CN117272031B - Multi-source-based coal flow balance self-adaptive control method - Google Patents

Abstract

The invention relates to the technical field of coal flow data processing, in particular to a coal flow balance self-adaptive control method based on multiple sources. The method comprises the following steps: acquiring coal flow data through a coal flow data acquisition device so as to obtain real-time coal flow data, wherein the real-time coal flow data comprises real-time coal mine data, real-time transfer station data and real-time coal processing data; carrying out multi-source data fusion on the real-time coal stream data so as to obtain coal stream fusion data; extracting coal flow change characteristics of the coal flow fusion data to obtain coal flow change characteristic data, and carrying out coal flow loss evaluation according to the coal flow change characteristic data and coal flow source data corresponding to the real-time coal flow data to obtain coal flow loss evaluation data; and acquiring historical coal flow data, and extracting time sequence coal flow change characteristics according to the historical coal flow data, so as to obtain time sequence coal flow change characteristic data. The invention can efficiently control and optimize the coal flow and reduce the coal flow loss.

Description

Multi-source-based coal flow balance self-adaptive control method

Technical Field

The invention relates to the technical field of coal flow data processing, in particular to a coal flow balance self-adaptive control method based on multiple sources.

Background

The multi-source-based coal flow balance self-adaptive control method is a control strategy for managing a coal mine production system. The automatic adjustment and control of the coal flow are realized by integrating information from a plurality of coal sources, so that the balance state of the production system is maintained. The general method often has uneven quality or errors of data acquired from different coal sources, and can influence the accuracy and stability of the control method, so that the coal flow loss is high.

Disclosure of Invention

The invention provides a multi-source-based coal flow balance self-adaptive control method for solving at least one technical problem.

The application provides a coal flow balance self-adaptive control method based on multiple sources, which comprises the following steps:

step S1: acquiring coal flow data through a coal flow data acquisition device so as to obtain real-time coal flow data, wherein the real-time coal flow data comprises real-time coal mine data, real-time transfer station data and real-time coal processing data;

step S2: carrying out multi-source data fusion on the real-time coal stream data so as to obtain coal stream fusion data;

step S3: extracting coal flow change characteristics of the coal flow fusion data to obtain coal flow change characteristic data, and carrying out coal flow loss evaluation according to the coal flow change characteristic data and coal flow source data corresponding to the real-time coal flow data to obtain coal flow loss evaluation data;

Step S4: acquiring historical coal flow data, and extracting time sequence coal flow change characteristics according to the historical coal flow data so as to obtain time sequence coal flow change characteristic data;

step S5: estimating the coal flow demand according to the time sequence coal flow characteristic data to obtain coal flow demand estimated data, and generating a multi-source coal flow control strategy according to the coal flow demand estimated data and the coal flow loss estimated data to obtain multi-source coal flow control strategy data.

According to the invention, through real-time coal flow data acquisition, the system can acquire data of links such as coal mines, transfer stations, coal treatment and the like in time, so that the control system can perform quick feedback and regulation according to real-time conditions, and the system can effectively cope with sudden conditions such as environmental changes, equipment faults and the like. The data of a plurality of sources such as the coal mine, the transfer station, the coal treatment and the like are fused, so that deviation and error among the data of different sources can be eliminated, the accuracy and reliability of the data are improved, and the system can know the whole coal flow process more accurately. Through extracting the coal flow change characteristics of the fused data, the information such as the change trend, the law and the like of the coal flow can be deeply known, and the system is facilitated to identify specific conditions or conditions causing loss. The time sequence change feature extraction is carried out based on the historical coal flow data, so that the system can carry out deep analysis on past coal flow behaviors, and a more basis is provided for current coal flow estimation. The coal flow demand is estimated through time sequence coal flow characteristic data, and the system can formulate a multi-source coal flow control strategy on the basis of knowing the coal flow demand in advance, so that the coal flow loss is reduced on the premise of ensuring the balance of supply and demand.

Preferably, step S1 is specifically:

step S11: the coal mine data acquisition device is connected with a coal mine data source to acquire real-time coal mine data;

step S12: the coal flow data acquisition device is connected with a transfer station data source to acquire real-time transfer station data;

step S13: the coal flow data acquisition device is connected with a coal processing data source to acquire real-time coal processing data;

step S14: data integration is carried out on the real-time coal mine data, the real-time transfer station data and the real-time coal processing data, so that preliminary real-time coal flow data are obtained;

step S15: and carrying out coal flow data quality screening on the preliminary real-time coal flow data, thereby obtaining real-time coal flow data.

In the invention, the steps S11 to S13 are connected with the coal mine, the transfer station and the coal processing data source through the coal flow data collector, so that the rapid acquisition of real-time data is realized, the system can acquire the data of the coal mine production condition, the transfer station transportation condition and the coal processing link in real time, the step S14 integrates the real-time coal mine, the transfer station and the coal processing data, thereby obtaining preliminary real-time coal flow data, ensuring that the data from different sources are reasonably integrated and uniformly processed, the step S15 performs the quality screening of the real-time coal flow data, and the screening can eliminate the abnormal or low-quality data and ensure the accuracy and reliability of the subsequent analysis and control. The system has the capability of real-time feedback and control, and can respond and adjust quickly according to the latest condition by acquiring and arranging the data in time, thereby ensuring the high-efficiency operation of the whole coal flow process. By acquiring and screening in real time, systematic errors caused by abnormal data or poor quality data can be reduced, so that loss of coal flow is reduced, and the overall efficiency and resource utilization efficiency of coal mine production are positively influenced.

Preferably, step S15 is specifically:

step S151: performing coal flow data accuracy verification on the preliminary real-time coal flow data according to preset coal flow transportation electronic map data, so as to obtain accuracy verification data;

step S152: performing coal flow data restoration on the primary real-time coal flow data according to the accuracy verification data, so as to obtain real-time coal flow restoration data;

step S153: and carrying out distributed transaction verification and data correction on the real-time coal flow repair data, thereby obtaining the real-time coal flow data.

The preset coal flow transportation electronic map data are used for verifying the accuracy of the preliminary real-time coal flow data, and the verification ensures that the source and the content of the data are consistent with the expected data, and ensures the quality and the accuracy of the data. In this step, the preliminary real-time coal stream data is repaired according to the accuracy verification data, and the system can automatically correct the error or inaccurate data according to the verification data, thereby improving the reliability and accuracy of the data. The real-time coal flow repair data is further subjected to distributed transactional verification, the distributed transactional verification is adopted, the problem of potential data loss/tampering caused by centralized calculation is avoided, the consistency and the reliability of the data are enhanced, and the existing data conflict or error is avoided. At the same time, the data is also corrected to ensure that it meets the requirements and specifications of the system.

Preferably, in step S152, the coal stream data is restored by performing restoration policy selection on restoration index data generated by a coal stream data restoration calculation formula, where the coal stream data restoration calculation formula specifically includes:

；

for restoring index data, < >>For preliminary real-time coal flow data, +.>For preliminary real-time coal flow sequence item data, < ->For preliminary real-time coal flow data, < > for>For coal flow data, +.>For coal flow quality data, +.>For coal flow velocity data, +.>Is coal flow temperature weight data, +.>For coal flow temperature data, +.>For coal stream moisture weight data, < >>For coal flow humidity data, < >>Is coal flow pressure weight data, +.>For coal flow pressure data, +.>For coal flow density weight data, +.>Is coal flow density data.

The invention constructs a coal flow data restoration calculation formula, which calculates and obtains restoration index data according to preliminary real-time coal flow data and parameters such as coal flow quality, speed, temperature, humidity, pressure, density and the like) This repair index data will be the basis for selection of a subsequent repair strategy. Wherein->Is a repair index, which indicates the difficulty and effect of coal flow data repair, and the larger R indicates the more difficult the repair and the worse the effect; / >Smaller indicates easier repair and better effect. />Is the number of coal stream data and represents the number of data participating in repair. />Is a logarithmic function representing the logarithmic transformation of coal stream data for narrowing the variance and range of the data. />Is a limit symbol indicating when->The limit value approaching 0 is used for processing abnormal values and missing values in coal flow data. />And->Is a function of two different dimensions representing coal flow data, such as weight and velocity of the coal flow, etc./>Is an exponential function, representing the temperature of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of temperature. />Is a natural logarithmic function representing the moisture +.>The effect on the repair index is that,is a constant and represents the weight coefficient of humidity. />Is an arctangent function representing the pressure of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of the pressure. />Is a hyperbolic cosine function representing the density of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of the density. The invention improves the accuracy and pertinence of data restoration, thereby making the restoration strategy more scientific and efficient.

