Method and system for detecting and locating conveyor belt metal foreign objects using array transient electromagnetic

Abstract

The application provides a conveyor metal foreign matter detection and positioning method and system of array transient electromagnetic, and belongs to the technical field of conveyor safety monitoring, the method comprises the following steps: outputting an alternating magnetic field according to a preset frequency period; collecting signals and performing pretreatment; calculating difference parameters, and extracting a region of interest; identifying foreign matters based on a BP neural network model; three-dimensional positioning of foreign matters; early warning output; the system: the array detection module comprises two symmetrical detection panels; the multi-frequency excitation module comprises one main frequency converter, two slave frequency converters and eight excitation coils; the signal processing module integrates a filtering submodule, a data storage submodule and a foreign matter intelligent identification module, and is connected with the array detection module, the multi-frequency excitation module and the positioning and display module respectively; the positioning and display module comprises an infrared sensor, a gravity sensor and an explosion-proof display screen. The application can accurately detect metal foreign matters on a conveyor belt in real time and timely perform early warning.

Description

Technical Field

[0001] This invention belongs to the field of conveyor safety monitoring technology, specifically relating to an array-type transient electromagnetic method and system for detecting and locating metallic foreign objects in conveyors. Background Technology

[0002] In underground coal mine production systems, belt conveyors are the core equipment for coal transportation, undertaking more than 90% of the coal transport tasks. Their operational stability directly determines the coal mine's production efficiency and operational safety. However, during coal mining and transshipment, large metal foreign objects such as anchor bolts used for underground roadway support, scraper conveyor parts, and blasting residue can easily get mixed into the transported coal, becoming a core hidden danger that can induce belt conveyor failures.

[0003] Current methods for detecting foreign metal objects in underground coal mines are insufficient to meet the demands for high-precision, full-coverage detection under complex working conditions, exhibiting the following shortcomings: First, it is difficult to balance detection range and false negative rate: the current mainstream detection method uses single-point electromagnetic sensors (such as single-coil detectors deployed under conveyor belts), whose effective detection radius is typically only 0.3–0.5 m, while the width of underground conveyor belts is mostly 1.2–2 m, requiring dozens of sensors to be densely arranged to achieve full coverage, resulting in a significant increase in equipment costs. Simultaneously, single-point sensors are susceptible to interference from underground dust and vibration, with a false negative rate exceeding 40% for small-sized metal foreign objects of 5–10 cm. Furthermore, manual inspection methods are limited by factors such as dim lighting, high dust concentration, and harsh working environments underground, making real-time continuous detection impossible and further exacerbating the risk of false negatives. Secondly, the foreign object identification capability is weak, resulting in a high false alarm rate: Existing detection technologies mostly use single-frequency electromagnetic excitation, which can only determine "whether there is a metallic foreign object," but cannot distinguish between "threatening foreign objects" (such as anchor bolts or scrapers longer than 30cm, which can easily cause conveyor belt tearing) and "harmless metals" (such as thin iron wires with a diameter of less than 1cm, which do not affect equipment operation). This results in a false alarm rate of up to 20% for the detection and positioning system, requiring maintenance personnel to frequently shut down the system to troubleshoot false alarms, which not only increases labor intensity but also seriously affects the normal production rhythm of the coal mine. Furthermore, the positioning accuracy is insufficient, leading to low fault handling efficiency: Traditional detection technologies can only roughly determine that the foreign object is located within the "head-to-tail" range, and cannot accurately locate the specific coordinates of the foreign object in the width and vertical height directions of the conveyor belt. When a foreign object alarm occurs, maintenance personnel need to disassemble and inspect the entire length of the conveyor belt, a process that takes 2-3 hours, further extending downtime and increasing economic losses.

[0004] In summary, existing technologies cannot simultaneously meet the core requirements of "full-area coverage detection, accurate type identification, and three-dimensional coordinate positioning," and are ill-suited to complex application scenarios in coal mines, such as dust, vibration, and wide conveyor belts. Therefore, there is an urgent need for a metal foreign object detection technology that can adapt to the harsh underground environment and possesses full-area, blind-spot-free detection, accurate identification of threatening foreign objects, and precise three-dimensional coordinate positioning capabilities. This has become a pressing technical problem to be solved in the field of coal mine transportation safety. Summary of the Invention

[0005] To address the problems of the existing technologies, this invention provides an array-type transient electromagnetic method for detecting and locating metallic foreign objects on conveyors. This method is simple to implement, low in cost, and highly accurate in detection and location. It can detect metallic foreign objects on the conveyor belt in real time and provide timely warnings, effectively reducing the risk of conveyor belt failures. The system has a simple structure, high level of intelligence, and low manufacturing cost. It can achieve rapid and accurate detection and location of foreign objects on the conveyor belt, significantly reducing the incidence of accidents such as conveyor belt tearing and improving coal mine production efficiency.

