An intelligent control system for conveyor belts based on coal washing

Abstract

The present invention relates to the technical field of data processing, and specifically to an intelligent control system for a conveyor belt based on coal washing and beneficiation, including: a data acquisition module, a state segmentation module, an operation monitoring module, an associated load module, and a fault processing module; by collecting the operation feature vectors and coal feature vectors of the conveyor belt operation, the conveyor belt is divided into multiple feature units, and multiple state recognition regions are set according to the feature thresholds of the coal feature vectors and operation feature vectors under each feature unit; data monitoring is performed on the multiple state recognition regions to generate multiple feature curves, and the association model between the coal feature vectors and operation feature vectors is identified; the fault types of the associated events under the degree of association are verified, the risk weights corresponding to the fault types are determined, and the processing solutions existing in the database are checked to obtain the fault processing solution; the efficiency and stability during coal transmission are achieved, and the risks and costs in the production process are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and specifically to an intelligent control system for a conveyor belt based on coal washing and beneficiation. Background Art

[0002] During the coal washing and beneficiation process, as one of the key equipment, the operating state of the conveyor belt directly affects the efficiency of the entire production line and the product quality. However, the traditional coal washing and beneficiation process often relies on manual monitoring and manual adjustment, which has problems such as low efficiency and easy errors. At the same time, due to the complexity and uncertainty in the coal washing and beneficiation process, fault warning and diagnosis also face great challenges. Therefore, it is of great significance to develop a control system for the coal washing and beneficiation conveyor belt based on intelligent technology.

[0003] As disclosed in Chinese Patent Publication No. CN115196275A, a speed adjustment method, device, equipment, storage medium and its multi-layer conveyor line are provided, including: obtaining the operating states of each unit; determining a target conveyor belt when it is detected that the operating state of at least one processing mechanism in each unit is abnormal; adjusting the initial speed of the target conveyor belt to a first speed according to a first instruction; the first instruction is generated after determining the target conveyor belt; the unit is composed of at least two identical processing mechanisms; the target conveyor belt is the conveyor belt corresponding to the processing mechanism with a normal operating state in the unit where the processing mechanism with an abnormal operating state is located; the first speed is less than the initial speed.

[0004] As disclosed in Chinese Patent Publication No. CN115611029A, an intelligent coal loading method and system are provided. During the process of loading coal into a carriage by a coal filling system, the filling feature vector and the discharge port state vector of the carriage are obtained; a set number of carriages are divided into a preset number of categories; the minimum height threshold, height peak threshold and discharge port state vector threshold under each category are obtained; the random factor of the filling feature vector of each carriage is obtained; the filling feature vector of the carriage is divided into normal samples and abnormal samples by using the random factor; a binary classification model is trained by using the normal samples and abnormal samples; the operating parameters of the coal filling system are controlled by using the trained binary classification model, the minimum height threshold, height peak threshold and discharge port state vector threshold under each category to load coal.

[0005] In the prior art, it is described that the speed can be adjusted through the operating states of multiple machines, and the height recognized by coal at multiple sampling moments is used to adjust the working process of coal transportation; however, in this processing method, it is difficult to discover the relationship between the equipment-related vector and the coal-related vector during coal transportation, and whether this relationship is related to the abnormal conditions that occur in the conveyor belt. Therefore, it is necessary to monitor and process the data related to the entire conveyor belt to improve the safety of coal transportation and processing. Summary of the Invention

[0006] To solve the above technical problems, the technical solution adopted by the present invention is: an intelligent control system for a conveyor belt based on coal washing and separation, including: a data acquisition module, which is used to collect the state information, coal flow rate, and conveyor belt speed during the operation of the conveyor belt as the operation feature vector of the conveyor belt operation; and the coal humidity and coal particle size distribution are used as the coal feature vector related to coal.

[0007] A state segmentation module, which is used to divide the conveyor belt into multiple feature units according to the coal feature vector and operation feature vector at different positions during coal washing and separation, and set multiple state recognition regions according to the feature thresholds of the coal feature vector and operation feature vector under each feature unit.

[0008] An operation monitoring module, which is used to monitor the data of multiple state recognition regions, generate multiple feature curves according to the coal feature vector and operation feature vector, analyze the feature curves, and identify the correlation model between the coal feature vector and the operation feature vector.

[0009] A correlation load module, which is used to determine the correlation degree during coal washing and separation according to the correlation model between the coal feature vector and the operation feature vector; verify the fault types of each correlation event under the corresponding correlation degree, and output the basic event combination corresponding to the fault type.

[0010] A fault handling module, which is used to determine the risk weight corresponding to the fault type according to the obtained fault type, check the processing solutions existing in the database, and obtain the fault handling solution corresponding to the current fault type.

[0011] The beneficial effects of the present invention are as follows: First, by using the state information, coal flow rate, and conveyor belt speed during coal operation as the operation feature vector of the conveyor belt operation, the present invention can understand the working conditions of each conveyor belt between corresponding working settings when the conveyor belt conveys coal, and can timely reflect the corresponding conveyor belt problems during coal transportation, which is convenient for subsequent timely targeted processing; at the same time, by using the coal humidity and coal particle size distribution as the coal feature vector related to coal, the present invention can understand the specific situation of coal during the washing and separation process, and prevent the problem of coal blockage during transportation.

[0012] Second, by using the coal feature vector and the operation feature vector, the present invention divides the conveyor belt into multiple state recognition regions, and conducts correlation analysis on the data on the state recognition regions, which can improve the correlation between the working-related information and the situation of the coal itself during data processing, and obtain a model data that can accurately feedback the correlation, ensuring the stability and consistency of the product quality during coal production.

[0013] III. By using the corresponding events in the association model and analyzing the corresponding associated events using a fault tree, the present invention can obtain the types of faults that are likely to occur under the current multiple data associations, and describe the fault types using risk weights and specific values, which can further identify the faults that occur during the production process, and the risks and costs during the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below with reference to the drawings and embodiments.

[0015] Figure 1 It is a system framework diagram of an intelligent control system for a conveyor belt based on coal washing and beneficiation.

[0016] Figure 2 It is a schematic flow diagram of a state segmentation module of an intelligent control system for a conveyor belt based on coal washing and beneficiation.