Preferably, step S2 is specifically:

step S21: carrying out multi-source weighted data fusion on the real-time coal stream data so as to obtain first coal stream fusion data;

step S22: performing multi-source clustering data fusion on the real-time coal stream data so as to obtain second coal stream fusion data;

step S23: and carrying out confidence voting calculation according to the first coal flow fusion data and the second coal flow fusion data, thereby obtaining the coal flow fusion data.

In the invention, step S21 utilizes the weight information of different data sources through multi-source weight data fusion, so that the real-time coal flow data can be combined more accurately, and the data error is reduced. Step S22 is used for multi-source clustering data fusion, similar data in different data sources can be gathered together, and the clustering accuracy of the data is improved. The step S23 utilizes confidence voting calculation, combines the confidence information of the first coal flow fusion data and the second coal flow fusion data, further improves the reliability and the accuracy of the data, and avoids the problems of inaccuracy or data deviation caused by single data fusion processing. Through different data fusion processes, the influence of errors or noise caused by a single data source on a final result can be reduced, and the overall data quality is improved. According to different real-time coal flow data sources and data characteristics, the method and the system can flexibly adjust weights and clustering strategies, so that the method and the system are suitable for data fusion requirements under different scenes and conditions. By introducing confidence voting calculation, the influence of abnormal values or errors of a single data source on a final result is avoided, and the stability and reliability of the system are enhanced.

Preferably, step S22 is specifically:

step S221: performing target detection on coal flow image data in the real-time coal flow data so as to obtain coal flow target detection data;

step S222: clustering calculation is carried out according to non-coal flow image data and coal flow target detection data in the real-time coal flow data, so that real-time coal flow clustering data is obtained;

step S223: performing similarity calculation on the real-time coal flow clustering data to obtain clustering similarity data;

step S224: and carrying out multi-source weighted data fusion on the real-time coal stream data according to the clustering similarity data, thereby obtaining second coal stream fusion data.

According to the invention, through target detection, the position and boundary of the coal flow can be accurately identified, so that accurate basic information is provided for subsequent clustering and data fusion. Through clustering, similar coal flow data can be classified to reduce the problem of overlarge calculation load caused by direct use of similarity calculation, and through multi-source weighted data fusion, information of different sources can be fully utilized, and the comprehensiveness and accuracy of the data are improved.

Preferably, the multi-source weighted data fusion is calculated by a multi-source coal stream data weighted calculation formula, the preliminary coal stream fused data comprises first coal stream fused data and second coal stream fused data, wherein the multi-source coal stream data weighted calculation formula specifically comprises:

；

For preliminary coal flow fusion data, +.>For real-time coal flow order item data, < + >>For real-time coal flow data, < > data>Is the firstWeight coefficient data of real-time coal stream data corresponding to individual coal stream data source data,/weight coefficient data of real-time coal stream data corresponding to individual coal stream data source data>Is->Real-time coal flow data corresponding to the individual coal flow data source data,/-the real-time coal flow data corresponding to the individual coal flow data source data>Is coal flow change rate data.

According to the invention, the information of different data sources can be comprehensively synthesized by introducing a multi-source coal flow data weighting calculation formula, and the contribution of each data source can be calculated according to the weight coefficient of the data source) The method can accurately reflect the data, so that the fused data is more representative and comprehensive. Wherein->Is first coal flow fusion data, and represents a weighted average value of multi-source coal flow data plus a limit term; />Is the number of sources of coal stream data, e.g. +.>=3 indicates that there are three different coal stream data sources; />Is->The weight coefficient of a personal coal stream data source, representing the credibility or importance of the data source, generally satisfies 0.ltoreq.o ≡>Is less than or equal to 1；/>Is->Real-time coal flow data of a coal flow data source represents the coal flow or mass measured by the data source; />Is a variable representing the rate of change or the degree of fluctuation of coal flow data; />Is a logarithmic function representing the nonlinear characteristics or complexity of the coal stream data; / >Is a root function, and represents the smoothness or stability of coal flow data; the invention has a certain degree of inhibition effect on abnormal values, and the influence on abnormal data generated due to various reasons is reduced in the data fusion process, thereby ensuring the stability and reliability of fusion results.

Preferably, step S23 is specifically:

step S231: normalizing the first coal flow fusion data and the second coal flow fusion data to obtain first coal flow fusion normalized data and second coal flow fusion normalized data;

step S232: performing self-attention calculation on the first coal flow fusion normalized data and the second coal flow fusion normalized data so as to obtain first coal flow fusion self-attention weight data and second coal flow fusion self-attention weight data;

step S233: performing self-attention weighted calculation and dimension reduction processing on the first coal flow fusion normalized data and the second coal flow fusion normalized data according to the first coal flow fusion self-attention weight data and the second coal flow fusion self-attention weight data, so as to obtain first coal flow fusion dimension reduction data and second coal flow fusion dimension reduction data;

Step S234: performing similarity matrix construction on the first coal flow fusion dimensionality reduction data and the second coal flow fusion dimensionality reduction data to obtain coal flow fusion similarity matrix data;

step S235: confidence calculation is carried out on the coal flow fusion similarity matrix data, so that coal flow fusion confidence data are obtained;

step S236: and carrying out coal flow data fusion on the first coal flow fusion data and the second coal flow fusion data according to the coal flow fusion confidence data, so as to obtain the coal flow fusion data.

The self-attention mechanism adopted in the invention can highlight important characteristics, thereby improving the key information of the data. The dimension reduction can reduce the dimension of data, improve the efficiency of subsequent calculation, and simultaneously keep key information. Through the construction of the similarity matrix, the similarity degree between data can be quantized, and meanwhile, the complexity brought by linear calculation can be reduced through matrix operation, so that the calculation load is reduced. The confidence reflects the credibility of the fusion data, and the contribution degree of each data source is effectively evaluated through analysis and calculation of the similarity. The final fusion operation is carried out on the first coal flow fusion data and the second coal flow fusion data by considering the confidence coefficient, so that the fusion result is ensured to be more representative and comprehensive, and a high-quality data basis is provided for the subsequent steps. The invention fully plays the roles of data processing and feature extraction, so that the coal flow fusion data has more representativeness and credibility.

Preferably, step S3 is specifically:

step S31: extracting coal flow variation characteristics of the coal flow fusion data by utilizing preset sliding window data, so as to obtain coal flow variation characteristic data;

step S32: performing characteristic association according to the coal flow characteristic data and the real-time coal flow data, so as to obtain coal flow associated characteristic data;

step S33: and carrying out coal flow loss evaluation on the coal flow associated characteristic data by utilizing a preset rule engine and coal flow source data corresponding to the real-time coal flow data, thereby obtaining coal flow loss evaluation data.

According to the invention, the sliding window technology is used for processing the coal flow fusion data, so that the dynamic change characteristics of the coal flow can be captured. By doing so, the sensitivity of the data can be improved, so that the response of the system to the coal flow change is more timely and accurate. By correlating the coal flow characteristic data with the real-time coal flow data, the inherent relationship between the coal flow characteristic data and the real-time coal flow data can be found, so that the system can better understand the actual meaning of the characteristics, and an accurate basis is provided for subsequent evaluation. The preset rule engine is a rule-based expert system capable of reasoning and making decisions based on specific conditions. By combining the real-time coal flow data with the coal flow source data, the coal flow associated characteristic data can be subjected to refined evaluation. The process has certain intelligence and can adapt to the wear evaluation requirements under different conditions.

Preferably, the multi-source coal flow control strategy data includes demand priority control strategy data and low-loss priority control strategy data, and step S5 is specifically:

step S51: constructing a coal flow demand estimation model according to the time sequence coal flow characteristic data, so as to obtain the coal flow demand estimation model;

step S52: carrying out coal flow demand estimation on the real-time coal flow data by utilizing a coal flow demand estimation model so as to obtain coal flow demand estimation data;

step S53: generating a demand priority control strategy according to the coal flow demand estimated data and the coal flow loss estimated data, so as to obtain demand priority control strategy data;

step S54: and generating a low-loss control strategy according to the estimated data of the coal flow demand and the estimated data of the coal flow loss so as to obtain low-loss priority control strategy data.