[0006] To achieve the above objectives, the present invention provides a method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array, characterized by comprising the following methods;

[0007] Step 1: System initialization and calibration; Collect the voltage, phase, and reactance reference values ​​of the array coil under foreign object-free conditions, and periodically and dynamically update the reference values ​​during the operation of the detection and positioning system;

[0008] Step 2: Multi-frequency magnetic field excitation; output an alternating magnetic field at a preset frequency cycle to generate eddy currents and secondary induced magnetic fields in the metal foreign objects on the conveyor belt;

[0009] Step 3: Signal Acquisition and Preprocessing; The induced signal is acquired through an array coil and preprocessed for noise reduction using a wavelet thresholding and Kalman filtering fusion algorithm;

[0010] Step 4: Anomaly detection and ROI extraction; Calculate the difference parameters between the real-time signal and the reference signal, identify suspected foreign object regions based on the difference parameters, perform cluster analysis on the suspected foreign object regions, and extract the region of interest corresponding to each suspected foreign object region;

[0011] Step 5: Foreign object identification; Based on physical rules, the material of the foreign object is quickly initially determined, and then the specific type is accurately identified through a BP neural network, forming a dual identification mechanism that complements rapid screening and precise judgment;

[0012] Step Six: 3D positioning of the foreign object; the planar coordinates are solved using a planar positioning algorithm, the height coordinates are solved based on the signal strength ratio of the upper and lower coils, and the planar coordinates and height coordinates are fused to obtain the 3D position;

[0013] Step 7: Early warning output; tiered early warnings are issued based on foreign object parameters, and external devices are activated in conjunction with these warnings.

[0014] Furthermore, in order to effectively improve the recognition accuracy of metals of different materials, in step two, the preset frequencies are 4kHz, 8kHz, and 12kHz, and the frequencies are output alternately in the order of 4kHz, 8kHz, and 12kHz, with a continuous excitation time of 0.1s for each frequency.

[0015] Furthermore, in order to effectively focus on valid data, reduce the overall computational load, and effectively improve processing speed, the anomaly detection and ROI extraction process in step four is as follows:

[0016] S41: Set voltage change threshold Voltage change for all coils Perform a comprehensive detection, first filtering out voltage changes. Exceeding the voltage change threshold Candidate abnormal coils are identified, and then the adjacency relationship of the candidate abnormal coils is checked. When there are no less than 3 adjacent candidate abnormal coils that are triggered synchronously, the group of adjacent coils is determined to be the trigger coil.

[0017] S42: A distance-based clustering algorithm is used to cluster the selected coil groups, and coil groups in the same connected domain are identified as signal regions triggered by the same foreign object, thus realizing the preliminary division of suspected foreign object regions.

[0018] S43: For each suspected foreign object region obtained by division, extract the coil signal data of the region and its surrounding preset range to form a ROI data set; the preset range is a 5×5 coil centered on the suspected foreign object region.

[0019] As a preferred option, the foreign object identification process in step five is as follows:

[0020] S51: Rapid material screening based on physical rules;

[0021] Using the change in reactance With phase shift The preliminary material determination is based on the following rules: when and satisfy At that time, it was initially determined to be a ferromagnetic foreign object; when and satisfy At that time, it was initially determined to be a non-ferromagnetic foreign object;

[0022] S52: Accurate type identification based on BP neural network;

[0023] voltage change Phase shift Reactance change The components are combined into a feature vector, which is then input into a pre-trained BP neural network model. The model then performs computations to identify and output the specific type of the foreign object and its corresponding recognition confidence level.

[0024] S53: Verification of recognition results fusion;

[0025] If the material properties corresponding to the specific type output by the BP neural network are consistent with the results of the rapid material screening based on physical rules, then the specific type output by the neural network shall prevail; if the two results contradict each other, it is determined that the current signal is affected by environmental interference or encounters an unknown new foreign object. The system automatically increases the sampling frequency for secondary confirmation and adopts the safety-oriented conclusion of the physical rules to ensure the safe operation of the conveyor belt.

