Material level control method for furnace top stock bin of calcium carbide furnace

Abstract

The invention provides a material level control method for furnace top bins of a calcium carbide furnace, and relates to the technical field of material level control, the method comprises the following steps: obtaining actually measured material level data and calculated material level data of each furnace top bin in a plurality of furnace top bins; fusing the actually measured material level data and the calculated material level data to generate fused material level data; processing the fused material level data based on a preset time sequence prediction model so as to predict the duration required by the material level in each furnace top stock bin to reach a preset low material level threshold value; acquiring a material supplementing rate and a material supplementing task time sequence of a material supplementing device; based on the material supplementing task time sequence and the material supplementing rate, processing the time length required by each furnace top stock bin to reach a preset low material level threshold value and a preset high material level threshold value so as to carry out task sorting and task insertion on the material supplementing task time sequence, and generating an optimized material supplementing task time sequence; and the multiple furnace top bins are supplemented with materials on the basis of the optimized material supplementing task time sequence. By means of the method, the technical effect of improving the balance of the material levels in the multiple furnace top bins is achieved.

Description

Technical Field

[0001] This application relates to the technical field of material level control, and in particular to a method for controlling the material level in the top silo of a calcium carbide furnace. Background Technology

[0002] A calcium carbide furnace is an industrial device that uses electric arcs and resistance heating to chemically react quicklime with carbonaceous raw materials in a high-temperature environment above 2000℃ to produce calcium carbide (calcium carbide). To ensure continuous production and accurate proportioning, modern calcium carbide furnaces typically employ a graded charging structure, mainly including raw material storage silos, weighing silos for achieving precise proportioning of lime and carbonaceous raw materials, and multiple furnace top silos located at the furnace top for receiving billets and distributing them circumferentially into the furnace.

[0003] In existing technologies, the management and distribution control of material levels in multiple furnace top silos generally employs independent control strategies based on fixed high and low threshold alarms or a simple sequential, rotating distribution mode. These methods are essentially passive response mechanisms, meaning that upstream weighing is only triggered to replenish a specific silo when its level drops to the low alarm limit; or they mechanically distribute batches of material according to a preset order, ignoring real-time differences in material levels across silos. Because the rate of material consumption in the furnace area corresponding to each silo fluctuates dynamically, these methods struggle to accurately, in real-time, and proactively maintain an overall balance of material levels in all furnace top silos. This easily leads to uneven material distribution within the furnace, subsequently causing process problems such as three-phase current imbalance, localized overheating, or furnace condition fluctuations in the calcium carbide furnace.

[0004] Therefore, how to improve the balance of material levels in multiple furnace top silos has become an urgent technical problem to be solved. Summary of the Invention

[0005] To improve the balance of material levels in multiple furnace top silos, this application provides a method for controlling the material level in the top silo of a calcium carbide furnace.

[0006] This application provides a method for controlling the material level in the top charge silo of a calcium carbide furnace, using the following technical solution: A method for controlling the material level in the top silo of a calcium carbide furnace, comprising the following steps: The system acquires measured and estimated material level data for each of the multiple furnace top silos; the estimated material level data is obtained from the weighing silo and the feeding device; the measured and estimated material level data are fused to generate fused material level data; the fused material level data is processed based on a preset time series prediction model to predict the time required for the material level in each furnace top silo to reach a preset low material level threshold; the feeding rate and feeding task sequence of the feeding device are acquired; based on the feeding task sequence and feeding rate, the time required for each furnace top silo to reach the preset low material level threshold and the preset high material level threshold are processed to sort and insert tasks in the feeding task sequence, generating an optimized feeding task sequence; the multiple furnace top silos are fed based on the optimized feeding task sequence.

[0007] By adopting the above technical solution, the measured data obtained by direct measurement is integrated with the indirect data based on material flow calculation, and a time series algorithm is used for forward prediction. Then, based on the prediction results, the predetermined material replenishment plan is dynamically reconstructed and optimized, realizing the transformation from passive response to active collaborative control. This helps to improve the overall balance of material level height and operational stability among multiple furnace top silos, and effectively avoids the three-phase imbalance and process fluctuation parameters in the furnace caused by local material levels that are too low or too high.

[0008] Optionally, the measured material level data and estimated material level data of each of the multiple furnace top silos are acquired. Specifically, this includes: periodically collecting the original material level signals of the multiple furnace top silos; processing the original material level signals to obtain the measured material level data; acquiring in real time the unloading timestamp and corresponding unloading weight data of each unloading completion from the weighing silo, and simultaneously acquiring the feeding timestamp of the feeding device into the multiple furnace top silos; calculating cumulatively based on the feeding timestamp and the preset single feeding volume to obtain the cumulative total volume of material entering the multiple furnace top silos; and outputting the estimated material level data of the multiple furnace top silos through a balance calculation model based on the principle of material conservation, according to the unloading weight data, the cumulative total volume of material entering the silos, and the preset material calibration bulk density.

[0009] By adopting the above technical solution, the dual data acquisition mechanism, which combines periodic signal acquisition with a balance calculation model based on logistics tracking, can not only capture instantaneous states through actual measurement, but also calculate material level changes that are not affected by short-term interference from the measurement environment through the principle of material conservation. This can improve the monitoring dimensions of the actual material inventory in the silo and the reliability of the data source.

[0010] Optionally, the measured material level data and the estimated material level data are fused to generate fused material level data. Specifically, this includes: calculating the first rate of change of the measured material level data in adjacent sampling periods and the second rate of change of the estimated material level data in adjacent calculation periods; comparing the first rate of change and the second rate of change with preset normal rate of change ranges and obtaining comparison results; dynamically determining the first confidence weight and the second confidence weight based on the comparison results; weighting the measured material level data of the current period according to the first confidence weight and weighting the estimated material level data of the current period according to the second confidence weight; and adding the weighted measured material level data and the weighted estimated material level data to obtain the fused material level data of the current period.

[0011] By adopting the above technical solution, this application introduces dynamic weighting based on the reliability assessment of data change rate, which can automatically adjust the confidence weight of the measured and estimated data according to their respective instantaneous performance, thereby filtering out abnormal data points caused by temporary sensor failure or instantaneous error of logistics model, effectively improving the accuracy, anti-interference ability and adaptability to complex and harsh working conditions of the final fused material level value.

[0012] Optionally, the fused material level data is processed based on a preset time series prediction model to predict the time required for the material level in each furnace top silo to reach a preset low material level threshold. Specifically, this includes: for each furnace top silo, extracting fused material level data for a first preset continuous time period in chronological order to determine a material level change sample sequence; inputting the material level change sample sequence into a trained time series prediction model, and predicting the material level value at a second preset continuous time point from the current moment by iteratively calling the time series prediction model to generate a material level prediction curve; comparing the material level prediction curve with the preset low material level threshold to determine the first timestamp when the material level prediction curve first falls below the preset low material level threshold; and generating the required time based on the time difference between the first timestamp and the current moment.

[0013] By adopting the above technical solutions and using the trained time series model to deeply mine and extrapolate the historical fusion material location data, it is possible to accurately predict the future time when each material silo will reach the safety lower limit. This elevates the judgment of replenishment demand from "whether there is a shortage of materials now" to the predictive level of "when there will be a shortage of materials in the future". This provides key time window information for realizing forward scheduling and resource pre-allocation, and can improve the foresight and initiative of production management.

[0014] Optionally, the feeding rate and feeding task sequence of the feeding device are obtained, specifically including: obtaining the feeding rate and production shift plan, and generating an initial feeding instruction set; parsing the initial feeding instruction set to generate standardized task units; wherein, the standardized task unit includes the furnace top hopper number, the planned execution time, and the rated feeding quantity; arranging all standardized task units in the order of the planned execution time to generate the feeding task sequence.

[0015] By adopting the above technical solution, the macro production plan is parsed and standardized into independent task units with clear attributes (target, time, quantity), and initially sorted according to time. This constructs a clear, structured task input queue that can be directly processed by a computer, providing a stable and orderly optimization framework for subsequent steps. This enables efficient and unambiguous connection between upper-level production instructions and lower-level dynamic execution.

[0016] Optionally, based on the timing of the replenishment tasks and the replenishment rate, the required time and preset high level threshold for each furnace top silo to reach a preset low level threshold are processed to sort and insert tasks in the replenishment task timing sequence, generating an optimized replenishment task timing sequence. Specifically, this includes: generating a first scheduling task queue based on the replenishment task timing sequence, creating a corresponding level prediction copy for each furnace top silo, and initializing the level prediction copy by fusing level data; setting an urgency index for the corresponding furnace top silo based on the required time; wherein the urgency index is negatively correlated with the required time; arranging the first scheduling task queue in descending order based on the urgency index to sort tasks and generate a second scheduling task queue; simulating the execution of the second scheduling task queue, and inserting tasks into the second scheduling task queue according to the required time and the preset high level threshold to generate an optimized replenishment task sequence.

