A method for deep mining and fusion processing of multi-source data information for performance preservation

Abstract

The present invention belongs to the field of data processing technology, and discloses a method for deep mining and fusion processing of multi-source data information for performance maintenance: using different data acquisition devices to collect multi-source heterogeneous data representing the operating status and performance status of terminal equipment, performing data fusion processing on the multi-source heterogeneous data, performing deep mining of the internal relationship of the fused data, determining the key factors affecting the performance status of the terminal equipment, and performing comprehensive diagnosis on the operating status of the terminal equipment and judging its change trend. The technical solution of the present invention can reliably evaluate the performance maintenance of the equipment, thereby improving the efficiency and accuracy of equipment management.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a multi-source data information deep mining and fusion processing method for performance preservation. Background Art

[0002] With the development of industrial automation and intelligence, maintaining equipment performance has become the key to equipment management. The operating status and performance status of these terminal devices are directly related to the stability and efficiency of the entire system. Therefore, continuous monitoring and evaluation of the performance of terminal devices and timely discovery and resolution of potential problems have become the key to ensuring the normal and efficient operation of the system.

[0003] However, in actual applications, the data generated by terminal devices often have the characteristics of multi-source, heterogeneity and massiveness. Multi-source means that the data comes from different collection devices and systems; heterogeneity means that there are differences in data formats, types and structures; massiveness means that the amount of data is huge and it is very complicated to process. These characteristics have brought great challenges to the performance monitoring and evaluation of terminal devices. Traditional equipment performance evaluation methods rely on a single data source and cannot fully reflect the performance status of the equipment. At the same time, due to differences in data formats and structures, data between different data sources are often difficult to effectively integrate and process, resulting in serious information islands and the inability to form a complete and coherent performance monitoring system.

[0004] Therefore, a method for deep mining and fusion processing of multi-source data information for performance preservation is provided. By deeply mining the internal relationships of the fused data, the key factors affecting the performance status of the terminal equipment are determined, providing a scientific basis for performance optimization and maintenance of the equipment. Summary of the invention

[0005] In view of the above-mentioned technical problems, the present invention provides a multi-source data information deep mining and fusion processing method for performance preservation, which comprises the following steps:

[0006] Step 1: Use different data collection devices to collect multi-source heterogeneous data representing the operating status and performance status of the terminal device;

[0007] Step 2: Perform data fusion processing on multi-source heterogeneous data;

[0008] Step 3: Conduct in-depth mining of the internal relationships of the fused data to determine the key factors that affect the performance status of the terminal equipment;

[0009] Step 4: Conduct a comprehensive diagnosis of the operating status of the terminal equipment and determine its changing trend.

[0010] Specifically, the multi-source heterogeneous data includes operation process data, operation status data and control instruction data; among which,

[0011] The operation process data includes the specific operation and process data of the terminal device during operation; the operation status data includes the status information data of the terminal device during operation; and the control instruction data includes the instruction data issued by the control system of the terminal device.

[0012] Specifically, step 2 includes:

[0013] Step 21: Convert data types of multi-source heterogeneous data;

[0014] Step 22, clearing noise from the multi-source heterogeneous data after data type conversion, performing inference calculation to fill in missing and incomplete data, correcting abnormal data, clearing duplicate data, and removing erroneous data;

[0015] Step 23: Classify and store the data processed in step 22.

[0016] Specifically, a machine learning algorithm is used to perform pattern recognition on the fused multi-source heterogeneous data to determine the key factors affecting the performance status of the terminal equipment, wherein the machine learning algorithm is one of a decision tree, a random forest, a support vector machine, and a neural network.

[0017] Specifically, step 4 includes:

[0018] A correlation matrix is ​​constructed based on key elements, where any row of the correlation matrix is ​​a data sequence of any key element, and any column represents a time point;

[0019] Use the grey correlation calculation formula to calculate the correlation between each key factor and other key factors, and calculate the weight value of any key factor based on the correlation;

[0020] Obtain a first data sequence of any key element, and repeat this step based on the standard deviation of the first data sequence until the standard deviations of all key elements are obtained;

[0021] The comprehensive performance evaluation value of the terminal equipment is calculated based on the weight value and standard deviation.