[0026] Furthermore, in order to quickly and accurately obtain the location information of the foreign object, the process of fusing the three-dimensional position in step six is ​​as follows:

[0027] S61: Planar coordinate positioning is performed using the centroid positioning method or the nonlinear least squares fitting method;

[0028] S61-1: Planar coordinate positioning using the centroid positioning method: Obtain the relative coordinates of the foreign object in the width and thickness directions of the conveyor belt according to formulas (1) and (2) respectively. , ;

[0029] (1);

[0030] (2);

[0031] In the formula, , The first The physical coordinates of each detection coil in the width and thickness directions; For the first The amount of voltage change sensed by each detection coil; The total number of valid detection coils involved in the calculation;

[0032] S61-2: Planar coordinate positioning using nonlinear least squares fitting method: First, establish the location of the foreign object based on the electromagnetic field propagation theory. The mathematical mapping relationship between it and the response signals of each coil Then, with the goal of minimizing the sum of squares of the residuals between the measured voltage change and the theoretical prediction, an optimization function is constructed according to formula (3); subsequently, a nonlinear optimization algorithm is used to gradually approach the optimal solution from the initial value.

[0033] (3);

[0034] S62: Absolute position fusion in the length direction; first, infrared sensors are used for speed monitoring and time recording to obtain the conveyor belt running speed. The time point at which a foreign object triggers a specific reference coil Then, through spatiotemporal information fusion and coordinate correction, the global absolute coordinates are obtained, as shown in formula (4).

[0035] (4);

[0036] In the formula, This refers to the longitudinal position of the foreign object in the conveyor belt. To detect the fixed distance from the starting end of the panel to the head of the conveyor;

[0037] S63: Determine the position in the width direction; combine the peak coil coordinates to perform an intuitive conversion of the position in the width direction, and obtain the absolute position of the foreign object in the conveying width direction according to formula (5). ;

[0038] (5);

[0039] In the formula, These are the column coordinates of the peak coil; The width of a single coil;

[0040] S64: Height coordinate positioning; first, target the already determined horizontal projection area. Based on formulas (6) and (7), summarize all effective voltage responses of the upper and lower detection panels respectively. , Then, the signal amplitude is standardized according to formula (8) to obtain the normalized intensity scaling factor. Then, the vertical distance between the foreign object and the conveyor belt surface, i.e., the height position, is obtained according to formula (9). ;

[0041] (6);

[0042] (7);

[0043] (8);

[0044] (9);

[0045] In the formula, This refers to the installation interval between the upper and lower detection panels.

[0046] This technical solution employs a combination of complementary algorithms, mathematical formula quantification, and multi-dimensional information fusion design, which combines positioning flexibility and accuracy. It can achieve full-dimensional coordinate calculation without additional sensors, adapt to the dynamic working conditions of belt conveyors, and has strong traceability. It can accurately locate foreign objects in three-dimensional space of length, width, and height, thus shortening the time for foreign object inspection.

[0047] Furthermore, in order to achieve effective group-based early warning and ensure the safe and stable operation of the conveyor, the process of graded early warning based on foreign object parameters and linkage with external equipment in step seven is as follows:

[0048] S71: Based on the type, size, and location of large ferromagnetic foreign objects, the warning level is determined according to the following priority: Level 1 warning corresponds to: any foreign object ≥ 50cm; or a ferromagnetic foreign object > 30cm located in the middle; Level 2 warning corresponds to: ferromagnetic foreign objects > 30cm and < 50cm located at the edge; ferromagnetic foreign objects ≥ 10cm and ≤ 30cm; non-ferromagnetic foreign objects ≥ 30cm and < 50cm; unknown foreign objects > 30cm; Level 3 warning corresponds to: ferromagnetic foreign objects < 10cm; non-ferromagnetic foreign objects < 30cm; unknown foreign objects ≤ 30cm; if a foreign object meets multiple conditions, the highest level is taken; unknown foreign objects are defined by an identification confidence level below 70%, and are no longer downgraded solely based on confidence level; when location information is missing, the warning level is determined according to the case where the foreign object is located in the middle area, ensuring that large ferromagnetic foreign objects are not missed due to misjudgment of location;

[0049] S72: The warning level results are displayed in real time on the screen. At the same time, the warning level is uploaded to the coal mine production monitoring system, and the on-site audible and visual alarms are triggered simultaneously. The first-level warning is indicated by a red light, the second-level warning by a yellow light, and the third-level warning by a blue light. In addition, in conjunction with the foreign object location linkage equipment control, if the foreign object is located near the machine head, the conveyor will be automatically slowed down to buy time for on-site handling.

