Method for monitoring abnormal raw material batching into kiln based on multi-source heterogeneous data

Abstract

The present application belongs to the technical field of abnormal monitoring of raw material batching into kiln, and discloses an abnormal monitoring method of raw material batching into kiln based on multi-source heterogeneous data. The present application dynamically sorts the abnormal traceability priority list based on the frequency of historical abnormal events, thereby avoiding the blindness in the traditional random investigation mode. In combination with the mapping relationship between the deviation degree and the abnormal confidence, the core reason is quickly locked, the abnormal response cycle is significantly shortened, and the risk of raw material quality fluctuation caused by continuous abnormality is reduced. The present application monitors the abnormal reason change trend in real time through the correlation confidence. When the measurement error proportion rises due to equipment aging, the priority list is automatically adjusted, so that the traceability order is synchronized with the actual production state. The present application solves the problem of insufficient adaptability of the traditional fixed process to the equipment life cycle and the raw material market fluctuation, and ensures that the high-efficiency traceability capability can be maintained when the production conditions change.

Description

Technical Field

[0001] This invention belongs to the field of monitoring technology for abnormal raw material batching in kilns, and relates to a method for monitoring abnormal raw material batching in kilns based on multi-source heterogeneous data. Background Technology

[0002] In the cement production process, the quality of the raw meal entering the kiln directly affects the calcination effect of cement clinker and the final product quality. The raw meal ratio, including the lime saturation coefficient, silica ratio, and alumina ratio, is a key indicator for measuring the rationality of the raw meal batching, and its stability is crucial for the smooth operation of the production process and the consistency of product quality.

[0003] However, the raw meal batching process involves the mixing of multiple raw materials and the coordinated operation of various equipment. Fluctuations in raw material composition, errors in metering equipment, and malfunctions in transmission equipment can all lead to abnormal raw meal yield values. If these abnormalities are not detected in a timely manner and their causes are not accurately traced, it can result in decreased production efficiency, energy waste, and even substandard product quality. Therefore, achieving accurate monitoring and efficient traceability of abnormal raw meal batching at the kiln has become a pressing technical problem to be solved in the cement production industry.

[0004] Specifically, traditional technical solutions still have the following problems: 1. The investigation of abnormal causes lacks a clear priority order, and often adopts random or fixed order detection, which may waste a lot of time on low probability causes and delay the time to solve the problem.

[0005] 2. Once the priority list for anomaly tracing is set, it remains unchanged for a long time and cannot be dynamically adjusted according to the occurrence pattern of anomalies in actual production, making it difficult to adapt to changes in production conditions.

[0006] 3. The lack of statistical analysis on the association confidence of various abnormal causes makes it impossible to reveal the intrinsic connection between different causes and abnormal events, thus affecting the accuracy of source tracing. Summary of the Invention

[0007] In view of this, in order to solve the problems mentioned in the background technology, a method for monitoring abnormal raw material batching in the kiln based on multi-source heterogeneous data is proposed.

[0008] The objective of this invention can be achieved through the following technical solution: a method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data, comprising: collecting raw meal rate data, including lime saturation coefficient, silicon content and aluminum content, using an X-ray fluorescence analyzer, and comparing them with preset raw meal rate related thresholds to determine whether there are any abnormal raw meal rate values.

[0009] When the raw material yield value is determined to be abnormal, the cause of the abnormality is traced according to the order in the pre-built abnormality tracing priority list, including raw material composition detection, metering equipment detection and transmission equipment detection, and the result of abnormality tracing is saved to the database.

[0010] Based on the analysis of the anomaly cause tracing results stored in the database, the correlation confidence of each anomaly cause is determined to determine whether the anomaly tracing priority list needs to be adjusted.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention avoids the blindness of the traditional random investigation mode by dynamically sorting the anomaly tracing priority list based on the frequency of historical anomalies. By combining the mapping relationship between deviation degree and anomaly confidence degree, the core cause can be quickly identified, the anomaly response cycle can be significantly shortened, and the risk of raw material quality fluctuation caused by continuous anomalies can be reduced.

