Orepass all-parameter intelligent detection device and method based on multi-sensor fusion

Abstract

The invention discloses a draw shaft full-parameter intelligent detection device and method based on multi-sensor fusion, and relates to the technical field of draw shaft detection, the draw shaft full-parameter intelligent detection device comprises a draw shaft, a sensing unit is arranged in the draw shaft, and a main control unit, a communication unit, a power supply unit and an auxiliary protection system are arranged outside the draw shaft; the method further comprises the steps of S1, detection starting, S2, data acquisition, S3, data preprocessing and fusion, S4, three-dimensional model construction, S5, health assessment, S6, anomaly detection, S7, graded early warning and detection report generation, and S8, linkage control. Through the sensing unit and the main control unit and by combining the steps S2 and S4, the technical limitation that manual inspection or a single sensor can only monitor a single parameter is avoided, and coverage monitoring and visual presentation of all parameters of draw shaft operation are achieved; through the communication unit and the main control unit and in combination with the steps S2, S7 and S8, a traditional intermittent detection mode of manual regular inspection is replaced.

Description

Technical Field

[0001] This invention relates to the field of well pass detection technology, and in particular to an intelligent detection device and method for all parameters of well passes based on multi-sensor fusion. Background Technology

[0002] Underground mine ore passes are the core vertical transportation channels connecting the mining area with the beneficiation plant, crushing station, and other links in the mining production system. Their main function is to realize the efficient transfer of ore from the underground mining area to the surface or subsequent processing stages. Their operating status directly determines the efficiency of ore transportation in the mine and is closely related to the safety of underground operations and the stable operation of equipment. They are key infrastructure to ensure continuous mine production. At present, the mining industry mainly relies on two traditional methods for the inspection of ore passes: one is manual inspection, and the other is single sensor detection.

[0003] Existing ore pass detection technologies suffer from several drawbacks. They rely on single sensors or manual methods, monitoring only basic parameters like material level and localized vibrations. They cannot simultaneously collect multi-dimensional data from multiple sensors, leading to blind spots in critical ore pass operation. Furthermore, they lack real-time capabilities, requiring periodic manual inspections with intermittent monitoring. Single sensors lack real-time data transmission and dynamic analysis mechanisms, often only detecting problems after ore runoff or blockages occur. They also exhibit poor anti-interference capabilities, lacking solutions for protection and data processing in high-dust, dark mining environments. Finally, data is isolated, with data collected by different detection methods operating independently without multi-source information fusion mechanisms, making it difficult for managers to integrate multi-sensor data. Therefore, a multi-sensor fusion-based intelligent ore pass detection device and method are needed to address these issues. Summary of the Invention

[0004] To achieve the above objectives, the present invention is implemented through the following technical solution: a multi-sensor fusion-based intelligent detection device for all parameters of a chute, comprising a chute, wherein a sensing unit is provided inside the chute, and a main control unit, a communication unit, a power supply unit, and an auxiliary protection system are provided outside the chute; The sensing unit includes a millimeter-wave radar sensor, an infrared thermal imager, a laser displacement sensor, a vibration sensor, and a gas sensor. The auxiliary protection system includes a high-pressure air curtain dustproof self-cleaning device and an anti-collision buffer mechanism; The output terminal of the power supply unit is connected to the input terminal of the main control unit and the communication unit. The output terminal of the sensing unit and the auxiliary protection system is connected to the input terminal of the main control unit. The output terminal of the main control unit is connected to the input terminal of the communication unit.

[0005] Preferably, the main control unit is an explosion-proof industrial computer used for multi-sensor data fusion analysis and decision control; The communication unit adopts an intrinsically safe 5G / Wi-Fi 6 module for mining, which is used to transmit data collected by multiple sensors to the monitoring center in real time; The power supply unit adopts a dual power system of explosion-proof lithium battery and solar power to extend the battery life to 72 hours.

[0006] Preferably, the main control unit is configured to execute a multi-sensor data fusion algorithm to generate a three-dimensional digital twin model of the ore pass, and output an early warning signal based on the anomaly detection model; The three-dimensional digital twin model is constructed based on Kalman filtering and fusion of millimeter-wave radar, laser displacement, and infrared thermal imaging data.

[0007] Preferably, the intelligent detection method for all parameters of a well pass based on multi-sensor fusion, applied to an intelligent detection device for all parameters of a well pass based on multi-sensor fusion, includes the following detection methods: S1. Startup Detection: Start the power supply unit and complete the self-test of each core unit, initialize the auxiliary protection system, and ensure that the device meets the detection operation conditions; S2. Data Acquisition: Multi-dimensional parameter data of the well operation are collected synchronously by multiple sensors in the sensing unit, and the collected data is transmitted to the main control unit in real time. S3. Data Preprocessing and Fusion: The main control unit performs noise reduction, calibration and time synchronization processing on the collected raw data, and adopts a multi-sensor fusion algorithm to achieve collaborative fusion of multi-source data, and removes invalid data to ensure data validity. S4. 3D Model Construction: The main control unit calls the Kalman filter algorithm to fuse the data from millimeter-wave radar sensor, infrared thermal imager, laser displacement sensor, vibration sensor and gas sensor to construct a real-time 3D digital twin model of the ore pass. S5. Health Assessment: Based on the three-dimensional digital twin model and the fused parameter data, a comprehensive analysis is conducted on the material level status, well wall structure, temperature distribution, gas concentration and ventilation efficiency, and ore runoff risk of the ore pass, and the health level of the ore pass is output. S6. Anomaly Detection: Based on the anomaly detection model, compare the deviation between real-time parameter data and health baseline data to identify the type and location of anomalies in the operation of the ore pass. S7. Tiered Early Warning and Detection Report Generation: Based on the anomaly detection data, if an anomaly is detected, an early warning level is determined according to the severity of the anomaly, and the corresponding early warning mechanism is triggered; if no anomaly is detected, a detection report containing detection data, health assessment results, and anomaly information is generated and transmitted to the monitoring center through the communication unit. S8. Linkage Control: Based on the early warning results and the type of anomaly, the main control unit sends control commands to the auxiliary protection system or related equipment of the ore pass to realize linkage control for anomaly handling and feeds back the control results to the monitoring center.