[0050] This invention is based on the principle of transient electromagnetic induction, operating within a closed-loop logic of alternating magnetic field excitation, metal eddy current induction, differential signal acquisition, and intelligent algorithm analysis. It utilizes the synergistic effect of excitation and receiving coils as the core for detecting and locating metallic foreign objects. First, the excitation coil outputs alternating currents at 4kHz, 8kHz, and 12kHz, forming a multi-frequency alternating magnetic field uniformly covering the entire width of the conveyor belt, providing a stable excitation source for foreign object induction. Then, under multi-frequency excitation, the eddy current responses of different metal materials differ significantly. The receiving coil array captures the superimposed signal of the primary magnetic field (directly generated by the excitation coil) and the secondary magnetic field (generated by metal eddy currents), providing key features for accurate identification of foreign object types. After processing with Kalman filtering and wavelet threshold fusion algorithms, environmental noise from underground dust and vibration is effectively removed, retaining only the core effective signal, thus better adapting to the complex environment of high dust and strong vibration underground. Subsequently, feature extraction is performed on the pre-processed signal, and the voltage change is calculated. Phase shift Reactance change Key parameters such as the foreign object are analyzed; further, a pre-trained BP neural network is used to intelligently analyze these differential parameters, thereby completing the determination of the presence of foreign objects and the identification of their specific types; at the same time, the centroid positioning and nonlinear least squares fitting algorithm are used to calculate the three-dimensional spatial coordinates of the foreign object, realizing the precise positioning of the foreign object and triggering risk warnings simultaneously, providing effective technical support for rapid disposal.

[0051] Compared with the prior art, the present invention has the following advantages:

[0052] 1. The detection capability has been comprehensively improved from missed detection to full coverage; the array coil layout combined with the dynamic calibration algorithm can adapt to the harsh working conditions of downhole with dust concentration of 100mg / m³ and vibration frequency of 5~10Hz. The detection rate of large metal foreign objects such as anchor bolts and scrapers is ≥99%, while the missed detection rate of traditional technology in this environment exceeds 30%, and the stability is significantly better than the traditional solution.

[0053] 2. The recognition accuracy has been significantly optimized from false alarms to accurate differentiation. Through the fusion of multi-frequency excitation and BP neural network, it can accurately distinguish between ferromagnetic and non-ferromagnetic foreign objects, as well as the specific type of foreign object. The accuracy of distinguishing between ferromagnetic and non-ferromagnetic foreign objects is ≥98%, and the accuracy of identifying specific types is ≥92%. The false alarm rate has been significantly reduced. Taking a coal mine application as an example, the false alarm rate was found to have dropped from 20% of the traditional technology to below 3%.

[0054] 3. Positioning efficiency has been significantly improved from blind to precise location; practical verification shows that the 3D positioning accuracy error is ≤10cm, allowing maintenance personnel to quickly and accurately locate foreign objects directly based on the positioning coordinates, reducing the traditional 2-3 hour investigation time to 10-15 minutes. In a coal mine application, the time for a single foreign object disposal was reduced from 120 minutes to 15 minutes, fully demonstrating the significant improvement in positioning efficiency.

[0055] 4. The operation and maintenance costs have been significantly reduced from high consumption to high efficiency; no additional zero-adjustment coils and dense single-point sensors are required, and the hardware cost is reduced by about 40% compared with the traditional solution; at the same time, through automatic calibration and remote early warning functions, the frequency of manual inspection is reduced from 3 times a day to 1 time, reducing labor costs and workload.

[0056] This method is simple to implement, low in cost, and highly accurate in detection and positioning. By effectively integrating array-type electromagnetic induction, multi-frequency excitation, and intelligent recognition algorithms, it constructs a complete solution for foreign object detection and positioning on conveyor belts. Ultimately, it achieves the goals of early warning, accurate positioning, and rapid processing, significantly reducing the probability of accidents such as conveyor belt tearing and greatly improving the production efficiency and safety factor of coal mines. It provides effective technical support for the safe and stable operation of belt conveyors and can be effectively used in complex underground production environments.

[0057] This invention also provides a conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics, used to implement a method for detecting and locating conveyor metal foreign objects using array-type transient electromagnetics, comprising:

[0058] An array-type detection module includes two symmetrically arranged detection panels, which are distributed opposite each other on the upper and lower sides of the conveyor belt carrying section. Each detection panel is equipped with several receiving coils arranged in a matrix to form a detection area that covers the entire area.

[0059] The multi-frequency excitation module includes one main frequency converter, two slave frequency converters, and eight excitation coils. The eight excitation coils are connected in parallel and evenly distributed around the four edges of the two detection panels, with a distance of 10cm between them and the detection panels. The multi-frequency excitation module is used to output 4kHz, 8kHz, and 12kHz alternating magnetic fields.

[0060] The signal processing module integrates a filtering submodule, a data storage submodule, and a foreign object intelligent identification module. The signal processing module is connected to the array detection module and the multi-frequency excitation module, respectively, and is used to realize signal denoising, difference parameter calculation and foreign object intelligent identification, as well as hierarchical early warning based on the identification results.

