Coking coal blending precision calculation method and implementation system
By monitoring the cumulative amount and rate of coal discharge from the silos in real time, identifying and processing abnormal data such as suspended material and collapse, the problem of data misjudgment in the coking coal blending process was solved, and the accurate calculation of blending precision and quality indicators was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JIANGSU SHAGANG HIGH-TECH INFORMATION TECH CO LTD
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-10
Smart Images
Figure CN122117158B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of computational technology in the metallurgical industry, specifically a method and implementation system for calculating the accuracy of coking coal blending. Background Technology
[0002] Currently, while the coking coal blending process in the metallurgical industry has achieved a high level of automation, existing coal blending accuracy calculation systems generally rely on the cumulative coal discharge from silos directly collected from databases, assuming this data is completely equivalent to the actual physical coal discharge. However, in actual production, due to the influence of raw coal moisture, particle size, and silo structure, "suspended material" or "bulging material" phenomena frequently occur. In this case, the flow meter or weighing signal may only reflect vibration or virtual flow, rather than the actual material displacement, leading to a serious disconnect between the collected "actual coal quantity" and the physical coal discharge quantity.
[0003] To address such data anomalies, existing technologies typically employ linear interpolation to smooth minute-level cumulative quantities. This method misinterprets spurious cumulative increments as "normal coal feeding rates" during the material suspension period and misidentifies sudden changes in actual flow rate at the moment of collapse as anomalies, discarding or smoothing them out. This data processing approach directly leads to "false compliance" or "false exceedance" in proportioning accuracy calculations: the system cannot distinguish whether deviations stem from improper process proportioning or equipment malfunction, frequently generating misleading alarms. More seriously, because the theoretical quality of the mixed coal depends entirely on the distorted actual coal feeding ratio, a systematic deviation occurs between the theoretical value of the quality plan and subsequent mixed coal testing results, causing quality accuracy indicators to lose their physical meaning and process guidance value.
[0004] Therefore, the present invention provides a method and system for calculating the accuracy of coking coal blending. Summary of the Invention
[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.
[0006] The technical solution adopted by this invention to solve its technical problem is: a method for calculating the accuracy of coking coal blending, comprising the following steps:
[0007] Step S10: Collect the cumulative amount of coal fed into each silo and the instantaneous rate in real time, monitor the fluctuation characteristics of the weighing signal simultaneously, identify and mark false data segments caused by suspended or overflowing material, as well as the rate change segments at the moment of collapse.
[0008] Step S20: For the marked false data segments, pause the use of the original cumulative amount for interpolation calculation, estimate the actual coal amount per minute in the period based on the historical coal feeding trend of adjacent normal periods, and retain the original rate value for the rate change segment at the moment of collapse.
[0009] Step S30: For the unlabeled normal data segment, interpolation calculation is performed using the original cumulative amount. The estimated value of the false data segment, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segment are merged in chronological order to form the corrected actual coal delivery sequence per minute.
[0010] Step S40: Based on the corrected actual coal feeding amount, calculate the actual coal feeding ratio per minute by coal type, and compare it with the planned ratio minute by minute to determine the ratio accuracy;
[0011] Step S50: Based on the corrected actual coal feeding ratio, calculate the planned theoretical value and upper and lower limits of the quality of each component, compare them with the test results of the mixed coal, and generate the quality accuracy index of each component.
[0012] As a further aspect of the present invention: in step S10, the specific method for identifying and marking spurious data segments is as follows:
[0013] The data acquisition server reads the instantaneous rate records of the silo for the most recent 120 seconds every 60 seconds, calculates its arithmetic mean and standard deviation; reads the cumulative total value of the first and last moments within the 120 seconds, and calculates the actual increase in the cumulative amount; if the standard deviation of the instantaneous rate in the current 120-second period exceeds the fluctuation threshold, and the actual increase in the cumulative amount in the period is lower than the lower limit of the expected increase, then the period is determined to be a false data segment of suspended material.
[0014] As a further aspect of the present invention: In step S10, the process of identifying the rate abrupt change segment at the moment of collapse is as follows:
[0015] After marking false data segments, for each time period that has been marked as a false data segment, continue to monitor the instantaneous rate data within 60 seconds after the end of the time period, and set the mutation detection window to 10 seconds, that is, continuously detect 10 instantaneous rate sampling points;
[0016] When any five consecutive sampling points within the mutation detection window have instantaneous rates below the lower limit reference value, followed by five consecutive sampling points with instantaneous rates above the upper limit reference value, and the cumulative increment within 10 seconds exceeds twice the normal cumulative increment within 10 seconds, it is determined to be a collapse event; the time range covered by the collapse event is marked as the collapse rate mutation segment.
[0017] As a further aspect of the present invention: In step S20, the specific method for estimating the actual coal output per minute within the false data segment is as follows:
[0018] Taking the start time of the false data segment as minute 1, read the average rate of the last 60 seconds of the preceding normal period and the average rate of the first 60 seconds of the following normal period, and take the arithmetic mean of the two as the estimated baseline rate; for minute t, the estimated coal delivery is equal to the estimated baseline rate multiplied by 1 minute plus the slope of the change trend multiplied by t minus the difference of the intermediate minute number.
[0019] As a further aspect of the present invention: in step S20, the method for retaining the original rate value for the abrupt change in collapse rate is as follows:
[0020] Read all second-level records marked as collapse mutation segments, without performing any interpolation calculations or estimation replacement operations, and directly output the original instantaneous rate value and the original cumulative value in the record to subsequent steps; the cumulative difference between adjacent second-level records within the collapse mutation segment is not smoothed, and the original mutation amplitude is maintained.