[0008] Preferably, step S1 specifically includes: S1-1: Start the explosion-proof lithium battery and solar dual power system, and check the remaining power of the lithium battery and the on / off status of the solar power supply circuit to ensure that the dual power switching function is normal. S1-2: Initialize the explosion-proof industrial computer of the main control unit and load the multi-sensor data fusion algorithm and anomaly detection model; S1-3: Detect the network connection status of the intrinsically safe 5G / Wi-Fi 6 module for mining in the communication unit and confirm that the communication link with the monitoring center is unobstructed; S1-4: Check the response signals of each sensor in the sensing unit one by one to ensure that they can output detection signals normally; S1-5: Initialize the auxiliary protection system, start the high-pressure air curtain dustproof self-cleaning device for trial operation, and reset the anti-collision buffer mechanism to the initial protection position; Step S2 specifically includes: S2-1: Each sensor collects data according to a preset synchronization cycle. The millimeter-wave radar sensor continuously scans the ore accumulation surface inside the ore pass to obtain material level data. The infrared thermal imager captures the temperature field distribution of the ore and the wall inside the ore pass in real time. The laser displacement sensor performs cyclic sampling at preset monitoring points on the wall of the ore pass to obtain deformation data. The vibration sensor continuously collects vibration signals of the wall of the ore pass to extract ore runoff characteristics. The gas sensor collects gas samples inside the ore pass in real time to detect CO and CH4 concentrations and evaluates the ventilation status in combination with the wind speed data of the ore pass ventilation opening. S2-2: All collected data is timestamped and transmitted to the main control unit in real time via an intrinsically safe 5G / Wi-Fi 6 module for mining in an encrypted manner. If data packet loss occurs during transmission, the retransmission mechanism of the communication unit is triggered.

[0009] Preferably, step S3 specifically includes: S3-1: In the preprocessing stage, wavelet denoising algorithm is used to filter noise in the vibration signal collected by the vibration sensor, moving average method is used to smooth the concentration data collected by the gas sensor, and combined with the temperature and humidity environmental parameters of the well site, the detection data of laser displacement sensor and millimeter-wave radar sensor are calibrated for error. S3-2: Time synchronization stage. Based on the system clock of the main control unit, the data with timestamps attached to each sensor are time-aligned to ensure the consistency of multi-source data in the time dimension. S3-3: In the data fusion stage, a multi-source information fusion framework is adopted. Different weights are assigned based on the detection accuracy of each sensor. The Kalman filter algorithm is used to achieve the initial fusion of material level, well wall deformation, and temperature data. Then, the vibration characteristic data and gas concentration data are fused through the DS evidence theory to finally form a unified ore pass operation parameter dataset. At the same time, the 3σ criterion is used to remove extreme outliers in the dataset. Step S4 specifically includes: S4-1: Based on the well wall structure parameters of the well design drawings, and combined with the well wall monitoring point data sampled by the laser displacement sensor, construct the initial three-dimensional contour model of the well wall; S4-2: The material level data collected and fused by the millimeter-wave radar sensor is converted into ore accumulation morphology parameters and superimposed onto the initial three-dimensional contour model of the well wall to generate a preliminary three-dimensional model of the ore chute containing the ore accumulation state. S4-3: Extract temperature field data collected by infrared thermal imager, color-code high-temperature anomaly areas in the preliminary 3D model, and form a temperature-visualized 3D digital twin model of the chute. The model is updated in real time according to the data acquisition cycle.

[0010] Preferably, step S5 specifically includes: S5-1: Establish a health assessment index system for ore passes, including indicators such as material level safety factor, well wall deformation rate, temperature anomaly rate, gas safety index, ventilation efficiency, and ore runoff risk value. S5-2: A weighted scoring method is used to quantify the scores of each indicator. The well wall deformation rate, gas safety index, and ore run risk value are given higher weights, while the material level safety factor, temperature anomaly ratio, and ventilation efficiency are given normal weights. The health level of the ore pass is divided into different categories based on the total score, and a health assessment report is generated. Step S6 specifically includes: S6-1: Based on historical normal operation data of ore passes and typical abnormal case data, train a machine learning anomaly detection model and establish a feature threshold library for different anomaly types. S6-2: Input the real-time parameter data fused in step S3 into the anomaly detection model, compare the deviation between the real-time data and the feature threshold library, and if the deviation exceeds the set range, it is determined to be an anomaly. S6-3: Locating abnormal locations using a three-dimensional digital twin model.