[0017] Figure 3 It is a schematic flow diagram of an operation monitoring module of an intelligent control system for a conveyor belt based on coal washing and beneficiation. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The embodiments of the present invention will be described in detail below. The following described embodiments are exemplary and are only used to explain the present invention and should not be construed as a limitation of the present invention. For those without specific technical or conditions noted in the embodiments, the techniques or conditions described in the literature in the art or according to the product description are followed.

[0019] Refer to Figure 1 , an intelligent control system for a conveyor belt based on coal washing and beneficiation, comprising: a data acquisition module, a state segmentation module, an operation monitoring module, an associated load module, and a fault handling module; wherein, the output end of the data acquisition module is connected to the state segmentation module, the output end of the state segmentation module is connected to the operation monitoring module, the output end of the operation monitoring module is connected to the associated load module, and the output end of the associated load module is connected to the fault handling module.

[0020] The data acquisition module is configured to collect the state information, coal flow rate, and conveyor belt speed during the operation of the conveyor belt as the operation feature vector of the conveyor belt operation; and use the coal humidity and coal particle size distribution as the coal feature vector related to coal.

[0021] The state segmentation module is configured to divide the conveyor belt into multiple feature units according to the coal feature vector and operation feature vector at different positions during coal washing and beneficiation, and set multiple state recognition regions according to the feature thresholds of the coal feature vector and operation feature vector under each feature unit.

[0022] The operation monitoring module is used to monitor data for multiple status recognition regions, generate multiple characteristic curves according to the coal characteristic vector and the operation characteristic vector, analyze the characteristic curves, and identify the association model between the coal characteristic vector and the operation characteristic vector.

[0023] The associated load module is used to determine the degree of association during coal washing according to the association model between the coal characteristic vector and the operation characteristic vector; verify the fault types of each associated event under the corresponding degree of association, and output the basic event combination corresponding to the fault type.

[0024] The fault handling module is used to determine the risk weight corresponding to the fault type according to the obtained fault type, and check the processing solutions existing in the database to obtain the fault handling solution corresponding to the current fault type.

[0025] The above-mentioned status information includes the conveyor belt temperature, the operation flag indicating whether the conveyor belt is running normally, and the flag indicating whether there is a blockage during the operation of the conveyor belt, so as to determine whether the current conveyor belt can transport coal normally when connecting to the equipment for coal washing.

[0026] Conveyor belt temperature: The temperature data of the conveyor belt can be obtained by installing temperature sensors. These sensors can monitor the temperature change of the conveyor belt in real time to ensure that the equipment operates within a safe temperature range.

[0027] Operation flag: By using a combination of a controller and sensors, it can be determined whether the conveyor belt is in a normal operation state. For example, by monitoring parameters such as the motor current and vibration of the conveyor belt, abnormalities can be judged. The obtained operation flag is mainly used to determine whether the current conveyor belt is operating normally and the states corresponding to startup and shutdown.

[0028] Blockage situation flag: By installing pressure sensors or weight sensors on the conveyor belt, the weight change of the coal on the conveyor belt can be monitored in real time to judge whether there is a blockage situation. When the weight exceeds the set threshold, the system can issue an alarm. At the same time, this flag is set by identifying whether there is excessive coal and overall blockage.

[0029] Coal flow: A flowmeter or a weighing sensor is used to measure the coal flow passing through the conveyor. By monitoring the speed and time of the coal flow, the coal quantity can be calculated to obtain the coal flow data. This unit indicates the corresponding weight of coal that can flow through within a fixed time.

[0030] Conveyor belt speed: High-precision speed measurement devices such as laser speedometers and radar speedometers are used to measure the speed of the conveyor belt in real time. These devices usually have the characteristics of fast response and high precision, which can ensure the accuracy of the measurement results.

[0031] Coal humidity: The measurement of coal humidity can be achieved through humidity sensors. These sensors can monitor the humidity changes of coal in real time, which is of great significance for ensuring the quality of coal storage and transportation.

[0032] Coal particle size distribution: Methods such as mechanical screening method, microscopic observation method, image processing method or laser particle size analysis method of coal are used to measure the particle size distribution of coal. These methods can qualitatively and quantitatively analyze coal according to the particle size, shape and distribution of coal particles.

[0033] Screening method: The coal sample is screened through a series of sieves with different pore sizes, and the coal particles are divided into parts with different particle sizes. The coal sample is placed on the topmost sieve, and by vibrating or shaking the sieve, the particles pass through the sieve holes in order of size. The residual particles on each sieve are collected and weighed. According to the mass of the residual particles on each sieve, the mass percentage of each part is calculated, so as to obtain the particle size distribution.

[0034] Laser particle size analysis method: Using the principle of laser scattering, the particle size distribution is calculated by measuring the scattered light intensity of coal particles in the laser beam. The coal sample is placed in a laser particle size analyzer, and the instrument emits a laser beam and measures the scattered light intensity. According to the distribution of the scattered light intensity, the particle size distribution is calculated. According to the two methods illustrated above, the particle size of the coal itself can be known, and the current state and situation of the coal can be more comprehensively understood by combining the humidity of the coal.

[0035] The acquisition of these parameters can monitor the operating state of the conveyor belt and the coal flow rate in real time, and potential production problems such as conveyor belt blockage and insufficient coal supply can be discovered and solved in time, so as to ensure the stable operation of the production line and improve production efficiency.

[0036] The implementation methods of the operation feature vector also include: According to the state information at each sampling moment, the obtained coal flow rate and conveyor belt speed are associated, and the first difference between the coal flow rate and the preset coal flow rate and the second difference between the conveyor belt speed and the preset conveyor belt speed are calculated respectively. After converting the first difference and the second difference into ratios, the product corresponding to the first difference and the second difference is obtained as the correlation value of the coal flow rate and the conveyor belt speed. The preset coal flow rate and the preset conveyor belt speed represent the average coal flow rate and the average conveyor belt speed when the current coal washing production line is producing coal normally. At this time, obtaining the first difference and the second difference is to judge whether there is a situation that does not conform to normal production when the current conveyor belt is transporting coal for washing, and combining the values of the state information in different situations to obtain a value representing the production state. The way of converting the first difference and the second difference into ratios is to compare the first difference and the second difference with the preset coal flow rate and the preset conveyor belt speed.