[0017] By adopting the above technical solution, the predicted "material shortage time" is quantified into an "urgency index," and this index is used as the core to re-prioritize the initial task queue. Simultaneously, virtual material location replicas are created for full-process simulation and rehearsal. This frees replenishment scheduling decisions from the constraints of fixed timetables, transforming them into a dynamic and intelligent optimization process closely aligned with the real-time consumption rate and urgency of each warehouse, thus improving the overall rationality of the control strategy.

[0018] Optionally, the second scheduling task queue is simulated and tasks are inserted into it according to the required duration and a preset high material level threshold to generate an optimized replenishment task sequence. Specifically, this includes: simulating execution based on the second scheduling task queue; calculating the task execution duration based on the rated replenishment quantity and replenishment rate; updating the material level prediction replicas based on the task execution duration; during the simulation, if the required duration for a furnace top silo is less than the preset threshold, a high-priority emergency replenishment task is generated for that silo; searching for an insertable gap in the second scheduling task queue for the emergency replenishment task; wherein, after insertion, the material level of all material level prediction replicas must be less than or equal to the preset high material level threshold at any time during the simulation; and outputting the second scheduling task queue from the simulation to generate an optimized replenishment task sequence.

[0019] By employing the above technical solution, the sorted task queue is "pre-executed," and during this process, the virtual material levels in each warehouse are dynamically monitored. If a new urgent need arises in the simulation, a high-priority task is immediately generated, and a compliant gap is searched in the queue for placement. This method, to a certain extent, ensures that the final optimized sequence not only prioritizes current urgent tasks but also accommodates future unexpected needs.

[0020] Optionally, an emergency index is set for the furnace top hopper corresponding to the required duration based on the required duration. Specifically, this includes: setting a baseline emergency value; inputting the required duration of each furnace top hopper into a preset index calculation function to generate an emergency value; wherein the index calculation function is configured such that the emergency value decreases monotonically as the required duration increases; and adding the baseline emergency value to the emergency value to determine the emergency index for each furnace top hopper.

[0021] Optionally, in the second scheduling task queue, a suitable gap is searched for the emergency replenishment task. This includes: scanning the time interval between every two adjacent tasks in the second scheduling task queue in task order; calculating the required execution time based on the rated replenishment amount and replenishment rate of the emergency replenishment task; filtering out time intervals greater than or equal to the required execution time from all scanned time intervals to generate a candidate gap set; and selecting the gap with the earliest start time from the candidate gap set as the suitable gap for the emergency replenishment task.

[0022] By adopting the above technical solution, the gap scanning and filtering mechanism finds placement positions for emergency tasks. It calculates the required duration of the task and matches it with idle time windows in the queue, prioritizing the earliest available gap. This strategy meets the requirements for rapid response to emergency tasks while minimizing disruption to the original planned sequence, optimizing equipment utilization, and improving the ability to handle emergencies in the production process in a timely and efficient manner.

[0023] Optionally, the method further includes: continuously acquiring the instantaneous material level drop rate during the unreplenished stage of the furnace top silo; if the instantaneous material level drop rate is lower than the predefined unobstructed discharge threshold based on the geometry of the furnace top silo for multiple consecutive monitoring cycles, it is determined that material flow obstruction has occurred in the furnace top silo; vibrating the furnace top silo according to a preset vibration time and preset vibration intensity, and monitoring the instantaneous material level drop rate after vibration ends; if the instantaneous material level drop rate is still less than or equal to the unobstructed discharge threshold, increasing the vibration intensity and / or extending the vibration time according to a preset strategy until the instantaneous material level drop rate is greater than the unobstructed discharge threshold; issuing an early warning if the instantaneous material level drop rate is still less than or equal to the unobstructed discharge threshold within a preset handling time window.

[0024] By adopting the above technical solution, the natural descent rate of the material level is continuously monitored during the feeding interval. Once an abnormal slowdown in the rate is detected and it remains below the theoretical threshold, it is determined to be material stagnation and triggers a graded vibration program. This mechanism transforms the guarantee of smooth material flow from relying on regular manual inspections or post-event handling to a real-time, automatic preventive maintenance measure, which can alleviate problems such as material sloshing and wall adhesion to a certain extent and ensure the expected smooth flow of materials.

[0025] The material level control method for the top charge silo of a calcium carbide furnace provided in this application has at least the following technical effects: 1. By integrating multi-source data and predicting future trends, and combining dynamic optimization scheduling, the system proactively maintains a balanced material level across multiple silos, changing the traditional passive response mode and enhancing the ability to combat fluctuations in furnace material consumption and maintain process stability, thus providing control assurance for the efficient and stable operation of the calcium carbide furnace.

[0026] 2. By adopting a dual data acquisition path that combines actual measurement and extrapolation, the problem of single sensors being prone to failure in harsh environments is effectively overcome, thereby improving the reliability and adaptability of the entire control system in complex industrial scenarios.

[0027] 3. By quantifying predicted demand into an emergency index for intelligent sorting and safely inserting emergency tasks through simulation technology, replenishment scheduling can respond flexibly and accurately to real-time changes in production demand, while strictly adhering to safety boundaries, thus achieving efficient and flexible allocation of production resources and controllable risks.

[0028] 4. The introduction of real-time monitoring and intelligent vibration intervention for material feeding can proactively prevent and handle blockage faults, ensuring the stability of the downstream material supply chain to a certain extent and providing a physical basis for precise material level control and prediction. Attached Figure Description

[0029] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a method for controlling the material level in the top silo of a calcium carbide furnace, provided in an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] This application provides a material level control method for the top charge bin of a calcium carbide furnace, which solves the following technical problem: how to improve the balance of material levels in multiple top charge bins.

[0032] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0033] Figure 1 A flowchart illustrating a method for controlling the material level in the top silo of a calcium carbide furnace, provided as an embodiment of this application. Figure 1 As shown in the embodiment of this application, a method for controlling the material level in the top silo of a calcium carbide furnace specifically includes the following steps: S1. Obtain the measured material level data and the estimated material level data of each of the multiple furnace top material bins; wherein, the estimated material level data is obtained by the weighing bin and the feeding device.

[0034] S1.1 Periodically collect the original material level signals from multiple furnace top silos.

[0035] Methods for detecting the material level in the furnace top silo fall into two main categories: direct measurement and indirect measurement. Direct measurement includes non-contact radar level gauges, ultrasonic level gauges, contact steel strip level gauges, and capacitive limit switches. Indirect measurement includes weighing measurement and pressure inference measurement.

[0036] However, due to the high dust, high temperature, and highly corrosive environment in the calcium carbide furnace, contact tools are easily damaged. Therefore, this application embodiment uses a non-contact radar level gauge or ultrasonic level gauge. Taking a non-contact radar level gauge as an example, the original level signal is detected by the non-contact radar level gauge.

[0037] For example, if the M-type calcium carbide furnace includes 12 furnace top silos, a radar level gauge is installed on the top of each furnace top silo to collect the original material level signal.

[0038] In a specific example, the M calcium carbide furnace is equipped with 12 furnace top hoppers (A, B, C, D, E, F, G, H, I, J, K, L). Each hopper is equipped with a non-contact radar level gauge, with a data acquisition cycle of 5 seconds. During the acquisition phase, each non-contact radar level gauge transmits microwave pulses to the material surface below (i.e., the material surface in the furnace top hopper) and receives the echoes. Based on the time difference between transmission and reception, a linear current signal is generated, which is the original material level signal.

[0039] Understandably, the method of acquiring the raw material level signal and the acquisition cycle can be set according to the actual working environment.

[0040] S1.2. Process the raw material level signal to obtain the measured material level data.

[0041] Because of environmental factors such as dust in the furnace top hopper and high temperature of the calcium carbide furnace, the original material level signal collected in step S1.1 has errors or fluctuations. Therefore, it is necessary to smoothly convert the original material level signal into measured material level data. Referring to step S1.1, the collected signal is obtained by a non-contact radar level gauge. Therefore, analog-to-digital conversion and digital filtering should be used to process the original material level signal to obtain measured material level data.