The invention uses the sliding window technology to process the coal flow fusion data, can capture the dynamic change characteristics of the coal flow, can improve the sensitivity of the data, and ensures that the response of the system to the coal flow change is more timely and accurate. By correlating the coal flow characteristic data with the real-time coal flow data, the inherent relationship between the coal flow characteristic data and the real-time coal flow data can be found, so that the system can better understand the actual meaning of the characteristics, and an accurate basis is provided for subsequent evaluation. The preset rule engine is a rule-based expert system capable of reasoning and making decisions based on specific conditions. By combining the real-time coal flow data with the coal flow source data, the coal flow associated characteristic data can be subjected to refined evaluation. The invention can make reasonable decisions in complex coal flow environments, and ensures the efficient transportation and utilization of coal flows, thereby obtaining remarkable economic and social benefits in practice.

The invention has the beneficial effects that: by integrating the real-time coal mine data, the real-time transfer station data and the real-time coal processing data, the multi-source data can be effectively fused and utilized, the comprehensiveness and accuracy of the data can be improved, and a solid foundation can be provided for subsequent coal flow analysis and control. By means of multi-source data fusion, information of different data sources is integrated together, comprehensive coal flow fusion data are obtained, and the system is guaranteed to be capable of controlling the whole coal flow system more comprehensively. By extracting the characteristics of the fused data, the system can understand the change rule of the coal flow more deeply. The time sequence characteristic extraction is carried out by utilizing the historical coal flow data, so that the evolution process of the coal flow can be known in the time dimension, and the knowledge of the coal flow characteristics is further enriched. By predicting the time sequence coal flow characteristic data, the system can know the trend of future coal flow demands in advance, so that the coal flow control strategy can be adjusted in a targeted manner. Based on the coal flow demand estimation data and the coal flow loss evaluation data, the system can make a multi-source coal flow control strategy, so that the coal flow loss can be reduced to the greatest extent while the supply is ensured.

Drawings

Other features, objects and advantages of the present application will become more apparent upon reading of the detailed description of non-limiting implementations made with reference to the following drawings in which:

FIG. 1 shows a flow chart of steps of a multi-source based coal stream balance adaptive control method according to an embodiment;

FIG. 2 shows a step flow diagram of step S1 of an embodiment;

FIG. 3 shows a step flow diagram of step S15 of an embodiment;

FIG. 4 shows a step flow diagram of step S2 of an embodiment;

fig. 5 shows a step flow diagram of step S22 of an embodiment.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

Referring to fig. 1 to 5, the present application provides a multi-source-based adaptive control method for coal flow balance, comprising the following steps:

step S1: acquiring coal flow data through a coal flow data acquisition device so as to obtain real-time coal flow data, wherein the real-time coal flow data comprises real-time coal mine data, real-time transfer station data and real-time coal processing data;

specifically, high-performance sensors and devices (such as Internet of things devices) are used for collecting data in real-time monitoring of coal yield of a coal mine, transportation conditions of a transfer station and coal treatment processes. The collected data is transmitted to a data processing center by a communication technology (e.g., 5G).

Specifically, for example, in a coal mine, a coal stream data collector comprising a sensor network and a data interface is used to obtain real-time coal stream data. The system monitors coal mine data, such as coal mine production, ore quality, and transport speed, in real time by means of sensors connected to the coal mine equipment. At the same time, the system acquires real-time transfer station data (e.g., traffic flow, vehicle status) and real-time coal processing data (e.g., coal concentration, processing efficiency) via sensors coupled to the transfer station and the coal processing facility.

Step S2: carrying out multi-source data fusion on the real-time coal stream data so as to obtain coal stream fusion data;

specifically, real-time data from coal mines, transfer stations and coal processing are integrated together, for example, using data fusion algorithms such as weighted fusion, clustered fusion, and the like. By means of the distributed system, real-time synchronization and fusion of a plurality of data sources are achieved.

Step S3: extracting coal flow change characteristics of the coal flow fusion data to obtain coal flow change characteristic data, and carrying out coal flow loss evaluation according to the coal flow change characteristic data and coal flow source data corresponding to the real-time coal flow data to obtain coal flow loss evaluation data;

Specifically, for example, signal processing techniques such as fourier transforms are applied to extract spectral features of the coal stream. The coal flow characteristics are classified and evaluated using a machine learning algorithm, such as a Support Vector Machine (SVM), or the like.

Specifically, extracting time sequence characteristics through a time sliding window technology, for example, to obtain coal flow characteristic data; and carrying out coal flow loss evaluation on the coal flow change characteristic data and the coal flow source data corresponding to the real-time coal flow data according to a preset expert rule, so as to obtain coal flow loss evaluation data. The method comprises the following steps of: if the coal flow profile data indicates that the coal quality is good and the source of the real-time coal flow is from a high quality coal mine, the coal flow loss is assessed as low. If the coal flow characteristic data indicates that the coal quality is general and the source of the real-time coal flow is from a medium quality coal mine, the coal flow loss is assessed as medium. If the coal flow profile data indicates that the coal quality is poor and the source of the real-time coal flow is from a low quality coal mine, the coal flow loss is assessed as high. Data: the coal flow characteristic data indicates that the coal quality is "good". The source of the real-time coal stream is from a "high quality coal mine".

Step S4: acquiring historical coal flow data, and extracting time sequence coal flow change characteristics according to the historical coal flow data so as to obtain time sequence coal flow change characteristic data;

Specifically, historical coal stream data is stored and managed, for example, using database technology, such as SQL or NoSQL. Time sequence analysis methods, such as sliding window analysis, are applied to extract timing features from the historical data.

Step S5: estimating the coal flow demand according to the time sequence coal flow characteristic data to obtain coal flow demand estimated data, and generating a multi-source coal flow control strategy according to the coal flow demand estimated data and the coal flow loss estimated data to obtain multi-source coal flow control strategy data.

Specifically, the coal flow demand estimated data is obtained by predicting the timing characteristics, for example, by means of a machine learning model, such as a Recurrent Neural Network (RNN) or a long short time memory network (LSTM). A multi-source coal flow control strategy is generated by utilizing an optimization algorithm, such as a genetic algorithm or a simulated annealing algorithm, so as to meet the coal flow requirement to the greatest extent and reduce the loss.

Specifically, for example, coal flow demand estimates: a linear regression model is preset, wherein time sequence coal flow characteristic data is taken as input, and coal flow requirements are taken as output. The regression model is trained using the historical data. And predicting time sequence coal flow change characteristics of the real-time coal flow data by using the trained regression model to obtain coal flow demand estimated data. Coal flow loss evaluation: and carrying out loss evaluation on the real-time coal flow data and the coal flow source data according to preset expert rules to obtain coal flow loss evaluation data. Multi-source coal flow control strategy generation: a series of control rules are set, which may be formulated based on the coal flow demand estimation data and the coal flow loss estimation data. For example, if demand is large and loss is low, certain strategies may be taken to prioritize the use of high quality coal mine flows.

According to the invention, through real-time coal flow data acquisition, the system can acquire data of links such as coal mines, transfer stations, coal treatment and the like in time, so that the control system can perform quick feedback and regulation according to real-time conditions, and the system can effectively cope with sudden conditions such as environmental changes, equipment faults and the like. The data of a plurality of sources such as the coal mine, the transfer station, the coal treatment and the like are fused, so that deviation and error among the data of different sources can be eliminated, the accuracy and reliability of the data are improved, and the system can know the whole coal flow process more accurately. Through extracting the coal flow change characteristics of the fused data, the information such as the change trend, the law and the like of the coal flow can be deeply known, and the system is facilitated to identify specific conditions or conditions causing loss. The time sequence change feature extraction is carried out based on the historical coal flow data, so that the system can carry out deep analysis on past coal flow behaviors, and a more basis is provided for current coal flow estimation. The coal flow demand is estimated through time sequence coal flow characteristic data, and the system can formulate a multi-source coal flow control strategy on the basis of knowing the coal flow demand in advance, so that the coal flow loss is reduced on the premise of ensuring the balance of supply and demand.