[0011] Preferably, step S7 specifically includes: S7-1: The early warning levels are divided into three levels: Level 1 warning, triggered by severe cracks in the well wall, gas concentration far exceeding the safety limit, or ore run-off; Level 2 warning, triggered by material level approaching the upper or lower limit, slight temperature abnormality, or gas concentration approaching the safety limit; Level 3 warning, triggered by slight fluctuations in sensor data and a medium health level in the ore pass. S7-2: The test report includes basic information, parameter data, health assessment results, abnormal information, and early warning records. The report is transmitted to the monitoring center in real time after it is generated. The S8 step specifically includes: S8-1: If the dust concentration inside the chute is too high, causing the sensor detection accuracy to decrease, the main control unit sends a start command to the high-pressure air curtain dust prevention and self-cleaning device, and dynamically adjusts the air curtain wind speed and running time according to the dust concentration until the sensor detection accuracy returns to normal. S8-2: If the millimeter-wave radar sensor detects that a large piece of ore is falling and poses a risk of impacting the well wall, the main control unit controls the anti-collision buffer mechanism to extend to the protective position to reduce the impact force of the ore. After the impact ends, the control mechanism is reset. S8-3: If the gas sensor detects that the CO or CH4 concentration exceeds the standard, the main control unit sends a control command to the chute ventilation system to increase the ventilation fan speed or turn on the standby fan until the gas concentration drops to a safe range. S8-4: If the material level is close to the upper limit, the main control unit sends a stop command to the chute feeding equipment; if the material level is close to the lower limit, it sends a start command.

[0012] In summary, this invention provides an intelligent detection device and method for all parameters of a well pass based on multi-sensor fusion, which has the following beneficial effects: 1. This invention, by adopting a hardware and algorithm collaborative structure of sensing units and main control units, and combining the S2 data acquisition and S4 3D model construction steps, avoids the technical limitations of manual inspection or a single sensor monitoring only a single parameter. It achieves comprehensive monitoring and visualization of all parameters of ore operation, completely eliminating the monitoring blind spots of key dimensions such as material level, well wall structure, and gas state, and providing comprehensive technical support for the operation status of ore.

[0013] 2. This invention, by adopting a real-time response structure of communication unit and main control unit, and combining S2 data acquisition, S7 hierarchical early warning and detection report generation and S8 linkage control steps, replaces the traditional intermittent detection mode of manual periodic inspection. It constructs a system of real-time data acquisition, real-time anomaly identification, real-time early warning triggering and real-time fault handling, avoiding sudden faults such as ore runoff, blockage, and well wall cracking, preventing the technical lag of being discovered only after they occur, and significantly improving the efficiency and timeliness of fault response and handling in ore chutes.

[0014] 3. This invention employs an anti-interference collaborative structure of an auxiliary protection system and a main control unit, combined with S1 start-up detection and S3 data preprocessing and fusion steps, to specifically resist the complex environmental interference of high dust, temperature and humidity fluctuations, and equipment vibration in mines. It avoids the technical defects of traditional ultrasonic level gauges, such as dust interference, failure of video monitoring in dark environments, and false alarms caused by single vibration sensors, thus significantly improving the accuracy of detection data and the stability of sensor operation.

[0015] 4. By adopting a data analysis and integration structure of the main control unit, and combining the S3 data preprocessing and fusion and the S5 health assessment steps, this invention avoids the technical limitations of independent data and lack of collaborative analysis in detection methods. It realizes in-depth integration and correlation analysis of multi-dimensional data of the ore pass, providing unified and reliable data support for the scientific assessment of the overall health status of the ore pass, and avoiding misjudgment or omission of the ore pass operation status due to isolated data. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the intelligent detection device and method for all parameters of a well pass based on multi-sensor fusion according to the present invention; Figure 2 This is a schematic diagram of the process architecture of the intelligent detection device and method for all parameters of a well pass based on multi-sensor fusion according to the present invention.

[0017] Explanation of reference numerals in the attached figures: 1. Passage; 2. Sensing Unit; 201. Millimeter-wave radar sensor; 202. Infrared thermal imager; 203. Laser displacement sensor; 204. Vibration sensor; 205. Gas sensor; 3. Main control unit; 4. Communication unit; 5. Power supply unit; 6. Auxiliary protection system; 601. High-pressure air curtain dustproof self-cleaning device; 602. Anti-collision buffer mechanism. Detailed Implementation

[0018] The following is in conjunction with the appendix Figure 1 -Appendix Figure 2 This application will be described in further detail below.

[0019] Example: Please see Figure 1 As shown, the present invention provides a technical solution: a multi-sensor fusion-based intelligent detection device for all parameters of a chute, including a chute 1, a sensing unit 2 is installed inside the chute 1, and a main control unit 3, a communication unit 4, a power supply unit 5 and an auxiliary protection system 6 are installed outside the chute 1. The sensing unit 2 includes a millimeter-wave radar sensor 201, an infrared thermal imager 202, a laser displacement sensor 203, a vibration sensor 204, and a gas sensor 205; The auxiliary protection system 6 includes a high-pressure air curtain dustproof self-cleaning device 601 and an anti-collision buffer mechanism 602; The output of the power supply unit 5 is connected to the input of the main control unit 3 and the communication unit 4. The outputs of the sensing unit 2 and the auxiliary protection system 6 are connected to the input of the main control unit 3. The output of the main control unit 3 is connected to the input of the communication unit 4.