[0022] Specifically, step 4 includes:

[0023] The indicator data of key elements are tested through the information system performance evaluation model to check the abnormal information of each indicator data, and the cause of the abnormal information is located;

[0024] The fault information is connected to the alarm platform for unified formatting, and the alarm information is mined and combined and compressed through the association mining strategy. Finally, the compressed alarm information is sent to relevant personnel for processing;

[0025] Predict the data value of each indicator at the next moment and perceive the performance status of the terminal equipment in advance.

[0026] Specifically, the information system performance evaluation model includes an anomaly detection module and a fault location module. The anomaly detection module is used to automatically detect the abnormal conditions of each performance indicator. If an indicator abnormality occurs, it is submitted to the fault location module for fault cause identification; the fault location module is used to perform causal inference based on the correlation between indicators and events, and finally lock the location of suspected fault events.

[0027] Specifically, the abnormal information of each indicator data is checked including a fixed threshold detection method and a dynamic threshold detection method;

[0028] Fixed threshold detection method is used for the first indicator data with stable fluctuations. By setting the threshold range on the server side, if the current value of any first indicator data exceeds the corresponding threshold range, it is considered to be abnormal;

[0029] The dynamic threshold detection method is used for the second indicator data with obvious trend and periodicity. The exponential smoothing method is used to learn the original trend of the data. The time series data of the performance indicator is segmented by period and a dynamic time window is set. The data in the same time window in different periods are then regarded as obeying the normal distribution. Different dynamic thresholds are set for the data in different time windows according to the Laida criterion.

[0030] Specifically, a time series prediction method is used to predict the data value of each indicator at the next moment, wherein the time series prediction method is one of simple smoothing, Holt linear trend model, damped trend model, simple seasonal model, and Winter model.

[0031] Specifically, in step 4, the comprehensive diagnosis also includes evaluating the current number of runs and the remaining number of runs, the current running time and the remaining running time, the current performance status and the remaining performance status.

[0032] The present invention discloses a multi-source data information deep mining and fusion processing method for performance preservation, and the beneficial effects are at least as follows:

[0033] 1. Collect multi-source heterogeneous data, effectively integrate and utilize data from different collection devices, more comprehensively reflect the operating status and performance status of terminal devices, and avoid information loss and deviation that may be caused by a single data source.

[0034] 2. Use machine learning and time series prediction technology to predict the data value of each indicator at the next moment, predict the performance trend of the equipment, perceive potential failure risks in advance, and provide a scientific basis for preventive maintenance.

[0035] 3. Through the analysis and weight calculation of key factors, while providing scientific data support for equipment maintenance and operation, it reduces the amount of calculation, improves the calculation speed, helps decision makers formulate more reasonable maintenance plans and operation strategies, reduces maintenance costs, and improves equipment utilization efficiency.

[0036] 4. Comprehensive diagnostic results include evaluation of current and remaining operation times, duration and performance status, providing important decision support for equipment maintenance and management. Based on the performance evaluation results, equipment can be optimized and scheduled to improve resource utilization efficiency and reduce energy consumption and costs. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0038] Figure 1 The present invention is a flowchart of a multi-source data information deep mining and fusion processing method for performance preservation. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the specific embodiments described herein are only used to explain the present invention and are part of the embodiments of the present invention, rather than 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 work belong to the scope of protection of the present invention.

[0040] It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0041] Figure 1FIG. 1 is a flow chart of an embodiment of a multi-source data information deep mining and fusion processing method for performance preservation provided by the present invention, and the flow chart specifically includes the following steps:

[0042] Step 1: Use different data acquisition devices to collect multi-source heterogeneous data representing the operating status and performance status of the terminal device.

[0043] Data acquisition equipment includes different monitoring devices, sensors, etc.

[0044] Specifically, the multi-source heterogeneous data includes operation process data, operation status data and control instruction data; among which,

[0045] The operation process data includes the specific operation and process data of the terminal device during operation; the operation status data includes the status information data of the terminal device during operation; and the control instruction data includes the instruction data issued by the control system of the terminal device.

[0046] In order to obtain the performance status of key components or parts in different equipment or systems, different monitoring equipment and sensors are used to collect and process data on the corresponding operating status and performance status. Since these data are obtained through different means and methods, there will be certain differences in source, type, format and structure. Some are operating process data, some are operating status data, some are control instruction data, and some are historical experience data.

