Coal conveying system energy efficiency optimization and pollution source intelligent tracking method based on digital twinning

Abstract

The invention relates to the technical field of coal conveying system intelligent management and control, in particular to a coal conveying system energy efficiency optimization and pollution source intelligent tracking method based on digital twinning, and the method comprises the steps: collecting a coal quality parameter and equipment state dynamic correction friction coefficient and system characteristics in real time through constructing a digital twinning model with a self-adaptive updating module; redundant sensor deployment and edge calculation are adopted to optimize data transmission, and 5G / optical fiber hybrid networking is combined to reduce delay; and energy efficiency optimization and pollution source tracking are realized through fuzzy control and a particle filter algorithm.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for coal conveying systems, specifically a method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins. Background Technology

[0002] As a core component of industries such as thermal power and mining, the coal conveying system's energy efficiency and environmental performance directly impact a company's operating costs and environmental compliance. However, existing technologies face two major bottlenecks: First, the prediction accuracy of digital twin models is insufficient. The physical processes of coal conveying systems involve complex characteristics such as belt tension distribution and material flow friction. In actual operation, fluctuations in coal moisture content (±5%) can lead to changes in the friction coefficient of 20%-30%, and differences in particle size distribution (0.5-50mm) further exacerbate model bias, resulting in energy consumption prediction errors as high as 10%-15%. At the same time, equipment aging (such as a 50% increase in rotational resistance after roller wear) gradually changes the dynamic characteristics of the system, but existing models lack real-time adaptive mechanisms, requiring manual parameter adjustments every quarter. This leads to a disconnect between the model and the actual system, and the optimization effect decays rapidly.

[0003] Secondly, sensor reliability and data transmission issues are prominent in harsh environments. Coal conveying corridors contain high levels of dust (PM2.5 concentrations can reach 2000 μg / m³), high humidity (humidity > 90%), and extreme temperatures (-40℃ to 80℃), leading to high failure rates for traditional sensors: dust sensors are prone to clogging, with data failure cycles as short as 1-2 days; excessive humidity can cause vibration sensor signal drift, increasing errors by more than 30%. Furthermore, signal transmission delays in long-distance coal conveying corridors (e.g., over 5km) often exceed 100ms, causing real-time control lag and further reducing the timeliness of energy efficiency optimization and pollution control.

[0004] In existing technologies, digital twin models mostly rely on static parameter inputs, failing to consider the coupled effects of dynamic changes in coal quality and equipment aging; sensor deployment lacks targeted protection, and data transmission does not incorporate edge computing and heterogeneous network optimization, making it difficult to meet the requirements of high-precision and high-reliability control. Therefore, there is an urgent need for an intelligent method that can achieve adaptive model updates and reliable perception in harsh environments to overcome the aforementioned technical bottlenecks. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies by providing a digital twin-based method for optimizing energy efficiency and intelligently tracking pollution sources in coal conveying systems. Through adaptive model correction, optimized sensor deployment, and efficient data transmission, it achieves precise energy efficiency optimization and pollution source control under all operating conditions.

[0006] The technical solution adopted by this invention to solve its technical problem is: a method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins, comprising the following steps: Step 1: Construct an adaptive digital twin model A three-dimensional model was established using EDEM discrete element method coupled with multibody dynamics, integrating a coal quality parameter library (moisture content 3%-12%, particle size 0.5-50mm) and an equipment characteristic library (idler stiffness, belt elastic modulus). Data was collected in real time using an online coal quality analyzer (moisture content detection accuracy ±0.5%) and vibration sensors (sampling frequency 1000Hz). The friction coefficient (correction period 5-10 minutes) and equipment aging parameters were dynamically corrected based on an LSTM neural network to ensure that the model prediction deviation was ≤5%.

[0007] Step 2: Optimized sensor deployment and data transmission in harsh environments In high dust areas (PM2.5 > 1000 μg / m³), anti-clogging lidar (IP67 protection level) and self-cleaning dust sensors are deployed; in high humidity areas (humidity > 90%), moisture-proof vibration and temperature sensors are used; in long corridors (> 2 km), edge computing nodes are set every 500 m, and data is transmitted through a hybrid network of 5G (latency < 20 ms) and fiber optic (bandwidth 10 Gbps) to achieve sensor fault self-diagnosis and automatic switching of backup nodes.