[0020] Main control unit 3 uses an explosion-proof industrial computer for multi-sensor data fusion analysis and decision control; Communication unit 4 adopts an intrinsically safe 5G / Wi-Fi 6 module for mining, which is used to transmit data collected by multiple sensors to the monitoring center in real time; Power supply unit 5 uses a dual power supply system of explosion-proof lithium battery and solar power to extend the battery life to 72 hours.

[0021] The main control unit 3 is configured to execute a multi-sensor data fusion algorithm, generate a three-dimensional digital twin model of the chute 1, and output an early warning signal based on the anomaly detection model; The three-dimensional digital twin model is constructed based on Kalman filtering and fusion of millimeter-wave radar, laser displacement, and infrared thermal imaging data.

[0022] Please see Figure 2 As shown, the intelligent detection method for all parameters of a well pass based on multi-sensor fusion is applied to an intelligent detection device for all parameters of a well pass based on multi-sensor fusion, and includes the following detection methods: S1. Start-up test: Start the power supply unit 5 and complete the self-test of each core unit, initialize the auxiliary protection system 6, ensure that the device meets the test operation conditions, complete the self-test of the core unit and the initialization of the auxiliary protection system 6 in advance, avoid interruption due to equipment failure during the test, and ensure the continuity of full parameter test of the chute 1 from the source.

[0023] Step S1 specifically includes: S1-1: Start the explosion-proof lithium battery and solar dual power system, and check the remaining power of the lithium battery and the on / off status of the solar power supply circuit to ensure that the dual power switching function is normal. By checking the status of the dual power system and the switching function, the explosion-proof lithium battery and solar power supply are stable. The 72-hour battery life can avoid the detection interruption caused by the power failure of a single power source, and significantly improve the power supply reliability and fault tolerance of the chute 1 detection process. S1-2: Initialize the explosion-proof industrial computer of the main control unit 3, load the multi-sensor data fusion algorithm and anomaly detection model, initialize the main control unit 3 in advance and load the core algorithm to avoid the delay of temporarily loading the algorithm during detection, so that the explosion-proof industrial computer can quickly have data processing capabilities and improve the overall detection efficiency. S1-3: Detect the network connection status of the intrinsically safe 5G / Wi-Fi 6 module for mining in communication unit 4, confirm that the communication link with the monitoring center is unobstructed, and ensure that the detection data of ore pass 1 can be transmitted to the monitoring center in real time and without interruption by verifying the intrinsically safe 5G / Wi-Fi 6 module link for mining, so as to avoid the problem of easy disconnection of communication without pre-check and ensure the timeliness of remote control. S1-4: Check the response signals of each sensor in the sensing unit 2 one by one to ensure that they can output detection signals normally. Check the devices in the sensing unit 2 one by one to avoid data loss due to the failure of a single sensor. Compared with the traditional no self-test mode, this ensures the integrity of the data source for the full parameter acquisition of the chute 1. S1-5: Initialize the auxiliary protection system 6, start the high-pressure air curtain dustproof self-cleaning device 601 for trial operation, reset the anti-collision buffer mechanism 602 to the initial protection position. By trial operation of the dustproof device and reset of the anti-collision mechanism, the protection function of the auxiliary protection system 6 can be activated in advance, which can reduce the impact of dust in the ore pass 1 on the sensor accuracy and the risk of ore impact damaging the equipment, and extend the sensor life.

[0024] S2. Data Acquisition: Multi-dimensional parameter data of the operation of chute 1 are collected synchronously by multiple sensors of sensing unit 2, and the collected data is transmitted to main control unit 3 in real time. Relying on the synchronous acquisition by multiple sensors of sensing unit 2, the multi-dimensional parameters of chute 1, such as material level, temperature and well wall deformation are covered, thereby improving the detection coverage.

[0025] Step S2 specifically includes: S2-1: Each sensor collects data according to a preset synchronization cycle. The millimeter-wave radar sensor 201 continuously scans the ore accumulation surface inside the ore pass 1 to obtain material level data. The infrared thermal imager 202 captures the temperature field distribution of the ore and the wall inside the ore pass 1 in real time. The laser displacement sensor 203 performs cyclic sampling at preset monitoring points on the wall of the ore pass 1 to obtain deformation data. The vibration sensor 204 continuously collects vibration signals of the wall of the ore pass 1 to extract ore runoff characteristics. The gas sensor 205 collects gas samples inside the ore pass 1 in real time to detect CO and CH4 concentrations and evaluates the ventilation status in combination with the ventilation speed data of the vent of the ore pass 1. Each sensor accurately collects key data of the ore pass 1 according to a preset cycle, such as the millimeter-wave radar measuring the material level and the laser displacement sensor 203 measuring the deformation, and captures the operating status in real time. Compared with manual intermittent collection, it reduces data lag and provides fine-grained raw data for evaluation. S2-2: All collected data are timestamped and transmitted in real time to the main control unit 3 via an intrinsically safe 5G / Wi-Fi 6 module for mining in an encrypted manner. If data packet loss occurs during transmission, the retransmission mechanism of the communication unit 4 is triggered. Through timestamping, encrypted transmission, and packet loss retransmission mechanism, the integrity and security of the detection data of the ore pass 1 are ensured, preventing data leakage and analysis errors caused by packet loss during transmission, and improving data reliability.

[0026] S3. Data Preprocessing and Fusion: The main control unit 3 performs noise reduction, calibration and time synchronization processing on the collected raw data. It adopts a multi-sensor fusion algorithm to achieve collaborative fusion of multi-source data, and removes invalid data to ensure data validity. The main control unit 3 performs noise reduction, calibration, synchronization and fusion on the raw data, removes invalid data, improves data quality, and provides a high-quality data foundation for the state analysis and model construction of the ore pass 1.