[0021] As a further aspect of the present invention: in step S30, the interpolation calculation for the unlabeled normal data segment is performed as follows:
[0022] Obtain the original cumulative values at the start and end times of the normal data segment, calculate the difference between the two as the total increment of the cumulative value, and then divide it by the total duration of the data segment to obtain the interpolated coal feeding amount per minute.
[0023] The interpolated coal quantity is assigned to all 60 second-level records within that minute, and the cumulative quantity is generated second by second in an equal increment manner.
[0024] As a further aspect of the present invention: in step S30, the method for merging to form the corrected sequence of actual coal delivery per minute is as follows:
[0025] Establish a continuous minute time axis covering the entire coal blending execution period. According to the priority rule that the slump retention value takes precedence over the estimated value, and the estimated value takes precedence over the interpolated normal value, the corresponding coal feeding value is sequentially taken from the temporary table of slump retention value, the temporary table of estimated value, and the temporary table of interpolated normal value and filled into the minute point to form the actual coal feeding sequence per minute.
[0026] Among them, the interpolation normal value temporary table is composed of the interpolated coal amount per minute of the normal data segment, the estimated value temporary table is composed of the estimated value per minute corresponding to the false data segment, and the collapse retention value temporary table is composed of the original instantaneous rate value per second corresponding to the collapse rate mutation segment, which is aggregated and stored by minute.
[0027] As a further aspect of the present invention: in step S40, the method for determining the proportioning accuracy is as follows:
[0028] The total coal discharge is obtained by summing the actual coal discharge amounts of all silos every minute. The actual coal discharge amount of each coal type is obtained by adding the coal discharge amounts of the corresponding silos. The ratio of the total coal discharge amount to the actual coal discharge amount of each coal type is calculated to obtain the actual coal discharge ratio of each coal type per minute.
[0029] Calculate the absolute difference between the actual coal ratio and the planned ratio. If it is greater than 5%, it is counted as one deviation. Add up the number of deviations for all coal types and for all minutes to obtain the ratio accuracy.
[0030] As a further aspect of the present invention: in step S50, the method for generating the quality accuracy index is as follows:
[0031] Calculate the product of each coal type's quality index and the actual coal feeding ratio per minute, and sum them to obtain the theoretical value of the quality plan per minute. Then multiply by the preset first coefficient 1.1 and second coefficient 0.9 to obtain the upper and lower limits. Compare the actual values of the mixed coal test with the corresponding upper and lower limits per minute, and count the proportion of qualified tests to the total number of tests to obtain the quality accuracy index of each component.
[0032] A coking coal blending accuracy calculation system includes the following modules:
[0033] The abnormal data identification and marking module collects the cumulative amount and instantaneous rate of coal feeding in each silo in real time, and simultaneously monitors the fluctuation characteristics of the weighing signal. It identifies and marks false data segments caused by suspended or overflowing material, as well as segments of sudden rate changes at the moment of collapse.
[0034] The false segment estimation and compensation module suspends the use of the original cumulative amount for interpolation calculation of the marked false data segments. Based on the historical coal feeding trend of adjacent normal periods, it estimates the actual coal feeding per minute in the period. For the rate change segment at the moment of collapse, the original rate value is retained.
[0035] The time-series data fusion and correction module uses the original cumulative amount to interpolate the unlabeled normal data segments, and merges the estimated value of the false data segments, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segments in chronological order to form a corrected sequence of actual coal output per minute.
[0036] The proportioning accuracy comparison module calculates the actual coal feeding ratio per minute based on the corrected actual coal feeding amount, combines the coal types to calculate the actual coal feeding ratio per minute, and compares it with the planned ratio minute by minute to determine the proportioning accuracy.
[0037] The quality index comparison and generation module calculates the planned theoretical value and upper and lower limits of the quality of each component based on the corrected actual coal feeding ratio, compares it with the test results of the mixed coal, and generates the quality accuracy index of each component.
[0038] The beneficial effects of this invention are as follows:
[0039] First, this invention establishes a dual-condition judgment rule based on standard deviation and cumulative increment by real-time monitoring of the fluctuation characteristics of instantaneous rate and cumulative amount. This allows for the accurate identification and differentiation of false data segments related to suspended material and abrupt changes in slump rate. Compared to existing technologies that treat all data as equivalent, this invention identifies data anomaly types at the source, effectively solving the problem of data decoupling from the physical coal quantity caused by suspended material.
[0040] Second, this invention suspends the use of the original cumulative amount for interpolation in the false data segment of suspended material, and instead estimates the coal delivery amount based on the historical trend of adjacent normal periods; for the collapse mutation segment, the original rate value is directly retained without smoothing and elimination. This differentiated processing mechanism avoids the dual defects of the existing linear interpolation method, which misinterprets false signals as "normal rates" and misjudges real mutations as anomalies.
[0041] Third, this invention combines the estimated value, slump retention value, and interpolated normal value in chronological order to form a corrected sequence of actual coal feeding per minute. This sequence recreates the actual coal feeding process, enabling the proportioning accuracy calculation to accurately distinguish between process deviations and equipment malfunctions, eliminating the "false compliance" or "false exceedance" problems caused by data distortion in existing technologies.
[0042] Fourth, this invention calculates the theoretical value and upper and lower limits of the quality plan based on the corrected actual coal feeding ratio, so as to establish a real physical correspondence between the theoretical value and the test results of mixed coal, thus solving the problem in the prior art that the quality accuracy index loses its process guidance value due to data distortion. Attached Figure Description
[0043] The invention will now be further described with reference to the accompanying drawings.
[0044] Figure 1 This is a flowchart illustrating the steps of a method for calculating the accuracy of coking coal blending according to an embodiment of the present invention.