[0037] Calculate the average value of the correlation values of coal flow rate and conveyor belt speed under the same state information, and use this average value as the eigenvalue of the operation feature vector; this eigenvalue can be used as part of the operation feature vector for subsequent monitoring, analysis, or decision support. For example, it can be used to identify potential problems in the production process, optimize production parameters, or predict equipment failures, etc.

[0038] For the same state information, it represents the average situation of the differences generated under the same conveyor belt temperature, the same operation flag, and the same blockage situation flag. The operation flag generally represents flags such as normal operation, shutdown, standby, deceleration operation, etc., to describe the operation situation. The temperature is the temperature of the conveyor belt when receiving and processing coal during its operation. The blockage situation flag is the flag indicating the presence or absence of blockage.

[0039] The implementation method of the coal feature vector also includes: according to the coal humidity and coal particle size distribution at each sampling moment, obtain the maximum coal humidity, the maximum coal particle size distribution, the average coal humidity, and the average coal particle size distribution, and calculate the humidity similarity index between the maximum coal humidity and the average coal humidity and the particle size similarity index between the maximum coal particle size distribution and the average coal particle size distribution respectively; use the weighted sum of the humidity similarity index and the particle size similarity index as the correlation value of the coal humidity and the coal particle size distribution.

[0040] Calculate the average value of the correlation value of coal humidity and coal particle size distribution under the same state information, and use this average value as the eigenvalue of the coal feature vector.

[0041] Here, the data at the sampling moment is analyzed to determine the stability and consistency of the coal feature vector under the corresponding time change. For the humidity similarity index and the particle size similarity index, calculation methods such as cosine similarity or Pearson correlation coefficient can be used to obtain these indexes, and then the weighted sum of the two is calculated to obtain a relative eigenvalue. At the same time, this weighted sum is used as part of the coal feature vector to improve the accuracy of subsequent classification processing of the coal feature vector.

[0042] In an embodiment of the present invention, the state segmentation module mainly divides according to the eigenvalues corresponding to the coal feature vector and the operation feature vector. The feature threshold is the value taken to divide the eigenvalue into multiple intervals, which represents what regions the coal feature vector and the operation feature vector should be divided into under different values, and whether corresponding problems will occur when the conveyor belt conveys coal in these regions.

[0043] Such as Figure 2As shown in the figure, the implementation method of the state segmentation module includes: determining the coal characteristic vectors and operation characteristic vectors on the feeding area, processing area, and discharging area in sequence according to the position of the conveyor belt, and dividing the conveyor belt into multiple characteristic units; here, according to the distribution position of the conveyor belt, the corresponding vectors are corresponded to the conveyor belt position to determine the changes of the characteristics corresponding to the coal on different conveyor belts during the coal washing process, and the data identified at these positions is linked to the data of the normal operation of the production line to discover whether there are abnormal situations in the current production process.

[0044] Feeding area: Here, the initial characteristics of the coal are mainly concerned, such as humidity, particle size distribution, etc.

[0045] Processing area: It includes the conveying sections between various processing devices, and it is necessary to monitor the changes in coal characteristics and the operating status of the devices.

[0046] Discharging area: Pay attention to the quality characteristics of the final product.

[0047] Identify the main component vectors of the coal characteristic vectors and operation characteristic vectors on each characteristic unit, and select the characteristic thresholds of the coal characteristic vectors and operation characteristic vectors according to the values of the main component vectors. At this time, the characteristic thresholds of the coal characteristic vectors and operation characteristic vectors will select the content selected by the main component vectors on different characteristic units, indicating the values required for the characteristic units to divide the state recognition area.

[0048] The main component vectors illustrate the main identified characteristics at the relative positions of the conveyor belt in the feeding area, processing area, and discharging area. For example, in the feeding area, the coal flow rate, coal humidity, and coal particle size distribution are mainly identified.

[0049] In the processing area, the conveyor belt speed, coal load, and foreign object detection are mainly identified. The coal load is identified by the coal flow rate of the coal entering the conveyor belt per unit time, and the foreign object detection is obtained through the blockage situation flag in the status information to determine whether there is a large blockage and whether the conveyor belt is overloaded.

[0050] In the discharging area, the coal humidity, coal particle size distribution, and the blockage situation flag in the status information are mainly identified to determine whether the coal will be affected by external conditions and its own quality.

[0051] These contents are the main component vectors on each feature unit. Then, according to the content to be recognized by these main component vectors, the feature threshold for judgment at this time is selected. For the landmark information, its value is marked in the form of 0, 1, 2, 3, etc. or 0, 1. 0 indicates no problem, and 1, 2, 3 represent the corresponding problem degrees, or 1 is used as the mark indicating the existence of a problem to describe whether there is a problem in the corresponding area of the current conveyor belt; other numerical features use the boundary values identified in historical data, or use the average value, minimum value, or maximum value of the corresponding features in the early warning, abnormal, and normal operating conditions to set the feature threshold at this time.

[0052] For example, the representation of the feeding area in the normal, early warning, and abnormal states is as follows.

[0053] Normal state: Set a reasonable coal flow range (for example, 100 to 200 tons per hour), control the coal humidity within a range that does not affect fluidity (such as below 8%), and the coal particle size distribution conforms to the preset standard (for example, the proportion of particles less than 5mm does not exceed 30%).

[0054] Early warning state: If the coal flow is close to the upper and lower limits (such as 90 or 210 tons / hour), the humidity is slightly higher than the recommended value (such as 8% - 10%), or the coal particle size distribution shows a slight deviation, an early warning is triggered to prompt the operator to pay attention.

[0055] Abnormal state: When the coal flow exceeds the safe range (below 80 or above 220 tons / hour), the humidity is too high (exceeding 10%), or the coal particle size distribution seriously deviates from the standard, the system should immediately alarm and take measures (such as slowing down the conveyor belt speed or stopping the feeding).

[0056] For example, the representation of the processing area in the normal, early warning, and abnormal states is as follows.