[0061] The positioning and display module includes an infrared sensor, a gravity sensor, and an explosion-proof display screen. The positioning and display module is connected to a signal processing module. The infrared sensor is installed at the drive shaft end of the drive roller to collect the running speed signal of the conveyor belt. The gravity sensor is installed at the bottom of the detection panel on the lower side to detect the weight signal of foreign objects. The explosion-proof display screen is used to display three-dimensional coordinates and warning information.

[0062] To improve detection accuracy, the receiving coils in the two detection panels of the array detection module are completely symmetrical, the distance between the two detection panels is 0.8 to 1.2 m, the size of a single receiving coil is 6 cm × 6 cm, the number of receiving coils in each detection panel is ≥500, and the sampling frequency is 100 Hz.

[0063] To effectively eliminate signal drift caused by temperature, vibration, and noise, the signal processing module triggers dynamic calibration every 2 hours, completing data acquisition of the foreign object-free area during a 0.5s excitation pause. The reference value is updated when the acquired data deviates from the reference value by more than ±3%. Preferably, the positioning and display module features an infrared sensor detection accuracy ≤0.01m / s, a gravity sensor weighing error ≤50g, and an explosion-proof display screen with an explosion-proof rating of Ex ib 1 Mb.

[0064] In this invention, two detection panels, positioned vertically, form a pincer detection layout on the conveyor belt. Each detection panel is equipped with several receiving coils arranged in a matrix, achieving full-width and full-length detection of the conveyor belt without blind spots, effectively eliminating missed detection zones. Eight excitation coils are divided into two groups and evenly distributed around the edges of the two detection panels, forming a double-layered, surrounding magnetic field structure. This ensures that the magnetic field strength distribution deviation across the conveyor belt cross-section is ≤5%, further eliminating detection blind spots in the central area. The annular arrangement of four coils in each group creates a region with a nearly uniform magnetic field within the detection panel area, ensuring more stable eddy current induction intensity for metallic foreign objects regardless of their location on the conveyor belt, avoiding missed detections due to magnetic field gradients. The signal processing module facilitates not only noise reduction preprocessing and difference parameter calculation of the acquired signals but also allows for accurate identification of foreign object material and type based on the difference parameters using the intelligent foreign object identification module. Furthermore, it enables tiered early warning based on the identification results. The positioning and display module is designed to facilitate the acquisition of basic parameters required for positioning using its infrared and gravity sensors, thereby enabling the signal processing module to quickly obtain positioning results. The explosion-proof display feature allows for the easy display of the three-dimensional coordinates of foreign objects and early warning information, enabling relevant personnel to intuitively observe positioning and warning information.

[0065] The system has a simple structure, a high degree of intelligence, and low manufacturing cost. It can quickly and accurately detect and locate foreign objects on the conveyor belt, which can significantly reduce the incidence of accidents such as conveyor belt tearing and improve coal mine production efficiency. Attached Figure Description

[0066] Figure 1 This is a schematic diagram of a metal positioning principle based on a transient electromagnetic array;

[0067] Figure 2 This is a schematic diagram showing the arrangement of an excitation coil and multiple receiving coils;

[0068] Figure 3 This is a flowchart of the method portion of this invention;

[0069] Figure 4 This is a principle block diagram of the system part of this invention;

[0070] Figure 5 This is a schematic diagram of the positioning system in this invention being assembled on a conveyor.

[0071] In the diagram: 1. Detection panel, 2. Conveyor belt, 3. Receiving coil, 4. Excitation coil, 5. Conveyor. Detailed Implementation

[0072] The invention will now be further described with reference to the accompanying drawings.

[0073] like Figures 1 to 3 As shown, the present invention provides a method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array, characterized by comprising the following methods;

[0074] Step 1: System initialization and calibration;

[0075] Two detection panels 1 in the array-type detection module are arranged five meters away from the head of the conveyor 5, with the two detection panels 1 arranged vertically opposite each other on the upper and lower sides of the carrying section of the conveyor belt 2, forming a detection area that clamps the conveyor belt 2 from both above and below. The distance between the edge of the detection panel 1 and the sides of the conveyor belt 2 is preferably 5 cm, and the spacing between the two detection panels 1 is preferably set to 1 m. Eight excitation coils 4 in the multi-frequency excitation module are connected in parallel and evenly installed on the four edges of the two detection panels 1. At the same time, a communication connection is established between the signal processing module and the array-type detection module, the multi-frequency excitation module and the explosion-proof display screen. Preferably, the signal processing module is connected to the array-type detection module, the multi-frequency excitation module and the positioning and display module through an explosion-proof cable.