[0008] Step 3: Dynamic Energy Efficiency Optimization Control Based on an improved fuzzy control algorithm, the input variables are real-time coal flow rate (0-500t / h), moisture content (3%-12%), and belt tension (10-30kN), and the output frequency converter frequency (15-50Hz): when the moisture content is >8% and the coal flow rate is <30% of the rated value, the frequency is reduced to 20Hz; combined with a genetic algorithm to optimize the idler roller spacing (adjustment range 1.2-1.8m), the idling energy consumption is reduced by 20%-35%.

[0009] Step 4: Intelligent tracking and control of pollution sources By integrating dust concentration (0-2000μg / m³) and equipment vibration spectrum as observed values, and employing an improved particle filtering algorithm (1500 particles), the particle weights are updated in real time through edge nodes to achieve a pollution source location accuracy of ±0.3 meters. The system is linked to a zoned spray system (droplet size 1-5μm), and high-frequency spraying is activated in high-concentration areas (>50μg / m³), achieving a dust reduction efficiency of >90%.

[0010] Specifically, in step 1, the LSTM neural network corrects the friction coefficient by inputting historical moisture content (3%-12%), particle size distribution (0.5-50mm), and measured friction coefficient, and outputting a corrected dynamic friction coefficient model with a correction error ≤0.02.

[0011] Specifically, the sensor fault diagnosis in step 2 includes: monitoring data fluctuations through variance analysis, and automatically switching to the backup sensor when the deviation is greater than 10% for three consecutive cycles, with a switching time of less than 50ms.

[0012] Specifically, in step 3, belt tension optimization involves adjusting the tension in real time using a hydraulic alignment device based on tension sensor data (accuracy ±1%FS) to avoid increased energy consumption caused by tension fluctuations greater than 5%.

[0013] Specifically, in step 4, pollution source type identification involves combining vibration spectrum (10-1000Hz) with dust composition analysis to distinguish between impact dust from falling materials (main frequency 50-200Hz) and equipment wear dust (main frequency 500-800Hz).

[0014] Specifically, in step 1, the equipment aging parameters are updated by identifying roller wear through the vibration signal kurtosis value (threshold > 6) and automatically correcting the roller rotation resistance coefficient in the model (correction range 0.01-0.05).

[0015] Specifically, the edge computing node preprocessing in step 2 includes: filtering and denoising the sensor data (using wavelet threshold denoising), compressing the data by ≥30%, and reducing the transmission bandwidth requirements.

[0016] Specifically, the fuzzy control rules in step 3 are as follows: when the coal flow rate is greater than 80% of the rated value and the tension is less than 25kN, the frequency is increased to 45-50Hz; when the moisture content is greater than 10% and the flow rate is less than 30%, the frequency is reduced to 15-20Hz.

[0017] Specifically, the method is applicable to working conditions where the coal moisture content fluctuates by ±5%, the temperature is -40℃ to 80℃, and the humidity is 20% to 95%.

[0018] Specifically, the system response indicators are: model update delay < 10 seconds, pollution source location response < 30 seconds, annual energy consumption reduction ≥ 25%, and dust emission concentration ≤ 10 μg / m³.

[0019] The beneficial effects of this invention are: Significantly improve the prediction accuracy and adaptability of digital twin models. A three-dimensional model is constructed by coupling EDEM discrete element method with multibody dynamics, and the friction coefficient and equipment aging parameters (such as idler roller wear resistance coefficient) are corrected in real time by combining LSTM neural network. The model prediction deviation is controlled within ≤5%, which solves the problem of large prediction error and the need for manual quarterly correction caused by coal quality fluctuations and equipment aging in traditional models. This ensures long-term dynamic matching between the model and the actual system.

[0020] Significantly improves sensor reliability and data transmission efficiency in harsh environments. For environments with high dust, high humidity, and extreme temperatures, it employs anti-clogging lidar, self-cleaning dust sensors, and moisture-proof vibration sensors. Combined with edge computing nodes for filtering, noise reduction, and data compression, the sensor failure rate is reduced from 30% to 12%, and the effective data acquisition time is extended from 1-2 days to over 30 days. 5G / fiber optic hybrid networking ensures real-time and continuous data transmission, with low latency and fast switching between faulty sensors.