[0057] Normal state: Set a suitable conveyor belt speed (such as 2 m / s), keep the coal load within the design capacity of the equipment (such as not exceeding 80% of the maximum load), and no foreign objects are detected.

[0058] Early warning state: When the conveyor belt speed slightly deviates from the set value (±0.2 m / s), the coal load is close to the upper limit (75% - 80%), or small-sized foreign objects are occasionally detected, a warning signal is issued.

[0059] Abnormal state: If the conveyor belt speed significantly deviates from the set value (more than ±0.5 m / s), the coal load exceeds the standard, or large foreign objects are frequently detected, immediate action is required to prevent equipment damage or blockage.

[0060] For example, the representation of the discharging area in the normal, early warning, and abnormal states is as follows.

[0061] Normal state: Ensure that the coal quality at discharge meets the requirements (such as specific moisture level and ideal particle size distribution), the conveyor belt runs smoothly without obstruction, and the environmental conditions are suitable.

[0062] Warning status: If the discharge quality is found to be slightly deteriorating (such as humidity increasing to 9%, particle size distribution slightly deviating from the ideal value), or if slight signs of poor transmission are observed, timely attention should be paid.

[0063] Abnormal state: Once the discharge quality obviously does not meet the standards (such as humidity exceeds 10%, particle size distribution deviates seriously), or serious transmission blockage occurs, the machine must be stopped immediately to check and correct the problem.

[0064] According to the thresholds described for the corresponding main component vectors at different positions, the areas divided by the feature thresholds will be defined as normal working areas, warning areas, and fault areas. Data in abnormal states will be considered as fault areas with fault risks, so as to facilitate the subsequent timely review of fault problems that may occur in the current conveyor belt during coal washing.

[0065] Using the characteristic thresholds of the coal feature vector and the operating feature vector, we define feature units, sequentially designated as normal operation zone, warning zone, and fault zone. These defined feature units serve as the output state recognition regions. The resulting state recognition regions represent the relative operating conditions of the conveyor belt at different locations, assisting in subsequent verification of the normal transport of coal on different conveyor belts.

[0066] At the same time, when dividing the state identification area, it can also be divided according to the characteristic values calculated by the coal characteristic vector and the operation characteristic vector. The characteristic values referred to in this part are the characteristic values calculated by the data acquisition module for the correlation between coal flow and conveyor belt speed, and the correlation between coal moisture and coal particle size distribution. If this characteristic value is used as the subsequent characteristic threshold, the current scenario considering the conveyor belt will become a scenario in which the correlation between coal flow, conveyor belt speed, coal moisture and coal particle size distribution is used as the main identification parameter. That is, it is divided into multiple state identification areas according to the correlation between the corresponding data of the coal characteristic vector and the operation characteristic vector. The state identification areas here will include normal transportation state area, flow overload state area, insufficient flow state area, humidity abnormal state area, particle size distribution abnormal state area and comprehensive fault state area; these areas will represent the state identification area in the normal working area, warning area and fault area according to the values in the main component vector, further explaining the type corresponding to the state identification area, so as to facilitate the subsequent generation of characteristic curves for corresponding problems.

[0067] Normal conveying state area: within this area, the coal flow rate, conveyor belt speed, coal humidity, and particle size distribution are all within the preset normal range; the conveyor belt operates stably, and the coal is conveyed continuously and evenly without the need for additional control or adjustment.

[0068] Flow overload state area: The coal flow rate exceeds the preset maximum value, which may cause the conveyor belt to be overloaded. It is necessary to increase the conveyor belt speed, start the standby conveyor belt, or adjust the upstream coal feeding volume to reduce the load, which may be accompanied by risks such as coal accumulation and conveyor belt slipping.

[0069] Flow insufficiency state area: The coal flow rate is lower than the preset minimum value, and the operating efficiency of the conveyor belt decreases. The conveyor belt speed can be reduced to save energy, or check whether the upstream coal feeding equipment is faulty. Long-term insufficient flow may cause the conveyor belt to run empty, increasing energy consumption and wear.

[0070] Humidity abnormal state area: The coal humidity is too high or too low, exceeding the preset range. Too high humidity may cause problems such as conveyor belt slipping and coal sticking; too low humidity may cause coal dust to fly and increase conveyor belt wear. It is necessary to take drying or humidifying measures to adjust the coal humidity.

[0071] Particle size distribution abnormal state area: The coal particle size distribution does not meet the preset requirements, with the particle size being too large or too small. Too large particle size may cause the conveyor belt to be blocked; too small particle size may increase conveyor belt wear and energy consumption. It is necessary to take screening or crushing measures to adjust the coal particle size distribution.

[0072] Comprehensive fault state area: When multiple parameters deviate from the normal range simultaneously, it may enter the comprehensive fault state area. At this time, it is necessary to comprehensively consider the influence of each parameter and take comprehensive measures for fault detection and handling, which may involve conveyor belt shutdown for maintenance, replacement of damaged components, etc.

[0073] When using the defined characteristic unit as the output state recognition area, it also includes: extracting the characteristic value corresponding to the correlation value of the coal flow rate and the conveyor belt speed on the defined characteristic unit as the first boundary condition; the characteristic value described here is the average value calculated from the correlation value of the coal flow rate and the conveyor belt speed under the same state information, and this average value represents the correlation between the coal flow rate and the conveyor belt speed on a characteristic unit area.

[0074] Extracting the characteristic value corresponding to the correlation value of the coal humidity and the coal particle size distribution on the defined characteristic unit as the second boundary condition; the characteristic value described here is the average value calculated from the coal humidity and the coal particle size distribution under the same state information.

[0075] According to the first boundary condition and the second boundary condition, the boundary of the state recognition region is processed to determine the distribution positions of the state recognition regions. At this time, the first boundary condition and the second boundary condition are further used to divide the current state recognition region to obtain regions with different associations between coal moisture, coal particle size distribution, coal flow rate, and conveyor belt speed under each feature unit, so as to perform more detailed feature recognition on the currently set state recognition unit, facilitating subsequent comprehensive viewing of these data.