[0027] Step S3 specifically includes: S3-1: In the preprocessing stage, wavelet denoising algorithm is used to filter noise from the vibration signal collected by vibration sensor 204, and moving average method is used to smooth the concentration data collected by gas sensor 205. Combined with the environmental parameters of temperature and humidity at the well site, the detection data of laser displacement sensor 203 and millimeter-wave radar sensor 201 are calibrated for error. Wavelet denoising and moving average algorithms are used in a targeted manner to reduce the impact of environmental interference on the data of vibration sensor 204 and gas sensor 205. Combined with temperature and humidity calibration of laser displacement and millimeter-wave radar data, the problem of large errors due to unprocessed data is avoided, and the accuracy is improved. S3-2: In the time synchronization stage, the data with timestamps attached to each sensor are time-aligned based on the system clock of the main control unit 3 to ensure the consistency of multi-source data in the time dimension. Data time synchronization is achieved based on the system clock of the main control unit 3 to avoid fusion deviation caused by time misalignment of multi-source data, ensure the synergy of multi-dimensional data of chute 1, and improve fusion accuracy. S3-3: In the data fusion stage, a multi-source information fusion framework is adopted. Different weights are assigned based on the detection accuracy of each sensor. The Kalman filter algorithm is used to achieve the initial fusion of material level, well wall deformation, and temperature data. Then, the vibration characteristic data and gas concentration data are fused using the DS evidence theory to finally form a unified dataset of operating parameters for Well 1. At the same time, the 3σ criterion is used to remove extreme outliers in the dataset. The data is weighted according to sensor accuracy, and the Kalman filter and DS evidence theory are used to fuse the data. The 3σ criterion is then used to remove outliers to form a unified dataset, providing scientific and unified data support for the analysis of Well 1.

[0028] S4. 3D Model Construction: The main control unit 3 uses the Kalman filter algorithm to fuse data from millimeter-wave radar sensor 201, infrared thermal imager 202, laser displacement sensor 203, vibration sensor 204, and gas sensor 205 to construct a real-time 3D digital twin model of ore pass 1. This multi-sensor data-driven model visually recreates the ore accumulation, temperature distribution, and other conditions within ore pass 1. The dynamic update function reflects changes in real time, enhancing visualization and control capabilities. Step S4 specifically includes: S4-1: Based on the well wall structure parameters of the design drawings of well 1, and combined with the well wall monitoring point data sampled by the laser displacement sensor 203, an initial three-dimensional contour model of the well wall of well 1 is constructed. The model is built by combining the design drawings of well 1 and the measured data of the laser displacement sensor 203 to avoid the problem of deviation between the pure design drawings and the actual well wall, and to make the initial contour model of the well wall more in line with the real structure. S4-2: The material level data collected and fused by the millimeter-wave radar sensor 201 is converted into ore accumulation morphology parameters and superimposed onto the initial three-dimensional contour model of the well wall to generate a preliminary three-dimensional model of the ore chute 1 containing the ore accumulation state. The material level data of the millimeter-wave radar sensor 201 is converted into ore morphology parameters and superimposed to model the model so that the model not only contains the well wall structure, but also presents the internal material state, improving the state restoration accuracy. S4-3: Extract temperature field data collected by infrared thermal imager 202, color-mark the high temperature anomaly areas in the preliminary 3D model, and form a temperature-visualized 3D digital twin model of chute 1. The model is updated in real time according to the data acquisition cycle. By color-marking the high temperature areas and updating the model in real time, the high temperature anomaly of chute 1 is made intuitively visible. There is no need for manual analysis of temperature data to find anomalies. Compared with static models, temperature risks can be quickly detected, and the timeliness of anomaly identification is improved.

[0029] S5. Health Assessment: Based on the three-dimensional digital twin model and the fused parameter data, a comprehensive analysis is conducted on the material level status, well wall structure, temperature distribution, gas concentration and ventilation efficiency, and ore runoff risk of ore pass 1, and the health level of ore pass 1 is output. Based on the three-dimensional model and fused data, the multi-dimensional status of the material level and well wall of ore pass 1 is comprehensively analyzed and the health level is output.

[0030] Step S5 specifically includes: S5-1: Establish a health assessment index system for ore pass 1. The indexes include material level safety factor, well wall deformation rate, temperature anomaly ratio, gas safety index, ventilation efficiency, and ore runoff risk value. Establish an assessment system covering multiple indicators such as material level safety factor and ore runoff risk value, without missing any key risks. Compared with assessments with fewer indicators, this provides a comprehensive and scientific framework for the health scoring of ore pass 1, ensuring that the assessment has no blind spots. S5-2: A weighted scoring method is used to quantify and score each indicator. The well wall deformation rate, gas safety index, and ore run risk value are given higher weights, while the material level safety factor, temperature anomaly ratio, and ventilation efficiency are given normal weights. The health level of ore pass 1 is classified according to the total score, and a health assessment report is generated. The key indicator of well wall deformation rate is given high weight to avoid the weakening of key risks due to equal weights. The health status of ore pass 1 is made clearer by classifying the level according to the total score and generating a report.