Preferably, step S1 is specifically:

step S11: the coal mine data acquisition device is connected with a coal mine data source to acquire real-time coal mine data;

specifically, for example, intelligent sensor networks such as vibration sensors, temperature sensors, humidity sensors, etc. are used to connect to coal mine equipment to monitor coal mine data in real time. And an industrial Internet of things protocol, such as Modbus or OPC UA, is used for connecting the sensor with the data collector so as to realize real-time data transmission and collection.

Step S12: the coal flow data acquisition device is connected with a transfer station data source to acquire real-time transfer station data;

specifically, for example, an RFID (radio frequency identification) technology is deployed, an RFID tag is attached to a transportation device, and position and status information of the transportation device is acquired in real time by an RFID reader.

Step S13: the coal flow data acquisition device is connected with a coal processing data source to acquire real-time coal processing data;

specifically, for example, sensors are deployed on a coal processing facility to monitor data such as water content, particle size distribution, etc. of the coal and transmit the data to a data collector via a communication protocol.

Step S14: data integration is carried out on the real-time coal mine data, the real-time transfer station data and the real-time coal processing data, so that preliminary real-time coal flow data are obtained;

Specifically, data acquired from coal mines, transfer stations and coal processing equipment are integrated, for example, using a real-time database system, to form a preliminary real-time coal flow dataset. Such as data integration and storage using a real-time database system (e.g., influxDB, mongoDB).

Step S15: and carrying out coal flow data quality screening on the preliminary real-time coal flow data, thereby obtaining real-time coal flow data.

Specifically, for example, the data quality evaluation model is utilized to screen the preliminary real-time coal flow data, and abnormal data is removed, so that the accuracy and reliability of the data are ensured. Such as using statistical methods or machine learning models to evaluate the quality of the data.

Specifically, the coal flow data of each data source is analyzed, for example, using basic statistical methods such as mean, standard deviation, and the like. According to the domain knowledge, a threshold is set, and data exceeding the threshold is marked as an outlier. For example, if the yield in the coal mine data exceeds a reasonable range, it may be marked as outliers. Such as calculating the mean and standard deviation of the coal flow. The following results were assumed to be obtained: mean (Mean) =1500 tons/hour, standard deviation (standard device) =200 tons/hour, set the threshold: and setting a reasonable range according to the domain knowledge and the statistical result. For example, the yield is considered to be normal within a range of plus or minus two standard deviations of the mean A kind of electronic device. Upper threshold =1900 tons/hour, lower threshold = =>=1100 tons/hour, where ∈>For mean value->The standard deviation is marked with outliers: for data exceeding the upper threshold or the lower threshold, it is marked as an outlier. For example, if the production rate exceeds 1900 tons/hr or falls below 1100 tons/hr within a certain hour, the data is marked as abnormal.

Specifically, for example, a time series of real-time coal flow data is analyzed to check whether there is a significant trend, seasonality, or the like. Additional threshold contrast is required for the time points of the anomalies, below or above which the anomaly data is marked.

In the invention, the steps S11 to S13 are connected with the coal mine, the transfer station and the coal processing data source through the coal flow data collector, so that the rapid acquisition of real-time data is realized, the system can acquire the data of the coal mine production condition, the transfer station transportation condition and the coal processing link in real time, the step S14 integrates the real-time coal mine, the transfer station and the coal processing data, thereby obtaining preliminary real-time coal flow data, ensuring that the data from different sources are reasonably integrated and uniformly processed, the step S15 performs the quality screening of the real-time coal flow data, and the screening can eliminate the abnormal or low-quality data and ensure the accuracy and reliability of the subsequent analysis and control. The system has the capability of real-time feedback and control, and can respond and adjust quickly according to the latest condition by acquiring and arranging the data in time, thereby ensuring the high-efficiency operation of the whole coal flow process. By acquiring and screening in real time, systematic errors caused by abnormal data or poor quality data can be reduced, so that loss of coal flow is reduced, and the overall efficiency and resource utilization efficiency of coal mine production are positively influenced.

Preferably, step S15 is specifically:

step S151: performing coal flow data accuracy verification on the preliminary real-time coal flow data according to preset coal flow transportation electronic map data, so as to obtain accuracy verification data;

specifically, for example, in a coal mine scenario, the position and movement track of the mine car are monitored and verified in real time by using a GPS positioning technology or a laser range finder.

Step S152: performing coal flow data restoration on the primary real-time coal flow data according to the accuracy verification data, so as to obtain real-time coal flow restoration data;

specifically, the missing values are filled in, for example, using an interpolation method. Common interpolation methods include linear interpolation, polynomial interpolation, spline interpolation, and the like. For example, the missing values are estimated using linear interpolation, which is performed between the missing value locations based on known data points.

Step S153: and carrying out distributed transaction verification and data correction on the real-time coal flow repair data, thereby obtaining the real-time coal flow data.

Specifically, for example, in the data processing process, a distributed database system such as MySQL Cluster, cockreach db, google Cloud Spanner and the like is adopted to ensure consistency and integrity of data. Real-time coal stream data is stored in a plurality of nodes in a scattered manner, and the data is copied at the same time. In the data processing process, a transaction mechanism is adopted to ensure the consistency of the data. When data needs to be updated, inserted or deleted, the operations are placed in a transaction, ensuring that the operations are either all successful or all failed. In each transaction, a validation step may be added to transactionally validate the data. This may include validity checks, scope checks, etc. of the data. If data inconsistencies or errors are found in the transaction, the correctness of the data may be ensured by rolling back the transaction or performing a corresponding corrective action.

Specifically, real-time coal stream data is stored on multiple nodes, for example using MySQL Cluster as a distributed database system. During data processing, if data needs to be updated, operations can be performed by opening a transaction. In a transaction, a verification step of the data may be included, such as checking whether the quality is within a reasonable range, whether the speed is satisfactory, etc. If the data is found to be abnormal, the correctness of the data can be ensured by rolling back the transaction or executing corresponding data correction operation.

The preset coal flow transportation electronic map data are used for verifying the accuracy of the preliminary real-time coal flow data, and the verification ensures that the source and the content of the data are consistent with the expected data, and ensures the quality and the accuracy of the data. According to the accuracy verification data, the preliminary real-time coal flow data is repaired, and the system can automatically correct the error or inaccurate data according to the verification data, so that the reliability and accuracy of the data are improved. The real-time coal flow repair data is further subjected to distributed transactional verification, the distributed transactional verification is adopted, the problem of potential data loss/tampering caused by centralized calculation is avoided, the consistency and the reliability of the data are enhanced, and the existing data conflict or error is avoided. At the same time, the data is also corrected to ensure that it meets the requirements and specifications of the system.

Preferably, in step S152, the coal stream data is restored by performing restoration policy selection on restoration index data generated by a coal stream data restoration calculation formula, where the coal stream data restoration calculation formula specifically includes:

；

for restoring index data, < >>For preliminary real-time coal flow data, +.>For preliminary real-time coal flow sequence item data, < ->For preliminary real-time coal flow data, < > for>For coal flow data, +.>For coal flow quality data, +.>For coal flow velocity data, +.>Is coal flow temperature weight data, +.>For coal flow temperature data, +.>For coal stream moisture weight data, < >>For coal flow humidity data, < >>Is coal flow pressure weight data, +.>For coal flow pressure data, +.>For coal flow density weight data, +.>Is coal flow density data.

Specifically, the following data are obtained, for example, from real-time monitoring: preliminary real-time coal flow quantity=50, coal stream quality dataCoal flow velocity data =120 kg/cubic meter ∈>Coal flow temperature weight data =2 m/s +.>Coal stream temperature data =0.1 ∈>Coal flow humidity weight data =60 degrees celsius +.>Coal flow humidity data =0.05 ∈>Coal flow pressure weight data =0.8->Coal flow pressure data =0.08->Coal flow density weight data =1.2 mpa +. >Coal flow density data =0.06>=1.5 megagrams/cubic meter. Calculating a repair index using the given coal flow data and the repair calculation formula>. According to repair index->Is compared with a preset threshold value, e.g. +.>Time selectionAnd (3) a high-efficiency repair strategy, wherein when R is less than or equal to 10, a standard repair strategy is selected. For example, if r=15 is calculated, an efficient repair policy is selected, and a corresponding data repair job is performed according to the selected repair policy. Standard repair strategy: when the repair index R is less than or equal to 10, a standard repair strategy is selected, wherein the standard repair strategy is a relatively simple repair method, and is generally used for repairing general data loss or damage by interpolation, smoothing and the like. Efficient repair strategy: when repairing index R>At 10, a high-efficiency repair strategy is selected, and the high-efficiency repair strategy usually adopts a data recovery algorithm based on deep learning, such as a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN), and some mathematical models constructed based on statistics or preset rules to perform data repair so as to recover missing or damaged data points more accurately, so that the high-efficiency repair strategy is suitable for repairing serious data loss or damage and needs more precise processing.