[0042] In a specific example, taking the top silo of furnace A as an example, the original material level signal of the top silo of furnace A is processed by moving average filtering to suppress instantaneous noise introduced by dust emission or electromagnetic interference. The processed original material level signal is then compared with the calibrated range of the non-contact radar level meter (e.g., 0-10 meters corresponds to 4-20 mA) to convert it into a unit of length. For example, the processed original material level signal corresponds to 12.5 mA, and the calculated height is 3.20 meters.

[0043] Understandably, 3.20 meters cannot be directly used as the actual measured material level data of the furnace top silo at the current moment. The total height of the furnace top silo should be taken into account. For example, if the total height of the furnace top silo is 8 meters, then the actual measured material level data is 4.80 meters.

[0044] S1.3. Real-time acquisition of the unloading timestamp and corresponding unloading weight data of each unloading completion from the weighing hopper, and synchronous acquisition of the feeding timestamp of the feeding device to multiple furnace top hoppers.

[0045] The unloading timestamp and unloading weight data refer to the moment when the upstream lime weighing silo or carbon material weighing silo completes the unloading of a batch of materials and the net weight of that batch. The feeding timestamp refers to the moment when the feeding device (such as a ring feeder) moves to the feed inlet of a specific furnace top silo and completes the tilting action.

[0046] In a specific example, the system monitors the "unloading complete" digital signal from the upstream batching unit in real time. Upon receipt, the current time is immediately recorded as the unloading timestamp, and the unloading weight data is read from the weighing sensor. Simultaneously, the system uses a combination of the rotary encoder and position switch of the ring feeder to determine which specific furnace top hopper (e.g., "B furnace top hopper") is fed into each batch of material, recording the time as the feeding timestamp.

[0047] S1.4. Based on the feeding timestamp and the preset single feeding volume, calculate cumulatively to obtain the total volume of materials entering multiple furnace top silos.

[0048] The total volume of materials entering the warehouse refers to the total volume of materials entering each specific furnace top silo, starting from the beginning of the calculation period.

[0049] In a specific example, an independent volume accumulator is maintained for each of the 12 top hoppers. The preset single-feed volume is a constant pre-calibrated based on the geometric volume of each hopper of the annular feeder (e.g., 0.18 cubic meters per hopper). Whenever a feeding timestamp is detected for a specific hopper (e.g., "Top Hopper F"), the accumulator value for Top Hopper F is increased by 0.18 cubic meters. Through continuous accumulation, the total cumulative material volume data of each hopper is updated in real time.

[0050] S1.5 Based on the unloading weight data, the total volume of materials entering the warehouse, and the preset material calibration bulk density, the estimated material level data of multiple furnace top silos is output through a balance calculation model based on the principle of material conservation.

[0051] Estimated material level data refers to the estimated material level height in the furnace top silo, calculated based on the balance between inflow and consumption. Estimated material level data represents the material level deduced from a logistics perspective and is independent of measured material level data.

[0052] Calculate the net incoming mass: Net incoming mass = (Cumulative incoming volume of this silo × Material bulk density) - Cumulative unloading weight allocated to this silo. The "Cumulative unloading weight allocated to this silo" needs to be allocated to the 12 silos according to certain rules (such as the historical consumption ratio of each silo) based on the production process model.

[0053] Calculate the current material inventory volume: Current inventory volume = Initial inventory volume + Net incoming mass / Material bulk density. The initial inventory volume is calculated based on measured material level data and silo geometry at startup.

[0054] Convert to estimated material level: Based on the known cross-sectional shape and size of each silo (e.g., the diameter of a cylindrical silo), convert the current inventory volume into the material stacking height within the silo.

[0055] For cylindrical silos, the formula is: Estimated material level height = Total silo height - (Current inventory volume / (π × silo radius squared)). The height value calculated in this way is the estimated material level data.

[0056] In a specific example: Taking the furnace top silo numbered A as an example, this silo is cylindrical, with a total height of 8 meters and a diameter of 2.5 meters.

[0057] S1.1 and S1.2: The output signals of the top radar level gauge are converted and filtered to obtain the measured level data of 5.10 meters (i.e., the material surface is 2.90 meters from the top of the silo).

[0058] S1.3: Between 10:00:00 and 10:05:00, a related event was recorded: the ring feeder fed material into the top hopper of furnace A at 10:03:22 (feeding timestamp).

[0059] S1.4: The top charge silo of furnace A in this shift has received a total of 85 feedings, with a single feeding volume of 0.18 cubic meters. Therefore, the total volume of materials entering the silo is 15.30 cubic meters.

[0060] S1.5: Running the equilibrium calculation model.

[0061] The current standard bulk density of the mixture is known to be 0.95 tons per cubic meter.

[0062] Calculate the cumulative weight of goods entering the warehouse: 15.30 cubic meters × 0.95 tons / cubic meter = 14.535 tons.

[0063] Calculate the cumulative mass consumed: Based on the production data, the model calculates that the total mass of material to be consumed by the top silo of furnace A in this shift is 12.635 tons.

[0064] Calculate current inventory: Set the initial inventory of the furnace top silo at the start of this shift to 11,000 tons.

[0065] Current inventory quality = Initial inventory + Cumulative inventory received - Cumulative inventory consumed = 11.000 + 14.535 - 12.635 = 12.900 tons.

[0066] Current inventory volume = Current inventory mass / Bulk density = 12.900 tons / 0.95 tons / cubic meter ≈ 13.579 cubic meters.

[0067] Converting to estimated material level height: Based on the cylinder volume formula (volume = base area × height), the estimated material level height = current inventory volume / silo base area.

[0068] The bottom area of ​​the silo = π × r × r = 3.14 × 1.25 × 1.25 ≈ 4.906 square meters.

[0069] Estimated material level = 13.579 cubic meters / 4.906 square meters ≈ 2.77 meters.

[0070] S2. Generate fused material level data based on measured material level data and estimated material level data.

[0071] S2.1 Calculate the first rate of change of the measured material level data in adjacent sampling periods, and calculate the second rate of change of the material level data in adjacent calculation periods.

[0072] The first rate of change refers to the rate of decrease or increase of the measured material level data, obtained by dividing the change in the measured material level data by the time interval between two adjacent sampling moments, measured in meters per second. The second rate of change refers to the corresponding rate of change of the estimated material level data between two adjacent calculation cycles.

[0073] In a specific example, a historical sequence of measured and estimated material level data is continuously recorded. For silo A, the current measured material level is 4.95 meters, and the previous period's was 5.00 meters (sampling interval 5 seconds). Therefore, the first rate of change is (4.95-5.00) / 5 = -0.01 meters / second. The current estimated material level is 4.60 meters, and the previous period's was 4.65 meters (calculation interval 5 seconds). Therefore, the second rate of change is also -0.01 meters / second.

[0074] S2.2. Compare the first rate of change and the second rate of change with the preset normal rate of change range and obtain the comparison results.

[0075] The normal rate of change range is a threshold interval pre-defined based on statistical analysis of the material discharge characteristics from the top hopper during normal production of the calcium carbide furnace, for example, [-0.030, -0.005] m / s. This range defines the reasonable rate of descent of the material level under normal operating conditions with good material flowability and no blockage.

[0076] In a specific example, the calculated rate of change of the measured data and the estimated rate of change of the data are compared with the preset normal rate of change range. The comparison results are used to qualitatively determine the reliability of each data source at the current moment, and the results are usually marked as "normal" (within the range), "too slow" (the absolute value of the rate is too small, indicating a problem), "too fast" (the absolute value of the rate is too large, indicating an anomaly), or "invalid".

[0077] S2.3. Dynamically determine the weights of the first confidence level and the second confidence level based on the comparison results.

[0078] The first confidence weight and the second confidence weight are confidence coefficients assigned to the measured material level data and the estimated material level data in the period, respectively. Both are real numbers between 0 and 1, and the sum of their weights is 1.

[0079] It is understandable that the values ​​and allocations of the first and second confidence weights are based on changes in the historical operating conditions of the calcium carbide furnace.

[0080] Specifically, the system assesses the level by continuously monitoring the rate of change of both measured and estimated material level data, and their consistency with historical trends. Specifically, when the measured material level data shows stable changes and conforms to the expected consumption model (i.e., the first rate of change is stable and within the preset normal rate of change range), the measured material level data is assigned a higher weight, indicating that it is in a reliable working state. Conversely, if the measured data shows drastic fluctuations, remains unchanged for a long period, or deviates significantly and continuously from the estimated value, it indicates that the measured material level data is being interfered with by high dust, high temperature, or "material sloshing" phenomena. In this case, the weight of the measured material level data is automatically reduced, and the estimated material level data calculated based on the weighing silo unloading data and material consumption is relied upon more heavily.