[0045] Figure 2 This is a flowchart of a coking coal blending accuracy calculation system according to an embodiment of the present invention. Detailed Implementation
[0046] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.
[0047] Example 1, please refer to Figure 1 As shown in the embodiment of the present invention, a method for calculating the accuracy of coking coal blending includes the following steps:
[0048] Step S10: Collect the cumulative amount of coal fed into each silo and the instantaneous rate in real time, monitor the fluctuation characteristics of the weighing signal simultaneously, identify and mark false data segments caused by suspended or overflowing material, as well as the rate change segments at the moment of collapse.
[0049] In step S10, the process of real-time acquisition of the cumulative coal discharge volume and instantaneous rate of each silo is as follows:
[0050] A solid flow meter is installed at the discharge chute of each silo. This flow meter continuously measures the instantaneous coal discharge rate at a sampling frequency of once per second, in tons per minute. The flow meter has a built-in integrator that integrates the instantaneous rate over time, simultaneously calculating the cumulative coal discharge volume from the start of this coal blending operation to the current moment, in tons. The instantaneous rate and cumulative total volume data streams are transmitted in real time to a data acquisition server. Each data record contains four fields: silo number, timestamp (accurate to the second), instantaneous rate value, and cumulative total volume value.
[0051] In step S10, the process of synchronously monitoring the fluctuation characteristics of the weighing signal is as follows:
[0052] The data acquisition server performs a fluctuation characteristic analysis on the data from each silo every 60 seconds. The specific steps are as follows: It reads all instantaneous rate records for the silo over the past 120 seconds from memory or a time-series database, totaling 121 data points (including the current moment). It calculates the arithmetic mean and standard deviation of these 121 data points. Simultaneously, it reads the cumulative total value at the first and last two moments within this 120 seconds, calculates the difference between the two, and obtains the actual increase in the cumulative amount over that 120 seconds.
[0053] In step S10, the process of identifying and marking false data segments caused by suspended or protruding material is as follows:
[0054] The calculated standard deviation of the time rate is compared with the preset fluctuation threshold;
[0055] If the instantaneous rate standard deviation exceeds the fluctuation threshold within the current 120-second period, and the actual cumulative increase within the period is lower than the expected lower limit of the increase, then the period is determined to be a false data segment with suspended data.
[0056] The fluctuation threshold is set as follows: take all normal time period data of the silo in the past 24 hours that have not been marked as abnormal, calculate the average value of the standard deviation of the instantaneous rate, and take twice the average value as the fluctuation threshold. The expected increment lower limit is calculated as follows: take the average instantaneous rate of the normal coal feeding period of the silo in the past 24 hours, multiply it by 120 seconds (i.e. 2 minutes), and then multiply it by a coefficient of 0.2.
[0057] Once a data segment is identified as false, all second-level records within the false data period are written with the number 1 in the marker field; the marker field is stored independently and does not overwrite the original collected data;
[0058] It should be noted that the calculation basis for setting the fluctuation threshold and the lower limit of the expected increment is as follows: the fluctuation threshold is twice the average standard deviation of the instantaneous rate during the normal coal feeding period of the silo in the past 24 hours. The reason is that during normal coal feeding, the instantaneous rate follows a random normal distribution around the mean. The probability of the standard deviation falling within ±1 times the mean is about 68%, and the probability of falling within ±2 times the mean is about 95%. Setting the threshold to twice the standard deviation can effectively eliminate random noise interference while covering 95% of normal fluctuations. If the threshold is set too low (e.g., once), normal random fluctuations will be misjudged as abnormal; if it is set too high (e.g., three times), abnormal fluctuation signals in the early stage of material suspension will be missed. The lower limit of the expected increment is the average instantaneous rate during the normal coal feeding period of the silo in the past 24 hours multiplied by 120 seconds and then multiplied by a coefficient of 0.2. The reason is that when material suspension occurs, the material in the silo forms an arch bridge structure, the feeding channel is blocked, and the actual discharge volume usually drops to less than 20% of the normal flow rate. The coefficient 0.2, or 20%, physically represents the maximum permissible residual material ratio under suspended material conditions. If the coefficient is set too high (e.g., 0.5), normal low-flow periods will be misjudged as suspended material; if set too low (e.g., 0.1), some suspended material will be missed. Multiplying by a 120-second time window is to accumulate sufficient statistical time to avoid misjudgments triggered by minute-level fluctuations.
[0059] In step S10, the process of identifying the rate abrupt change segment at the moment of collapse is as follows:
[0060] After marking false data segments, for each time period marked as a false data segment, continue monitoring the instantaneous rate data within 60 seconds after the end of that time period. Set the mutation detection window to 10 seconds, meaning continuously monitor 10 instantaneous rate sampling points.
[0061] The lower limit reference value is calculated as 50% of the average instantaneous rate during the normal coal feeding period of the silo, and the upper limit reference value is calculated as 150% of the average instantaneous rate. When the instantaneous rate of any 5 consecutive sampling points in the sudden change detection window is lower than the lower limit reference value, and the instantaneous rate of the next 5 consecutive sampling points is higher than the upper limit reference value, and the cumulative increase in volume within these 10 seconds exceeds twice the normal cumulative increase in volume within 10 seconds, it is judged as a collapse event.