[0021] Achieve dynamic optimization and significant improvement in the energy efficiency of the coal conveying system. By dynamically adjusting the inverter frequency through an improved fuzzy control algorithm and optimizing the idler spacing using a genetic algorithm, idling energy consumption is reduced by 20%-35%, and total annual energy consumption is reduced by ≥25%. For example, for high moisture content and low flow conditions, the frequency is reasonably reduced to decrease additional energy consumption; and the belt tension fluctuation is controlled by a hydraulic belt alignment device to avoid energy consumption increases caused by abnormal tension.

[0022] Achieve precise tracking and efficient control of pollution sources. Based on an improved particle filter algorithm, using dust concentration and vibration spectrum as observation values, the accuracy of pollution source location is ±0.3 meters. Combined with the dominant vibration frequency, the type of pollution source is accurately identified. The linkage zoned spray system achieves dust reduction efficiency >90% in high-concentration areas, and controls dust emission concentration to ≤10μg / m³, which not only meets environmental compliance requirements but also avoids the resource waste of traditional large-area spraying.

[0023] This system comprehensively enhances its intelligent management and control capabilities as well as its economic efficiency. Through sensor fault self-diagnosis, automatic backup node switching, and real-time edge computing processing, it reduces manual maintenance costs. The synergistic effect of energy efficiency optimization and pollution control not only lowers the company's operating energy costs but also reduces the risk of fines due to exceeding environmental standards, achieving both economic and environmental benefits. Attached Figure Description

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0025] Figure 1 A flowchart illustrating the digital twin-based method for optimizing energy efficiency and intelligently tracking pollution sources in a coal conveying system, provided by this invention. Figure 2 The flowchart illustrates the construction of an adaptive digital twin model in the digital twin-based coal conveying system energy efficiency optimization and intelligent pollution source tracking method provided by this invention. Figure 3 The flowchart illustrates the sensor optimization deployment and data transmission in harsh environments within the digital twin-based coal conveying system energy efficiency optimization and pollution source intelligent tracking method provided by this invention. Figure 4 The flowchart illustrates the dynamic energy efficiency optimization control in the digital twin-based coal conveying system energy efficiency optimization and intelligent pollution source tracking method provided by this invention. Figure 5 The flowchart illustrates the intelligent tracking and control of pollution sources in the energy efficiency optimization and intelligent pollution source tracking method for coal conveying systems based on digital twins provided by this invention. Detailed Implementation

[0026] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0027] like Figures 1-5 As shown, the energy efficiency optimization and intelligent pollution source tracking method for coal conveying systems based on digital twins of the present invention includes the following steps: Adaptive Digital Twin Model Construction: Using EDEM discrete element simulation and multibody dynamics coupling technology, a 3D model including equipment such as belts, idlers, and feed pipes is established, integrating a basic parameter library (coal moisture content 3%-12%, particle size 0.5-50mm; idler stiffness 200-500N / mm, belt elastic modulus 1000-2000MPa). An online coal quality analyzer (installed at the feeder outlet, detection frequency 1 time / minute, moisture content accuracy ±0.5%) and a high-frequency vibration sensor (installed on the idler bearing seat, sampling frequency 1000Hz) are deployed to collect real-time coal quality and equipment status data. A parameter correction model is constructed based on an LSTM neural network, dynamically updating the friction coefficient (correction formula: μ = ...) every 5-10 minutes based on measured data. ×(1+0.03×Δw), where Δw is the moisture content deviation) and the roller rotation resistance coefficient (the resistance coefficient increases from 0.01 to 0.05 after wear), to ensure that the model prediction deviation is ≤5%.

