production characteristics of the n-th similar production component;

[0062] The kernel density estimation method is used to smooth the distribution of sample data of production characteristics using the kernel function. The kernel density estimation formula of production characteristics is: ,in, is the probability density estimate when the sample data of the j-th production feature in the n-th similar production component is x, j = 1, 2, 3, ..., J, J is a positive integer, j is the number of the production feature in the same production component, m The number of the sample data of the j-th production feature in the same production component, m = 1, 2, 3, ..., i, i is a positive integer, h is the bandwidth parameter, K is the Gaussian kernel function, ;

[0063] It should be noted that similar production components have the same production characteristics. By analyzing the sample data of production characteristics, we can obtain the performance of similar production components under different production characteristics. Based on the probability density of similar production components under different production characteristics, we can determine the distribution trend of different production components in the same production components under a certain production characteristic.

[0064] The bandwidth parameter is set by professional staff. If the bandwidth is too large, the estimation result will be too smooth and lose data details; if the bandwidth is too small, the estimation result will be overfitted and the noise will be amplified. The bandwidth can be selected by rule method. According to Silverman rule, in, is the standard deviation of the sample data.

[0065] Determine the objective function of the production feature and label the objective function of the production feature as: , where j = 1, 2, 3, ..., J, J is a positive integer, and j is the number of the production feature of the same type of production component;

[0066] It should be noted that the objective function of production characteristics is set by staff in professional fields. The design of the objective function is mainly determined based on the relationship between production needs and characteristics. It can reflect the preference of production tasks for specific characteristics. By combining production process requirements, historical data, equipment protection needs and other factors, the objective function can be flexibly adjusted in different production scenarios, enabling the system to effectively evaluate and optimize the performance of components.

[0067] For example, if the production characteristic is welding error, and the production requirement is to make the welding error as small as possible, you can define: Where k is the welding error, It is the adjustment parameter for production characteristic welding error.

[0068] Calculate the feature attention weight coefficient. The calculation formula of the feature attention weight coefficient is: ,in, is the feature attention weight coefficient of the j-th production feature of the n-th similar production component, is the maximum value in the sample data of the jth production feature of the nth similar production component, is the minimum value in the sample data of the jth production feature of the nth similar production component, is the probability density estimate of the j-th production feature of the n-th similar production component, is the objective function of the j-th production feature of the n-th similar production component;

[0069] It should be noted that the feature attention weight coefficient reflects the contribution of the feature to the task under different feature values, which helps to determine which feature values ​​are most meaningful for the task. Feature values ​​with high density and that meet task requirements will have higher weights. The feature weights can be normalized by integration to ensure that the sum of the weights of all features is 1.

[0070] Calculate the attention score coefficient, the calculation formula is: ;in, is the attention score coefficient of the nth similar production component, It is the specific weight coefficient of different production characteristics in the same production component.

[0071] It should be noted that the specific weight coefficients of different production characteristics in the same type of production components reflect the relative importance of different production characteristics. The larger the specific weight coefficient of the characteristic, the greater the impact of its performance on the final evaluation result, and the influence of the characteristic with a smaller weight is relatively weaker. The specific weight coefficient can be determined by the hierarchical analysis method and is set by staff in professional fields. The attention score coefficient reflects the overall production adaptability and performance of the component through historical data, and more accurately evaluates the performance of similar production components under different characteristics, thereby helping factories make better decisions when selecting and optimizing production equipment. The larger the attention score coefficient, the better the performance of the component among similar production components.

[0072] By determining the stability information and difference information of similar production components based on the real-time production feature data of similar production components during operation, after collection, the stability information of similar production components is represented by the stability difference coefficient, and the difference information of similar production components is represented by the portrait similarity coefficient.

[0073] It should be noted that real-time production characteristic data provides the actual performance of production components in the current production task. Since the production environment and working conditions may change, real-time production characteristic data reflects the current operating status of the component. By comparing real-time data with historical data, discrepancies are identified. Based on this discrepancy information, the factory can determine whether the component is currently in normal operation or whether there is any deviation from the ideal state.