[0081] In one specific embodiment, the comparison result of furnace top silo A is "normal", so the first confidence weight is determined to be 0.6 and the second confidence weight is determined to be 0.4. If the comparison result of furnace top silo C is "abnormal" (change rate -0.06 m / s), then the first confidence weight is automatically adjusted to 0.2 and the second confidence weight is adjusted to 0.8.

[0082] S2.4. Weight the measured material level data of the current period according to the first confidence level weight, and weight the estimated material level data of the current period according to the second confidence level weight.

[0083] In a specific embodiment, referring to S2.1 to S2.3, when the comparison result of the furnace top hopper A is "normal", the first confidence weight is determined to be 0.6, the second confidence weight is determined to be 0.4, and then the measured material level data and the estimated material level data are weighted.

[0084] S2.5 Add the weighted measured material level data with the weighted estimated material level data to obtain the fused material level data for the current period.

[0085] The merged material level data is the final material level estimate synthesized after dynamic reliability weighting.

[0086] In a specific example, the two weighted values ​​obtained in the previous step are added together, and the sum is the fusion material level data for the current cycle.

[0087] In a complete embodiment of this step, it is assumed that the Nth fusion cycle is being performed on the aforementioned A furnace top silo (cylindrical, 8 meters in total height), and the following background data is known: The measured material level for the current cycle is 4.95 meters, and the previous cycle was 5.00 meters (sampling interval 5 seconds).

[0088] The current cycle's estimated material level is 4.60 meters, and the previous cycle's was 4.65 meters (calculation interval 5 seconds).

[0089] The preset normal rate of change range is [-0.03, -0.005] m / s.

[0090] S2.1 and S2.2: The measured rate of change is (5.10 - 5.15) / 5 = -0.01 m / s. This value falls within the normal range, and the comparison result is "normal". The calculated rate of change is (4.60 - 4.65) / 5 = -0.01 m / s. This value also falls within the normal range, and the comparison result is also "normal".

[0091] S2.3: Since the rate of change is consistent and normal, according to the preset rules, the direct measurement data is tended to be assigned a slightly higher confidence level.

[0092] Therefore, the weight of the first confidence level is determined to be 0.6, and the weight of the second confidence level is determined to be 0.4.

[0093] S2.4 and S2.5: Weighted measured value = 4.95 meters × 0.6 = 2.970 meters; Weighted estimated value = 4.60 meters × 0.4 = 1.840 meters; The combined material level data is 2.970 meters + 1.840 meters = 4.810 meters.

[0094] S3. Based on a preset time series prediction model, process the fused material level data to predict the time required for the material level in each furnace top hopper to reach the preset low material level threshold.

[0095] S3.1 For each furnace top hopper, extract the fused material level data for the first preset continuous time in chronological order to determine the material level change sample sequence.

[0096] A material level change sample sequence refers to a continuous set of fused material level data extracted in chronological order from the historical database of a furnace top silo, reflecting the actual consumption behavior and trend of the silo in a recent period.

[0097] In a specific example, starting from the current moment, a pre-defined time length (i.e., a first preset continuous time, such as "the past 30 minutes") is traced back. Then, all fused material level data recorded within this time window at fixed sampling intervals (e.g., one point every 5 seconds) are extracted. These data points are then strictly sorted by timestamp to form the material level change sample sequence used for this prediction.

[0098] For example, for the top silo of furnace A, a sequence containing 360 data points (corresponding to the past 30 minutes) is obtained, showing the process of the material level gradually decreasing from 5.10 meters to 4.40 meters.

[0099] S3.2 Input the material level change sample sequence into the trained time series prediction model, and predict the material level value at the second preset continuous time point from the current moment by iteratively calling the time series prediction model, so as to generate the material level prediction curve.

[0100] A time series forecasting model refers to a mathematical model or algorithm that predicts future values ​​by analyzing patterns in historical data sequences. A material level prediction curve refers to the trend line output by the aforementioned model, consisting of a series of future time points (intervals of 5 seconds) and their corresponding predicted material level values.

[0101] In a specific example, a predictive model is built-in or invoked, pre-trained using a large amount of historical production data. The material level change sample sequence prepared in S3.1 is fed into the model as input. The model first analyzes the inherent patterns of the sequence (e.g., determining whether it is a linear or exponential decrease, and whether the rate of decrease is stable).

[0102] Then, the model begins multi-step forward prediction: first, it predicts the material level at the next time point (e.g., 5 seconds later), then incorporates this predicted value into the sequence, and then predicts the time point after that, iterating in this way. This process continues until the prediction covers a second preset continuous time period extending into the future from the current moment (e.g., "the next 120 minutes"). Finally, by connecting all the predicted future time points with the material level values, a material level prediction curve extending into the future from the current moment is formed.

[0103] S3.3. Compare the material level prediction curve with the preset low material level threshold to determine the first timestamp when the material level prediction curve first falls below the preset low material level threshold.

[0104] The preset low material level threshold refers to the safe lower limit of the material level set to prevent material shortages and ensure production continuity. It is a fixed height value (unit: meters). The first timestamp refers to the future moment when the predicted material level value on the material level prediction curve first equals or falls below the preset low material level threshold.

[0105] In a specific example, the preset low material level threshold set in the furnace top silo is read (for example, for a silo with a total height of 8 meters, the low material level threshold is set to 1.5 meters, meaning that the risk of material shortage is extremely high when the material level is below this height). Then, along the material level prediction curve generated in S3.2, the system scans forward point by point from the current moment, comparing the predicted material level at each point with the threshold. When the first point where the predicted material level value is less than or equal to 1.5 meters is scanned, the corresponding future time is recorded; this time is the first timestamp. For example, it is determined that the prediction curve will first reach 1.5 meters "95 minutes after the current moment".

[0106] S3.4. Generate the required duration based on the time difference between the first timestamp and the current time.

[0107] The required duration refers to the estimated time from the current moment until the predicted material level will drop to the low material level threshold.

[0108] In a specific example, the current time is obtained, and the difference between it and the first timestamp determined in S3.3 is calculated. This difference is the required duration. For example, if the current time is 10:00:00 and the first timestamp is the predicted 11:35:00, then the required duration is 95 minutes.

[0109] In a specific embodiment of this step: taking the top silo of furnace A (total height 8 meters) as an example, the current time is 10:00.

[0110] S3.1: Extract the fusion material level data from the top silo of furnace A within the first preset continuous time period (30 minutes) from 09:30 to 10:00, one point per minute, for a total of 361 points. This sequence shows that the material level steadily decreases from 4.50 meters to 4.00 meters, exhibiting a good linear downward trend.

[0111] S3.2: Input the sequence of these 361 data points into a trained time series prediction model (e.g., a linear regression-based predictor). The model identifies a downward trend and, starting from the current material level of 4.00 meters, iteratively predicts the material level per minute over a second preset continuous time period (120 minutes). The final result is a material level prediction curve showing that the material level will linearly decrease from 4.00 meters, approaching 0 meters after 125 minutes.

[0112] S3.3: The preset low material level threshold of the top silo of furnace A is known to be 1.50 meters. The predicted material level curve is compared with the 1.50-meter line. Through calculation, it is determined that the predicted curve will first drop to 1.50 meters in approximately 85 minutes (i.e., around 11:25). This future moment (11:25) is the first timestamp.

[0113] S3.4: Calculate the time difference between the current time (10:00) and the first timestamp (11:25) to obtain the required duration of 85 minutes.

[0114] S4. Obtain the feeding rate and feeding task sequence of the feeding device.

[0115] S4.1 Obtain the replenishment rate and production shift plan, and generate an initial set of replenishment instructions.

[0116] The feeding rate refers to the maximum capacity of a feeding device (e.g., a fixed belt scale for conveying mixtures to a furnace top silo, a mobile feeding vehicle, or other devices with feeding capabilities) to stably convey materials per unit time under rated operating conditions. The feeding rate is an inherent performance parameter determined by the equipment itself.

[0117] A production shift plan refers to a production target issued by the manufacturing execution or production management department of a calcium carbide production enterprise, targeting a specific calcium carbide furnace for a defined period of time (usually a production shift, such as 8 hours). It specifies the planned output, the required proportions of each raw material (lime, carbon material), and the total material consumption budget for that period.

[0118] The initial replenishment instruction set refers to a preliminary list of instructions on "when to replenish which bin and how much material" generated, based solely on the total material demand of the production shift plan, the operating efficiency of the replenishment device, and the principle of balanced production, without considering the real-time material level differences of each furnace top bin.

[0119] In a specific example, the rated conveying capacity parameters of the feeding device are read and loaded into memory as a constant feeding rate value. Simultaneously, the currently active production shift plan is obtained from upper-level production management via a standard industrial data interface (such as OPCUA).