[0028] Deployment and data transmission optimization for harsh environments: For high dust areas (PM2.5 > 1000 μg / m³), anti-clogging lidar (lens with compressed air purging device, IP67 protection rating) and self-cleaning dust sensors (built-in ultrasonic cleaning module, cleaning cycle 30 minutes) are selected; for high humidity areas (humidity > 90%), moisture-proof vibration and temperature sensors (housing seal rating IP68, built-in heating and defogging module) are used. In long corridors (> 2 km), edge computing nodes (computing power ≥ 2 TOPS) are set every 500 m to filter and denoise the data (wavelet threshold denoising, signal-to-noise ratio improved by 20 dB) and compress it (compression rate ≥ 30%). The data is transmitted to the digital twin platform via a hybrid network of 5G (coverage blind spots, latency < 20 ms) and fiber optic (backbone transmission, bandwidth 10 Gbps). An integrated sensor health monitoring module is used to achieve fault self-diagnosis through variance analysis (fault determination based on data deviation > 10% for three consecutive cycles), and automatic switching of backup sensors (switching time < 50 ms).

[0029] Dynamic energy efficiency optimization control: An improved fuzzy control algorithm is constructed, with input variables being real-time coal flow rate (0-500t / h), moisture content (3%-12%), and belt tension (10-30kN), and the output frequency of the inverter (15-50Hz). Core rules: When coal flow rate > 80% of rated value and tension < 25kN, the frequency is increased to 45-50Hz; when moisture content > 10% (increased viscosity) and flow rate < 30%, the frequency is reduced to 15-20Hz (to avoid material sticking and idling). Combined with a genetic algorithm to optimize the idler roller spacing (adjustment range 1.2-1.8m), frictional energy consumption caused by belt sagging is reduced, resulting in a reduction of idling energy consumption of over 30%.

[0030] Intelligent pollution source tracking and control: An improved particle filtering algorithm is employed, using dust concentration (0-2000 μg / m³) and equipment vibration spectrum (10-1000 Hz) as observation values, to initialize 1500 particles covering the coal conveying corridor. Pollution source types are identified through vibration spectrum: impact dust from falling materials has a dominant frequency of 50-200 Hz, while equipment wear dust has a dominant frequency of 500-800 Hz. Combined with a dust diffusion model, particle weights are iteratively updated to achieve a positioning accuracy of ±0.3 meters. A coordinated zoned spray system is implemented; in high-concentration areas (>50 μg / m³), high-frequency spraying (droplet size 1-5 μm, flow rate 5-10 L / min) is activated, achieving a dust reduction efficiency of >90% within 5 minutes.

[0031] Example 1: High Moisture Content Coal Conditions This embodiment verifies the adaptive correction and energy efficiency optimization effects of the model in scenarios where the moisture content of coal fluctuates significantly.

[0032] Operating parameters: Coal moisture content is 8%-13%, within the 3%-12% fluctuation range specified in the claims, with actual fluctuations reaching ±5%; particle size distribution is 10-30mm, belonging to common medium-particle coal; ambient temperature is stable at 20-25℃, humidity is 60%-70%, and coal conveying capacity dynamically varies between 100-400t / h depending on production demand. Under these conditions, high moisture content easily leads to increased coal viscosity, significant changes in belt friction coefficient, and traditional models are prone to large prediction biases.

[0033] System deployment details: An online coal quality analyzer is installed at the coal feeder outlet to collect coal moisture content data in real time. The sampling interval is set to 1 minute to ensure data timeliness. A high-frequency vibration sensor is deployed at the bearing housing of the head roller of the coal conveyor belt, with a sampling frequency set to 1000Hz, to simultaneously monitor belt tension fluctuations and equipment operating status. Sensor data is transmitted via shielded cable to the nearest edge computing node, which is equipped with an industrial-grade processor to support real-time data preprocessing and model interaction.

[0034] Model adaptive correction process: When the coal moisture content increases from an initial 8% to 13%, the online coal quality analyzer provides real-time feedback on the moisture content change, and the digital twin platform inputs the data into the LSTM neural network model. Based on historical training data showing the correlation between friction coefficient and moisture content, the model dynamically corrects the material friction coefficient: the initial friction coefficient is set at 0.3, and it is gradually adjusted as the moisture content increases, eventually reaching 0.345. By comparing the model predictions before and after correction with actual operating data, the belt tension prediction deviation decreased from 12% in the traditional model to 4.2%, meeting the requirement of "model prediction deviation ≤ 5%" in the claims.