[0047] Operation process data records the specific operations and processes of the equipment during operation, including information such as the start, stop, operating speed, and working cycle of the equipment. Operation status data refers to the status information of the equipment during operation, such as real-time monitoring parameters such as temperature, pressure, vibration, current, and voltage. Control instruction data refers to the instructions issued by the equipment control system, including operation instructions such as start, stop, and adjust parameters. Historical experience data is data accumulated based on the experience of previous equipment operation and maintenance, and may include historical information such as fault records, maintenance records, and replacement parts records.

[0048] Step 2: Perform data fusion processing on multi-source heterogeneous data.

[0049] Multi-source heterogeneous data comes from various sources, types and formats, and needs to be integrated and analyzed through specific data preprocessing and fusion processing technologies in order to accurately evaluate the performance of the equipment.

[0050] Specifically, step 2 includes:

[0051] Step 21: Convert data types of multi-source heterogeneous data;

[0052] Step 22, clearing noise from the multi-source heterogeneous data after data type conversion, performing inference calculation to fill in missing and incomplete data, correcting abnormal data, clearing duplicate data, and removing erroneous data;

[0053] Step 23: Classify and store the data processed in step 22.

[0054] Step 3: Conduct in-depth mining of the internal relationships of the fused data to determine the key factors that affect the performance status of the terminal equipment.

[0055] By deeply mining the internal relationships of the fused data, the key factors that affect the performance status of the terminal equipment can be more accurately identified, which helps to focus resources and attention on the most critical factors and improve the pertinence of performance optimization and fault prevention.

[0056] After determining the key factors, targeted maintenance and management strategies are formulated to avoid blindly processing and analyzing all data, provide more accurate data support for decision makers, improve decision-making efficiency, and reduce maintenance costs.

[0057] Specifically, a machine learning algorithm is used to perform pattern recognition on the fused multi-source heterogeneous data to determine the key factors affecting the performance status of the terminal equipment, wherein the machine learning algorithm is one of a decision tree, a random forest, a support vector machine, and a neural network.

[0058] Step 4: Conduct a comprehensive diagnosis of the operating status of the terminal equipment and determine its changing trend.

[0059] Specifically, step 4 includes:

[0060] A correlation matrix is ​​constructed based on key elements, where any row of the correlation matrix is ​​a data sequence of any key element, and any column represents a time point;

[0061] Use the grey correlation calculation formula to calculate the correlation between each key factor and other key factors, and calculate the weight value of any key factor based on the correlation;

[0062] Obtain a first data sequence of any key element, and repeat this step based on the standard deviation of the first data sequence until the standard deviations of all key elements are obtained;

[0063] The comprehensive performance evaluation value of the terminal equipment is calculated based on the weight value and standard deviation.

[0064] Exemplarily, the calculation formula for the weight value of any key factor is:

[0065] ,

[0066] Among them, w iis the weight value of the i-th key factor, r i is the correlation degree of the i-th key factor, i is a positive integer from 1 to I, and I is the total number of key factors.

[0067] The calculation formula for the comprehensive performance evaluation value of the terminal equipment is:

[0068] ,

[0069] Among them, E is the comprehensive performance evaluation value, SD i is the current measurement value of the i-th key element, S i is the standard value of the i-th key element, is the standard deviation of the ith key factor.

[0070] By constructing a correlation matrix, the interrelationships between different key elements can be clearly displayed. Based on the correlation analysis, the weight of each performance parameter can be determined, providing a scientific basis for system performance evaluation. By calculating the standard deviation, the volatility and stability of key elements can be better reflected, thereby improving the accuracy of the comprehensive performance evaluation value. This makes the performance evaluation not only rely on the average value, but also considers the discreteness of the data. At the same time, by analyzing the standard deviation of key elements, it is possible to identify which elements have greater volatility, which can then be used as an early warning indicator for potential faults, helping to take timely measures to prevent faults from occurring and enhancing fault early warning capabilities.

[0071] Specifically, step 4 includes:

[0072] The indicator data of key elements are tested through the information system performance evaluation model to check the abnormal information of each indicator data, and the cause of the abnormal information is located;

[0073] The fault information is connected to the alarm platform for unified formatting, and the alarm information is mined and combined and compressed through the association mining strategy. Finally, the compressed alarm information is sent to relevant personnel for processing;

[0074] Predict the data value of each indicator at the next moment and perceive the performance status of the terminal equipment in advance.