[0012] (2) This invention monitors the changing trends of abnormal causes in real time by associating confidence levels. When the proportion of measurement errors increases due to equipment aging, the priority list is automatically adjusted to synchronize the traceability order with the actual production status. This solves the problem of insufficient adaptability of traditional fixed processes to equipment life cycle and raw material market fluctuations, ensuring that efficient traceability capabilities are maintained even when production conditions change.

[0013] (3) This invention integrates multi-source heterogeneous information such as X-ray fluorescence analysis data, raw material testing data, and equipment operating parameters to construct a full-link data association system. By mining hidden correlations of anomalies through historical data, the depth of source tracing is improved. The accumulated source tracing results in the database can feed back into threshold optimization, continuously improving the system's intelligence level. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram illustrating the implementation steps of the method of the present invention.

[0016] Figure 2 A flowchart for determining and handling abnormal raw material yield values ​​corresponding to one embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1As shown, the present invention provides a method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data, including: collecting raw meal rate data using an X-ray fluorescence analyzer, including lime saturation coefficient, silicon content and aluminum content, and comparing them with preset raw meal rate related thresholds to determine whether there is an abnormality in the raw meal rate.

[0019] In a preferred embodiment of the present invention, the specific setting process of the raw material rate value related threshold is as follows: the raw material rate value related threshold includes the lime saturation coefficient deviation threshold, the silicon ratio deviation threshold, and the aluminum ratio deviation threshold.

[0020] Collect raw material yield data during historical production processes under normal operating conditions, including continuous records of lime saturation coefficient, silicon content, and aluminum content, ensuring that the data covers different production shifts, raw material batches, and equipment operating cycles.

[0021] It should be noted that the raw material yield data collected during normal production operations must cover continuous records of lime saturation coefficient, silicon content, and aluminum content. The data must also include different production shifts, raw material batches, and equipment operating cycles to ensure the comprehensiveness and representativeness of the data and avoid deviations in threshold settings due to limited data.

[0022] Calculate the statistical center value of the lime saturation coefficient data under normal production conditions. Based on the statistical center value of the lime saturation coefficient, calculate the relative deviation of all valid data from the center value, statistically analyze the distribution of the relative deviation, and determine the maximum value of the difference under normal fluctuation conditions as the lime saturation coefficient deviation threshold.

[0023] Based on the method used to process lime saturation coefficient data, the silicon ratio deviation threshold and aluminum ratio deviation threshold were calculated from the historical data of silicon ratio and aluminum ratio respectively.

[0024] It should be noted that the raw material yield threshold is the benchmark for determining whether the raw material yield value is abnormal, and its setting is based on actual production needs. During production, the raw material yield value will fluctuate normally due to factors such as shift, raw material batch, and equipment cycle time; a reasonable fluctuation range needs to be clearly defined. By collecting historical normal data and calculating the statistical center value and relative deviation distribution of each rate value, using the maximum normal fluctuation as the threshold, we can accurately distinguish between normal fluctuations and anomalies, avoiding the subjectivity of experience-based thresholds. This provides an objective basis for subsequent anomaly judgment and tracing, ensuring the scientific nature and reliability of the monitoring method.

[0025] In a preferred embodiment of the present invention, the specific method for determining whether there is an abnormality in the raw material rate value is as follows: the relative deviation of the lime saturation coefficient, silicon ratio and aluminum ratio is compared with the preset reference lime saturation coefficient, reference silicon ratio and reference aluminum ratio to obtain the lime saturation coefficient deviation, silicon ratio deviation and aluminum ratio deviation.

[0026] The deviations of the lime saturation coefficient, silicon ratio, and aluminum ratio are compared with the corresponding deviation thresholds of the lime saturation coefficient, silicon ratio, and aluminum ratio, respectively.