[0074] Stability information refers to the determination of whether the performance is consistent over time by generating real-time production characteristic data of components. For example, if the performance parameters (such as temperature and vibration) of a component fluctuate greatly, it may indicate that the component is unstable, which may indicate future failures or performance degradation. Stability is usually measured by analyzing the volatility or rate of change of real-time data. If the fluctuation range of real-time data is small, it means that the component is relatively stable. If the fluctuation range is large, it may indicate instability.

[0075] The logic for obtaining the stable difference coefficient is: obtaining the real-time production feature data of the production components within the monitoring interval, and marking the real-time production feature data of the production components within the monitoring interval as: , where q = 1, 2, 3, ..., Q, Q is a positive integer, and q is the number of the production feature data in the monitoring interval of the j-th production feature;

[0076] By performing discrete continuous wavelet transform on discrete real-time production characteristic data, the changes in the time series of production characteristic data in the monitoring interval are determined, and the peak coefficient and valley coefficient of the production characteristic data in the monitoring interval are identified;

[0077] The expression of the peak coefficient of real-time production characteristic data is: Where, is the peak coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, is the local peak value of the translation position;

[0078] The expression of the valley coefficient of real-time production characteristic data is: Where, is the valley coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, is the local valley value of the translation position;

[0079] Calculate the peak-to-valley difference coefficient. The calculation formula for the peak-to-valley difference coefficient is: in, is the peak-valley difference coefficient of the j-th production feature, g = 1, 2, 3, ..., G, G is a positive integer, and g is the number of the peak and valley values ​​captured in the monitoring interval;

[0080] Calculate the stable difference coefficient, the calculation formula is: in, is the coefficient of stable variation.

[0081] It should be noted that the monitoring interval is a time series. The production characteristic data in the monitoring interval are sorted according to the time series. The monitoring interval is set by professional staff. When using wavelet transform (CWT) for time series analysis, the scale a and translation position are usually set by professional staff or researchers based on data characteristics, analysis objectives and application scenarios, and try to meet the requirements of determining the maximum and minimum values ​​of production characteristic data at the same scale. The larger the stable difference coefficient, the greater the fluctuation of the characteristic data when the production component is currently running.

[0082] Difference information refers to the similarity between the characteristic data of the current production component and the characteristic data of the historical production component. By comparing the similarity between the two, the difference in characteristic performance between the current component and the historical component is judged, and then the stability and performance of the current component are evaluated to see whether they meet expectations, or whether there are potential abnormalities. If the similarity is high, it means that the performance of the current production component is consistent with that of the historical component, which may indicate that the production process is stable and meets expectations; if the similarity is low, it may indicate that the behavior of the current component is different from the history, and there may be anomalies or the status has changed.

[0083] The logic for obtaining the portrait similarity coefficient is as follows: based on the real-time production feature data and historical production feature data of the production component within the monitoring interval, the historical production feature data of the production component is marked as: Determine the portrait similarity coefficient using similarity calculation method;

[0084] It should be noted that the historical production feature data is set based on the historical data of the production components and is set by professional staff. The similarity calculation methods usually include cosine similarity, Manhattan distance, and Euclidean distance. This embodiment uses Euclidean distance to obtain the portrait similarity coefficient.

[0085] Calculate the portrait similarity coefficient using the following formula: in, is the image similarity coefficient.

[0086] It can be seen from the formula that the smaller the portrait similarity coefficient is, the higher the similarity between the characteristic data of the current production component and the historical data is, there is no obvious abnormality or failure, and the task can be effectively completed within the expected working range.

[0087] The stable difference coefficient and the portrait similarity coefficient are used to construct a real-time evaluation model to generate a real-time evaluation coefficient. The expression of the real-time evaluation coefficient is: ;in, is the real-time evaluation coefficient, is the proportional coefficient of the stable difference coefficient, is the proportional coefficient of the image similarity coefficient, and Both are greater than 0.