[0075] Specifically, the information system performance evaluation model includes an anomaly detection module and a fault location module. The anomaly detection module is used to automatically detect the abnormal conditions of each performance indicator. If an indicator abnormality occurs, it is submitted to the fault location module for fault cause identification; the fault location module is used to perform causal inference based on the correlation between indicators and events, and finally lock the location of suspected fault events.

[0076] The anomaly detection module is used to automatically detect the anomalies of each performance indicator. If an indicator is abnormal, it will be submitted to the fault location module for fault cause identification. Since the characteristics of performance indicator data of different categories from different data sources are different, and the data fluctuations are also different, different methods need to be set for detection. Generally, there are fixed threshold detection method and dynamic threshold detection method.

[0077] Specifically, the abnormal information of each indicator data is checked including a fixed threshold detection method and a dynamic threshold detection method;

[0078] Fixed threshold detection method is used for the first indicator data with stable fluctuations. By setting the threshold range on the server side, if the current value of any first indicator data exceeds the corresponding threshold range, it is considered to be abnormal;

[0079] The dynamic threshold detection method is used for the second indicator data with obvious trend and periodicity. The exponential smoothing method is used to learn the original trend of the data. The time series data of the performance indicator is segmented by period and a dynamic time window is set. The data in the same time window in different periods are then regarded as obeying the normal distribution. Different dynamic thresholds are set for the data in different time windows according to the Laida criterion.

[0080] Specifically, a time series prediction method is used to predict the data value of each indicator at the next moment, wherein the time series prediction method is one of simple smoothing, Holt linear trend model, damped trend model, simple seasonal model, and Winter model.

[0081] Simple Smoothing is the most basic forecasting method. It assumes that future data values ​​will be close to the most recent observed values. Simple smoothing predicts future values ​​by taking the average of the most recent data points. This method is not suitable for data with large random fluctuations. Holt's Linear Trend Model is a variant of simple exponential smoothing. It takes into account the trend component in the data. The Holt model estimates the baseline (level) and trend (slope) of the data and predicts their future changes, thereby providing more accurate forecasts. The Damped Trend Model is an extension of the Holt Linear Trend Model. It allows the trend to gradually weaken over time. This model is suitable for scenarios where long-term trends will gradually disappear. The Simple Seasonal Model takes into account the seasonal changes in the data, that is, the regular changes in the data within a fixed period (such as a year). The Simple Seasonal Model makes predictions by identifying and simulating this periodic pattern. Winter's Model is a more complex time series forecasting method that combines trend and seasonal factors. This model is suitable for data that has both trend changes and seasonal fluctuations. In practical applications, the choice of time series forecasting method depends on the characteristics of the data and the forecasting goal. For example, if the data has obvious seasonal changes, then it may be more appropriate to use a simple seasonal model or a Winter model. If the data mainly shows trend changes, then the Holt linear trend model or a damped trend model may be more applicable. Simple smoothing is suitable for relatively stable data without obvious trends or seasonality.

[0082] Predict the data value of each indicator at the next moment, predict the performance trend of the equipment, perceive potential failure risks in advance, and provide a scientific basis for preventive maintenance.

[0083] Preferably, time series analysis is used to predict future changes in equipment performance, wherein the time series analysis includes but is not limited to autoregressive moving average model (ARMA), autoregressive integrated moving average model (ARIMA) or seasonally decomposed time series forecasting (STL), etc.

[0084] Specifically, in step 4, the comprehensive diagnosis also includes evaluating the current number of runs and the remaining number of runs, the current running time and the remaining running time, the current performance status and the remaining performance status.

[0085] "Remaining performance status" can be understood as the performance level that the equipment can maintain relative to its initial state after a certain period of use. This is usually used to evaluate the degree of performance degradation of the equipment and predict its future performance trend. Specifically, "remaining performance status" may include: 1. Performance degradation, the degree to which the performance of the equipment decreases over time and use; 2. Performance maintenance, the performance level that the equipment can maintain under specific conditions; 3. Performance prediction, based on the current performance status, predicting the performance of the equipment in the future. This indicator is of great significance for equipment maintenance, life prediction and performance optimization. It can help managers understand the current performance status of the equipment and take corresponding measures to extend the service life of the equipment and maintain its performance level.