[0076] Within 30 minutes of the detection and positioning system's startup, the voltage, phase, and reactance reference values ​​of the array coils in a foreign object-free state are collected. During system operation, these reference values ​​are periodically and dynamically updated. Specifically, a standard frequency magnetic field is output using a multi-frequency excitation module, and the voltage, phase, and reactance parameters of all coils are synchronously collected by an array-type detection module as initial reference values. During real-time operation, dynamic calibration is triggered every 2 hours via a signal processing module. During dynamic calibration, multi-frequency excitation is paused for 0.5 seconds, and coil data from the current foreign object-free area is collected and compared with the initial reference value. If the deviation exceeds ±3%, the reference value is updated (without downtime and without affecting production). This eliminates the need for a dedicated zero-adjustment coil in traditional detection technologies, enabling automatic calibration solely through software algorithms, reducing hardware complexity and adapting to complex environmental conditions such as downhole vibration and temperature variations.

[0077] Step 2: Multi-frequency magnetic field excitation;

[0078] The multi-frequency excitation module outputs an alternating magnetic field at a preset frequency cycle, causing eddy currents and secondary induced magnetic fields to be generated in the metal foreign objects on the conveyor belt 2.

[0079] Step 3: Signal Acquisition and Preprocessing;

[0080] The induced signals of each receiving coil 3 at different frequencies are collected by the array coil, and then the noise is preprocessed by the wavelet threshold and Kalman filter fusion algorithm.

[0081] Specifically, the upper and lower array coils synchronously acquire induced voltage signals. During the acquisition process, the voltage, phase, and reactance parameters of each receiving coil 3 are recorded in real time and then transmitted to the signal processing module. During data transmission, the coil data of the suspected foreign object area (approximately 50 to 100 coils, accounting for 5% to 10% of the total coils) are packaged and transmitted to the PLC to reduce the amount of data transmitted and improve the processing speed. After receiving the data, the signal processing module performs front-end amplification and filtering.

[0082] Step 4: Anomaly detection and ROI extraction;

[0083] The signal processing module is used to calculate the difference parameters between the real-time signal and the reference signal. Based on the difference parameters, suspected foreign object regions are identified. After cluster analysis of the suspected foreign object regions, the region of interest (ROI) corresponding to each suspected foreign object region is extracted.

[0084] Specifically, the real-time acquired signal is compared with the reference signal, and three key difference parameters are calculated, namely the voltage change. Phase shift Reactance change Among them, voltage change = Actual voltage - Reference voltage (threshold set to ±5%, exceeding this value indicates a suspected foreign object signal); Phase offset = Actual phase - Reference phase; Ferromagnetic foreign objects cause a phase lag of 5°–10°, while non-ferromagnetic foreign objects cause a phase lead of 3°–8°; Reactance change = Actual reactance - Reference reactance; The reactance change of ferromagnetic foreign objects is positive, and the reactance change of non-ferromagnetic foreign objects is negative;

[0085] Step 5: Foreign Object Identification;

[0086] The signal processing module is equipped with a pre-trained BP neural network model. After obtaining the difference parameters, it quickly makes a preliminary judgment on the material of the foreign object based on physical rules, and then accurately identifies the specific type through the BP neural network, forming a dual identification mechanism that complements fast screening and precise judgment; Step 6: Three-dimensional localization of the foreign object;

[0087] The planar coordinates are solved by a planar positioning algorithm, the height coordinates are solved based on the signal strength ratio of the upper and lower coils, and the three-dimensional position is obtained by fusing the planar coordinates and the height coordinates.

[0088] Step 7: Early warning output;

[0089] The system provides graded early warnings based on foreign object parameters and links them to external devices.

[0090] As a preferred method, in step two, the preset frequencies are 4kHz, 8kHz, and 12kHz, and the output is cyclically alternated in the order of 4kHz-8kHz-12kHz, with a continuous excitation time of 0.1s for each frequency. Because the "eddy current effect" of different metallic foreign objects varies significantly at specific frequencies—for example, ferromagnetic foreign objects (anchor rods) have the strongest induced signal at 4kHz, while non-ferromagnetic foreign objects (copper connectors) show the most obvious signal difference at 12kHz—multi-frequency excitation can effectively improve the identification accuracy of different metal materials.

[0091] In order to effectively focus on valid data, reduce the overall computational load, and effectively improve processing speed, the anomaly detection and ROI extraction process in step four is as follows:

[0092] S41: Set voltage change threshold Voltage change for all coils Perform a comprehensive detection, first filtering out voltage changes. Exceeding the voltage change threshold Candidate abnormal coils are identified, and then the adjacency relationship of the candidate abnormal coils is checked. When there are no less than 3 adjacent candidate abnormal coils that are triggered synchronously, the group of adjacent coils is determined to be the trigger coil.