[0076] For example, if a feature unit has a situation such as a flow rate shortage state region, a humidity anomaly state region, a particle size distribution anomaly state region, etc. after being defined, then at this time, the data on the associations between coal flow rate and conveyor belt speed, as well as coal moisture and coal particle size distribution in these situations are further described, and data descriptions are supplemented at the positions of these already marked regions, facilitating improving the accuracy when setting the association model for the overall data.

[0077] In an embodiment of the present invention, the operation monitoring module monitors the data in the normal working area, early warning area, and fault area existing in the state recognition region, and converts the coal feature vector and the operation feature vector into feature curves. This feature curve is generated according to the main recognized features in each region. For example, multiple different forms of feature curves are generated according to coal flow rate, coal moisture, coal particle size distribution, coal load, and conveyor belt speed, etc. Among them, coal flow rate, coal moisture, coal load, and conveyor belt speed can directly display the variation of relevant data over time. For coal particle size distribution, considering the particle size of coal after processing such as crushing, the proportion of coal particle size distribution at the corresponding time point is used to generate the feature curve corresponding to coal particle size distribution. Then, the correlations of these several different forms of feature curves are verified according to their variation trends to find out whether when the data on one feature curve is abnormal, the variation forms of other feature curves will simultaneously show abnormalities, or whether the abnormalities will be distributed in a certain time interval form. Then, the frequent item sets of these feature curves when abnormalities occur are found to identify the association patterns when the conveyor belt has abnormalities and serve as the subsequent output association model. At the same time, when generating these several feature curves, other data in the coal feature vector and the operation feature vector are also marked to identify the performance of other less important data such as corresponding flag information in the state recognition region.

[0078] As Figure 3 shown, the implementation method of the operation monitoring module includes: standardizing the coal feature vector and the operation feature vector on the state recognition region according to the type of the state recognition region to obtain at least one standard sample information, and generating multiple feature curves according to the values of each standard sample information.

[0079] At this time, the type of the status recognition area indicates the location where the current status recognition area is located, that is, which position on the feeding area, the processing area, and the discharging area, and which label among the normal working area, the warning area, and the fault area is set for the data at this position. This label represents the type of the status recognition area.

[0080] At this time, standardization processing is performed to remove duplicates and invalid values from the corresponding data in the coal characteristic vector and the operation characteristic vector, and then set dimensionless values on the vector to facilitate subsequent comprehensive calculation of the relationships between multiple characteristic curves. The standard sample information obtained later represents the content after standard processing, and finally multiple characteristic curves are generated.

[0081] Construct a relationship list for each characteristic curve. The relationship list is a list that combines the data in each characteristic curve in pairs and calculates their support degrees. Use the relationship list to extract candidate frequent item sets; candidate frequent item sets are item combinations extracted from a transaction database or event log with a support degree higher than a certain threshold; support degree refers to the frequency of a specific item set appearing in the entire dataset. For example, during coal processing, if it is observed that high humidity and small particle size often occur simultaneously, then the item set composed of these two characteristics is a candidate frequent item set. The support degree set at this time will be set according to the support degree of these data appearing under normal historical data conditions. When it exceeds this support degree, it indicates that there may be related problems with the currently associated data.

[0082] Match the candidate frequent item sets with the preset association rules, obtain the matching results of the candidate frequent item sets and the preset association rules, and generate a matching polynomial corresponding to the matching results. Set the association model according to the matching polynomial. The preset association rules refer to rules defined in advance based on domain knowledge or experience for identifying patterns in data. For example, in a coal conveyor belt system, there may be a preset association rule: "If the coal humidity exceeds 8%, and the proportion of particles smaller than 5mm in the particle size distribution exceeds 30%, then blockage may occur"; here, "the coal humidity exceeds 8%" and "the proportion of particles smaller than 5mm exceeds 30%" are the conditions (antecedents), and "blockage may occur" is the result (consequent); then when such data is matched with the candidate frequent item sets, if the candidate frequent item set satisfies all the conditions of the preset association rule, it is considered that this frequent item set matches this rule successfully. For example, assume that a frequent item set is found from the data: "The coal humidity is 9%, and the proportion of particles smaller than 5mm is 32%", which exactly meets the above preset association rule, so this item set matches the rule successfully.

[0083] The simultaneous matching polynomial is used to quantify the importance or influence degree of the matching result, which can be quantified by the product of the ratio of the corresponding elements in the matching result exceeding the preset data in the preset association rule, indicating the relative importance existing after the matching, and using the data corresponding to the matching polynomial to set the association model to predict the operation situation during coal transportation.

[0084] When obtaining the matching result of the candidate frequent item set and the preset association rule, it also includes: when the candidate frequent item set meets all the conditions of the preset association rule, it is considered that the frequent item set matches the preset association rule successfully, and it is judged whether there is a neighboring candidate frequent item set that matches successfully in the corresponding state recognition region. The neighboring candidate frequent item set means that there is a value in the item set that matches the current successful match. For example, if the successfully matched candidate frequent item set is the coal humidity and the coal particle size distribution, then the neighboring candidate frequent item set contains one of the two. At the same time, if this neighboring candidate frequent item set also matches successfully through the preset association rule, the corresponding data association existing when an abnormality occurs can be discovered, and this part of the data is output together. If it exists, the neighboring candidate frequent item sets corresponding to the current candidate frequent item set are merged and output as the matching result of the candidate frequent item set and the preset association rule.

[0085] When generating the matching polynomial corresponding to the matching result, the data represented by all the matching results on the corresponding state recognition region can also be set as a concept graph, the successfully matched data is expressed using node relationships, and according to the connection relationships of the corresponding data on the graph for this state region, the output matching polynomial is set; or the corresponding data successfully matched on each characteristic curve is comprehensively output to obtain the current relationship model.

[0086] Therefore, the implementation method of generating the matching polynomial corresponding to the matching result includes: based on the preset association rule in the matching result, converting the matching result into matching nodes, and each matching node represents a specific concept or entity described in the preset association rule, such as "high humidity environment", "fine particle material", etc., which directly describe the content of the corresponding data on the matching node. The edges between the matching nodes represent the confidence between the matching nodes, which is used to describe the relationship strength between two matching nodes.