[0120] Subsequently, the scheduling engine performs preliminary calculations based on the total material requirements in the shift plan, the total number of the 12 furnace top silos, and the overall safety stock level required to maintain continuous production. The calculations typically follow the principle of "total distribution and periodic replenishment".

[0121] For example, if a shift plan requires the consumption of 150 tons of mixed feedstock over the next two hours, then the plan is to replenish all 12 silos once within those two hours. The total of 150 tons is evenly distributed across 12 replenishment operations, resulting in an initial replenishment amount for each operation (e.g., 12.5 tons). Then, based on the production rhythm and equipment operating efficiency, an initial, equally spaced planned execution time is assigned to these 12 replenishment operations (e.g., starting from the current moment, a replenishment is scheduled every 10 minutes). The resulting 12 replenishment instructions constitute the initial replenishment instruction set.

[0122] S4.2. Parse the initial feeding instruction set to generate standardized task units; wherein, the standardized task unit includes the furnace top hopper number, the planned execution time, and the rated feeding quantity.

[0123] A standardized task unit is a minimal structured data object defined to facilitate computer processing and scheduling, used to describe an independent replenishment operation.

[0124] The furnace top silo number refers to a unique code that identifies one of the 12 furnace top silos, such as the letter AL.

[0125] The planned execution time refers to the expected start time initially arranged for this replenishment operation in the initial replenishment instruction set.

[0126] Rated replenishment quantity refers to the planned material replenishment quantity set for a single replenishment operation based on initial calculations.

[0127] In a specific example, each text or structured instruction in the initial replenishment instruction set is parsed line by line. For each instruction, key information is extracted and instantiated into a data object with a uniform format, i.e., a standardized task unit.

[0128] For example, an instruction "Replenish 12.5 tons of material to the top charge silo of furnace A at 10:30" will be parsed and generated into the following task unit: {“Hopper Number”: “A”, “Planned Execution Time”: “10:30:00”, “Rated Replenishment Quantity”: 12.5} Perform this operation for each instruction in the instruction set, thereby generating a set of standardized task units that are equal in number and have the same format.

[0129] S4.3 Arrange all standardized task units in chronological order of planned execution time to generate a replenishment task sequence.

[0130] The replenishment task sequence refers to an ordered task list formed by sorting all standardized task units according to their internal "planned execution time" field from morning to night. It represents a raw replenishment operation schedule based on fixed-time triggers.

[0131] All generated standardized task units are loaded into a list or queue data structure. Then, a sorting algorithm is used to sort the entire set using the "planned execution time" of each task unit as the sort key. After sorting, a task queue is obtained, where the task at the top of the queue is the one with the earliest planned execution time.

[0132] For example, the sorted queue order is: [Task A (10:30), Task B (10:40), Task F (10:50), Task D (11:00), ...]. This queue is the initial replenishment task sequence. It is important to emphasize that this sequence does not yet incorporate any information about the real-time urgency of each silo (reflected by the "required duration" predicted in step S3). Therefore, there is an unreasonable situation where "the silo most in need of replenishment is ranked last," which is precisely the core problem that subsequent optimization steps aim to solve.

[0133] In a specific example: Assuming the current time is 10:00 AM, start executing S4.

[0134] S4.1: Query the database to determine that the rated feeding rate of the feeding belt scale serving the M-type calcium carbide furnace is 120 tons / hour. The received production shift plan shows that production needs to be guaranteed between 10:00 and 12:00, with an estimated consumption of approximately 144 tons of mixed material.

[0135] To ensure two hours of production, it is planned to replenish each of the 12 silos once during this period. The initial rated replenishment amount is set at 144 tons / 12 = 12 tons. The first replenishment will be scheduled at 10:10, with a 10-minute interval, generating an initial replenishment instruction set containing 12 instructions.

[0136] S4.2: For example, the instruction for the top charge hopper of furnace A is converted into a standardized task unit: {Number: "A", Time: "10:10:00", Recharge amount: 12}.

[0137] The instruction for hopper B is converted to: {Number: "B", Time: "10:20:00", Replenishment Quantity: 12}.

[0138] This process continues until the top hopper of furnace L is reached.

[0139] S4.3: Sort these 12 task units according to "planned execution time" to form the initial replenishment task sequence: [Task A (10:10), Task B (10:20), Task C (10:30), Task D (10:40), Task E (10:50), Task F (11:00), Task G (11:10), Task H (11:20), Task I (11:30), Task J (11:40), Task K (11:50), Task L (12:00)].

[0140] S5. Based on the feeding task timing and feeding rate, process the required time and preset high material level threshold for each furnace top hopper to reach the preset low material level threshold, so as to sort and insert the feeding task timing and generate an optimized feeding task timing.

[0141] S5.1 Based on the timing of the material replenishment task, a first scheduling task queue is generated, and a corresponding material level prediction copy is created for each furnace top silo. The material level prediction copy is initialized by merging material level data.

[0142] The first scheduling task queue refers to loading the replenishment task sequence (i.e., a series of standardized task units) generated in S4.3 and ordered by the initial planned execution time into a data structure (such as a list or queue) that can be processed sequentially by the scheduling algorithm. This queue reflects the original replenishment plan without optimization adjustments.

[0143] The predicted material level copy refers to a virtual material level state variable maintained separately for each furnace top silo within the scheduling system. It is used to track and predict the material level changes of the silo at various future times during simulated replenishment tasks, and is the core basis for scheduling decisions and conflict checking.

[0144] Initialization refers to assigning an initial value to each furnace top silo's predicted material level copy before scheduling optimization begins. This initial value uses the latest calculated fused material level data at the current moment, ensuring that the starting point of the simulation prediction is consistent with the actual state.

[0145] In a specific example, the current time is 10:00 AM. The initial replenishment task sequence has been generated according to step S4, containing 12 standardized task units, ordered by time: Standardized Task Unit A (10:10), Standardized Task Unit B (10:20), Standardized Task Unit C (10:30), Standardized Task Unit D (10:40), Standardized Task Unit E (10:50), Standardized Task Unit F (11:00), Standardized Task Unit G (11:10), Standardized Task Unit H (11:20), Standardized Task Unit I (11:30), Standardized Task Unit J (11:40), Standardized Task Unit K (11:50), and Standardized Task Unit L (12:00). This list is first loaded to form the first scheduling task queue. Next, 12 independent material level prediction copies are created for each of the 12 furnace top silos (A to L).

[0146] Then, the latest calculated fused material level data for each silo in step S2 is read: Silo A (top silo) is 4.401 meters, Silo B is assumed to be 5.200 meters, Silo C is assumed to be 3.800 meters, Silo D is assumed to be 4.600 meters, Silo E is assumed to be 5.000 meters, Silo F is assumed to be 4.100 meters, Silo G is assumed to be 4.900 meters, Silo H is assumed to be 4.300 meters, Silo I is assumed to be 4.700 meters, Silo J is assumed to be 4.000 meters, Silo K is assumed to be 4.500 meters, and Silo L is assumed to be 3.900 meters. These data are then used to initialize the corresponding material level prediction replicas, i.e., the value of "Prediction Replica-A" is set to 4.401 meters, the value of "Prediction Replica-B" is set to 5.200 meters, and so on, thereby establishing a baseline state for simulating future material level changes.

[0147] S5.2 Set an emergency index for the corresponding furnace top silo based on the required duration; wherein, the emergency index is negatively correlated with the required duration.

[0148] The required duration refers to the time (e.g., minutes, seconds) required for the material level in a certain furnace top silo to drop to a preset low material level threshold from the current moment, as predicted by step S3. The shorter the duration, the closer the silo is to a material shortage state, and the more urgent the demand.

[0149] The urgency index is a numerical indicator used to quantify the urgency of the need for replenishing the top silo. This index is negatively correlated with the required time; that is, the shorter the required time, the higher the urgency index, indicating that the silo should receive higher priority in scheduling.

[0150] S5.2.1 Set the baseline emergency value.

[0151] The baseline emergency value refers to a preset constant used as the basis or offset for calculating the emergency index, ensuring that the emergency index of all silos is calculated from a reasonable baseline level, which facilitates subsequent comparison and sorting.

[0152] S5.2.2 Input the required duration of each furnace top hopper into a preset exponential calculation function to generate an emergency value; wherein, the exponential calculation function is configured such that the emergency value decreases monotonically as the required duration increases.

[0153] An exponential calculation function refers to a predefined mathematical relationship or mapping rule, whose input is the required duration (T) and output is the urgent value (U). This function is designed to be monotonically decreasing, meaning that as T increases, U decreases.