[0027] If any of the parameters—lime saturation coefficient deviation, silicon ratio deviation, and aluminum ratio deviation—exceeds the corresponding threshold or the sum of all parameters exceeds the preset cumulative deviation threshold, it is determined that there is an abnormality in the raw material rate value; otherwise, it is determined that there is no abnormality in the raw material rate value.

[0028] It should be noted that if any one of the deviations in the lime saturation coefficient, silica ratio, or aluminum ratio exceeds its corresponding threshold, it is directly considered an anomaly. This is because a significant deviation in any one rate value may individually affect the raw meal quality and should be prioritized for identification. If none of the three deviations exceed their respective thresholds, but the sum exceeds the preset cumulative deviation threshold, it is still considered an anomaly. This takes into account the cumulative effect of small deviations in multiple parameters. A single parameter deviation may be within the normal range, but when multiple parameters deviate from the baseline value simultaneously, their combined effect may lead to an overall anomaly in the raw meal batching.

[0029] It should be noted that the lime saturation coefficient, silica ratio, and alumina ratio were chosen as the criteria for anomaly detection because these three are core indicators of raw meal quality. The lime saturation coefficient reflects the ratio of calcium oxide to other components, affecting clinker mineral formation; the silica ratio relates to clinker burnability and strength; and the alumina ratio is associated with setting time and strength development. These three factors collectively determine the rationality of raw meal proportioning, and any anomaly or synergistic deviation will affect production. Using these indicators allows for comprehensive detection of raw meal quality fluctuations, providing precise and crucial indicators for anomaly detection, ensuring the targeted and effective nature of monitoring.

[0030] When the raw material yield value is determined to be abnormal, the cause of the abnormality is traced according to the order in the pre-built abnormality tracing priority list, including raw material composition detection, metering equipment detection and transmission equipment detection, and the result of abnormality tracing is saved to the database.

[0031] It's important to note that testing the raw material composition, metering equipment, and transmission equipment is crucial because these three elements significantly impact the raw material yield. Raw material composition directly determines the basic structure of the raw meal, and fluctuations in this composition can lead to abnormal yield values. Metering equipment is responsible for precise control of raw material proportions; errors can alter the actual proportions. The operational status of transmission equipment affects the stability of raw material transport; malfunctions can cause uneven mixing. These three elements are linked to the raw material yield from the perspectives of the raw materials themselves, proportioning accuracy, and transport stability, respectively. Testing them comprehensively covers key aspects of raw material batching, accurately pinpoints the root cause of anomalies, provides a basis for subsequent processing, and ensures stable production.

[0032] In a preferred embodiment of the present invention, the specific construction process of the anomaly tracing priority list is as follows: retrieve the recorded raw material yield value anomaly event dataset from the production history database, extract the final confirmed tracing cause of each anomaly event, count the frequency of occurrence of various tracing causes in historical events, and establish a correspondence table between anomalies and tracing causes, which includes anomaly event number, occurrence time, and specific tracing cause.

[0033] Based on the frequency statistics of various causes in the historical abnormal event dataset, the three types of factors—raw material composition mutation, metering equipment error, and transmission equipment failure—are initially sorted in descending order of frequency to generate an initial abnormality tracing priority list, with the most frequent tracing factor listed first.

[0034] It should be noted that this invention avoids the blindness of traditional random investigation methods by dynamically sorting the anomaly tracing priority list based on the frequency of historical anomalies. By combining the mapping relationship between deviation degree and anomaly confidence degree, the core cause can be quickly identified, significantly shortening the anomaly response cycle and reducing the risk of raw material quality fluctuations caused by continuous anomalies.

[0035] In a preferred embodiment of the present invention, the specific analysis method for tracing the cause of the anomaly is as follows: extracting the detection results of raw material component detection, metering equipment detection, and transmission equipment detection.