[0062] The time range covered by the collapse event (from the first sampling point when the rate starts to fall below the lower reference value to the last sampling point when the rate recovers to below the upper reference value) is marked as the collapse rate mutation segment; the number 2 is written into the marking field. If a second-level record simultaneously meets the conditions of the false data segment marking (number 1) and the collapse mutation segment marking (number 2), the number 2 is taken as the final marking value, that is, the collapse marking has higher priority than the false data marking;
[0063] The lower and upper reference values for collapse detection are set as follows: the lower reference value is 50% of the normal average instantaneous rate, and the upper reference value is 150% of the normal average instantaneous rate. This is based on the fact that before a collapse, suspended material causes the instantaneous rate to remain consistently low; a value 50% below the normal value effectively distinguishes between a suspended material state and a normal low-load state. During a collapse, the accumulated material surges out instantly, causing the instantaneous rate to increase sharply to over 150% of the normal value. This increase is significantly higher than the rate fluctuation range under normal operating conditions (usually not exceeding ±30% of the average). The symmetrical threshold settings of 50% and 150% ensure a high detection rate for collapse events while avoiding misjudging normal start-ups or short-term fluctuations as collapses. The setting for a cumulative increase exceeding twice the normal value is based on the following: the cumulative increase in 10 seconds equals the normal average instantaneous rate multiplied by 10 seconds and then divided by 60 (converting tons / minute to tons / 10 seconds). When a collapse occurs, the material accumulated during the suspension period is released in a concentrated manner, with the cumulative increase typically exceeding twice the normal value. Setting twice the value as the threshold is based on the following: under normal operating conditions, the random fluctuation range of the cumulative increase is generally within ±30%. A threshold of twice the value significantly exceeds the normal fluctuation range, uniquely indicating a collapse event. If this threshold is set too low (e.g., 1.5 times), normal load fluctuations may be misjudged as collapses; if set too high (e.g., 3 times), some collapse events may be missed.
[0064] Understandably, the significance of step S10 lies in the fact that, in actual coking coal blending production, materials in the silo are prone to suspension and bulging due to factors such as moisture and particle size, leading to false fluctuations in the weighing signal. Furthermore, sudden rate changes during collapse are often misjudged and rejected by traditional smoothing algorithms as outliers. Step S10, by real-time acquisition of instantaneous rate and cumulative quantity, synchronously monitors fluctuation characteristics and establishes a dual-condition judgment rule based on standard deviation and cumulative quantity increment. This allows for accurate differentiation between three states: normal coal feeding, false suspension, and sudden collapse. The significance of this step lies in identifying data anomaly types at the source, providing a reliable basis for subsequent segmented processing, avoiding misattributing equipment operating conditions to coal blending process deviations, and thus ensuring the accuracy and validity of the data foundation for coal blending accuracy calculations.
[0065] Step S20: For the marked false data segments, pause the use of the original cumulative amount for interpolation calculation, estimate the actual coal amount per minute in the period based on the historical coal feeding trend of adjacent normal periods, and retain the original rate value for the rate change segment at the moment of collapse.
[0066] In step S20, the method for suspending the use of the original cumulative amount for interpolation calculation for the marked spurious data segments is as follows:
[0067] The data acquisition server reads the marker field of each silo. When multiple consecutive second-level records have a marker value of 1 (suspicious data segment), a pause interpolation command is triggered. The specific condition is: within a continuous time range of the same silo, the number of records with a marker value of 1 reaches 180 (i.e., 3 consecutive minutes). Once this condition is met, the system pauses the execution of the original cumulative linear interpolation calculation program for all second-level records within that suspicious data segment. During the pause, the original cumulative data continues to be collected and stored, but is not used in the coal blending accuracy calculation process. When the marker value returns to 0 (normal segment) or becomes 2 (collapse segment), the interpolation calculation automatically resumes.
[0068] In step S20, the method for determining adjacent normal time periods is as follows:
[0069] Using the marked spurious data segment as a baseline, search for normal data records with a marker value of 0 both before and after the timeline. The preceding normal time period consists of at least 120 consecutive records (2 minutes) with all marker values of 0, preceding the start time of the spurious data segment. The following normal time period consists of at least 120 consecutive records with all marker values of 0, following the end time of the spurious data segment. If the condition of 120 consecutive normal records cannot be met before or after the timeline, the search continues in that direction until at least 60 consecutive normal records are found; if this condition is still not met, only one side of the normal time period is used for estimation.
[0070] In step S20, the historical coal-falling trend of adjacent normal periods is calculated as follows:
[0071] Read the instantaneous rate values of the last 60 seconds within the preceding normal period, calculate the arithmetic mean of these 60 values, and denot it as the forward rate mean V_front. Read the instantaneous rate values of the first 60 seconds within the following normal period, calculate the arithmetic mean of these 60 values, and denot it as the backward rate mean V_back. Take the arithmetic mean of V_front and V_back again to obtain the estimated baseline rate V_base within this spurious data segment. Simultaneously, read the cumulative increment every 10 seconds within the last 60 seconds of the preceding normal period and calculate its slope; read the cumulative increment every 10 seconds within the first 60 seconds of the following normal period and calculate its slope; take the average of the two slopes as the cumulative change trend slope K within this spurious data segment.
[0072] In step S20, the specific process for estimating the actual coal output per minute during the time period is as follows:
[0073] The false data segment is divided into minutes, with each minute segment containing 60 records at the second level. For minute t (t=1,2,3…, with the start time of the false data segment as minute 1), the estimated coal quantity Q_t is calculated using the formula: Q_t=V_base×1 minute+K×(t-T_mid); where T_mid is the middle minute number of the false data segment;
[0074] If both preceding and following normal time periods exist within the spurious data segment, then V_base is dynamically adjusted using linear interpolation.
[0075] V_base(t) = V_front × (T_end - t) / (T_end - T_start) + V_back × (t -T_start) / (T_end - T_start); where T_start and T_end are the start and end minute numbers of the dummy data segment, respectively;
[0076] The calculated Q_t is used as the basis for assigning the instantaneous estimated rate value for each second within this minute segment. That is, all 60 second-level records within this minute segment are assigned the same instantaneous estimated rate value, which is equal to Q_t divided by 60 and then multiplied by 60 (i.e., Q_t itself, in tons / minute).