The invention constructs a coal flow data restoration calculation formula, which calculates and obtains restoration index data according to preliminary real-time coal flow data and parameters such as coal flow quality, speed, temperature, humidity, pressure, density and the like) This repair index data will be the basis for selection of a subsequent repair strategy. Wherein->Is a repair index, which indicates the difficulty and effect of coal flow data repair, and the larger R indicates the more difficult the repair and the worse the effect; />Smaller indicates easier repair and better effect. />Is the number of coal stream data and represents the number of data participating in repair. />Is a logarithmic function, which means that the coal flow data is subjected to logarithmic transformation byTo narrow down the variance and range of data. />Is a limit symbol indicating when->The limit value approaching 0 is used for processing abnormal values and missing values in coal flow data. />And->Is a function of two different dimensions representing coal flow data, such as weight and velocity of the coal flow, etc. />Is an exponential function, representing the temperature of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of temperature. />Is a natural logarithmic function representing the moisture +.>Influence on the repair index- >Is a constant and represents the weight coefficient of humidity. />Is an arctangent function representing the pressure of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of the pressure. />Is a hyperbolic cosine function representing the density of the coal stream data +.>Influence on the repair index->Is a constant and represents the weight coefficient of the density. The invention improves the accuracy and pertinence of data restoration, thereby making the restoration strategy more scientific and efficient.

Preferably, step S2 is specifically:

step S21: carrying out multi-source weighted data fusion on the real-time coal stream data so as to obtain first coal stream fusion data;

specifically, for example, each data source is assigned a weight coefficient, for example: coal mine data weightTransfer station data weight =0.4>=0.3, coal treatment data weight +.>=0.3. And generating a weighted result according to the weight and the real-time data. F1 =Wherein->For coal mine yield,/->Transport speed, & gt>Is coal quality information.

Step S22: performing multi-source clustering data fusion on the real-time coal stream data so as to obtain second coal stream fusion data;

specifically, for example, target detection is performed on real-time coal stream data to obtain target detection data, clustering calculation is performed on the real-time coal stream data and the target detection data to obtain clustering calculation data, similarity calculation is performed on the clustering calculation data to obtain similarity data, and multi-source weighted data fusion is performed on the real-time coal stream data according to the similarity data to obtain second coal stream fusion data, such as target detection: using image processing techniques, coal flow targets in the real-time coal flow are identified. Clustering calculation: similar targets are grouped into one class using the K-means algorithm. Similarity calculation: and calculating the similarity between the clustering results. And carrying out multi-source weighted data fusion on the real-time coal stream data according to the similarity data, wherein the higher the similarity data is, the higher the weight of the weighted combination of the corresponding data is, and the lower the weight is otherwise, so that the error problem caused by different source data is reduced. Wherein the target detection technique is a deep learning based approach, such as using Convolutional Neural Networks (CNNs) in combination with modern detection frameworks such as YOLO (You Only Look Once) or fast R-CNN.

Specifically, for example, real-time coal stream data from three different sources, each source providing the mass (mass value range 0 to 100) and velocity (velocity value range 1 to 10) of the coal stream: source 1: coal flow 1 mass: 60. coal stream 1 velocity: 5. coal stream 2 mass: 70. coal stream 2 velocity: 6, source 2: coal flow 1 mass: 50. coal stream 1 velocity: 4. coal stream 2 mass: 80. coal stream 2 velocity: 7, source 3: coal flow 1 mass: 65. coal stream 1 velocity: 6. coal stream 2 mass: 75. coal stream 2 velocity: 8, carrying out cluster fusion on the data, and adopting a K-means clustering algorithm. In this example, the data is clustered into two categories. Initializing a clustering center: assume that the initial cluster center is selected as: cluster center 1: (coal flow mass=55, coal flow velocity=5), clustering center 2: (coal flow mass=70, coal flow velocity=7), the distance of the sample to the cluster center is calculated: calculating for each sample to two cluster centersEuclidean distance. Assign samples to the nearest cluster center: each sample is assigned to the nearest cluster center. For example, the first sample (mass 60, speed 5) is at a distance from cluster center 1 ofDistance from cluster center 2 is And hence to cluster center 2. Updating a clustering center: for each cluster center, the average of all samples assigned to the class is calculated as the new cluster center. For cluster center 1: ((50+65)/2, (4+6)/2) = (57.5,5), for cluster center 2: ((60+70+70+80+75)/5, (5+6+7+8+6)/5) = (71,6.4), iterating: the steps (assigning samples to the nearest cluster center) and (updating the cluster center) are repeated until no more changes occur in the cluster center or a certain number of iterations is reached. Obtaining second coal flow fusion data: finally, according to the clustering result, second coal flow fusion data can be obtained. Category 1: (57.5,5) (coal stream 1 from source 1 and coal stream 1 from source 2), category 2: (71,6.4) (coal stream 2 of source 1, coal stream 2 of source 2, coal stream 1 of source 3 and coal stream 2).

Step S23: and carrying out confidence voting calculation according to the first coal flow fusion data and the second coal flow fusion data, thereby obtaining the coal flow fusion data.

In particular, for example, two different data sources (a and B) provide quality information of the coal stream: coal stream quality of data source a: [60,70,80,90], mass of coal flow of data source B: 65,75,85,95 a more accurate coal flow quality value is calculated by confidence voting. Calculating the confidence of each data source: for each data source, some metrics (e.g., reliability of the data, historical accuracy, etc.) may be used to assign a confidence score. Assuming a confidence level of 0.8 for data source A, a confidence level of 0.7 for data source B. Combining confidence with the data: multiplying the coal stream quality provided by each data source by its corresponding confidence to obtain a weighted value: data source a weighted coal stream mass: [48,56,64,72], data source B weighted coal stream mass: [45.5,52.5,59.5,66.5], vote: voting is carried out on the weighted value of each data source, and a simple voting mode can be adopted, namely, the average value of the values provided by all the data sources is taken: final coal flow mass value: [46.75,54.25,61.75,69.25].

In the invention, step S21 utilizes the weight information of different data sources through multi-source weight data fusion, so that the real-time coal flow data can be combined more accurately, and the data error is reduced. Step S22 is used for multi-source clustering data fusion, similar data in different data sources can be gathered together, and the clustering accuracy of the data is improved. The step S23 utilizes confidence voting calculation, combines the confidence information of the first coal flow fusion data and the second coal flow fusion data, further improves the reliability and the accuracy of the data, and avoids the problems of inaccuracy or data deviation caused by single data fusion processing. Through different data fusion processes, the influence of errors or noise caused by a single data source on a final result can be reduced, and the overall data quality is improved. According to different real-time coal flow data sources and data characteristics, the method and the system can flexibly adjust weights and clustering strategies, so that the method and the system are suitable for data fusion requirements under different scenes and conditions. By introducing confidence voting calculation, the influence of abnormal values or errors of a single data source on a final result is avoided, and the stability and reliability of the system are enhanced.

Preferably, step S22 is specifically:

step S221: performing target detection on coal flow image data in the real-time coal flow data so as to obtain coal flow target detection data;

specifically, real-time coal flow target detection is performed, for example, using a deep learning model such as YOLO (You Only Look Once) network. The model can simultaneously identify a plurality of objects, has high speed and is suitable for real-time application. For example, after a real-time coal flow image is detected, the system identifies three target objects: a is that ₁ (coal briquette), B ₁ (rock), C ₁ (other objects).

Step S222: clustering calculation is carried out according to non-coal flow image data and coal flow target detection data in the real-time coal flow data, so that real-time coal flow clustering data is obtained;

specifically, the target detection data and the non-coal flow image data are clustered, for example, using a clustering algorithm such as K-means. As will A ₂ 、B ₂ 、C ₂ Clustering three target objects with some non-coal flow image data to obtain two clusters, one cluster containing A ₂ 、B ₂ Two coal flow targets, the other cluster containing C ₂ And some other non-coal stream objects.