[0086] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0087] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the above-mentioned program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0088] The above embodiments only express the preferred implementation mode of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A multi-source data information deep mining and fusion processing method for performance preservation, characterized in that: The steps include: Step 1: Use different data acquisition devices to collect multi-source heterogeneous data representing the operation status and performance status of the terminal device, wherein the multi-source heterogeneous data includes operation process data, operation status data and control instruction data, wherein the operation process data includes specific operation and process data of the terminal device during operation, and the specific operation and process data include the start, stop, operation speed and working cycle of the terminal device; the operation status data includes the state information data of the terminal device during operation, and the state information data includes temperature, pressure, vibration, current and voltage; the control instruction data includes instruction data issued by the control system of the terminal device, and the instruction data includes start, stop and adjustment parameters; Step 2: performing data fusion processing on the multi-source heterogeneous data; Step 3: perform in-depth mining of the internal relationships of the fused data to determine the key factors that affect the performance status of the terminal device; Step 4: Comprehensively diagnose the operating status of the terminal equipment and determine its changing trend; The step 4 comprises: Constructing a correlation matrix based on the key elements, wherein any row of the correlation matrix is ​​a data sequence of any key element, and any column represents a time point; Use the grey correlation calculation formula to calculate the correlation between each key factor and other key factors, and calculate the weight value of any key factor based on the correlation; Obtain a first data sequence of any of the key elements, and repeat this step based on the standard deviation of the first data sequence until the standard deviations of all the key elements are obtained; A comprehensive performance evaluation value of the terminal device is calculated based on the weight value and the standard deviation.

2. The method according to claim 1, characterized in that The step 2 comprises: Step 21: converting the data types of the multi-source heterogeneous data; Step 22, clearing noise, inferring and calculating to fill in missing and incomplete data, correcting abnormal data, clearing duplicate data and removing erroneous data from the multi-source heterogeneous data after data type conversion; Step 23: Classify and store the data processed in step 22.

3. The method according to claim 1, characterized in that A machine learning algorithm is used to perform pattern recognition on the fused multi-source heterogeneous data to determine key factors affecting the performance status of the terminal device, wherein the machine learning algorithm includes a decision tree, a random forest, a support vector machine or a neural network.

4. The method according to claim 1, characterized in that: The step 4 comprises: The index data of the key elements are tested by the information system performance evaluation model to obtain abnormal information of each index data, and the cause of the abnormal information is located; The fault information is connected to the alarm platform for unified formatting, and the alarm information is mined and merged and compressed through the association mining strategy, and finally the compressed alarm information is sent to relevant personnel for processing; The data value of each indicator at the next moment is predicted to perceive the performance status of the terminal device in advance.

5. The method according to claim 4, characterized in that The information system performance evaluation model includes an anomaly detection module and a fault location module. The anomaly detection module is used to automatically detect the abnormality of each performance indicator. If an indicator is abnormal, it is submitted to the fault location module for fault cause identification; The fault location module is used to perform causal deduction based on the correlation between indicators and events, and ultimately locate the location of the suspected fault event.

6. The method according to claim 5, characterized in that The abnormal information of each indicator data is checked by a fixed threshold detection method and a dynamic threshold detection method; The fixed threshold detection method is used for the first indicator data with stable fluctuations. By setting the threshold range on the server side, if the current value of any first indicator data exceeds the corresponding threshold range, it is considered to be abnormal; The dynamic threshold detection method is used for the second indicator data with obvious trend and periodicity. The exponential smoothing method is used to learn the original trend of the data, and then the time series data of the performance indicator is segmented by period, and a dynamic time window is set. Then, the data in the same time window in different periods are regarded as obeying the normal distribution, and different dynamic thresholds are set for the data in different time windows according to the Laida criterion.

7. The method according to claim 4, characterized in that The data value of each indicator at the next moment is predicted by using a time series prediction method, wherein the time series prediction method includes simple smoothing, Holt linear trend model, damped trend model, simple seasonality and / or Winter model.

8. The method according to claim 1, characterized in that In step 4, the comprehensive diagnosis also includes evaluating the current number of runs and the remaining number of runs, the current running time and the remaining running time, the current performance status and the remaining performance status.

Images