[0077] In step S20, the process of preserving the original rate value for the rate abrupt change segment at the moment of collapse is as follows:
[0078] Read all second-level records marked with a value of 2 (collapse rate abrupt change segment). For these records, the system does not perform any interpolation calculations or estimation replacement operations, but directly outputs the original instantaneous rate value and original cumulative value in the record as the actual coal feeding data at that moment to subsequent steps. At the same time, for the cumulative difference between adjacent second-level records within the collapse abrupt change segment, no smoothing processing is performed, and its original abrupt change amplitude is maintained. If the collapse abrupt change segment and the spurious data segment overlap in time, the mark value of 2 for the collapse abrupt change segment is used, and all original rate values in the overlapping part are retained, without using the estimation logic of the spurious data segment. The start time of the collapse abrupt change segment is defined as the sampling point when the rate first falls below the lower limit reference value, and the end time is defined as the sampling point when the rate recovers to below the upper limit reference value and remains stable for 3 consecutive seconds. All second-level records from the start to the end are marked with 2 and the original value is retained;
[0079] Understandably, the significance of step S20 lies in the following: for marked false data segments, continuing to use the original cumulative amount for interpolation calculations would smooth the false signals to a "normal" coal feeding rate, leading to distortion in the proportioning accuracy calculation; while for abrupt collapse segments, smoothing would lose the true physical impact characteristics. Step S20 adopts differentiated processing for the two types of anomalies: interpolation is paused for suspended material segments, and the coal feeding amount is estimated based on the historical trend of adjacent normal periods; the original rate value is directly retained for collapsed segments. The significance of this step is that it establishes a classification and repair mechanism for abnormal data, which not only fills the data gaps during the suspended material period but also completely preserves the physical authenticity of the collapse moment, providing a reliable data sequence for subsequent proportion calculations.
[0080] Step S30: For the unlabeled normal data segment, interpolation calculation is performed using the original cumulative amount. The estimated value of the false data segment, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segment are merged in chronological order to form the corrected actual coal delivery sequence per minute.
[0081] In step S30, the process of interpolating the unlabeled normal data segments using the original cumulative values is as follows:
[0082] Read all second-level records where the value in the marker field is 0. These records correspond to unmarked normal data segments. For each normal data segment, obtain the original cumulative value C_start at the start time and the original cumulative value C_end at the end time, as well as the start timestamp T_start and end timestamp T_end. The total duration of this data segment is D = T_end - T_start, in minutes. The total increment of the cumulative amount within this data segment is ΔC = C_end - C_start. For each minute i (i=1,2,3,…,D) within this data segment, the interpolated coal amount Q_normal_i is calculated using the formula: Q_normal_i = ΔC / D. This means that the total increment of the cumulative amount within this data segment is evenly distributed over minutes, resulting in the same interpolated coal amount for each minute. Q_normal_i is assigned to all 60 second-level records within that minute as the instantaneous coal feeding rate value per second for that minute, in tons per minute. The cumulative value within that minute is generated second by second, incrementing by Q_normal_i / 60 per second.
[0083] In step S30, the estimated value of the false data segment, the original value of the instantaneous rate change segment during collapse, and the interpolation result of the normal data segment are obtained as follows:
[0084] Read the estimated value Q_t per minute corresponding to the false data segment from the output of step S20, where each Q_t corresponds to a minute timestamp. Read the raw instantaneous rate value per second corresponding to the collapse mutation segment (marked as 2) from the raw collected data of step S10, and aggregate it by minute: take the arithmetic mean of 60 raw instantaneous rate values within the same minute as the raw coal discharge amount Q_collapse_i for that minute. Read the interpolated coal discharge amount Q_normal_i per minute of the normal data segment from the calculation results of the first part of step S30. The above three types of data are stored independently in a temporary data table. Each type of data contains four fields: silo number, minute timestamp, coal discharge amount value, and data source identifier, where the data source identifiers are "estimated value", "collapse retention value", and "interpolated normal value", respectively.
[0085] In step S30, the process of forming the corrected sequence of actual coal delivery per minute is as follows:
[0086] Using minutes as the smallest time unit and silo number as the primary key, a continuous minute timeline covering the entire coal blending execution period (from the start to the end of coal blending) is established. Each minute point on the timeline corresponds to a unique timestamp. Each minute point is traversed in ascending order of timestamps, and the following checks are performed sequentially: If the minute timestamp falls within the range of the collapse abrupt change segment marker, Q_collapse_i is retrieved from the "collapse retention value" temporary table and filled into the minute point; if the minute timestamp falls within the range of the false data segment marker, Q_t is retrieved from the "estimated value" temporary table and filled into the minute point; if the minute timestamp falls within the range of the normal data segment marker, Q_normal_i is retrieved from the "interpolated normal value" temporary table and filled into the minute point. For minute points that simultaneously satisfy multiple marker ranges (i.e., different markers overlap at the minute boundary), values are selected according to the priority rule of "collapse retention value takes precedence over estimated value, estimated value takes precedence over interpolated normal value," forming a sequence of actual coal output per minute.
[0087] Understandably, the significance of step S30 lies in the following: After processing in steps S10 and S20, the original data is divided into three categories: normal segments, spurious estimation segments, and slump retention segments. These need to be re-integrated into a unified time series for subsequent calculations. Step S30 merges the processing results of the three types of data in chronological order and establishes priority rules (slump retention values take precedence over estimation values, and estimation values take precedence over interpolated normal values) to form a corrected sequence of actual coal output per minute. The significance of this step is that it achieves seamless splicing of heterogeneous data sources, ensuring that each minute on the timeline has a clear data source. At the same time, through deduplication of boundary minutes, it eliminates data jumps that may be introduced by segmented processing, providing continuous, complete, and traceable input data for coal blending accuracy calculation.