[0031] S6. Anomaly Detection: Based on the anomaly detection model, the deviation between real-time parameter data and health baseline data is compared to identify the anomaly type and location in the operation of chute 1. By comparing real-time and baseline data with the trained model, the anomaly type of chute 1 is automatically identified and located, which greatly shortens the anomaly identification time and solves the problems of easy omission and difficulty in location by manual methods.

[0032] Step S6 specifically includes: S6-1: Based on the historical normal operation data and typical abnormal case data of chute 1, train a machine learning anomaly detection model, establish a feature threshold library for different anomaly types, train the model based on the historical normal data and abnormal cases of chute 1, so that the model can be adapted to the actual working conditions of chute 1, avoid the problem of high misjudgment rate of general models, and the established feature threshold library makes anomaly judgment more in line with reality. S6-2: Input the real-time parameter data fused in step S3 into the anomaly detection model, compare the deviation between the real-time data and the feature threshold library. If the deviation exceeds the set range, it is judged as an anomaly. Input the high-quality fused data after processing in S3. Compared with judging anomalies using the original data, it reduces the false judgment caused by interference and quickly identifies anomalies through deviation comparison. S6-3: Locating abnormal locations using a three-dimensional digital twin model. The abnormal location of well 1 can be intuitively located using a three-dimensional digital twin model, eliminating the need for manual inspection down the well, avoiding high-risk operations, and significantly improving the efficiency and safety of abnormal location.

[0033] S7. Tiered Early Warning and Detection Report Generation: Based on the abnormality detection data, if an abnormality is detected, an early warning level is assigned according to the severity of the abnormality, triggering the corresponding early warning mechanism; if no abnormality is detected, a detection report containing detection data, health assessment results, and abnormality information is generated and transmitted to the monitoring center through communication unit 4. Tiered early warning is issued according to the severity of the abnormality to avoid over-response or under-response. When no abnormality is detected, a complete report is generated and transmitted to the monitoring center. Compared with the single early warning and no-report mode, this makes the control more precise and efficient and ensures comprehensive information.

[0034] Step S7 specifically includes: S7-1: The early warning levels are divided into three levels: Level 1 warning, triggered by severe cracks in the well wall, gas concentration far exceeding the safety limit, or ore run-out; Level 2 warning, triggered by material level approaching the upper or lower limit, slight temperature abnormality, or gas concentration approaching the safety limit; Level 3 warning, triggered by slight fluctuations in sensor data, or the health level of ore pass 1 being medium. Level 1 warning corresponds to emergency situations such as cracks in the well wall, while Level 3 warning corresponds to slight fluctuations. Different warnings trigger different mechanisms to avoid delayed response to emergency situations or over-handling of minor issues. S7-2: The inspection report includes basic information, parameter data, health assessment results, abnormal information, and early warning records. After the report is generated, it is transmitted to the monitoring center in real time. The report covers all dimensions of basic information and parameter data and is transmitted to the monitoring center in real time, so that the management party does not need to retrieve data in a piecemeal manner, and can fully grasp the status of well 1, thereby improving the integrity of management.

[0035] S8. Linkage Control: Based on the early warning results and the type of anomaly, the main control unit 3 sends control commands to the auxiliary protection system 6 or the associated equipment of the ore 1 to realize linkage control for anomaly handling and feeds back the control results to the monitoring center. The main control unit 3 automatically sends commands to the auxiliary protection system 6 or the associated equipment without manual intervention. Compared with manual handling, it significantly shortens the anomaly response time, resolves the risk of ore 1 in a timely manner, and improves the timeliness and safety of anomaly handling.

[0036] Step S8 specifically includes: S8-1: If the dust concentration inside the chute 1 is too high, causing the sensor detection accuracy to decrease, the main control unit sends a start command to the high-pressure air curtain dust prevention and self-cleaning device 601. The air curtain wind speed and running time are dynamically adjusted according to the dust concentration until the sensor detection accuracy returns to normal. The high-pressure air curtain parameters are dynamically adjusted according to the dust concentration to quickly restore the sensor accuracy, avoid detection errors caused by dust accumulation, eliminate the need for manual disassembly and cleaning of the sensor, and improve detection stability and maintenance efficiency. S8-2: If the millimeter-wave radar sensor 201 detects that a large piece of ore is falling and poses a risk of impacting the well wall, the main control unit 3 controls the anti-collision buffer mechanism 602 to extend to the protective position to reduce the impact force of the ore. After the impact ends, the control mechanism is reset and the anti-collision buffer mechanism 602 is extended in advance to reduce the impact of the ore, protect the well wall of the ore pass 1 from damage, extend the service life of the ore pass 1, reduce maintenance costs, and ensure transportation safety. S8-3: If the gas sensor 205 detects that the CO or CH4 concentration exceeds the standard, the main control unit 3 sends a control command to the ventilation system of the chute 1 to increase the speed of the ventilation fan or turn on the standby fan until the gas concentration drops to a safe range. The ventilation system is automatically adjusted to reduce the CO and CH4 concentration, quickly eliminating the hidden danger of toxic gas. Compared with manually turning on the fan, this avoids the safety accident caused by the response delay and improves the safety of the working environment of the chute 1. S8-4: If the material level is close to the upper limit, the main control unit sends a stop command to the feeding equipment of chute 1. If the material level is close to the lower limit, it sends a start command and automatically controls the feeding equipment to adjust the material level of chute 1 to avoid blockage due to excessive material level or idling due to excessive material level, reduce the risk of transportation interruption, ensure smooth ore transmission, and improve operational stability.

[0037] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be covered within the scope of protection of this application.