[0093] Preferably, Set as voltage change ±5%, of course, the voltage change threshold. It can also be dynamically adjusted according to the real-time noise of the detection environment to adapt to the detection needs in different environments, eliminate the interference of environmental noise on the judgment of foreign objects, and improve the accuracy of the judgment of suspected foreign objects.

[0094] S42: A distance-based clustering algorithm is used to cluster the selected coil groups. The clustering algorithm selected is either the DBSCAN algorithm or the connected component analysis algorithm. The coil groups in the same connected component are determined to be the signal regions triggered by the same foreign object, so as to realize the preliminary division of the suspected foreign object region and avoid misjudgment caused by a single abnormal coil or a small number of non-adjacent abnormal coils.

[0095] S43: For each suspected foreign object area obtained, extract the coil signal data of the area and its surrounding preset range to form a ROI data set; the preset range is a 5×5 coil centered on the suspected foreign object area. The ROI data set is used for subsequent fine positioning of foreign objects, and the overall computation is reduced by focusing on the effective data range.

[0096] In this way, by quickly determining whether there are foreign objects and extracting ROI data before entering the computationally intensive localization stage, the amount of subsequent computation can be effectively reduced.

[0097] As a preferred option, the foreign object identification process in step five is as follows:

[0098] S51: Rapid material screening based on physical rules;

[0099] Using the change in reactance With phase shift The preliminary material determination is based on the following rules: when and satisfy Time (i.e., phase lag) (including boundary values), initially identified as a ferromagnetic foreign object; when and satisfy Time (i.e., phase lead) (Including boundary values), initially determined to be a non-ferromagnetic foreign object. This rule is directly based on the physical principle of electromagnetic induction, and has the characteristics of low computational load and fast response speed, used to support the validity verification of input data for subsequent neural network recognition;

[0100] S52: Accurate type identification based on BP neural network;

[0101] voltage change Phase shift Reactance change The features are combined into a feature vector, which is then input into a pre-trained BP neural network model. The model then identifies and outputs the specific type of foreign object (such as anchor bolt, scraper, wire) and the corresponding recognition confidence level.

[0102] S53: Verification of recognition results fusion;

[0103] If the material properties corresponding to the specific type output by the BP neural network (e.g., anchor rods correspond to ferromagnetic materials) are consistent with the results of the rapid material screening based on physical rules, then the specific type output by the neural network shall prevail. If the two results contradict each other (e.g., the network outputs copper blocks but the physical rules point to ferromagnetic materials), then it is determined that the current signal is affected by environmental interference or has encountered an unknown new type of foreign object. The system automatically increases the sampling frequency for secondary confirmation and prioritizes the safety-oriented conclusion of the physical rules (i.e., issuing an early warning based on ferromagnetic foreign objects) to ensure the safe operation of conveyor belt 2.

[0104] As a preferred option, the training samples used to train the BP neural network include 1,000 sets of data from 20 common metal objects such as anchor rods, scrapers, keys, and wires. The recognition confidence level for anchor rods can be set to 92%. An example of the model output format is: anchor rod, confidence level 92%. If the output confidence level is <70%, it is determined to be an unknown foreign object, triggering a high-level warning.

[0105] When collecting 1000 sets of data, first ensure that there are no foreign objects on conveyor belt 2, and then collect data on 20 common metal foreign objects. After collection, the data is stored in the data storage submodule (EEPROM). During the training of the BP neural network, it is necessary to ensure that the recognition accuracy of the trained BP neural network module is ≥92%.

[0106] To quickly and accurately obtain the location information of the foreign object, the process of fusing the three-dimensional position in step six is ​​as follows:

[0107] S61: Planar coordinate positioning is performed using the centroid positioning method or nonlinear least squares fitting method; wherein, during the planar positioning process, the coil at the location of the foreign object... Maximum, centered on the matrix coordinates of the peak coil, combined with the adjacent coils. Distribution (e.g., the peak coil is in row 10, column 8, and the surrounding coils are in rows 9-11 and columns 7-9) (decreasing sequentially) to fit the planar contour of the foreign object, providing auxiliary basis for accurate solution of planar coordinates;

[0108] S61-1: Planar coordinate positioning is performed using the centroid localization method: The relative coordinates of the foreign object in the width and thickness directions of the conveyor belt 2 are obtained according to formulas (1) and (2) respectively. , ;

[0109] (1);

[0110] (2);

[0111] In the formula, , The first The physical coordinates of each detection coil in the width and thickness directions; For the first The voltage change sensed by each detection coil is usually taken as an absolute value or a square value to ensure that the weight is positive. The total number of valid detection coils involved in the calculation;

[0112] S61-2: Planar coordinate positioning using nonlinear least squares fitting method: First, establish the location of the foreign object based on the electromagnetic field propagation theory. The mathematical mapping relationship between it and the response signals of each coil Then, with the goal of minimizing the sum of squares of the residuals between the measured voltage change and the theoretical prediction, an optimization function is constructed according to formula (3); subsequently, mature nonlinear optimization algorithms such as the Gauss-Newton method or the Levenberg-Marquardt method are used to gradually approach the optimal solution from the initial value.