[0088] Step S40: Based on the corrected actual coal feeding amount, calculate the actual coal feeding ratio per minute by coal type, and compare it with the planned ratio minute by minute to determine the ratio accuracy;
[0089] In step S40, the process of determining the mixing ratio accuracy is as follows:
[0090] From the sequence of actual coal discharge per minute, read the actual coal discharge value for each silo per minute. For each minute timestamp t, perform the following operations:
[0091] Sum the actual coal discharge amounts of all silos for that minute to obtain the total coal discharge amount Total_t = Σ(actual coal discharge amount of each silo). Then, group and summarize by coal type: add the actual coal discharge amounts of all silos corresponding to the same coal type for that minute to obtain the actual coal discharge amount of that coal type for that minute Amount_coal_t = Σ(actual coal discharge amount of each silo for that coal type); calculate the actual coal discharge ratio of that coal type for that minute Ratio_actual_coal_t = Amount_coal_t / Total_t;
[0092] Read the planned proportion of each coal type from the coal blending plan table. The coal blending plan table contains the following fields: coal blending order number, coal type name, planned proportion (percentage), start execution time, and end execution time.
[0093] For each coal type and each minute timestamp t, calculate the deviation value, i.e., deviation value = actual coal delivery ratio - planned ratio;
[0094] The absolute value of the deviation is compared with a preset deviation threshold (set to 5% in this embodiment). The judgment condition is: if |deviation value| > 5%, then the coal type is judged to be out of tolerance for that minute; otherwise, it is judged to be qualified. All coal types and all minutes are iterated through, and the number of minutes of deviation for each coal type is recorded.
[0095] The blending accuracy is measured in terms of the number of deviations, calculated as follows: Blending Accuracy (Number of Deviations) = Σ (Number of Minutes with Deviations for Each Coal Type). For example, for each coal type, count the total number of minutes during which the absolute value of the deviation exceeds 5% throughout the entire blending period. Then, add up the number of minutes with deviations for all coal types to obtain the final blending accuracy value. For instance, if coking coal has a deviation of 10 minutes, fat coal 5 minutes, one-third coke 3 minutes, and lean coking coal 2 minutes, then the blending accuracy is 10 + 5 + 3 + 2 = 20 times.
[0096] Understandably, the significance of step S40 lies in the fact that the corrected actual coal feeding sequence is minute-level data at the silo level, while the coal blending process focuses on the overall proportion of each coal type. Step S40, through the mapping relationship between silos and coal types, merges the coal feeding quantities of multiple silos by coal type, calculates the actual coal feeding ratio of each coal type per minute, and compares it with the planned ratio minute by minute. The number of minutes with deviations exceeding 5% is counted as the blending accuracy. The significance of this step is that it realizes the conversion from equipment-level data to process-level indicators, refines the time resolution of the coal blending process to the minute level, can accurately locate the specific time and coal type where the deviation occurs, and provides operators with quantitative basis for out-of-tolerance alarms and attribution analysis, thereby effectively guiding the real-time adjustment of the coal blending process.
[0097] Step S50: Based on the corrected actual coal feeding ratio, calculate the planned theoretical value and upper and lower limits of the quality of each component, compare them with the test results of the mixed coal, and generate the quality accuracy index of each component.
[0098] In step S50, the process of generating the quality accuracy index of each component is as follows:
[0099] Retrieves the quality index value used for each coal type during the current coal blending order execution from the single coal quality table. The single coal quality table contains fields: coal type name, mine name, test date, sulfur content, ash content, and volatile matter. For each coal type, the most recent test result before the start time of the coal blending order execution is taken as the quality index value, denoted as Quality_coal;
[0100] For each quality component, calculate the theoretical value of that component's quality plan, Theory_t. The calculation formula is: Theory_t = Σ(Quality_coal of that component for each coal type × actual coal feeding ratio for that coal type);
[0101] For each quality component, the upper and lower limits for that minute are calculated based on the theoretical value of the quality plan per minute, Theory_t. The formulas for calculating the upper and lower limits are: Upper limit = Theory_t × (1 + preset percentage), Lower limit = Theory_t × (1 - preset percentage). In this embodiment, the preset percentage is 10%, i.e., Upper limit = Theory_t × 1.1, Lower limit = Theory_t × 0.9;
[0102] Read all the test records of the blended coal during the execution of the current coal blending sheet from the blended coal quality table. The blended coal quality table contains the following fields: coal blending sheet number, test time, sulfur content, ash content, and volatile content. The test time is accurate to the minute. Each test corresponds to a blended coal sample, which represents the average quality of the blended coal during the period from the previous test to the current test. Use the test time as the timestamp of the test result, denoted as Test_t, and the corresponding test value as Actual_t (corresponding to three independent values for sulfur, ash content, and volatile content respectively);
[0103] For each test record, obtain the upper and lower limits of the quality plan (Upper_t and Lower_t) for the minute where the test timestamp is located. Determine whether the actual value Actual_t of this test satisfies: Lower_t ≤ Actual_t ≤ Upper_t. If it is satisfied, this test is judged as qualified; if not, it is judged as unqualified (for the three quality components: sulfur content, ash content, and volatile content, judge independently);
[0104] For each quality component, count the total number of all blended coal tests during the execution of the coal blending sheet, denoted as N_total, and count the number of qualified tests, denoted as N_pass. The calculation formula for the quality accuracy index is: Quality accuracy = N_pass / N_total; for example, if the total number of ash content tests is 50 times, and the number of qualified times is 47 times, then the ash content accuracy = 47 / 50 × 100% = 94%;
[0105] It can be understood that the significance of step S50 is as follows: The ultimate goal of coal blending is to ensure that the quality of the blended coal meets the requirements of coke production, and the quality indicators of single coal and the coal blending ratio jointly determine the theoretical quality of the blended coal. Based on the actual coal feeding ratio per minute calculated in step S40, combined with the sulfur content, ash content, and volatile content data of single coal, calculate the theoretical value of the quality plan per minute and its upper and lower limits of ±10%, and then compare with the actual test results of the blended coal, and count the qualified rate as the quality accuracy index. The significance of this step is to establish a computable mapping relationship from the quality of single coal to the quality of blended coal, convert the proportion deviation in the coal blending process into the fluctuation range of quality indicators, enabling operators to intuitively evaluate the potential impact of the current coal blending plan on the final coke quality, and achieving dual control of proportion accuracy and quality accuracy.