Claims

1. A multi-sensor fusion-based intelligent detection device for all parameters of a chute, comprising a chute (1), characterized in that: The chute (1) is equipped with a sensing unit (2) inside and a main control unit (3), a communication unit (4), a power supply unit (5) and an auxiliary protection system (6) outside. The sensing unit (2) includes a millimeter-wave radar sensor (201), an infrared thermal imager (202), a laser displacement sensor (203), a vibration sensor (204), and a gas sensor (205). The auxiliary protection system (6) includes a high-pressure air curtain dustproof self-cleaning device (601) and an anti-collision buffer mechanism (602). The output of the power supply unit (5) is connected to the input of the main control unit (3) and the communication unit (4). The output of the sensing unit (2) and the auxiliary protection system (6) is connected to the input of the main control unit (3). The output of the main control unit (3) is connected to the input of the communication unit (4).

2. The intelligent detection device for all parameters of a well pass based on multi-sensor fusion as described in claim 1, characterized in that: The main control unit (3) adopts an explosion-proof industrial computer for multi-sensor data fusion analysis and decision control; The communication unit (4) adopts a mining intrinsically safe 5G / Wi-Fi 6 module, which is used to transmit the data collected by multiple sensors to the monitoring center in real time; The power supply unit (5) adopts a dual power supply system of explosion-proof lithium battery and solar energy to extend the battery life to 72 hours.

3. The intelligent detection device for all parameters of a well pass based on multi-sensor fusion as described in claim 1, characterized in that: The main control unit (3) is configured to execute a multi-sensor data fusion algorithm to generate a three-dimensional digital twin model of the chute (1) and output an early warning signal based on the anomaly detection model; The three-dimensional digital twin model is constructed based on Kalman filtering and fusion of millimeter-wave radar, laser displacement, and infrared thermal imaging data.

4. A method for intelligent detection of all parameters of a well pass based on multi-sensor fusion, applied to the intelligent detection device for all parameters of a well pass based on multi-sensor fusion as described in any one of claims 1-3, characterized in that: The following detection methods are included: S1. Start-up test: Start the power supply unit (5) and complete the self-test of each core unit, initialize the auxiliary protection system (6), and ensure that the device meets the test operation conditions; S2. Data acquisition: Multi-dimensional parameter data of the well (1) operation are collected synchronously by multiple sensors of the sensing unit (2), and the collected data is transmitted to the main control unit (3) in real time. S3. Data preprocessing and fusion: The main control unit (3) performs noise reduction, calibration and time synchronization processing on the collected raw data, and adopts a multi-sensor fusion algorithm to achieve collaborative fusion of multi-source data, and removes invalid data to ensure data validity. S4. Three-dimensional model construction: The main control unit (3) calls the Kalman filter algorithm to fuse the fused data of the millimeter-wave radar sensor (201), infrared thermal imager (202), laser displacement sensor (203), vibration sensor (204) and gas sensor (205) to construct a real-time three-dimensional digital twin model of the chute (1); S5. Health assessment: Based on the three-dimensional digital twin model and the fused parameter data, the material level status, well wall structure, temperature distribution, gas concentration and ventilation efficiency, and ore run risk of the ore pass (1) are comprehensively analyzed, and the health level of the ore pass (1) is output. S6. Anomaly detection: Based on the anomaly detection model, compare the deviation between real-time parameter data and health baseline data to identify the anomaly type and location of the anomaly in the operation of the chute (1); S7. Graded Early Warning and Detection Report Generation: Based on the data of anomaly detection, if an anomaly is generated, an early warning level is divided according to the severity of the anomaly, and the corresponding early warning mechanism is triggered; If no abnormality is found, a test report containing test data, health assessment results, and abnormality information is generated and transmitted to the monitoring center via the communication unit (4); S8. Linkage control: The main control unit (3) sends control commands to the auxiliary protection system (6) or the associated equipment of the chute (1) according to the early warning results and the type of abnormality, so as to realize the linkage control of abnormality handling and feed back the control results to the monitoring center.

5. The intelligent detection method for all parameters of a well pass based on multi-sensor fusion according to claim 4, characterized in that: Step S1 specifically includes: S1-1: Start the explosion-proof lithium battery and solar dual power system, and check the remaining power of the lithium battery and the on / off status of the solar power supply circuit to ensure that the dual power switching function is normal. S1-2: Initialize the explosion-proof industrial computer of the main control unit (3) and load the multi-sensor data fusion algorithm and anomaly detection model; S1-3: Detect the network connection status of the intrinsically safe 5G / Wi-Fi 6 module for mining in the communication unit (4) and confirm that the communication link with the monitoring center is smooth; S1-4: Detect the response signals of each sensor in the sensing unit (2) one by one to ensure that all of them can output detection signals normally; S1-5: Initialize the auxiliary protection system (6), start the high-pressure air curtain dustproof self-cleaning device (601) for trial operation, and reset the anti-collision buffer mechanism (602) to the initial protection position; Step S2 specifically includes: S2-1: Each sensor collects data according to the preset synchronization cycle. The millimeter-wave radar sensor (201) continuously scans the ore accumulation surface inside the ore pass (1) to obtain material level data. The infrared thermal imager (202) captures the temperature field distribution of the ore and well wall inside the ore pass (1) in real time. The laser displacement sensor (203) performs cyclic sampling of the preset monitoring points on the well wall of the ore pass (1) to obtain deformation data. The vibration sensor (204) continuously collects the vibration signal of the well wall of the ore pass (1) to extract the ore run-out characteristics. The gas sensor (205) collects the gas sample inside the ore pass (1) in real time to detect the CO and CH4 concentrations, and evaluates the ventilation status in combination with the wind speed data of the vent of the ore pass (1). S2-2: All collected data are timestamped and transmitted to the main control unit (3) in real time via an intrinsically safe 5G / Wi-Fi 6 module for mining in an encrypted transmission method. If data packet loss occurs during transmission, the retransmission mechanism of the communication unit (4) is triggered.