Step S223: performing similarity calculation on the real-time coal flow clustering data to obtain clustering similarity data;

Specifically, for example, assume that the similarity of two clusters is 0.85 after similarity calculation, indicating that their clustering results are relatively similar.

Step S224: and carrying out multi-source weighted data fusion on the real-time coal stream data according to the clustering similarity data, thereby obtaining second coal stream fusion data.

Specifically, for example, the multi-source weighted data fusion is used as a weighted calculation, and the weight data in the preset multi-source weighted data fusion is adjusted according to the clustering similarity data. For example, two data sources (sourceA and sourceB) provide clustering similarity information of real-time coal stream data, as follows: cluster similarity data for SourceA: cluster similarity data for SourceB [0.9,0.8,0.7,0.6 ]: [0.8,0.7,0.6,0.5], herein the similarity value refers to the similarity of two coal stream data samples in the clustering space, with higher values indicating that the samples are more similar. Two real-time coal stream data samples: coal flow sample 1: [70,80,90,100], coal stream sample 2: [75,85,95,105], the following steps are as follows: given confidence and weight: for each data source, a confidence score is assigned that indicates the degree of trust in its data. Let sourceA be 0.8 confidence and sourceB be 0.7 confidence. At the same time, each data source is assigned a weight that indicates its importance in the fusion. Let sourceA weight be 0.6 and sourceB weight be 0.4. Combining similarity with weights: multiplying the cluster similarity value provided by each data source with the weight corresponding to the cluster similarity value to obtain a weighted similarity value: sourceA weighted similarity value: [0.54,0.48,0.42,0.36], similarity value after sourceB weighting: [0.32,0.28,0.24,0.20], weighted fusion: the weighted similarity values of the two data sources are weighted and fused, and a simple weighted average mode can be adopted: final similarity value: 0.43,0.38,0.33,0.28, performing data fusion according to the fused similarity;

Finally, the fused similarity value can be used for weighting and fusing the real-time coal flow data: weighted pre-fusion second coal stream fusion data: [ (700.43+750.38), (800.43+850.38), (900.43+950.38), (1000.43+1050.38) ], the fused second coal stream fusion data were weighted: [76.85,83.55,90.25,96.95].

According to the invention, through target detection, the position and boundary of the coal flow can be accurately identified, so that accurate basic information is provided for subsequent clustering and data fusion. Through clustering, similar coal flow data can be classified to reduce the problem of overlarge calculation load caused by direct use of similarity calculation, and through multi-source weighted data fusion, information of different sources can be fully utilized, and the comprehensiveness and accuracy of the data are improved.

Preferably, the multi-source weighted data fusion is calculated by a multi-source coal stream data weighted calculation formula, the preliminary coal stream fused data comprises first coal stream fused data and second coal stream fused data, wherein the multi-source coal stream data weighted calculation formula specifically comprises:

；

for preliminary coal flow fusion data, +.>For real-time coal flow order item data, < + >>For real-time coal flow data, < > data >Is the firstWeight coefficient data of real-time coal stream data corresponding to individual coal stream data source data,/weight coefficient data of real-time coal stream data corresponding to individual coal stream data source data>Is->Real-time coal flow data corresponding to the individual coal flow data source data,/-the real-time coal flow data corresponding to the individual coal flow data source data>Is coal flow change rate data.

Specifically, for example, the real-time coal flow data (m) is 3, the real-time coal flow data source weight coefficient data [ ],/>,) 0.2,0.3,0.5, real-time coal flow data (/ -for)>,/>,/>) 10,15,20, coal flow rate of change data (q): 0.1, f=16.5+0.105≡ 16.605.

Specifically, for example, for coal mine a: the weight coefficient is=0.6, real-time coal stream dataIs->=[10,15,12,14]Data for coal mine B: the weight coefficient is->=0.4, real-time coal stream data is +.>=[12,14,10,13]，F≈[14.4,20,15.2,18.4]。

According to the invention, the information of different data sources can be comprehensively synthesized by introducing a multi-source coal flow data weighting calculation formula, and the contribution of each data source can be calculated according to the weight coefficient of the data source) The method can accurately reflect the data, so that the fused data is more representative and comprehensive. Wherein->Is first coal flow fusion data, and represents a weighted average value of multi-source coal flow data plus a limit term; />Is the number of sources of coal stream data, e.g. +.>=3 indicates that there are three different coal stream data sources; />Is->The weight coefficient of a personal coal stream data source, representing the credibility or importance of the data source, generally satisfies 0.ltoreq.o ≡ >Is less than or equal to 1；/>Is->Real-time coal flow data of a coal flow data source represents the coal flow or mass measured by the data source; />Is a variable representing the rate of change or the degree of fluctuation of coal flow data; />Is a logarithmic function representing the nonlinear characteristics or complexity of the coal stream data; />Is a root function, and represents the smoothness or stability of coal flow data; the invention has a certain degree of inhibition effect on abnormal values, and the influence on abnormal data generated due to various reasons is reduced in the data fusion process, thereby ensuring the stability and reliability of fusion results. />

Preferably, step S23 is specifically:

step S231: normalizing the first coal flow fusion data and the second coal flow fusion data to obtain first coal flow fusion normalized data and second coal flow fusion normalized data;

specifically, for example, linear normalization or other common normalization methods such as Min-Max Scaling or Z-Score normalization, etc. are used. Min-Max Scaling is performed, for example, using the MinMaxscaler class of the scikit-learn library in Python.

Step S232: performing self-attention calculation on the first coal flow fusion normalized data and the second coal flow fusion normalized data so as to obtain first coal flow fusion self-attention weight data and second coal flow fusion self-attention weight data;

Specifically, weights are calculated by learning correlations between different locations in the data, for example using a self-attention mechanism (Attention Mechanism) in a transducer model. Such as using a deep learning framework like the self-attention module in TensorFlow or pyrerch.

Step S233: performing self-attention weighted calculation and dimension reduction processing on the first coal flow fusion normalized data and the second coal flow fusion normalized data according to the first coal flow fusion self-attention weight data and the second coal flow fusion self-attention weight data, so as to obtain first coal flow fusion dimension reduction data and second coal flow fusion dimension reduction data;

specifically, the normalized data is weighted summed, for example using self-attention weights, and then the data dimension may optionally be reduced using a dimension reduction method such as Principal Component Analysis (PCA). Principal component analysis is performed, for example, using the PCA class of the scikit-learn library in Python.

Step S234: performing similarity matrix construction on the first coal flow fusion dimensionality reduction data and the second coal flow fusion dimensionality reduction data to obtain coal flow fusion similarity matrix data;

specifically, for example, a similarity matrix is constructed using various similarity calculation methods such as euclidean distance, cosine similarity, and the like. Similarity calculation is implemented, for example, using scikit-learn library in Python or NumPy.

Step S235: confidence calculation is carried out on the coal flow fusion similarity matrix data, so that coal flow fusion confidence data are obtained;

specifically, for example, confidence calculations may determine confidence by thresholding or using other statistical methods based on data in a similarity matrix. Confidence calculations are implemented, for example, using NumPy in Python or a statistical library.

Step S236: and carrying out coal flow data fusion on the first coal flow fusion data and the second coal flow fusion data according to the coal flow fusion confidence data, so as to obtain the coal flow fusion data.

Specifically, the first and second coal stream fusion data are weighted fused, for example, according to confidence data. Data fusion is performed, for example, using a NumPy library in Python.

Specifically, for example, first coal stream fusion data: [50,60,70,80], second coal stream fusion data: [45,55,65,75], coal flow fusion confidence data: [0.9,0.8,0.7,0.6], where confidence data represents a degree of trust for each fused data, a higher value represents a higher degree of trust for the data. The following steps are as follows: given confidence and weight: let confidence be the weight, namely: weight of first coal stream fusion data: [0.9,0.8,0.7,0.6], weight of second coal stream fusion data: [0.9,0.8,0.7,0.6], weighted fusion: multiplying the first coal flow fusion data and the second coal flow fusion data with their corresponding weights respectively to obtain weighted data: after the first coal flow fusion data are weighted: [45,48,49,48], after the second coal stream fusion data weighting: [40.5,44,45.5,45] fusing the weighted data: the final fused data can be obtained by a simple weighted average method, namely adding the weighted data of the corresponding positions and dividing the weighted data by the sum of the weights: final fusion data: [ (45+40.5)/(0.9+0.9), (48+44)/(0.8+0.8), (49+45.5)/(0.7+0.7), (48+45)/(0.6+0.6) ]), final fusion data: [44.7368, 46.6667, 47.1429, 46.1538].