[0035] Energy efficiency optimization implementation: The fuzzy controller receives real-time data on coal flow rate, moisture content, and belt tension. When the coal flow rate is detected to be 200 t / h (within the 30%-80% rated range) and the moisture content reaches 11% (exceeding the 10% threshold), the controller determines it to be a "high moisture content + medium flow rate" condition and outputs a frequency adjustment command to the inverter, reducing the belt speed from the original fixed speed of 4 m / s to the speed corresponding to 25 Hz (approximately 2.5 m / s), thus reducing additional energy consumption caused by increased material viscosity. Simultaneously, by optimizing the idler roller layout using a genetic algorithm, the spacing between idlers near the material drop point is shortened from 1.5 m to 1.2 m, reducing frictional resistance caused by belt sagging. Actual measurements show that this adjustment reduces energy consumption per unit length by 8%.

[0036] Pollution source tracing and treatment effectiveness: High-moisture coal is prone to generating dust due to impact during the feeding process. Monitoring showed that the dust concentration at the feeding pipe suddenly increased to 80 μg / m³. The digital twin platform, through signal spectrum analysis of vibration sensors, identified the dominant vibration frequency in this area as 150 Hz, which matches the characteristic frequency range of impact dust generated by coal feeding. The particle filtering algorithm, using dust concentration and vibration signal as input, initialized 1500 particles covering a 3m radius around the feeding pipe. Through three iterations of particle weight updates, the final location deviation of the pollution source was only ±0.2 meters. The system immediately activated the corresponding area's spray device, setting the droplet size to 3 μm and the spray flow rate to 5 L / min. After 3 minutes, the dust concentration in this area dropped to 8 μg / m³, meeting environmental emission standards.

[0037] Overall Results: In this embodiment, the adaptive correction of the model significantly improves the accuracy of energy consumption prediction, and the overall energy efficiency is 32% higher than that of traditional control methods. The dust control response time is controlled within 3 minutes, which verifies the effectiveness of the method under high moisture content conditions.

[0038] Example 2: Equipment aging (roller wear) condition This embodiment is for a coal conveying system that has been in operation for many years, to verify the model's adaptability to equipment aging and its energy efficiency optimization effect.

[0039] Operating parameters: The coal conveying system has been running continuously for 3 years, with an average wear rate of 30% for the idler rollers; the coal has a total moisture content of 5%-7% and a particle size of 5-20mm, classifying it as low-moisture, medium-particulate coal; the ambient temperature is 15-20℃, the humidity is 50%-60%, and the coal conveying rate is stable at 100-250t / h. Equipment aging has led to increased rotational resistance of the idler rollers, and the traditional model, due to the lack of real-time parameter updates, has significantly increased energy consumption prediction errors.

[0040] System deployment details: Vibration sensors are evenly deployed at key locations on the idler roller assembly, with one sensor installed every 10 idler rollers. The sampling frequency is set to 1000Hz, focusing on monitoring changes in the kurtosis value of the idler roller vibration. Tension sensors are installed at the belt tensioning device to collect belt tension data in real time. All sensor data is aggregated to an edge computing node, which has a built-in equipment health analysis module that generates an equipment status report every 5 minutes.

[0041] Equipment aging monitoring and model updating: Vibration sensor data showed that the vibration kurtosis value of the worn idler roller increased from an initial 3 to 7.2, exceeding the set threshold of 6, and the system determined that the idler roller was severely worn. Based on this data, the digital twin platform automatically triggered the equipment parameter update mechanism, correcting the idler roller rotation resistance coefficient from an initial 0.01 to 0.035, and simultaneously adjusting the contact friction model between the belt and the idler roller. Before the correction, the model's prediction error for the drive motor power reached 14%; after the correction, the error decreased to 4.8%, meeting the requirement of "model prediction deviation ≤ 5%" in the claims.

[0042] Energy efficiency optimization implementation: To address the "low flow + high resistance" operating conditions, the fuzzy controller dynamically adjusts its control strategy: when the coal conveying rate is 100t / h (below 30% of the rated value), the output inverter frequency drops to 18Hz to reduce idling energy consumption; simultaneously, a genetic algorithm is activated to optimize the idler roller spacing, shortening the idler roller spacing in severely worn areas from the original 1.8m to 1.4m, and reducing the frictional contact area between the belt and the idlers by increasing support points. Actual measurements show that after the adjustment, the belt running resistance decreased by 22%, and the unit energy consumption decreased significantly.