[0088] The larger the real-time evaluation coefficient is, the larger the stability difference coefficient and the portrait similarity coefficient are, which means that the real-time operating status of the production component is more unstable and does not conform to the operating status of the historical production component. The production component cannot be adjusted based on the production component and the production experience of the production component. Conversely, the smaller the real-time evaluation coefficient is, the smaller the stability difference coefficient and the portrait similarity coefficient are, which means that the real-time operating status of the production component is more stable and conforms to the operating status of the historical production component. The production component can be adjusted based on the production component and the production experience of the production component.

[0089] Set the real-time evaluation coefficient threshold and the attention score coefficient threshold, compare the real-time evaluation coefficient and the attention score coefficient of the production component with the real-time evaluation coefficient threshold and the attention score coefficient threshold, and generate the following situation:

[0090] If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is greater than the attention score coefficient threshold, no warning signal is generated, indicating that the current operating status of the production component is good and the production component does not need to be adjusted;

[0091] If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is less than the attention score coefficient threshold, an adjustment signal is generated, indicating that the operating status of the current production component meets the expected operating status, but the production performance of the current production component is poor. The production component can be adjusted first, and the current production component can be replaced with a similar production component.

[0092] If the real-time evaluation coefficient is greater than the real-time evaluation coefficient threshold, an early warning signal is generated, indicating that the current production component is operating abnormally and needs to be replaced with a similar production component as soon as possible.

[0093] The present invention analyzes the distribution of historical production characteristic data of similar production components, quantifies production components with better performance among similar production components, and collects real-time production characteristic data of production components. By comparing with the historical production characteristic data and analyzing the stability of the real-time production characteristic data, the historical status of the current production component is evaluated. The present invention helps to accurately evaluate the performance and stability of the component and optimize decision-making in the production process.

[0094] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0095] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0096] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0097] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0098] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0100] 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 application, 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 and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0101] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A big data BI analysis system, characterized in that: It includes a generation component extraction module, a historical data analysis module, a scoring module, a real-time data analysis module, an evaluation module, and a threshold comparison module, with signal connections between the modules; Generate a component extraction module to identify the same production components or production components with similar functions in the production process, mark them as similar production components, and obtain production feature data of similar production components; A historical data analysis module is used to use the production characteristic data of the production component in the historical operation to smooth the distribution of the production characteristic data using a kernel function using a kernel density estimation method to determine the probability distribution of the production characteristic data; The scoring module is used to obtain the historical feature information of the production component by obtaining the probability density of each production feature of the production component and the objective function of each feature, and generate an attention scoring coefficient; The real-time data analysis module is used to collect real-time production characteristic data of similar production components during operation. The peak and valley values ​​extracted after continuous wavelet transform are used to indirectly measure the fluctuation amplitude, thereby obtaining stability information of similar production components. The module also obtains difference information of similar production components by comparing the similarity with historical characteristic data. The evaluation module is used to comprehensively analyze the stability information and difference information of similar production components to obtain the real-time evaluation coefficient of the production components; The threshold comparison module is used to compare the attention score coefficient and real-time evaluation coefficient of the production component with the pre-set threshold, and selectively replace and adjust similar production components.

2. A big data BI analysis system according to claim 1, characterized in that: Obtain historical characteristic information of production components, including: The historical characteristic information of the production components is represented by the attention score coefficient; The logic for obtaining the attention score coefficient is as follows: determine the production characteristics of similar production components, obtain a sample data set of the production characteristics of each similar production component based on the historical data of each similar production component, and mark the sample data set of the production characteristics of each similar production component as: TZ n , among which, TZ n ={x1, x2, x3, ..., x i }, n=1, 2, 3, ..., N, N is a positive integer, n is the number of the same type of production components, x1, x2, x3, ..., x i is the sample data of the production characteristics of the n-th similar production component, and i is the number of sample data of the production characteristics of the n-th similar production component; The kernel density estimation method is used to smooth the distribution of sample data of production characteristics using the kernel function. The kernel density estimation formula of production characteristics is: in, is the probability density estimate when the sample data of the j-th production feature in the n-th similar production component is x, j = 1, 2, 3, ..., J, J is a positive integer, j is the number of the production feature in the same production component, m is the number of the sample data of the j-th production feature in the n-th similar production component, m = 1, 2, 3, ..., i, i is a positive integer, h is the bandwidth parameter, K is the Gaussian kernel function, Determine the objective function of the production feature and mark the objective function of the production feature as: g j (x), where j = 1, 2, 3, ..., J, where J is a positive integer and j is the number of the production feature of the same type of production component; Calculate the feature attention weight coefficient. The calculation formula of the feature attention weight coefficient is: Among them, AT n,j is the feature attention weight coefficient of the j-th production feature of the n-th similar production component, x max is the maximum value in the sample data of the jth production feature of the nth similar production component, x min is the minimum value in the sample data of the jth production feature of the nth similar production component, is the probability density estimate of the jth production feature of the nth similar production component, g n,j (x) is the objective function of the jth production feature of the nth similar production component; Calculate the attention score coefficient, the calculation formula is: Among them, PF n,j is the attention score coefficient of the nth similar production component, θ j It is the specific weight coefficient of different production characteristics in the same production component.