The self-attention mechanism adopted in the invention can highlight important characteristics, thereby improving the key information of the data. The dimension reduction can reduce the dimension of data, improve the efficiency of subsequent calculation, and simultaneously keep key information. Through the construction of the similarity matrix, the similarity degree between data can be quantized, and meanwhile, the complexity brought by linear calculation can be reduced through matrix operation, so that the calculation load is reduced. The confidence reflects the credibility of the fusion data, and the contribution degree of each data source is effectively evaluated through analysis and calculation of the similarity. The final fusion operation is carried out on the first coal flow fusion data and the second coal flow fusion data by considering the confidence coefficient, so that the fusion result is ensured to be more representative and comprehensive, and a high-quality data basis is provided for the subsequent steps. The invention fully plays the roles of data processing and feature extraction, so that the coal flow fusion data has more representativeness and credibility.

Preferably, step S3 is specifically:

step S31: extracting coal flow variation characteristics of the coal flow fusion data by utilizing preset sliding window data, so as to obtain coal flow variation characteristic data;

specifically, the timing analysis algorithm or timing feature extraction is implemented, for example, using a common signal processing tool such as the Scipy library in MATLAB, python, or the like.

Specifically, the frequency domain information is acquired using fourier transform, or the change in the time domain is acquired using differential operation, for example.

Step S32: performing characteristic association according to the coal flow characteristic data and the real-time coal flow data, so as to obtain coal flow associated characteristic data;

specifically, for example, feature association is performed using a machine learning model such as a Support Vector Machine (SVM), a neural network, or the like, and feature association is performed by a preset machine learning model.

Specifically, for example, the coal flow correlation feature data is obtained by time-series alignment of the coal flow feature data and the real-time coal flow data, and calculating correlation coefficients between features or using model prediction correlation.

Step S33: and carrying out coal flow loss evaluation on the coal flow associated characteristic data by utilizing a preset rule engine and coal flow source data corresponding to the real-time coal flow data, thereby obtaining coal flow loss evaluation data.

Specifically, for example, using a rule engine such as Drools, etc., the coal flow loss evaluation may also be performed using a professional coal flow loss model, thereby obtaining the coal flow loss evaluation data.

Specifically, for example, real-time coal stream data: coal stream temperature: 60 ℃ and coal flow humidity: 0.8, coal flow pressure: 1.2 megapascals, coal flow density: 1.5 megagrams/cubic meter, rules engine: rule 1: if the coal flow temperature exceeds 70 degrees celsius, the coal flow loss increases by 20%. Rule 2: if the coal flow humidity exceeds 0.9, the coal flow loss increases by 10%. Rule 3: if the coal flow pressure exceeds 1.5 mpa, the coal flow loss increases by 15%. Rule 4: if the coal flow density is less than 1.0 megagram/cubic meter, the coal flow loss increases by 25%. These rules will now be applied to evaluate the loss of coal flow: according to the real-time coal flow data and rules of the rule engine: rule 1 does not apply because the coal stream temperature is 60 degrees celsius and does not exceed 70 degrees celsius. Rule 2 does not apply because the coal stream moisture is 0.8, not exceeding 0.9. Rule 3 does not apply because the coal stream pressure is 1.2 megapascals, not exceeding 1.5 megapascals. Rule 4 is also not applicable because the coal flow density is 1.5 megagrams per cubic meter, higher than 1.0 megagrams per cubic meter. Therefore, no additional adjustment of coal flow loss is required based on the evaluation of the rule engine. Final coal flow loss evaluation: according to the evaluation of the rule engine, the coal flow loss remained unchanged at 10%.

According to the invention, the sliding window technology is used for processing the coal flow fusion data, so that the dynamic change characteristics of the coal flow can be captured. By doing so, the sensitivity of the data can be improved, so that the response of the system to the coal flow change is more timely and accurate. By correlating the coal flow characteristic data with the real-time coal flow data, the inherent relationship between the coal flow characteristic data and the real-time coal flow data can be found, so that the system can better understand the actual meaning of the characteristics, and an accurate basis is provided for subsequent evaluation. The preset rule engine is a rule-based expert system capable of reasoning and making decisions based on specific conditions. By combining the real-time coal flow data with the coal flow source data, the coal flow associated characteristic data can be subjected to refined evaluation. The process has certain intelligence and can adapt to the wear evaluation requirements under different conditions.

Preferably, the multi-source coal flow control strategy data includes demand priority control strategy data and low-loss priority control strategy data, and step S5 is specifically:

step S51: constructing a coal flow demand estimation model according to the time sequence coal flow characteristic data, so as to obtain the coal flow demand estimation model;

Specifically, for example, a Scikit-learn library in a Python programming language is used to construct a linear regression model, and taking linear regression as an example, the model may involve calculation of coefficients, for example, y=ax+b, where y is output data of a coal flow demand estimation model, generally a coal flow demand pre-estimated value, a is multidimensional weight data, b is multidimensional adjustment data, and x is a vector form of time-series coal flow change feature data.

Step S52: carrying out coal flow demand estimation on the real-time coal flow data by utilizing a coal flow demand estimation model so as to obtain coal flow demand estimation data;

specifically, for example, the constructed prediction model is utilized to input real-time coal flow data into the model, so as to obtain a coal flow demand estimation result. Such as coal flow demand estimation using predictive functions in Scikit-learn libraries in the Python programming language.

Step S53: generating a demand priority control strategy according to the coal flow demand estimated data and the coal flow loss estimated data, so as to obtain demand priority control strategy data;

specifically, for example, a priority control strategy is designed based on the coal flow demand estimation data and the coal flow loss estimation data. For example, a rule may be formulated to preferentially satisfy the coal flows in the high demand areas.

Step S54: and generating a low-loss control strategy according to the estimated data of the coal flow demand and the estimated data of the coal flow loss so as to obtain low-loss priority control strategy data.

Specifically, for example, a low-loss priority control strategy is designed according to the coal flow demand estimation data and the coal flow loss estimation data. For example, a rule may be formulated to prioritize low loss paths for coal flow scheduling.

Specifically, for example, coal flow demand estimation data: high demand area a:1500 tons/hour, low demand zone B:800 ton/hr coal flow loss evaluation data: high demand area a:15%, low demand area B:10%, the coal stream source provides data of 1000 tons/hour; rule 1: if the coal flow demand of the high demand area A is estimated to be higher than that of the low demand area B, the coal flow of the high demand area A is preferentially satisfied. Examples: according to rule 1, the system will choose to preferentially satisfy the coal stream in high demand area A because of its greater demand. Rule 2: if the coal flow loss evaluation in the low demand area B is lower than that in the high demand area A, the low loss path is preferentially selected for coal flow scheduling. Examples: according to rule 2, the system will choose to prefer a low loss path for coal flow scheduling because the loss of coal flow is relatively low in the case of low demand region B. And (3) carrying out weighted calculation according to the rule 1 and the rule 2 so as to obtain composite control strategy data, and respectively carrying out coal flow supply operation on the areas A and B.

The invention uses the sliding window technology to process the coal flow fusion data, can capture the dynamic change characteristics of the coal flow, can improve the sensitivity of the data, and ensures that the response of the system to the coal flow change is more timely and accurate. By correlating the coal flow characteristic data with the real-time coal flow data, the inherent relationship between the coal flow characteristic data and the real-time coal flow data can be found, so that the system can better understand the actual meaning of the characteristics, and an accurate basis is provided for subsequent evaluation. The preset rule engine is a rule-based expert system capable of reasoning and making decisions based on specific conditions. By combining the real-time coal flow data with the coal flow source data, the coal flow associated characteristic data can be subjected to refined evaluation. The invention can make reasonable decisions in complex coal flow environments, and ensures the efficient transportation and utilization of coal flows, thereby obtaining remarkable economic and social benefits in practice.