[0043] Effect verification: The proportion of additional energy consumption due to equipment aging decreased from 12% to 5%, and the operating current of the drive motor stabilized at 12A from 15A. The replacement cycle of the idler rollers was extended from 3 years to 4 years, and the average annual maintenance cost was reduced by 25%. The model continuously tracks vibration data and updates the resistance coefficient every 10 minutes to ensure that the optimization strategy always adapts to the equipment status, thus verifying the dynamic adaptation capability to equipment aging.

[0044] Example 3: Optimized Data Transmission Scenario for Long-Distance Coal Conveying Corridors This embodiment verifies the reliability of sensors and data transmission efficiency under harsh conditions in a 5km long coal conveying corridor.

[0045] Operating parameters: The coal conveying corridor is 5km long. The humidity in the middle section is consistently maintained at 70%-85%, and some sections generate high dust levels due to coal dryness, with PM2.5 peak values ​​reaching 1200μg / m³. The total moisture content of the coal is 6%-9%, and the particle size is 10-30mm. The ambient temperature is -10 to 5℃, presenting harsh conditions with the combined effects of low temperature and high humidity. The coal conveying capacity is 150-350t / h. Long-distance transmission can easily lead to signal delays, and high humidity and high dust levels can easily cause sensor malfunctions.

[0046] System deployment details: One edge computing node is set up every 500m along the corridor, with a total of 10 nodes deployed. The nodes are connected by a fiber optic backbone network to supplement 5G communication modules in coverage blind spots. In the high humidity section (humidity > 80%), a moisture-proof vibration temperature sensor is selected with a sealed housing and a built-in heating and defogging module. In the high dust section (PM2.5 > 800μg / m³), anti-clogging lidar and self-cleaning dust sensors are deployed. The lidar lens is equipped with a compressed air blowing device, and the dust sensor has a built-in ultrasonic cleaning module with a cleaning cycle set to 30 minutes.

[0047] Sensor reliability assurance: The moisture-proof sensor maintains an internal temperature above 15°C through a heated defogging module, preventing condensation from affecting the signal. Actual measurements show that data drift has decreased from 30% of traditional sensors to 6%. The self-cleaning dust sensor, through high-frequency cleaning, extends the effective data acquisition time from 1-2 days in traditional equipment to over 30 days, achieving a data validity rate of 99%. Edge nodes preprocess the acquired data, employing wavelet threshold denoising algorithms to reduce noise interference, and then use data compression algorithms to reduce transmission volume by 35%, alleviating bandwidth pressure.

[0048] Data transmission and energy efficiency optimization: The fiber optic backbone network ensures core data transmission bandwidth of up to 10Gbps, 5G modules cover signal blind spots, and end-to-end data transmission latency is controlled within 18ms, meeting real-time control requirements. The digital twin platform adjusts energy efficiency strategies based on real-time data: during low-flow periods (150t / h), the belt speed is reduced from 3.5m / s to 2.8m / s, combined with optimized idler spacing (adjusted from 1.5m to 1.6m), reducing idling energy consumption by 30%; in the high-humidity mid-section, tension sensors monitor the risk of belt slippage and dynamically adjust tension to avoid increased energy consumption due to slippage.

[0049] Effect verification: The sensor failure rate has been reduced from 30% in traditional deployments to 12%, and the data interruption time is no more than 10 minutes per month; the long-distance transmission latency is controlled within 20ms, meeting the needs of real-time optimization; the overall energy efficiency has been improved by 28%, verifying the reliable sensing and efficient transmission capabilities of long-distance corridors.

[0050] Example 4: Extreme Low Temperature Conditions This embodiment focuses on coal conveying systems in frigid regions, verifying the model's adaptability and the effectiveness of pollution source control under extreme low-temperature environments.

[0051] Operating parameters: Ambient temperature -40℃ to -30℃, falling within the extreme range of -40℃ to 80℃ as defined in the claims; humidity 20%-30%, air dry; coal moisture content 3%-5%, particle size 5-20mm, dry coal is prone to dust generation; coal conveying capacity 150-300t / h. Low temperatures cause belt hardening and a decrease in elastic modulus, making traditional sensors prone to failure due to low temperatures, and significantly increasing model prediction bias.