[0036] The deviation degree of each detection item corresponding to the detection that is determined to be abnormal is compared with the detection item deviation degree-raw material rate value abnormal event occurrence confidence relationship pre-constructed based on historical abnormal cases, and the raw material rate value abnormal event occurrence confidence degree corresponding to each detection that is determined to be abnormal is obtained.

[0037] In a preferred embodiment, a specific process for constructing the mapping relationship between the deviation of detection items and the confidence level of abnormal events in raw material yield is provided: collecting records of abnormal events in raw material batching over a period of time. These records need to include the deviation of each detection item when the abnormality occurred, whether the corresponding raw material yield value is abnormal, and related basic production information. Then, cases with missing data or interference from special external factors are removed, and valid case data is retained.

[0038] From the valid case data, we screen out the test items that are directly related to the raw material yield value, determine the range of test items to be analyzed, and then divide the deviation of each test item into different intervals according to a certain numerical interval to ensure that each interval has a sufficient number of cases to support the analysis.

[0039] For each deviation range of each test item, the proportion of cases where the raw material yield value is abnormal within that range to the total number of cases in that range is calculated. This proportion is the confidence level of the occurrence of abnormal raw material yield value events for that test item in the corresponding deviation range.

[0040] Each test item, its defined deviation range, and corresponding confidence level are organized into a structured mapping table. New abnormal case data are periodically included, the confidence level of each range is recalculated, and the mapping table is updated and improved to adapt to changes in production conditions.

[0041] The confidence levels of the abnormal raw material yield values ​​corresponding to each test that is determined to be abnormal are compared and arranged in descending order. An abnormal cause tracing list is output, and the confidence levels of the abnormal raw material yield values ​​are marked.

[0042] It should be noted that this mapping relationship is constructed to quantify the correlation between the deviation of a detection item and the probability of an anomaly in the raw material yield value. When an abnormal deviation is detected, the corresponding confidence level can be directly obtained through this mapping relationship, thus clarifying the probability that each abnormal detection item causes an anomaly in the raw material yield value. This provides an objective basis for tracing the cause of anomalies, enabling the determination of the most likely leading cause from multiple abnormal detection items based on their confidence levels. This avoids the subjectivity of relying solely on experience, ensuring a more accurate and efficient tracing process and supporting subsequent targeted handling.

[0043] In a preferred embodiment of the present invention, the specific analytical method for detecting the raw material composition is as follows: several samples of various raw materials required for the raw material batching into the kiln are extracted from the raw material storage area according to a prescribed method.

[0044] The content of active ingredients in various raw material samples is detected using testing equipment.

[0045] The deviation of the effective components in various raw material samples is obtained by comparing the content of the effective components in the corresponding standard components.

[0046] The deviation of the effective components of various raw materials is compared with a pre-set threshold. If the deviation of the effective components of any raw material is greater than the corresponding threshold, it is determined that the raw material components are abnormal; otherwise, it is determined that the raw material components are not abnormal.

[0047] It should be noted that different raw materials have different properties and production requirements, and therefore require their own corresponding effective component deviation thresholds. During testing, the relative deviation of the effective component of the actual raw material sample from the standard value is calculated and then compared with the specific threshold for that raw material. If the deviation of any raw material exceeds its threshold, it indicates that the fluctuation of the raw material composition has exceeded the normal range, which may directly lead to an abnormal raw material yield value, thus determining that the raw material composition is abnormal. Conversely, if the deviation of all raw materials is within the corresponding threshold, it indicates that the raw material composition is within the normal fluctuation range, and the influence of raw material factors on the abnormal raw material yield value can be ruled out.

[0048] Specifically, common raw materials used in cement production include limestone, clay, and iron powder. Taking limestone as an example, its main active ingredient is calcium oxide. Under normal circumstances, the standard calcium oxide content of high-quality limestone is about 50%-55%. If a limestone sample is tested and found to have a calcium oxide content of only 45%, the deviation is calculated to be 10%. If the pre-set deviation threshold for limestone calcium oxide content is 8%, then the composition of this limestone raw material is abnormal.