[0154] The emergency value refers to an intermediate value calculated using an exponential function, which is simply converted from the required duration and reflects the urgency of the duration itself.

[0155] In a specific example, the preset baseline emergency value is 10. For the exponential calculation function, an inverse proportional function is used: emergency value U = 100 / (T + 5), where T is the required duration (minutes). This function satisfies the monotonically decreasing characteristic that "as T increases, U decreases". Assume that the required durations for each silo predicted by S3 are as follows: Silo A (top furnace) 85 minutes, Silo B 150 minutes, Silo C 45 minutes, Silo D 110 minutes, Silo E 130 minutes, Silo F 40 minutes, Silo G 120 minutes, Silo H 70 minutes, Silo I 100 minutes, Silo J 55 minutes, Silo K 90 minutes, and Silo L 140 minutes. Substitute these values ​​into the function for calculation: A furnace top silo: U A =100 / (85+5)≈1.11; B furnace top silo: U B =100 / (150+5)≈0.65; C Furnace Top Bin: U C =100 / (45+5)=2.00; F Furnace top charge hopper (as an emergency case): U F =100 / (40+5)≈2.22.

[0156] Understandably, the constant 100 is used to adjust the dimensions and range of the emergency value, and the constant 5 is used to prevent the denominator from being too small when the required duration T approaches zero. Both are adjustable parameters preset according to the actual scheduling sensitivity requirements.

[0157] The calculated values ​​are the emergency values ​​for each silo. It can be seen that silos C (45 minutes) and F (40 minutes), which have shorter required times, have obtained higher emergency values.

[0158] S5.2.3 Add the baseline emergency value to the emergency value to determine the emergency index for each furnace top hopper.

[0159] In a specific example, continuing from the previous one, the baseline emergency value of 10 is added to the emergency values ​​of each silo to obtain the final emergency index: Emergency index of furnace top silo A = 10 + 1.11 = 11.11; Emergency index of top charge hopper B = 10 + 0.65 = 10.65; Emergency index of furnace top charge bin C = 10 + 2.00 = 12.00; Emergency index of furnace top silo = 10 + 2.22 = 12.22; ... (Other silos are calculated accordingly).

[0160] At this point, each silo has been assigned a quantified urgency index. Silo F has the highest index (12.22), indicating that its replenishment needs are the most urgent; Silo B has the lowest index (10.65), indicating that its needs are relatively the least urgent.

[0161] S5.3. Arrange the first scheduling task queue in descending order based on the urgency index to sort the tasks and generate the second scheduling task queue.

[0162] Task sequencing refers to the process of readjusting the order of tasks in the queue based on the urgency index of the target silo corresponding to each replenishment task.

[0163] The second scheduling task queue refers to the new task queue generated after sorting in descending order based on the urgency index. In this queue, tasks are arranged from highest to lowest urgency according to the target silo, so that the most urgent replenishment tasks are prioritized.

[0164] In a specific example, each task unit in the first scheduling task queue is scanned, and its "bin number" is associated with the urgency index calculated in S5.2 for that bin. Then, all tasks are sorted according to the rule of urgency index from high to low (i.e., descending order). After sorting, the task order becomes: F (urgency index 12.22), C (12.00), A (11.11), J (assumed to be 11.05), H (assumed to be 10.98), ..., B (10.65). Tasks from other bins are also inserted into their corresponding positions according to their respective indices. The resulting new queue is the second scheduling task queue. At this time, the order of tasks no longer depends on the initial planned time (e.g., bin F was originally scheduled for 11:00), but is placed at the front because it is the most urgent.

[0165] S5.4 Simulate the execution of the second scheduling task queue, and insert tasks into the second scheduling task queue according to the required duration and the preset high material level threshold to generate an optimized replenishment task sequence.

[0166] Simulated execution refers to the process of virtually executing each material replenishment task in the order of the second scheduling task queue before generating actual control instructions, and dynamically updating the status of each material location prediction copy to evaluate the effectiveness of the scheduling scheme in the future time period.

[0167] The preset high material level threshold is a fixed upper limit for the safe material level set to prevent the hopper from overflowing or affecting the uniformity of the fabric. During the simulation, it must be ensured that the predicted material level of any hopper does not exceed this threshold at any time.

[0168] Task insertion refers to the process during simulation execution where, when an emergency situation in the material silo is identified (based on the required duration) requiring immediate intervention, a new, high-priority emergency replenishment task is inserted into the existing second scheduling task queue at a suitable location.

[0169] Optimized replenishment task sequence: refers to the final list of replenishment tasks that can be actually executed after a series of optimization operations such as simulation execution, task sorting and task insertion, which takes into account the urgency of each warehouse and meets the high material level constraint.

[0170] S5.4.1. Simulate execution based on the second scheduling task queue, calculate the task execution time according to the rated replenishment amount and replenishment rate, and update the material level prediction copy based on the task execution time.

[0171] Task execution time refers to the estimated time required to complete one material replenishment task. The calculation formula is: Task execution time = Rated material replenishment amount / Material replenishment rate.

[0172] Updating the material level prediction copy refers to calculating the changes in the material level of the silo during and after the execution of a certain replenishment task, based on the target silo, replenishment amount, and execution duration of the task, and modifying the value of its material level prediction copy accordingly.

[0173] In a specific example, the first task in the second scheduling task queue is processed: replenishing silo F. The rated replenishment capacity for this task is 12 tons, and the replenishment rate is 120 tons / hour. The task execution time is calculated as: 12 tons / 2 tons / minute = 6 minutes. Assume the simulation starts at virtual time 10:00, and the initial predicted level replica of silo F is 4.100 meters. The simulation executes replenishment of silo F from 10:00 to 10:06. During this period, the level in silo F rises due to replenishment. Assuming that, based on the silo's cross-sectional area, the replenishment rate corresponds to a level rise rate of approximately 0.5 meters / minute, then when replenishment is completed at 10:06, the predicted level replica of silo F will be updated to 4.100 + 0.5 * 6 = 7.100 meters.

[0174] During the simulation, real-time constraint checks immediately revealed that the updated predicted value of 7.100 meters exceeded the preset high material level threshold of 7.0 meters. This indicates that continuing with the original rated replenishment amount would result in a safety violation.

[0175] Therefore, while ensuring that the replenishment task can alleviate the emergency, the maximum allowable replenishment volume for this replenishment was recalculated. Based on the threshold, this replenishment can raise the material level to a maximum of 7.0 meters, or a maximum allowable lift of 2.9 meters. Based on an ascent rate of 0.5 meters per minute, the maximum allowable execution time is 5.8 minutes, corresponding to a maximum allowable replenishment volume of approximately 11.6 tons.

[0176] The task for silo F is dynamically replaced from the original task (12 tons, 6 minutes) to the adjusted task (11.6 tons, 5.8 minutes), and the analog clock and level prediction copy are updated. The analog clock advances to 10:05:48, and then the next task in the queue is processed.

[0177] S5.4.2 During the simulation, if the required time for the top hopper is less than the preset threshold, a high-priority emergency replenishment task will be generated for that top hopper.

[0178] The preset threshold refers to a time threshold (e.g., 30 minutes) used to determine whether a silo is in an "emergency" state. If the required time for a silo is less than this threshold, it means that it will soon face the risk of material shortage and needs to be replenished immediately.

[0179] Emergency replenishment tasks refer to replenishment instruction units that are dynamically generated to deal with the above-mentioned emergency situations. Their contents (target silo, rated replenishment quantity) are related to the silo identified as emergency and are marked with a high priority.

[0180] In a specific example, the required time for all furnace top silos was continuously monitored during the simulation. The preset emergency threshold was 30 minutes. At virtual time 10:25, it was found that silo H (whose required time was 70 minutes and was not initially at the top of the immediate execution queue) had, after approximately 25 minutes of simulation consumption, seen its re-predicted required time, based on the latest simulation status, shorten to only 28 minutes, less than the preset 30-minute threshold. Silo H was immediately declared to be in an emergency state. Subsequently, a high-priority emergency replenishment task was generated, targeting silo H, with a standard replenishment amount of 12 tons, and marked with the priority "emergency".

[0181] S5.4.3 In the second scheduling task queue, search for an insertable gap for the emergency replenishment task; wherein, after the gap is inserted, during the simulation execution, the material level of all material level prediction copies is less than or equal to the preset high material level threshold at any time.

[0182] An insertable gap refers to the time interval between two adjacent tasks in the existing second scheduling task queue, where the interval is sufficient to accommodate the time required for the execution of a new emergency task, and the insertion does not cause the predicted material level of any silo to exceed its preset high material level threshold.