[0106] Example 2, please refer to Figure 2 As shown, a coking coal blending accuracy calculation system described in an embodiment of the present invention includes the following modules:
[0107] The abnormal data identification and marking module collects the cumulative amount and instantaneous rate of coal feeding in each silo in real time, and simultaneously monitors the fluctuation characteristics of the weighing signal. It identifies and marks false data segments caused by suspended or overflowing material, as well as segments of sudden rate changes at the moment of collapse.
[0108] The false segment estimation and compensation module suspends the use of the original cumulative amount for interpolation calculation of the marked false data segments. Based on the historical coal feeding trend of adjacent normal periods, it estimates the actual coal feeding per minute in the period. For the rate change segment at the moment of collapse, the original rate value is retained.
[0109] The time-series data fusion and correction module uses the original cumulative amount to interpolate the unlabeled normal data segments, and merges the estimated value of the false data segments, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segments in chronological order to form a corrected sequence of actual coal output per minute.
[0110] The proportioning accuracy comparison module calculates the actual coal feeding ratio per minute based on the corrected actual coal feeding amount, combines the coal types to calculate the actual coal feeding ratio per minute, and compares it with the planned ratio minute by minute to determine the proportioning accuracy.
[0111] The quality index comparison and generation module calculates the planned theoretical value and upper and lower limits of the quality of each component based on the corrected actual coal feeding ratio, compares it with the test results of the mixed coal, and generates the quality accuracy index of each component.
[0112] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for calculating the accuracy of coking coal blending, characterized in that: Includes the following steps: Step S10: Collect the cumulative amount of coal fed into each silo and the instantaneous rate in real time, monitor the fluctuation characteristics of the weighing signal simultaneously, identify and mark false data segments caused by suspended or overflowing material, as well as the rate change segments at the moment of collapse. The specific methods for identifying and labeling spoofed data segments are as follows: The data acquisition server reads the instantaneous rate records of the silo for the most recent 120 seconds every 60 seconds, calculates its arithmetic mean and standard deviation; reads the cumulative total value of the first and last moments within the 120 seconds, and calculates the actual increase in the cumulative amount; if the standard deviation of the instantaneous rate in the current 120-second period exceeds the fluctuation threshold, and the actual increase in the cumulative amount in the period is lower than the expected lower limit of the increase, then the period is determined to be a false data segment of suspended material. The process of identifying the rate abrupt change segment at the moment of collapse is as follows: After marking false data segments, for each time period that has been marked as a false data segment, continue to monitor the instantaneous rate data within 60 seconds after the end of the time period, and set the mutation detection window to 10 seconds, that is, continuously detect 10 instantaneous rate sampling points; When any five consecutive sampling points within the mutation detection window have instantaneous rates below the lower limit reference value, followed by five consecutive sampling points with instantaneous rates above the upper limit reference value, and the cumulative increment within 10 seconds exceeds twice the normal cumulative increment within 10 seconds, it is determined to be a collapse event; the time range covered by the collapse event is marked as the collapse rate mutation segment. Step S20: For the marked false data segments, pause the use of the original cumulative amount for interpolation calculation, estimate the actual coal amount per minute in the period based on the historical coal feeding trend of adjacent normal periods, and retain the original rate value for the rate change segment at the moment of collapse. The specific method for estimating the actual coal output per minute within the spurious data segment is as follows: Taking the start time of the false data segment as minute 1, read the average rate of the last 60 seconds of the preceding normal period and the average rate of the first 60 seconds of the following normal period, and take the arithmetic mean of the two as the estimated baseline rate; for minute t, the estimated coal output is equal to the estimated baseline rate multiplied by 1 minute plus the slope of the trend multiplied by t minus the difference of the middle minute number. Step S30: For the unlabeled normal data segment, interpolation calculation is performed using the original cumulative amount. The estimated value of the false data segment, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segment are merged in chronological order to form the corrected actual coal delivery sequence per minute. Step S40: Based on the corrected actual coal feeding amount, calculate the actual coal feeding ratio per minute by coal type, and compare it with the planned ratio minute by minute to determine the ratio accuracy; Step S50: Based on the corrected actual coal feeding ratio, calculate the planned theoretical value and upper and lower limits of the quality of each component, compare them with the test results of the mixed coal, and generate the quality accuracy index of each component.
2. The method for calculating the accuracy of coking coal blending according to claim 1, characterized in that: In step S20, the method for retaining the original rate value for the abrupt collapse rate segment is as follows: Read all second-level records marked as collapse mutation segments, without performing any interpolation calculations or estimation replacement operations, and directly output the original instantaneous rate value and the original cumulative value in the record to subsequent steps; the cumulative difference between adjacent second-level records within the collapse mutation segment is not smoothed, and the original mutation amplitude is maintained.