[0049] In a preferred embodiment of the present invention, the specific analysis method for the metering equipment detection is as follows: retrieve the operating data records of each metering equipment within the time period corresponding to the abnormal raw material rate value, continuously collect a number of set flow values ​​and actual feedback flow values ​​corresponding to a large number of moments, and construct set flow time series data and actual feedback flow time series data.

[0050] The set flow time series data and the actual feedback flow time series data are compared to calculate the instantaneous relative error at each moment, and the average instantaneous relative error and maximum relative error of each metering device are statistically calculated.

[0051] The average instantaneous relative error and the maximum relative error are compared with a preset threshold. If any parameter of any metering device is greater than the threshold, the metering device is determined to be abnormal; otherwise, the metering device is determined not to be abnormal.

[0052] It should be noted that the average instantaneous relative error reflects the overall stability of the metering equipment during abnormal periods. If it exceeds the preset threshold, it indicates that the equipment is in an unstable state for a long time, which may cause the raw material ratio to deviate continuously from the set value. The maximum relative error reflects the degree of extreme deviation of the equipment. If it exceeds the standard, it indicates that the equipment is at risk of sudden and large deviation, which may instantly disrupt the balance of the raw material ratio.

[0053] The preset thresholds are based on the equipment's factory accuracy standards and historical normal operation data. When either the average instantaneous relative error or the maximum relative error of any metering device exceeds the corresponding threshold, the device is considered abnormal. This covers both the latent problem of persistent equipment deviation and the explicit fault of sudden, large deviations. Conversely, if both parameters are within the thresholds, it indicates that the equipment's metering accuracy meets the requirements, and its influence on abnormal raw material yield values ​​can be ruled out.

[0054] In a preferred embodiment of the present invention, the specific analysis method for detecting the transmission equipment is as follows: using equipment such as a speed measuring instrument, vibration sensor, and tension meter, the conveying speed, vibration amplitude of the transmission components, and belt tension and other operating parameters of the transmission equipment operating section are detected.

[0055] It should be noted that the tachometer is used to measure the actual conveying speed of the transmission equipment in real time. This parameter is directly related to the stability of the raw material conveying volume. An abnormality will cause the raw material conveying volume per unit time to deviate from the set value, thus affecting the raw material ratio. Vibration sensors are installed on the transmission components of the transmission equipment to monitor their vibration amplitude. During normal operation, the vibration amplitude is within a stable range. If the components are worn, loose, or unbalanced, the vibration amplitude will increase significantly. Such abnormalities may lead to unstable equipment operation and indirectly cause fluctuations in raw material conveying. Tension gauges are mainly used for belt conveyor equipment to detect belt tension. Too low tension will cause the belt to slip, while too high tension may aggravate equipment wear or even breakage, both of which will disrupt the continuity of conveying. Tension gauges can quantify the tension and determine whether it is within the standard range.

[0056] The detected operating parameters are compared with the standard operating parameters of the corresponding transmission equipment, and the relative deviation of each parameter is calculated.

[0057] The relative deviation of each operating parameter is compared with a preset threshold. If the deviation of any parameter is greater than the corresponding threshold, the transmission device is determined to be abnormal; otherwise, the transmission device is determined to be normal.

[0058] It should be noted that when the relative deviation of any operating parameter exceeds the corresponding threshold, it indicates that the parameter has exceeded the normal fluctuation range and may directly affect transmission stability, thus indicating an anomaly in the transmission equipment. Conversely, if the deviations of all parameters are within the thresholds, it indicates that the equipment is operating stably, and its influence on abnormal raw material yield values ​​can be ruled out. This logic of determining anomalies based on a single parameter exceeding the standard can comprehensively capture potential equipment problems, avoid missing hidden faults caused by the superposition of multiple parameter slight deviations, and provide a clear basis for tracing the source of transmission equipment anomalies.