[0052] System deployment details: All sensors are externally equipped with heat-insulating jackets, extending the operating temperature range to -50℃ to 80℃, ensuring normal start-up at low temperatures; temperature sensors are installed near the belt drive unit to monitor the ambient temperature of the equipment in real time; the digital twin platform integrates a low-temperature correction module, with a preset correlation curve between the belt's elastic modulus and temperature.

[0053] Model low-temperature adaptive correction: When the ambient temperature drops to -40℃, the belt's elastic modulus decreases from 1500MPa at room temperature to 1200MPa. The model automatically calls the correction curve based on temperature sensor data and adjusts the belt tension calculation model accordingly. The LSTM neural network, combined with coal moisture content (3%-5%) and low-temperature parameters, corrects the friction coefficient from 0.28 at room temperature to 0.31, reducing the tension prediction error from 11% to 3.5%, ensuring the optimization strategy is adapted to low-temperature characteristics.

[0054] Energy efficiency optimization and equipment protection: The fuzzy controller outputs a frequency of 30Hz to the inverter based on the "low temperature + medium flow (200t / h)" operating condition, preventing belt embrittlement caused by high-speed operation. Simultaneously, a belt preheating device is activated, raising the belt surface temperature to above -10℃ via electric heating. Although energy consumption increases by 5%, the belt breakage failure rate is reduced by 80%. The idler roller assembly uses low-temperature grease to reduce rotational resistance, resulting in a 15% reduction in idling energy consumption compared to traditional lubrication methods.

[0055] Pollution source tracing and control: Gaps appeared in the feed chute seal due to low-temperature hardening, causing the dust concentration to rise to 60 μg / m³. The vibration sensor detected a signal frequency of 600 Hz, indicating dust generation due to equipment wear. The particle filtering algorithm, under low-temperature signal noise interference, improved robustness by increasing the particle count to 2000, achieving a positioning deviation of ±0.35 meters within 30 seconds. The spray system employs a heating and anti-freezing design to ensure proper droplet atomization. With a droplet diameter of 3 μm, the dust concentration dropped to 7 μg / m³ within 5 minutes of startup, verifying the pollution control effectiveness under extreme low-temperature conditions.

[0056] Example 5: Extreme working conditions with high dust This embodiment is designed for high-dust extreme environments to verify the ability of redundant sensor deployment and precise pollution source control.

[0057] Operating parameters: Coal moisture content 3%-4%, particle size 0.5-10mm; fine-particle dry coal easily generates a large amount of dust; PM2.5 concentration peak reaches 2000μg / m³, far exceeding environmental standards; ambient temperature 30-35℃, humidity 30%-40%; high temperature and dryness exacerbate dust diffusion; coal conveying capacity 200-400t / h. High dust levels easily clog sensors and attenuate signals, making it difficult for traditional tracking methods to accurately locate pollution sources.

[0058] System deployment details: Three redundant anti-clogging dust sensors were deployed in a triangular arrangement with a spacing of 5m in the high dust area (PM2.5 > 1000 μg / m³) to ensure cross-validation of data; the lidar lens was equipped with a high-pressure air purging device, and the purging frequency was increased to once every 15 minutes; the edge node collected data once every 2 seconds and transmitted it to the digital twin platform through optical fiber to avoid the attenuation of wireless signal in high dust.

[0059] Sensor reliability assurance: Redundant sensors eliminate outlier data through cross-validation. When the data deviation of a single sensor exceeds 10%, two other sets of data are automatically activated, maintaining a data effectiveness rate of 98% (compared to only 30% for traditional single sensors). The self-cleaning module starts at high frequency, preventing significant dust accumulation on the LiDAR lens and maintaining stable scanning accuracy. Edge nodes perform real-time filtering of the data to remove signal fluctuations caused by dust interference, ensuring the accuracy of the data input to the model.

[0060] Pollution source identification and tracing: The system detected two high-concentration areas: a dust concentration of 1200 μg / m³ near the material discharge pipe and a vibration frequency of 180 Hz, identified as impact dust from material discharge; and a concentration of 800 μg / m³ near the idler roller assembly and a vibration frequency of 700 Hz, identified as dust from equipment wear. The particle filtering algorithm initialized particle swarms for both pollution sources, with a particle coverage range of 3m in the material discharge pipe area and 2m in the idler roller area. After five iterations, the positioning deviation of the pollution source in the material discharge pipe was ±0.25 meters, and the deviation in the idler roller area was ±0.3 meters.