[0087] Connect the matching nodes with each other and output the shortest paths of the connected matching nodes. Here, convert the confidence levels on each matching node to negative numbers, and use Dijkstra's algorithm to find the shortest paths of each matching node after connection. This path represents the path with the highest confidence level. When all matching nodes are connected and all meet the condition of the shortest path, output the corresponding connected content. This output content will be used for the calculation of the subsequent matching polynomial. Take the product or weighted sum of the ratios of these values exceeding the preset association rules as the value of the matching polynomial, and then form the association model under the current analysis.

[0088] If the matching is successful and the coal feature vector combined into the matching polynomial is , and the operation feature vector is , where m represents the number of elements in the coal feature vector, and n represents the number of elements in the operation feature vector; these data are mapped from the data after the matching process to judge the association model between the coal feature vector and the operation feature vector at this time; the output of the association model is , where Z represents the number of state recognition regions. At this time, the data in the characteristic curve are processed according to the state recognition regions they belong to, and the data output by each state recognition region is used to judge the association between the current coal feature vector and the operation feature vector, and find the possible faults under this association.

[0089] At this time, the association model is expressed as ; where represents the output value of the association model, represents the regression coefficient matrix, represents the error term, represents the polynomial basis function, , and the polynomial basis function will include the constant term 1, the values of single coal feature vectors and single operation feature vectors, the squared values of single coal feature vectors and single operation feature vectors, the interactions between two elements in the coal feature vector and the operation feature vector, and the interactions of self-elements in the coal feature vector and the operation feature vector. During regression, continuously adjust the output result of the association model and simultaneously change the data in the calculated polynomial basis function so that when the association model outputs data, its own error term can reach the minimum value.

[0090] For the regression coefficient matrix, it is expressed as: , where represents the transpose matrix of the polynomial basis function, represents the regularization coefficient, which is used to prevent overfitting; Denote the identity matrix. The elements on the main diagonal of the identity matrix are all 1, while the elements in other positions are all 0. Then, when the error term is minimized, the final output value of the current correlation model is obtained.

[0091] In an embodiment of the present invention, the associated load model mainly processes according to the value output by the correlation model, and extracts the correlation events in the preset correlation rules corresponding to the value. The correlation event represents an event that may occur after meeting the preset correlation rules. For example, if the coal moisture exceeds 8% and the proportion of particles smaller than 5 mm in the particle size distribution exceeds 30%, then blockage may occur. The data described above are the conditions of the preset correlation rules, and the possible blockage indicates the correlation event that may occur at this time. At this time, each correlation event is fitted, and the fault that is most likely to occur after fitting these correlation events is found, and after outputting this type using a fault tree, the processing set of relevant data under the correlation degree is completed.

[0092] The implementation method of the associated load module includes: extracting the correlation degree output by the correlation model of the coal characteristic vector and the operation characteristic vector, and obtaining each correlation event corresponding to the correlation degree. When outputting the correlation degree, because the number of data contained in each matching polynomial is different, when calculating its correlation degree, it will include multiple correlation events, and the correlation degree will output multiple values macroscopically; at this time, the specific fault type that occurs is used as the top event, each correlation event is used as the basic event, and the basic events are connected to form a logic gate to complete the establishment of the fault tree, and finally the correlation event corresponding to the current basic event is found.

[0093] The correlation event described here is the corresponding content that is dynamically merged or not merged after the operation monitoring module generates the correlation model by matching the candidate frequent item set and the preset correlation rules.

[0094] If the frequent item set indicates: high humidity and fine particles appear simultaneously (support degree > 0.7); the conveyor belt speed fluctuation and the sudden drop in coal flow occur simultaneously.

[0095] The preset correlation rules indicate: Rule 1: IF humidity > 8% AND the proportion of fine particles > 30% → predict blockage (confidence level = 0.9); Rule 2: IF temperature > 60°C AND speed fluctuation > 10% → predict motor overheating (confidence level = 0.85).

[0096] If the correlation events in adjacent state regions are triggered simultaneously (such as abnormal humidity in the feeding area + abnormal fine particles in the processing area), they are merged into a global correlation event.

[0097] As shown in Table 1, its correlation events can be expressed as the following content.

[0098] Table 1 Schematic Table of Correlation Events

[0099]

[0100] In the content shown in Table 1, point out the data example corresponding to the combination of the corresponding preset association rules for this frequent item set. This data example will be represented as the content represented by the matching polynomial, and after the subsequent processing is completed, the positions on the coal characteristic vector and the operation characteristic vector corresponding to this data will be used. The coal characteristic vector and the operation characteristic vector will use the association model to output relevant content to obtain the final desired association degree value and the data situation processed by the association model.

[0101] After using the fault tree for analysis, the basic events for analysis are extracted from the corresponding content of this associated event. When the association model outputs this association degree value, the associated events it can contain are regarded as basic events, and the basic events are connected to represent the content output by the fault tree.

[0102] Calculate the occurrence probabilities of each basic event in the fault tree, and use logic gates to calculate the fault probabilities of the top-level events layer by layer. Select the top-level event with the maximum fault probability and output the combination of basic events corresponding to this top-level event to obtain the output fault type.

[0103] The occurrence probabilities of each basic event in the fault tree are expressed as: ; where represents the occurrence probability of the basic event in the fault tree, j represents the index of the basic event, and the value range of j is from 1 to J; k represents the slope parameter of the logic function, which controls the steepness of the probability curve. The larger this value, the steeper the transition from low to high probability, and the higher the sensitivity near represents the association degree value of the i-th state recognition region in the association model. This value represents the output result of the association model, and the value range of i is from 1 to Z; represents the trigger threshold of the basic event . At this time, the trigger threshold of the basic event is the calculated value of the corresponding matching polynomial, which represents the relative importance corresponding to this basic event. represents the exponential constant.