[0059] Based on the analysis of the anomaly cause tracing results stored in the database, the correlation confidence of each anomaly cause is determined to determine whether the anomaly tracing priority list needs to be adjusted.

[0060] In a preferred embodiment of the present invention, the specific analysis method for analyzing the correlation confidence of each abnormal cause is as follows: retrieve the recorded raw material yield value abnormal event dataset from the production history database, and count the frequency of occurrence of the abnormal cause corresponding to each raw material yield value abnormal event within the preset monitoring period.

[0061] The correlation confidence of each abnormal cause is calculated by comparing the frequency of occurrence of each abnormal cause with the frequency of occurrence of abnormal events in the raw material yield value.

[0062] It should be noted that analyzing the association confidence of each anomaly cause aims to determine the comprehensive impact of multiple anomaly causes on the abnormal raw material yield value. First, records with two or more coexisting anomaly causes are extracted from historical anomaly cases to clarify the combination forms of these coexisting anomaly causes. Second, for each combination of anomaly causes, the actual number of occurrences of the abnormal raw material yield value event when that combination appears is counted, and compared with the total number of occurrences of that combination in historical cases to obtain the association confidence of that combination. Finally, multiple anomaly causes identified in real-time detection are combined and matched, and the association confidence of the corresponding combinations in historical cases is used as the basis for judging their synergistic impact.

[0063] In a preferred embodiment of the present invention, the specific analysis method for determining whether the priority list for anomaly tracing needs to be adjusted is as follows: the causes of each anomaly are arranged in descending order of their correlation confidence to obtain the actual anomaly tracing list within the monitoring period.

[0064] The anomaly tracing priority list is matched with the actual anomaly tracing priority list during the monitoring period to determine whether the anomaly tracing priority needs to be adjusted.

[0065] If any of the following conditions exist, it is determined that the priority of anomaly tracing needs to be adjusted; otherwise, it is determined that the priority of anomaly tracing does not need to be adjusted: Condition 1: There is a difference in the order of the anomaly causes between the actual anomaly tracing list and the anomaly tracing priority list.

[0066] Condition 2: The difference between the actual association confidence level and the historical association confidence level for any abnormal cause exceeds the preset range.

[0067] It should be noted that if the actual frequency of a certain type of abnormal cause in the recent period deviates from the ranking criteria of that type of cause in the current priority list by more than a preset proportion, it indicates that its probability of occurrence has significantly deviated from the historical pattern, and the priority needs to be adjusted to match the current situation; if the frequency of a certain type of abnormal cause shows a significant upward or downward trend in multiple consecutive statistical periods, it indicates that its occurrence pattern has undergone a continuous change, the original priority ranking is no longer applicable, and it is necessary to follow up on the new trend through regulation.

[0068] It should be noted that this invention monitors the changing trends of anomaly causes in real time through correlation confidence levels. When the proportion of measurement errors increases due to equipment aging, the priority list is automatically adjusted to synchronize the traceability order with the actual production status. This solves the problem of insufficient adaptability of traditional fixed processes to equipment lifecycles and raw material market fluctuations, ensuring that efficient traceability capabilities are maintained even when production conditions change.

[0069] It should be noted that this invention integrates multi-source heterogeneous information such as X-ray fluorescence analysis data, raw material testing data, and equipment operating parameters to construct a full-chain data association system. Hidden correlations in the causes of anomalies are mined from historical data, improving the depth of source tracing. The accumulated source tracing results in the database can feed back into threshold optimization, continuously improving the system's intelligence level.