3. A big data BI analysis system according to claim 2, characterized in that: Obtain stability information for similarly produced components, including: The stability information of similar production components is expressed by the stability difference coefficient; The acquisition logic of the stable difference coefficient is: obtain the real-time production feature data of the production components in the monitoring interval, and mark the real-time production feature data of the production components in the monitoring interval as: SSTZ jq , where q = 1, 2, 3, ..., Q, Q is a positive integer, and q is the number of the production feature data in the monitoring interval of the j-th production feature; By performing discrete continuous wavelet transform on discrete real-time production characteristic data, the changes in the time series of production characteristic data in the monitoring interval are determined, and the peak coefficient and valley coefficient of the production characteristic data in the monitoring interval are identified; The expression of the peak coefficient of real-time production characteristic data is: Where, is the peak coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, b max is the local peak value of the translation position; The expression of the valley coefficient of real-time production characteristic data is: Where, is the valley coefficient of the real-time production feature data in the j-th production feature, is the conjugate of the wavelet function, a is the scale, b min is the local valley value of the translation position; Calculate the peak-to-valley difference coefficient. The calculation formula for the peak-to-valley difference coefficient is: in, is the peak-valley difference coefficient of the j-th production feature, g = 1, 2, 3, ..., G, G is a positive integer, and g is the number of the peak and valley values ​​captured in the monitoring interval; Calculate the stable difference coefficient, the calculation formula is: Among them, WD cy is the coefficient of stable variation.

4. A big data BI analysis system according to claim 3, characterized in that: Obtain information on differences between similarly produced components, including: The difference information of similar production components is expressed by the portrait similarity coefficient; The logic for obtaining the similarity coefficient of the portrait is as follows: based on the real-time production feature data and historical production feature data of the production component within the monitoring interval, the historical production feature data of the production component is marked as: LSTZ j , use similarity calculation method to determine the portrait similarity coefficient; Calculate the portrait similarity coefficient using the following formula: Among them, HX xs is the image similarity coefficient.

5. A big data BI analysis system according to claim 4, characterized in that: Get real-time evaluation coefficients for production components, including: The stable difference coefficient and the portrait similarity coefficient are used to construct a real-time evaluation model to generate a real-time evaluation coefficient. The expression of the real-time evaluation coefficient is: Among them, PG ss is the real-time evaluation coefficient, β1 is the proportional coefficient of the stable difference coefficient, and β2 is the proportional coefficient of the image similarity coefficient. Both β1 and β2 are greater than 0.

6. A big data BI analysis system according to claim 1, characterized in that: Compare the production component's attention score coefficient and real-time evaluation coefficient with pre-set thresholds, including: Set the real-time evaluation coefficient threshold and the attention score coefficient threshold, compare the real-time evaluation coefficient and the attention score coefficient of the production component with the real-time evaluation coefficient threshold and the attention score coefficient threshold, and generate the following situation: If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is greater than the attention score coefficient threshold, no warning signal is generated; If the real-time evaluation coefficient is less than the real-time evaluation coefficient threshold, and the attention score coefficient is less than the attention score coefficient threshold, an adjustment signal is generated; If the real-time evaluation coefficient is greater than the real-time evaluation coefficient threshold, an early warning signal is generated.

Images