6. The intelligent detection method for all parameters of a well pass based on multi-sensor fusion according to claim 4, characterized in that: The S3 step specifically includes: S3-1: In the preprocessing stage, the vibration signal collected by the vibration sensor (204) is filtered for noise using the wavelet denoising algorithm, and the concentration data collected by the gas sensor (205) is smoothed using the moving average method. Combined with the on-site temperature and humidity environmental parameters of the chute (1), the detection data of the laser displacement sensor (203) and the millimeter-wave radar sensor (201) are calibrated for error. S3-2: During the time synchronization phase, the system clock of the main control unit (3) is used as a reference to perform time alignment on the data with timestamps attached to each sensor to ensure the consistency of multi-source data in the time dimension. S3-3: In the data fusion stage, a multi-source information fusion framework is adopted, and different weights are assigned based on the detection accuracy of each sensor. The Kalman filter algorithm is used to achieve the initial fusion of material level, well wall deformation, and temperature data. Then, the vibration characteristic data and gas concentration data are fused through the DS evidence theory to finally form a unified well chute (1) operating parameter dataset. At the same time, the 3σ criterion is used to remove extreme outliers in the dataset. Step S4 specifically includes: S4-1: Based on the well wall structure parameters of the design drawings of the chute (1), and combined with the well wall monitoring point data sampled by the laser displacement sensor (203) in a cyclic sampling, an initial three-dimensional contour model of the well wall of the chute (1) is constructed. S4-2: The material level data collected and fused by the millimeter-wave radar sensor (201) is converted into ore accumulation morphology parameters and superimposed onto the initial three-dimensional contour model of the well wall to generate a preliminary three-dimensional model of the ore chute (1) containing the ore accumulation state. S4-3: Extract the temperature field data collected by the infrared thermal imager (202), color-mark the high temperature anomaly area in the preliminary three-dimensional model, and form a temperature-visualized chute (1) three-dimensional digital twin model. The model is updated in real time according to the data acquisition cycle.

7. The intelligent detection method for all parameters of a well pass based on multi-sensor fusion according to claim 4, characterized in that: Step S5 specifically includes: S5-1: Establish a health assessment index system for ore pass (1), including the material level safety factor, well wall deformation rate, temperature anomaly ratio, gas safety index, ventilation efficiency, and ore run-off risk value. S5-2: The weighted scoring method is used to quantify the scores of each indicator. Among them, the well wall deformation rate, gas safety index and ore run risk value are given higher weights, while the material level safety factor, temperature anomaly ratio and ventilation efficiency are given conventional weights. The health level of the ore pass (1) is divided into three categories according to the total score, and a health assessment report is generated. Step S6 specifically includes: S6-1: Based on the historical normal operation data and typical abnormal case data of the chute (1), train the machine learning anomaly detection model and establish a feature threshold library for different anomaly types; S6-2: Input the real-time parameter data fused in step S3 into the anomaly detection model, compare the deviation between the real-time data and the feature threshold library, and if the deviation exceeds the set range, it is determined to be an anomaly. S6-3: Locating abnormal locations using a three-dimensional digital twin model.

8. The intelligent detection method for all parameters of a well pass based on multi-sensor fusion according to claim 4, characterized in that: The S7 step specifically includes: S7-1: The warning level is divided into three levels: Level 1 warning, the triggering conditions are severe cracking of the well wall, gas concentration far exceeding the safety limit, and ore run-off; Level 2 warning, the triggering conditions are that the material level is close to the upper or lower limit, the temperature is slightly abnormal, and the gas concentration is close to the safety limit; Level 3 warning, the triggering conditions are that the sensor data fluctuates slightly, and the health level of the ore pass (1) is medium. S7-2: The test report includes basic information, parameter data, health assessment results, abnormal information, and early warning records. The report is transmitted to the monitoring center in real time after it is generated. The S8 step specifically includes: S8-1: If the dust concentration inside the chute (1) is too high, causing the sensor detection accuracy to decrease, the main control unit sends a start command to the high-pressure air curtain dust prevention self-cleaning device (601) and dynamically adjusts the air curtain wind speed and running time according to the dust concentration until the sensor detection accuracy returns to normal. S8-2: If the millimeter-wave radar sensor (201) detects that a large piece of ore is falling and there is a risk of impacting the well wall, the main control unit (3) controls the anti-collision buffer mechanism (602) to extend to the protective position to reduce the impact force of the ore. After the impact ends, the control mechanism is reset. S8-3: If the gas sensor (205) detects that the CO or CH4 concentration exceeds the standard, the main control unit (3) sends a control command to the ventilation system of the chute (1) to increase the speed of the ventilation fan or turn on the standby fan until the gas concentration drops to a safe range. S8-4: If the material level is close to the upper limit, the main control unit sends a stop command to the feed equipment of the chute (1). If the material level is close to the lower limit, a start command is sent.

Images