[0070] Please see Figure 2 As shown, a process for determining and handling anomalies in raw material yield values ​​is provided. First, an anomaly determination step is triggered. The system relies on multi-source heterogeneous data and employs a preset algorithm to accurately determine whether an anomaly exists in the raw material yield value. If the determination result is negative, meaning the raw material yield value is in a normal and stable state, the process ends directly without additional intervention. If the determination result is positive, the process enters the anomaly cause tracing stage. Using data correlation analysis and historical case matching, potential factors causing the anomaly are identified from multiple dimensions, including raw material composition fluctuations, metering equipment deviations, and production process disturbances. After tracing the cause, an anomaly tracing priority list adjustment analysis is initiated. Based on indicators such as the impact of the anomaly cause, its frequency of occurrence, and handling costs, the tracing priority is evaluated and adjusted, providing clear guidance for subsequent efficient anomaly handling and optimized production control. The process ends after all steps are completed.

[0071] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data, characterized in that, include: X-ray fluorescence analyzer was used to collect raw material yield data, including lime saturation coefficient, silicon content and aluminum content, and compared with preset raw material yield thresholds to determine whether there are any abnormal raw material yield values. When the raw material yield value is determined to be abnormal, the cause of the abnormality is traced according to the order in the pre-built abnormality tracing priority list, including raw material composition detection, metering equipment detection and transmission equipment detection, and the abnormality cause tracing results are saved to the database. Based on the analysis of the anomaly cause tracing results stored in the database, the correlation confidence of each anomaly cause is determined to determine whether the anomaly tracing priority list needs to be adjusted. The specific construction process of the anomaly tracing priority list is as follows: retrieve the recorded raw material yield value anomaly event dataset from the production history database, extract the final confirmed tracing cause of each anomaly event, count the frequency of occurrence of various tracing causes in historical events, and establish a correspondence table between anomalies and tracing causes. This table includes the anomaly event number, occurrence time, and specific tracing cause. Based on the frequency statistics of various tracing causes in the historical anomaly event dataset, initially sort the three types of factors—raw material composition mutation, metering equipment error, and transmission equipment failure—in order of frequency from high to low, and generate an initial anomaly tracing priority list, in which the tracing factor with the highest frequency is placed at the top of the list. The specific analysis method for tracing the cause of the anomaly is as follows: extract the detection results of raw material component detection, metering equipment detection and transmission equipment detection; compare the deviation degree of each detection item corresponding to the anomaly with the confidence level of the anomaly event of the raw material rate value based on the pre-constructed mapping relationship of the deviation degree of the detection item and the anomaly event of the raw material rate value based on historical anomaly cases, and match to obtain the confidence level of the anomaly event of the raw material rate value based on the anomaly. The confidence levels of the abnormal raw material yield values ​​corresponding to each test that is determined to be abnormal are compared and arranged in descending order. An abnormal cause tracing list is output, and the confidence levels of the abnormal raw material yield values ​​are marked.

2. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 1, characterized in that: The specific process for setting the relevant threshold for the raw material yield value is as follows: The relevant thresholds for raw material yield include the lime saturation coefficient deviation threshold, the silicon content deviation threshold, and the aluminum content deviation threshold. Collect raw material yield data during normal operation in historical production processes, covering continuous records of lime saturation coefficient, silicon content, and aluminum content, ensuring that the data covers different production shifts, raw material batches, and equipment operating cycles; Calculate the statistical center value of the lime saturation coefficient data under normal production conditions. Based on the statistical center value of the lime saturation coefficient, calculate the relative deviation of all valid data from the center value. Statistically analyze the distribution of the relative deviation and determine the maximum value of the difference under normal fluctuation conditions as the lime saturation coefficient deviation threshold. Based on the method used to process lime saturation coefficient data, the silicon ratio deviation threshold and aluminum ratio deviation threshold were calculated from the historical data of silicon ratio and aluminum ratio respectively.

3. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 2, characterized in that: The specific method for determining whether there is an abnormal raw material yield value is as follows: The relative deviations of the lime saturation coefficient, silicon ratio, and aluminum ratio are compared with the preset reference lime saturation coefficient, reference silicon ratio, and reference aluminum ratio to obtain the lime saturation coefficient deviation, silicon ratio deviation, and aluminum ratio deviation, respectively. The deviation of lime saturation coefficient, silicon ratio, and aluminum ratio are compared with the corresponding deviation thresholds of lime saturation coefficient, silicon ratio, and aluminum ratio, respectively. If any of the parameters—lime saturation coefficient deviation, silicon ratio deviation, and aluminum ratio deviation—exceeds the corresponding threshold or the sum of all parameters exceeds the preset cumulative deviation threshold, it is determined that there is an abnormality in the raw material rate value; otherwise, it is determined that there is no abnormality in the raw material rate value.

4. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 1, characterized in that: The specific analytical methods for detecting the raw material components are as follows: Samples of various raw materials required for kiln raw material batching were extracted from the raw material storage area according to the prescribed method. The content of active ingredients in various raw material samples is detected using testing equipment; The deviation of the effective components in various raw material samples is obtained by comparing the content of the corresponding standard components with the content of the effective components in the raw materials. The deviation of the effective components of various raw materials is compared with a pre-set threshold. If the deviation of the effective components of any raw material is greater than the corresponding threshold, it is determined that the raw material components are abnormal; otherwise, it is determined that the raw material components are not abnormal.

5. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 4, characterized in that: The specific analysis method for the testing by the metering equipment is as follows: Retrieve the operating data records of each metering device during the time period corresponding to the abnormal raw material rate value, continuously collect a number of set flow values ​​and actual feedback flow values ​​corresponding to a large number of moments, and construct set flow time series data and actual feedback flow time series data. The set flow time series data and the actual feedback flow time series data are compared to calculate the instantaneous relative error at each moment, and the average instantaneous relative error and maximum relative error of each metering device are statistically calculated. The average instantaneous relative error and the maximum relative error are compared with a preset threshold. If any parameter of any metering device is greater than the threshold, the metering device is determined to be abnormal; otherwise, the metering device is determined not to be abnormal.

6. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 5, characterized in that: The specific analysis method for detecting the transmission equipment is as follows: Using a speedometer, vibration sensor, and tension meter, the conveying speed, vibration amplitude of transmission components, and belt tension of the operating section of the transmission equipment are detected. The detected operating parameters are compared with the standard operating parameters of the corresponding transmission equipment, and the relative deviation of each parameter is calculated. The relative deviation of each operating parameter is compared with a preset threshold. If the deviation of any parameter is greater than the corresponding threshold, the transmission device is determined to be abnormal; otherwise, the transmission device is determined to be normal.

7. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 1, characterized in that: The specific analysis method for the association confidence of each abnormal cause is as follows: Retrieve the recorded raw material yield value abnormal event dataset from the production history database, and count the frequency of occurrence of the abnormal causes corresponding to each raw material yield value abnormal event within the preset monitoring period; The correlation confidence of each abnormal cause is calculated by comparing the frequency of occurrence of each abnormal cause with the frequency of occurrence of abnormal events in the raw material yield value.

8. The method for monitoring abnormal raw meal batching in kiln based on multi-source heterogeneous data as described in claim 7, characterized in that: The specific analysis method for determining whether the anomaly tracing priority list needs to be adjusted is as follows: The actual anomaly source list for the monitoring period is obtained by arranging the causes of each anomaly in descending order of their correlation confidence. The anomaly tracing priority list is matched with the actual anomaly tracing priority list during the monitoring period to determine whether the anomaly tracing priority needs to be adjusted. If any of the following conditions exist, it is determined that the priority of anomaly tracing needs to be adjusted; otherwise, it is determined that the priority of anomaly tracing does not need to be adjusted: Condition 1: There is a difference in the order of the causes of each anomaly between the actual anomaly source list and the anomaly source priority list; Condition 2: The difference between the actual association confidence level and the historical association confidence level for any abnormal cause exceeds the preset range.

Images