[0183] S5.4.3.1 Scan the time interval between every two adjacent tasks in the second scheduling task queue in task order.

[0184] In a specific example, the current virtual time is 10:25. The next few tasks in the second scheduling task queue are, in order: Task F (executed, ending at 10:06), Task C (simulation estimated start at 10:06, ending at 10:12), Task A (simulation estimated start at 10:12, ending at 10:18), Task J (simulation estimated start at 10:18, ending at 10:24), Task H (originally scheduled, simulation estimated start at 10:30...)... The queue is scanned in task order, and the time interval between the end and start of each two adjacent tasks is calculated. For example, the interval between the end time of Task J (10:24) and the start time of the next task (H, the original task in the silo) (10:30) is 6 minutes.

[0185] S5.4.3.2 Calculate the required execution time based on the rated replenishment quantity and replenishment rate of the emergency replenishment task.

[0186] In a specific example, an emergency replenishment task is generated for silo H, with a rated replenishment volume of 12 tons and a replenishment rate of 2 tons / minute. The required execution time for this emergency task is calculated as: 12 tons ÷ 2 tons / minute = 6 minutes.

[0187] S5.4.3.3 From all the scanned time intervals, filter out time intervals that are greater than or equal to the required execution time to generate a candidate gap set.

[0188] In a specific example, all time intervals scanned from S5.4.3.1 are compared with the required execution time (6 minutes) of the urgent task. Intervals greater than or equal to 6 minutes are selected. Suppose the scan finds a 6-minute interval between task J (ending at 10:24) and task H (originally scheduled to start at 10:30). This interval is added to the candidate gap set.

[0189] S5.4.3.4 Select the gap with the earliest start time from the candidate gap set as the insertable gap for the emergency replenishment task.

[0190] In a specific example, examine the candidate gap set, which includes gap 1 (10:24–10:30). Select the gap with the earliest start time, gap 1 (starting at 10:24), as the insertable gap for this emergency replenishment task. This means that the emergency replenishment task for silo H is scheduled to begin at 10:24 and last for 6 minutes until 10:30.

[0191] S5.4.4 Output the second scheduling task queue through the simulated execution process to generate an optimized replenishment task timing.

[0192] In a specific example, the entire simulation execution process was completed, including processing all tasks in the original queue and successfully inserting the emergency replenishment task for silo H at 10:24. Throughout the simulation, continuous verification was performed to ensure that after the insertion of the emergency task, the material level prediction copies of all silos did not exceed their respective preset high material level thresholds (7.0 meters) at any simulation time. After successful verification, the final verified task sequence list containing the inserted emergency task was output. This list is an optimized replenishment task sequence based on the original plan, incorporating real-time urgency sorting and dynamic emergency insertion. Its order is similar to: F (10:00-10:06), C (10:06-10:12), A (10:12-10:18), J (10:18-10:24), H (emergency insertion, 10:24-10:30), ... (subsequent tasks are the original H and other tasks).

[0193] S6. Replenish multiple furnace top hoppers based on the optimized replenishment task timing.

[0194] In a specific example, after obtaining the optimized replenishment task sequence generated in step S5, the scheduler translates it into executable control instructions. At 10:00 AM, an instruction is first sent to the replenishment device to replenish silo F. Then, at 10:06, 10:12, and 10:18, replenishments are executed for silos C, A, and J, respectively. Crucially, at 10:24, instead of immediately executing the next task in the original plan, the emergency replenishment task for silo H is prioritized based on the optimized sequence. Afterward, subsequent optimized tasks continue to be executed. By following the optimized replenishment task sequence, proactive and predictive replenishment scheduling is achieved, ensuring that the material levels in all furnace top silos remain balanced while meeting production needs. This effectively addresses sudden emergencies such as those involving silo H, thereby achieving the technical effect of improving the balance of material levels in multiple furnace top silos.

[0195] This application also includes the following steps: A1. During the unreplenished stage of the top silo, continuously obtain the instantaneous rate of material level decline; A2. If the instantaneous material level drop rate is lower than the predefined unobstructed material discharge threshold based on the geometry of the furnace top silo for multiple consecutive monitoring cycles, it is determined that material flow obstruction has occurred in the furnace top silo. A3. Vibrate the furnace top hopper according to the preset vibration time and preset vibration intensity, and monitor the instantaneous material level drop rate after vibration. If the instantaneous material level drop rate is still less than or equal to the material unobstructed threshold, then increase the vibration intensity and / or extend the vibration time according to the preset strategy until the instantaneous material level drop rate is greater than the material unobstructed threshold. A4. If the instantaneous material level drop rate is still less than or equal to the material discharge smoothness threshold within the preset processing time window, an early warning will be issued.

[0196] A1: The "no-replenishment phase" refers to a period in which a furnace top silo is neither receiving material from the replenishment device nor in a waiting state where its predicted material level is marked as "about to rise" due to a planned replenishment task. In other words, it's a period of pure consumption where the material level should be continuously decreasing.

[0197] Instantaneous level drop rate: During the non-replenishment phase, the decrease in the material level height in the furnace top hopper per unit time, typically measured in meters per second or meters per minute. This rate is obtained by sampling and fusing level data (from step S2) at high frequency and calculating the difference in their changes.

[0198] In a specific example, let's take the top silo of furnace A as an example. At the current time, the top silo of furnace A is not in the task queue to be executed in the optimized replenishment task sequence (generated in S5) and is in the unreplenished stage. The latest fused material level data of the top silo of furnace A is read every 5 seconds. Assume that the material level is read as 4.20 meters at time t1 and as 4.18 meters at time t2 (5 seconds later). Then the instantaneous material level drop rate during this time period is calculated as: (4.18-4.20) meters / (10 / 60) minutes = -0.12 meters / minute (the negative sign indicates a drop).

[0199] A2: The unobstructed material flow threshold refers to a minimum acceptable rate of material descent, pre-set based on the specific geometric structure of the furnace top silo (such as outlet size, silo wall angle, etc.) and material characteristics (such as particle size and moisture content), through theoretical calculations or statistical analysis of historical normal operation data. When the actual descent rate consistently falls below this threshold, it indicates that material flow is obstructed, with issues such as "material buildup," wall adhesion, or localized blockage.

[0200] Multiple consecutive monitoring cycles: A judgment condition set to eliminate instantaneous fluctuation interference, requiring the descent rate to be below the threshold for several consecutive monitoring calculation cycles (e.g., five consecutive 5-second cycles).

[0201] Material flow obstruction refers to the judgment result, that is, confirming that the material feeding process of the furnace top hopper has become abnormally slow or nearly stopped.

[0202] In a specific case, for the top silo of furnace A (conical discharge port, diameter known), based on its design parameters and material characteristics, the predefined unobstructed material flow threshold is -0.10 m / min (i.e., the normal descent rate should be faster than 0.10 m / min), which translates to approximately -0.0083 m per 5 seconds. Continuous monitoring of the top silo of furnace A over five consecutive monitoring cycles (25 seconds in total) revealed instantaneous material level descent rates of -0.05, -0.03, -0.04, -0.06, and -0.02 m / min. All of these values ​​are below (i.e., their absolute values ​​are less than) the threshold of -0.10 m / min. Therefore, it is determined that material flow obstruction has occurred in the top silo of furnace A.

[0203] A3: The preset rapping time and intensity refer to the initial settings of the rapper's action parameters for this type of resistance, such as "rapping intensity 50%, continuous rapping for 15 seconds". The rapping intensity corresponds to the energy output percentage or frequency of the pneumatic or electromagnetic rapper.

[0204] The preset strategy refers to the escalation logic followed when the initial slapping is ineffective, such as "increasing the intensity by 20% each time" or "extending the time by 5 seconds each time", or a combination of both.

[0205] In a specific case, after determining that the top hopper of furnace A is blocked, a command is immediately sent to the programmable logic controller (PLC) controlling the hopper wall vibrator to execute a preset initial vibration scheme: the vibration intensity is set to 70%, and the vibration time is set to 20 seconds. After vibration, the material level is allowed to stabilize (e.g., 30 seconds), and then the monitoring process of A1 is restarted to obtain the instantaneous material level drop rate after vibration. Assuming the new rate obtained is -0.07 m / min, which is still less than or equal to the threshold of -0.10 m / min, it indicates that the initial vibration effect is poor. The preset strategy is then activated, this time increasing the vibration intensity: increasing the intensity to 85% and extending the vibration time to 30 seconds, and vibrating again. After vibration, monitoring is performed again. Assuming the instantaneous material level drop rate obtained this time recovers to -0.15 m / min, this value is greater than (i.e., the absolute value is greater than) the threshold of -0.10 m / min. It is determined that the blockage has been resolved, the vibration sequence is stopped, and the handling log is recorded.