By integrating the real-time coal mine data, the real-time transfer station data and the real-time coal processing data, the multi-source data can be effectively fused and utilized, the comprehensiveness and accuracy of the data can be improved, and a solid foundation can be provided for subsequent coal flow analysis and control. By means of multi-source data fusion, information of different data sources is integrated together, comprehensive coal flow fusion data are obtained, and the system is guaranteed to be capable of controlling the whole coal flow system more comprehensively. By extracting the characteristics of the fused data, the system can understand the change rule of the coal flow more deeply. The time sequence characteristic extraction is carried out by utilizing the historical coal flow data, so that the evolution process of the coal flow can be known in the time dimension, and the knowledge of the coal flow characteristics is further enriched. By predicting the time sequence coal flow characteristic data, the system can know the trend of future coal flow demands in advance, so that the coal flow control strategy can be adjusted in a targeted manner. Based on the coal flow demand estimation data and the coal flow loss evaluation data, the system can make a multi-source coal flow control strategy, so that the coal flow loss can be reduced to the greatest extent while the supply is ensured.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. The self-adaptive control method for the coal flow balance based on multiple sources is characterized by comprising the following steps of:

step S1: acquiring coal flow data through a coal flow data acquisition device so as to obtain real-time coal flow data, wherein the real-time coal flow data comprises real-time coal mine data, real-time transfer station data and real-time coal processing data;

Step S2: carrying out multi-source data fusion on the real-time coal stream data so as to obtain coal stream fusion data;

step S3: extracting coal flow change characteristics of the coal flow fusion data to obtain coal flow change characteristic data, and carrying out coal flow loss evaluation according to the coal flow change characteristic data and coal flow source data corresponding to the real-time coal flow data to obtain coal flow loss evaluation data;

step S4: acquiring historical coal flow data, and extracting time sequence coal flow change characteristics according to the historical coal flow data so as to obtain time sequence coal flow change characteristic data;

step S5: estimating the coal flow demand according to the time sequence coal flow characteristic data to obtain coal flow demand estimated data, and generating a multi-source coal flow control strategy according to the coal flow demand estimated data and the coal flow loss estimated data to obtain multi-source coal flow control strategy data.

2. The method according to claim 1, wherein step S1 is specifically:

step S11: the coal mine data acquisition device is connected with a coal mine data source to acquire real-time coal mine data;

step S12: the coal flow data acquisition device is connected with a transfer station data source to acquire real-time transfer station data;

Step S13: the coal flow data acquisition device is connected with a coal processing data source to acquire real-time coal processing data;

step S14: data integration is carried out on the real-time coal mine data, the real-time transfer station data and the real-time coal processing data, so that preliminary real-time coal flow data are obtained;

step S15: and carrying out coal flow data quality screening on the preliminary real-time coal flow data, thereby obtaining real-time coal flow data.

3. The method according to claim 2, wherein step S15 is specifically:

step S151: performing coal flow data accuracy verification on the preliminary real-time coal flow data according to preset coal flow transportation electronic map data, so as to obtain accuracy verification data;

step S152: performing coal flow data restoration on the primary real-time coal flow data according to the accuracy verification data, so as to obtain real-time coal flow restoration data;

step S153: and carrying out distributed transaction verification and data correction on the real-time coal flow repair data, thereby obtaining the real-time coal flow data.

4. The method of claim 3, wherein the repairing of the coal stream data in step S152 is performed by performing a repairing policy selection for repairing index data generated by a coal stream data repairing calculation formula, wherein the coal stream data repairing calculation formula specifically includes:

；

For restoring index data, < >>For preliminary real-time coal flow data, +.>For preliminary real-time coal flow sequence item data, < ->For preliminary real-time coal flow data, < > for>For coal flow data, +.>For coal flow quality data, +.>For coal flow velocity data, +.>Is coal flow temperature weight data, +.>For coal flow temperature data, +.>For coal stream moisture weight data, < >>For coal flow humidity data, < >>Is the pressure of coal flowWeight data->For coal flow pressure data, +.>For coal flow density weight data, +.>Is coal flow density data.

5. The method according to claim 1, wherein step S2 is specifically:

step S21: carrying out multi-source weighted data fusion on the real-time coal stream data so as to obtain first coal stream fusion data;

step S22: performing multi-source clustering data fusion on the real-time coal stream data so as to obtain second coal stream fusion data;

step S23: and carrying out confidence voting calculation according to the first coal flow fusion data and the second coal flow fusion data, thereby obtaining the coal flow fusion data.

6. The method according to claim 5, wherein step S22 is specifically:

step S221: performing target detection on coal flow image data in the real-time coal flow data so as to obtain coal flow target detection data;

Step S222: clustering calculation is carried out according to non-coal flow image data and coal flow target detection data in the real-time coal flow data, so that real-time coal flow clustering data is obtained;

step S223: performing similarity calculation on the real-time coal flow clustering data to obtain clustering similarity data;

step S224: and carrying out multi-source weighted data fusion on the real-time coal stream data according to the clustering similarity data, thereby obtaining second coal stream fusion data.

7. The method of claim 6, wherein the multi-source weighted data fusion is calculated by a multi-source weighted calculation formula for the coal stream data, the preliminary coal stream fusion data comprising first coal stream fusion data and second coal stream fusion data, wherein the multi-source weighted calculation formula for the coal stream data is:

；

for preliminary coal flow fusion data, +.>For real-time coal flow order item data, < + >>For real-time coal flow data, < > data>Is->Weight coefficient data of real-time coal stream data corresponding to individual coal stream data source data,/weight coefficient data of real-time coal stream data corresponding to individual coal stream data source data>Is->Real-time coal flow data corresponding to the individual coal flow data source data,/-the real-time coal flow data corresponding to the individual coal flow data source data>Is coal flow change rate data.

8. The method according to claim 5, wherein step S23 is specifically:

Step S231: normalizing the first coal flow fusion data and the second coal flow fusion data to obtain first coal flow fusion normalized data and second coal flow fusion normalized data;

step S232: performing self-attention calculation on the first coal flow fusion normalized data and the second coal flow fusion normalized data so as to obtain first coal flow fusion self-attention weight data and second coal flow fusion self-attention weight data;

step S233: performing self-attention weighted calculation and dimension reduction processing on the first coal flow fusion normalized data and the second coal flow fusion normalized data according to the first coal flow fusion self-attention weight data and the second coal flow fusion self-attention weight data, so as to obtain first coal flow fusion dimension reduction data and second coal flow fusion dimension reduction data;

step S234: performing similarity matrix construction on the first coal flow fusion dimensionality reduction data and the second coal flow fusion dimensionality reduction data to obtain coal flow fusion similarity matrix data;

step S235: confidence calculation is carried out on the coal flow fusion similarity matrix data, so that coal flow fusion confidence data are obtained;

step S236: and carrying out coal flow data fusion on the first coal flow fusion data and the second coal flow fusion data according to the coal flow fusion confidence data, so as to obtain the coal flow fusion data.

9. The method according to claim 1, wherein step S3 is specifically:

step S31: extracting coal flow variation characteristics of the coal flow fusion data by utilizing preset sliding window data, so as to obtain coal flow variation characteristic data;

step S32: performing characteristic association according to the coal flow characteristic data and the real-time coal flow data, so as to obtain coal flow associated characteristic data;

step S33: and carrying out coal flow loss evaluation on the coal flow associated characteristic data by utilizing a preset rule engine and coal flow source data corresponding to the real-time coal flow data, thereby obtaining coal flow loss evaluation data.

10. The method according to claim 1, wherein the multi-source coal flow control strategy data includes demand priority control strategy data and low-loss priority control strategy data, and step S5 is specifically:

step S51: constructing a coal flow demand estimation model according to the time sequence coal flow characteristic data, so as to obtain the coal flow demand estimation model;

step S52: carrying out coal flow demand estimation on the real-time coal flow data by utilizing a coal flow demand estimation model so as to obtain coal flow demand estimation data;

step S53: generating a demand priority control strategy according to the coal flow demand estimated data and the coal flow loss estimated data, so as to obtain demand priority control strategy data;

Step S54: and generating a low-loss control strategy according to the estimated data of the coal flow demand and the estimated data of the coal flow loss so as to obtain low-loss priority control strategy data.