3. The method for calculating the accuracy of coking coal blending according to claim 1, characterized in that: In step S30, the interpolation calculation for the unlabeled normal data segment is performed as follows: Obtain the original cumulative values at the start and end times of the normal data segment, calculate the difference between the two as the total increment of the cumulative value, and then divide it by the total duration of the data segment to obtain the interpolated coal feeding amount per minute. The interpolated coal quantity is assigned to all 60 second-level records within that minute, and the cumulative quantity is generated second by second in an equal increment manner.
4. The method for calculating the accuracy of coking coal blending according to claim 1, characterized in that: In step S30, the method for merging to form the corrected sequence of actual coal output per minute is as follows: Establish a continuous minute time axis covering the entire coal blending execution period. According to the priority rule that the slump retention value takes precedence over the estimated value, and the estimated value takes precedence over the interpolated normal value, the corresponding coal feeding value is sequentially taken from the temporary table of slump retention value, the temporary table of estimated value, and the temporary table of interpolated normal value and filled into the minute point to form the actual coal feeding sequence per minute. Among them, the interpolation normal value temporary table is composed of the interpolated coal amount per minute of the normal data segment, the estimated value temporary table is composed of the estimated value per minute corresponding to the false data segment, and the collapse retention value temporary table is composed of the original instantaneous rate value per second corresponding to the collapse rate mutation segment, which is aggregated and stored by minute.
5. The method for calculating the accuracy of coking coal blending according to claim 1, characterized in that: In step S40, the method for determining the mixing ratio accuracy is as follows: The total coal discharge is obtained by summing the actual coal discharge amounts of all silos every minute. The actual coal discharge amount of each coal type is obtained by adding the coal discharge amounts of the corresponding silos. The ratio of the total coal discharge amount to the actual coal discharge amount of each coal type is calculated to obtain the actual coal discharge ratio of each coal type per minute. Calculate the absolute difference between the actual coal ratio and the planned ratio. If it is greater than 5%, it is counted as one deviation. Add up the number of deviations for all coal types and for all minutes to obtain the ratio accuracy.
6. The method for calculating the accuracy of coking coal blending according to claim 1, characterized in that: In step S50, the method for generating the quality accuracy index is as follows: Calculate the product of each coal type's quality index and the actual coal feeding ratio per minute, and sum them to obtain the theoretical value of the quality plan per minute. Then multiply by the preset first coefficient 1.1 and second coefficient 0.9 to obtain the upper and lower limits. Compare the actual values of the mixed coal test with the corresponding upper and lower limits per minute, and count the proportion of qualified tests to the total number of tests to obtain the quality accuracy index of each component.
7. A coking coal blending accuracy calculation system, characterized in that: Includes the following modules: The abnormal data identification and marking module collects the cumulative amount and instantaneous rate of coal feeding in each silo in real time, and simultaneously monitors the fluctuation characteristics of the weighing signal. It identifies and marks false data segments caused by suspended or overflowing material, as well as segments of sudden rate changes at the moment of collapse. The specific methods for identifying and labeling spoofed data segments are as follows: The data acquisition server reads the instantaneous rate records of the silo for the most recent 120 seconds every 60 seconds, calculates its arithmetic mean and standard deviation; reads the cumulative total value of the first and last moments within the 120 seconds, and calculates the actual increase in the cumulative amount; if the standard deviation of the instantaneous rate in the current 120-second period exceeds the fluctuation threshold, and the actual increase in the cumulative amount in the period is lower than the expected lower limit of the increase, then the period is determined to be a false data segment of suspended material. The process of identifying the rate abrupt change segment at the moment of collapse is as follows: After marking false data segments, for each time period that has been marked as a false data segment, continue to monitor the instantaneous rate data within 60 seconds after the end of the time period, and set the mutation detection window to 10 seconds, that is, continuously detect 10 instantaneous rate sampling points; When any five consecutive sampling points within the mutation detection window have instantaneous rates below the lower limit reference value, followed by five consecutive sampling points with instantaneous rates above the upper limit reference value, and the cumulative increment within 10 seconds exceeds twice the normal cumulative increment within 10 seconds, it is determined to be a collapse event; the time range covered by the collapse event is marked as the collapse rate mutation segment. The false segment estimation and compensation module suspends the use of the original cumulative amount for interpolation calculation of the marked false data segments. Based on the historical coal feeding trend of adjacent normal periods, it estimates the actual coal feeding per minute in the period. For the rate change segment at the moment of collapse, the original rate value is retained. The specific method for estimating the actual coal output per minute within the spurious data segment is as follows: Taking the start time of the false data segment as minute 1, read the average rate of the last 60 seconds of the preceding normal period and the average rate of the first 60 seconds of the following normal period, and take the arithmetic mean of the two as the estimated baseline rate; for minute t, the estimated coal output is equal to the estimated baseline rate multiplied by 1 minute plus the slope of the trend multiplied by t minus the difference of the middle minute number. The time-series data fusion and correction module uses the original cumulative amount to interpolate the unlabeled normal data segments, and merges the estimated value of the false data segments, the original value of the instantaneous rate change segment of collapse, and the interpolation result of the normal data segments in chronological order to form a corrected sequence of actual coal output per minute. The proportioning accuracy comparison module calculates the actual coal feeding ratio per minute based on the corrected actual coal feeding amount, combines the coal types to calculate the actual coal feeding ratio per minute, and compares it with the planned ratio minute by minute to determine the proportioning accuracy. The quality index comparison and generation module calculates the planned theoretical value and upper and lower limits of the quality of each component based on the corrected actual coal feeding ratio, compares it with the test results of the mixed coal, and generates the quality accuracy index of each component.