[0061] Precision governance results: To address dust generated by material impact, a high-pressure spray system was activated, with droplet size of 5μm (wide coverage) and a flow rate of 10L / min. For dust generated by roller abrasion, a precision spray system was employed, with droplet size of 1μm (strong adhesion) and a flow rate of 5L / min. After 10 minutes, the dust concentration in the material discharge pipe area decreased to 10μg / m³, and in the roller area it decreased to 8μg / m³. The overall dust emission concentration was controlled below 10μg / m³, validating the dust control capability under extremely high dust conditions.

[0062] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins, characterized in that, Includes the following steps: Step 1: Construct an adaptive digital twin model: A three-dimensional model is established using EDEM discrete element and multibody dynamics coupling technology, integrating coal quality parameter library and equipment characteristic library, covering parameters such as moisture content, particle size, idler stiffness, and belt elastic modulus; data is collected in real time through an online coal quality analyzer and vibration sensor, and the friction coefficient and equipment aging parameters are dynamically corrected based on LSTM neural network to control the model prediction deviation; Step 2: Sensor Optimization Deployment and Data Transmission in Harsh Environments: Deploy anti-clogging lidar and self-cleaning dust sensors in high-dust areas; use moisture-proof vibration and temperature sensors in high-humidity areas; set up edge computing nodes at set intervals in long corridors, and transmit data through a hybrid 5G and fiber optic network to achieve sensor fault self-diagnosis and automatic switching of backup nodes. Step 3, Dynamic Energy Efficiency Optimization Control: Based on an improved fuzzy control algorithm, the input variables are real-time coal flow rate, moisture content, and belt tension, and the output is the inverter frequency; combined with a genetic algorithm, the idler group spacing is optimized to reduce idling energy consumption; Step 4, Intelligent tracking and control of pollution sources: An improved particle filtering algorithm is used to establish a probability distribution model of pollution source location based on dust concentration and equipment vibration spectrum as observation values; pollution source location is achieved by iteratively updating particle weights, and precise control is achieved by linking the zoned spray system.

2. The method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins as described in claim 1, characterized in that: In step 1, the LSTM neural network corrects the friction coefficient by: inputting historical moisture content, particle size distribution and measured friction coefficient, outputting a corrected dynamic friction coefficient model, and controlling the correction error.

3. The method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins according to claim 1, characterized in that: In step 2, sensor fault diagnosis includes: monitoring data fluctuations through variance analysis, and triggering automatic switching of backup sensors after determining a fault.

4. The method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins according to claim 1, characterized in that: In step 3, belt tension optimization: based on tension sensor data, the tension is adjusted in real time through a hydraulic adjustment device to control the tension fluctuation range.

5. The method for energy efficiency optimization and intelligent pollution source tracking of coal conveying systems based on digital twins according to claim 1, characterized in that: In step 4, pollution source type identification: combining vibration spectrum and dust composition analysis, distinguish between impact dust from falling materials and dust from equipment wear.

6. The method for energy efficiency optimization and intelligent pollution source tracking of a coal conveying system based on digital twins as described in claim 1, characterized in that: In step 1, the equipment aging parameters are updated: the wear of the idler rollers is identified by the kurtosis value of the vibration signal, and the rotational resistance coefficient of the idler rollers in the model is automatically corrected.

7. The method for energy efficiency optimization and intelligent pollution source tracking of a coal conveying system based on digital twins as described in claim 1, characterized in that: In step 2, the edge computing node preprocessing includes filtering and denoising the sensor data and compressing it to reduce the transmission bandwidth requirements.

8. The method for energy efficiency optimization and intelligent pollution source tracking of a coal conveying system based on digital twins according to claim 1, characterized in that: In step 3, the fuzzy control rule is as follows: when the coal flow rate exceeds the rated value and the tension is lower than the set value, the frequency is increased; when the moisture content exceeds the set value and the flow rate is lower than the rated value, the frequency is decreased.

Images