[0104] At this time, the probability values that each basic event divided within the fault tree can represent can be obtained. This probability value will represent the final calculation result generated from the data recognized on the current conveyor belt. When using logic gates to calculate the fault probability of the top-level event, the logic gates are divided into AND gates and OR gates, and the results of the two calculations will be different; for the top-level event of the AND gate, the basic events connected by the AND gate under this top-level event are used to represent the conditional probability that these basic events are satisfied simultaneously by multiplying; the OR gate represents the union of two basic events. For example, subtract the occurrence probability of the basic events connected by the OR gate from 1, multiply the subtracted values, and then subtract the multiplied value from 1 again to represent the probability of the top-level event of all basic events connected by the OR gate; then select the part with the largest value among the top-level events of all basic event combinations under these two calculations to obtain a maximum fault probability. This fault probability will indicate that the basic events connected by the logic gates are most likely to fail, and according to the fault type corresponding to this maximum fault probability, it can be known the most likely faults of the current conveyor belt. Finally, relevant preparations can be made for the current conveyor belt based on this fault type to reduce the losses caused by faults.

[0105] For example, when analyzing the top-level event of the fault tree, conveyor belt breakage is selected under a certain situation.

[0106] The intermediate events are represented as: long-term wear (triggered by the basic event E5: cumulative operating time > 1000h); instantaneous overload (triggered by the basic event AE1: blockage warning event).

[0107] The logic gate connections are represented as: conveyor belt breakage = long-term wear AND (instantaneous overload OR material fatigue); material fatigue = temperature anomaly event E3 AND speed fluctuation > 20%.

[0108] After that, the input data are: E5: cumulative operating time = 1200h, AE1: blockage warning event (triggered), E3: temperature = 65°C, and speed fluctuation = 25%.

[0109] After that, calculate and perform calculations on the corresponding data. For example, the long-term wear probability: P(E5) = 1 (because the operating time > 1000h).

[0110] The instantaneous overload probability: P(AE1) = 0.9.

[0111] The material fatigue probability: P(E3) × P(speed fluctuation) = 0.98 × 0.95 = 0.93.

[0112] Finally, the probability of its top-level event is obtained: P(breakage) = P(long-term wear) × [P(instantaneous overload) + P(material fatigue) - P(instantaneous overload)P(material fatigue)] = 1 × [0.9 + 0.93 - 0.9 × 0.93] = 0.993.

[0113] Based on the probability of the top event output by this final selection, its main basic events are discovered. After combining these basic events, they are output together with the fault type described by this top event. After judging these contents, the processing method required for the current conveyor belt is obtained.

[0114] Finally, a relevant table of the basic event combination is obtained, as shown in Table 2.

[0115] Table 2 Schematic Table of Basic Event Combinations

[0116]

[0117] In Table 2, the output situation of the combination of its basic events after using the fault tree is described, and it is explained that when the correlation degree of the coal characteristic vector and the operation characteristic vector corresponding to the correlation model and the correlation events corresponding to the correlation degree are output, the possible relevant problem combinations under the correlation degree may exist. And according to this problem combination, the relevant contents are processed subsequently to improve the safety during coal washing.

[0118] In an embodiment of the present invention, the risk weight of the fault type is judged according to the relative connection relationship between each basic event in the fault tree for the currently identified fault type, and then according to this risk weight, a solution search is performed in the database to find the corresponding fault handling solution.

[0119] The implementation method of the fault handling module includes obtaining the basic event combination corresponding to the fault type and calculating the cycle probability and expected loss value of each basic event in the basic event combination in the database.

[0120] For the cycle probability of each basic event, it represents extracting the occurrence times from the database and using the occurrence times divided by the total number of faults during the maintenance cycle to obtain the cycle probability of the occurrence of this basic event, which represents whether this basic event will have corresponding faults under different maintenance conditions and usage times.

[0121] For the expected loss value, it represents the mean square error of the cycle probability of the corresponding basic event under multiple maintenance cycles, which represents the distribution of the occurrence probabilities of each basic event under different cycles.

[0122] Set the risk weight of the current fault type according to the cycle probability and expected loss value of each basic event.

[0123] For the risk weight, according to the connection relationship of each basic event, the basic event combination is divided into multiple minimum cut sets, and then the risk weight of the minimum cut sets is set to obtain the risk weight of the finally output fault type.

[0124] The implementation method of setting the risk weight of the current fault type includes: using the period probability and expected loss value of each basic event to set the risk weight of each basic event. The risk weight of each basic event is expressed as the product of the corresponding period probability and expected loss value.

[0125] Obtain the minimum cut set of basic event combinations for the current fault type. Each minimum cut set represents a set of basic events connected via AND gates from the top-level event to a specific basic event. Their combined occurrence triggers the top-level event. Calculate the risk weight of the current minimum cut set using the risk weight of each basic event. The risk weight within the minimum cut set is calculated by multiplying the risk weights of each basic event. Set the sum of the risk weights of the minimum cut set as the risk weight for the fault type.

[0126] Use the risk weight of the current fault type to search the database for solutions and obtain the fault handling solution.

[0127] When searching for solutions, the risk weight of the fault type is used as the first calculation method. First, the correlation coefficient between the risk weight of the fault type and the preset risk weight in the database is calculated. Then, the correlation coefficient is calculated based on the risk weight of the minimum cut set under the current fault type and the preset risk weight as the second calculation method. Finally, the correlation coefficient is calculated based on the preset risk weight of each basic event in the minimum cut set as the third calculation method. The correlation coefficient calculated at this time is calculated using the Pearson correlation coefficient. Then, the correlation coefficients corresponding to the first, second, and third calculation methods are sorted from large to small. When sorting at this time, if the correlation coefficient values in the second calculation method are the same, the correlation coefficient values of the third calculation method are used for sorting. Then, when the correlation coefficients of the first, second, and third calculation methods are all maximum values, the corresponding processing solution is output as the fault processing solution.

[0128] The preset risk weights for the description will be set according to the fault described by the fault type and the fault described by each basic event to determine which specific processing method the current fault type prefers. Ultimately, the conveyor belt can be quickly identified and processed to improve the conveying efficiency of the conveyor belt.

[0129] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered by the scope of protection of the present invention.