[0206] A4: The preset treatment time window refers to the maximum time range (e.g., 5 minutes) allowed for automatic rapping treatment from the initial detection of obstruction. This is to prevent infinite loops of ineffective rapping, protect the equipment, and prompt for manual intervention.

[0207] The early warning refers to the alarm information sent to the upper-level monitoring, operator workstation or mobile terminal when the automatic handling timeout or failure occurs. The content includes the silo number where the blockage occurred, the duration, the handling measures already attempted and the current status, and suggests that manual inspection or other measures (such as manual material removal) are required.

[0208] In a specific case, the obstruction handling procedure was initiated for the top hopper of furnace A. The preset handling time window was 3 minutes. Within 3 minutes, a total of 3 rounds of progressively stronger vibration were performed according to the strategy (e.g., 70% / 20 seconds, 85% / 30 seconds, 100% / 40 seconds). However, after the third round of vibration, monitoring showed that the instantaneous material level drop rate was still only -0.04 m / min, less than or equal to the threshold. At this point, the total handling time had exceeded the 3-minute window. Automatic vibration was immediately stopped, and an early warning message was generated, prompting a message via audible and visual alarms and a pop-up window on the central control screen: "Material flow obstruction in the top hopper of furnace A; automatic vibration handling unsuccessful. Operators, please immediately check and handle the situation on-site!"

[0209] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

Claims

1. A method for controlling the material level in the top silo of a calcium carbide furnace, wherein the calcium carbide furnace includes a weighing silo, a feeding device, and multiple top silos, characterized in that, include: The measured material level data and the estimated material level data of each of the plurality of furnace top material bins are obtained; wherein, the estimated material level data is obtained by the weighing bin and the feeding device. The measured material level data and the estimated material level data are combined to generate fused material level data; The fused material level data is processed based on a preset time series prediction model to predict the time required for the material level in each furnace top silo to reach a preset low material level threshold. Obtain the feeding rate and feeding task sequence of the feeding device; Based on the feeding task timing and feeding rate, the required time and preset high material level threshold for each furnace top hopper to reach the preset low material level threshold are processed to sort and insert the feeding task timing and generate an optimized feeding task timing. The multiple furnace top hoppers are replenished based on the optimized replenishment task timing.

2. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 1, characterized in that, Obtaining the measured material level data and estimated material level data for each of the multiple furnace top silos, specifically including: The original material level signals of the multiple furnace top silos are periodically collected; The original material level signal is processed to obtain the measured material level data; The unloading timestamp and corresponding unloading weight data of each unloading completion from the weighing hopper are obtained in real time, and the feeding timestamp of the feeding device feeding materials to multiple furnace top hoppers are obtained simultaneously. Based on the feeding timestamp and the preset single feeding volume, the total volume of materials entering the multiple furnace top hoppers is calculated cumulatively to obtain the total volume of materials entering the hoppers. Based on the unloading weight data, the total volume of materials entering the warehouse, and the preset material calibration bulk density, the estimated material level data of multiple furnace top silos is output through a balance calculation model based on the principle of material conservation.

3. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 1, characterized in that, The measured material level data and the estimated material level data are combined to generate the combined material level data, specifically including: Calculate the first rate of change of the measured material level data in adjacent sampling periods, and the second rate of change of the estimated material level data in adjacent calculation periods, respectively. The first rate of change and the second rate of change are compared with a preset normal rate of change range, and the comparison results are obtained. The first confidence weight and the second confidence weight are dynamically determined based on the comparison results. The measured material level data for the current period is weighted according to the first confidence weight, and the estimated material level data for the current period is weighted according to the second confidence weight. The weighted measured material level data and the weighted estimated material level data are added together to obtain the fused material level data for the current period.

4. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 1, characterized in that, The fused material level data is processed based on a preset time series prediction model (suggested to be modified here to: preset time series prediction model) to predict the time required for the material level in each furnace top silo to reach a preset low material level threshold, specifically including: For each furnace top hopper, the fused material level data is extracted in chronological order for a first preset continuous time period to determine the material level change sample sequence; The material level change sample sequence is input into the trained time series prediction model, and the material level value at the second preset continuous time point from the current moment is predicted by iteratively calling the time series prediction model, so as to generate a material level prediction curve. The material level prediction curve is compared with the preset low material level threshold to determine the first timestamp when the material level prediction curve first falls below the preset low material level threshold; The required duration is generated based on the time difference between the first timestamp and the current time.

5. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 1, characterized in that, The acquisition of the feeding rate and feeding task timing of the feeding device specifically includes: Obtain the replenishment rate and production shift plan, and generate an initial set of replenishment instructions; The initial feeding instruction set is parsed to generate standardized task units; wherein, the standardized task unit includes the furnace top hopper number, the planned execution time, and the rated feeding quantity; All the standardized task units are arranged in chronological order of their planned execution times to generate the material replenishment task sequence.

6. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 5, characterized in that, The process of processing the required time and preset high level threshold for each furnace top silo to reach a preset low level threshold based on the feeding task timing and feeding rate, in order to sort and insert tasks in the feeding task timing and generate an optimized feeding task timing, specifically includes: Based on the timing of the material replenishment task, a first scheduling task queue is generated, and a corresponding material level prediction copy is created for each furnace top silo. The material level prediction copy is then initialized using the fused material level data. An emergency index is set for the furnace top hopper based on the required time; wherein the emergency index is negatively correlated with the required time. The first scheduling task queue is arranged in descending order based on the urgency index to sort the tasks and generate the second scheduling task queue. The second scheduling task queue is simulated and tasks are inserted into the second scheduling task queue according to the required duration and the preset high material level threshold to generate the sequence of optimized replenishment tasks.

7. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 6, characterized in that, Simulate the execution of the second scheduling task queue, and insert tasks into the second scheduling task queue according to the required duration and the preset high material level threshold to generate the sequence of optimized replenishment tasks, specifically including: Simulation execution is performed based on the second scheduling task queue. The task execution time is calculated according to the rated replenishment amount and the replenishment rate. The material level prediction copy is updated based on the task execution time. During the simulation, if the required time for the top hopper is less than a preset threshold, a high-priority emergency replenishment task is generated for that top hopper. In the second scheduling task queue, an insertable gap is searched for the emergency replenishment task; wherein, after the insertable gap meets the preset insertion conditions, during the simulation execution, the material level of all material level prediction copies is less than or equal to the preset high material level threshold at any time; The second scheduling task queue output through the simulated execution process will be used to generate an optimized replenishment task timing.

8. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 7, characterized in that, The emergency index set based on the required time duration for the corresponding furnace top hopper specifically includes: Set the baseline emergency value; The required duration of each furnace top hopper is input into a preset exponential calculation function to generate an emergency value; wherein, the exponential calculation function is configured such that the emergency value decreases monotonically as the required duration increases. The baseline emergency value is added to the emergency value to determine the emergency index for each furnace top hopper.

9. The method for controlling the material level in the top charge silo of a calcium carbide furnace according to claim 7, characterized in that, In the second scheduling task queue, a gap is searched for the emergency replenishment task to be inserted, specifically including: Scan the time interval between every two adjacent tasks in the second scheduling task queue in task order; Calculate the required execution time based on the rated replenishment amount and the replenishment rate of the emergency replenishment task; From all the scanned time intervals, select time intervals that are greater than or equal to the required execution time to generate a candidate gap set; The gap with the earliest start time is selected from the candidate gap set as the insertable gap for the emergency replenishment task.

10. The method for controlling the material level in the top silo of a calcium carbide furnace according to claim 1, characterized in that, The method further includes: During the unreplenished stage of the top silo, the instantaneous rate of material level decline is continuously acquired; If the instantaneous material level drop rate is lower than the material flow unobstructed threshold predefined according to the geometry of the furnace top silo for multiple consecutive monitoring cycles, it is determined that material flow obstruction has occurred in the furnace top silo. The furnace top hopper is vibrated according to a preset vibration time and preset vibration intensity. After the vibration is completed, the instantaneous material level drop rate is monitored. If the instantaneous material level drop rate is still less than or equal to the unobstructed material discharge threshold, the vibration intensity is increased and / or the vibration time is extended according to a preset strategy until the instantaneous material level drop rate is greater than the unobstructed material discharge threshold. An early warning will be issued if the instantaneous material level drop rate is still less than or equal to the unobstructed material discharge threshold within the preset processing time window.

Images