Claims

1. An intelligent control system for a conveyor belt based on coal washing and dressing, characterized in that, Including: A data acquisition module, which is used to collect the status information, coal flow rate, and conveyor belt speed during the operation of the conveyor belt as the operation feature vector of the conveyor belt operation; and use the coal humidity and coal particle size distribution as the coal feature vector related to coal; A status segmentation module, which is used to divide the conveyor belt into multiple feature units according to the coal feature vector and operation feature vector at different positions during coal washing and dressing, and set multiple status recognition regions according to the feature thresholds of the coal feature vector and operation feature vector under each feature unit; An operation monitoring module, which is used to monitor the data of multiple status recognition regions, generate multiple feature curves according to the coal feature vector and operation feature vector, analyze the feature curves, and identify the correlation model between the coal feature vector and operation feature vector; An associated load module, which is used to determine the degree of association during coal washing and dressing according to the correlation model between the coal feature vector and operation feature vector; verify the fault types of each associated event under the corresponding degree of association, and output the basic event combination corresponding to the fault type; A fault handling module, which is used to determine the risk weight corresponding to the fault type according to the obtained fault type, and check the processing solutions existing in the database to obtain the fault handling solution corresponding to the current fault type.

2. The intelligent control system for a conveyor belt based on coal washing and dressing according to claim 1, wherein The implementation method of the operation feature vector also includes: According to the status information at each sampling moment, associate the obtained coal flow rate and conveyor belt speed, calculate the first difference between the coal flow rate and the preset coal flow rate and the second difference between the conveyor belt speed and the preset conveyor belt speed respectively, and after converting the first difference and the second difference into ratios, obtain the product corresponding to the first difference and the second difference as the association value of the coal flow rate and the conveyor belt speed; Calculate the average value of the association value of the coal flow rate and the conveyor belt speed under the same status information, and use this average value as the eigenvalue of the operation feature vector.

3. The intelligent control system for a conveyor belt based on coal washing and dressing according to claim 1, wherein The implementation method of the coal feature vector also includes: According to the coal humidity and coal particle size distribution at each sampling moment, obtain the maximum coal humidity, maximum coal particle size distribution, average coal humidity, and average coal particle size distribution, and calculate the humidity similarity index between the maximum coal humidity and the average coal humidity and the particle size similarity index between the maximum coal particle size distribution and the average coal particle size distribution respectively; use the weighted sum of the humidity similarity index and the particle size similarity index as the association value of the coal humidity and the coal particle size distribution; Calculate the average value of the association value of the coal humidity and the coal particle size distribution under the same status information, and use this average value as the eigenvalue of the coal feature vector.

4. The intelligent control system for a conveyor belt based on coal washing and dressing according to claim 1, wherein, The implementation method of the status segmentation module includes: Sequentially determine the coal feature vector and operation feature vector in the feeding area, processing area, and discharging area according to the position of the conveyor belt, and divide the conveyor belt into multiple feature units; Identify the main component vectors of the coal feature vector and operation feature vector on each feature unit, and select the feature thresholds of the coal feature vector and operation feature vector according to the values of the main component vectors; Use the feature thresholds of the coal feature vector and operation feature vector to define the feature units, and define them as the normal working area, early warning area, and fault area in sequence; use the defined feature units as the output status recognition regions.

5. An intelligent control system for a conveyor belt based on coal washing and dressing according to claim 4, characterized in that, When the defined feature unit is used as the output state recognition area, it further includes: Extracting the eigenvalue corresponding to the correlation value of the coal flow rate and the conveyor belt speed on the defined feature unit as the first boundary condition; Extracting the eigenvalue corresponding to the correlation value of the coal humidity and the coal particle size distribution on the defined feature unit as the second boundary condition; According to the first boundary condition and the second boundary condition, performing boundary processing on the state recognition area to determine the distribution positions of each state recognition area.

6. The intelligent control system for conveyor belt based on coal washing and dressing according to claim 1, wherein, The implementation method of the operation monitoring module includes: According to the type of the state recognition area, standardizing the coal feature vector and the operation feature vector on the state recognition area to obtain at least one standard sample information, and generating multiple feature curves according to the values of each standard sample information; Constructing a relationship list for each feature curve and extracting candidate frequent item sets using the relationship list; Matching the candidate frequent item sets with the preset association rules, obtaining the matching results of the candidate frequent item sets and the preset association rules, generating a matching polynomial corresponding to the matching results, and setting the association model according to the matching polynomial.

7. An intelligent control system for a conveyor belt based on coal washing and dressing according to claim 6, characterized in that When obtaining the matching results of the candidate frequent item sets and the preset association rules, it further includes: When the candidate frequent item set meets all the conditions of the preset association rule, it is considered that the frequent item set matches the preset association rule successfully. Determine whether there is a matching adjacent candidate frequent item set for the current candidate frequent item set in the corresponding state recognition area. If so, merge the adjacent candidate frequent item sets corresponding to the current candidate frequent item set and output them as the matching results of the candidate frequent item set and the preset association rule.

8. An intelligent control system for a conveyor belt based on coal washing and dressing according to claim 1, characterized in that, The implementation method of the association load module includes: Extracting the degree of association output by the association model of the coal feature vector and the operation feature vector, obtaining each association event corresponding to the degree of association, using the fault type as the top event, using each association event as the basic event, and connecting the basic events to form a logic gate to complete the establishment of the fault tree; Calculating the occurrence probability of each basic event in the fault tree, using the logic gate to calculate the fault probability of the top event layer by layer, selecting the top event with the maximum fault probability, and outputting the combination of basic events corresponding to the top event to obtain the output fault type.

9. The intelligent control system for conveyor belt based on coal washing and dressing according to claim 1, wherein The implementation method of the fault handling module includes obtaining the combination of basic events corresponding to the fault type and calculating the cycle probability and expected loss value of each basic event in the combination of basic events in the database; Setting the risk weight of the current fault type according to the cycle probability and expected loss value of each basic event; Using the risk weight of the current fault type to retrieve the database for a fault handling solution.

10. The intelligent control system of a conveyor belt based on coal washing and dressing according to claim 9, wherein, The implementation method of setting the risk weight of the current fault type includes: Using the cycle probability and expected loss value of each basic event to set the risk weight of each basic event, and the risk weight of each basic event is expressed as the product of the corresponding cycle probability and expected loss value; Obtain the minimal cut sets of the basic event combinations for the current fault type, and use the risk weights of each basic event to calculate the risk weights of the current minimal cut sets. The risk weights within the minimal cut sets are obtained based on the product of the risk weights of each basic event; set the sum of the risk weights of the minimal cut sets as the risk weight of the fault type.

Images