[0113] (3);

[0114] S62: Absolute position fusion in the length direction; first, infrared sensors are used for speed monitoring and time recording to obtain the running speed of conveyor belt 2. The time point at which a foreign object triggers a specific reference coil Then, through spatiotemporal information fusion and coordinate correction, the global absolute coordinates are obtained, as shown in formula (4).

[0115] (4);

[0116] In the formula, This represents the longitudinal position of the foreign object in conveyor belt 2; The fixed distance from the starting end of the detection panel 1 to the head of the conveyor 5;

[0117] S63: Determine the position in the width direction; combine the peak coil coordinates to perform an intuitive conversion of the position in the width direction, and obtain the absolute position of the foreign object in the conveying width direction according to formula (5). ;

[0118] (5);

[0119] In the formula, These are the column coordinates of the peak coil; The width of a single coil;

[0120] S64: Height coordinate positioning; achieved by comparing signals from the upper and lower detection panels 1 coils combined with normalization calculation, providing double assurance of positioning accuracy; first, targeting the determined horizontal projection area. Based on formulas (6) and (7), summarize all effective voltage responses of the upper and lower detection panels 1 respectively. , Then, the signal amplitude is standardized according to formula (8) to obtain the normalized intensity scaling factor. To eliminate the influence of polarity, the vertical distance between the foreign object and the surface of conveyor belt 2, i.e., the height position, is then obtained according to formula (9). ;

[0121] (6);

[0122] (7);

[0123] (8);

[0124] (9);

[0125] In the formula, The installation interval between the upper and lower detection panels 1.

[0126] This technical solution employs a combination of complementary algorithms, mathematical formula quantification, and multi-dimensional information fusion design, which combines positioning flexibility and accuracy. It can achieve full-dimensional coordinate calculation without additional sensors, is compatible with the dynamic working conditions of belt conveyors, and has strong traceability. It can accurately locate foreign objects in three-dimensional space of length, width, and height, thus shortening the time for foreign object inspection.

[0127] To achieve effective group warning and ensure the safe and stable operation of conveyor 5, the process of graded warning based on foreign object parameters and linkage with external equipment in step seven is as follows:

[0128] S71: Based on the type of foreign object (ferromagnetic, non-ferromagnetic, unknown), size (unit: cm, boundary values ​​use left-closed, right-open to avoid overlap), and location of large ferromagnetic foreign objects (the central 50% width of the conveyor belt bearing surface is the middle, the rest is the edge), the warning level is determined according to the following priority: Level 1 warning corresponds to: any foreign object ≥ 50cm (regardless of type); or ferromagnetic foreign objects > 30cm located in the middle; Level 2 warning corresponds to: ferromagnetic foreign objects > 30cm and < 50cm, but located on the edge; ferromagnetic foreign objects ≥ 10cm and ≤ 30cm; ferromagnetic foreign objects ≥ 30cm and... Non-ferromagnetic foreign objects <50cm; unknown foreign objects >30cm; Level 3 warning corresponds to: ferromagnetic foreign objects <10cm; non-ferromagnetic foreign objects <30cm; unknown foreign objects ≤30cm; if a foreign object meets multiple conditions, the highest level is used; unknown foreign objects are defined by a recognition confidence level below 70%, and the warning level is no longer solely based on confidence level; when location information is missing, the warning level is determined according to the case where the foreign object is located in the middle area, ensuring that large ferromagnetic foreign objects are not missed due to misjudgment of location; this mechanism covers all size ranges, eliminating problems such as no warning, overlapping boundaries, and risk inversion in the original logic;

[0129] S72: The warning level results are displayed in real time on the screen. At the same time, the warning level is uploaded to the coal mine production monitoring system, and the on-site audible and visual alarms are triggered simultaneously. The first-level warning is indicated by a red light, the second-level warning by a yellow light, and the third-level warning by a blue light. In addition, combined with the control of the foreign object location linkage equipment, if the foreign object is located near the machine head, preferably less than 10cm away from the machine head, it is determined to be near the machine head, and the conveyor 5 is automatically slowed down to buy time for on-site handling.

[0130] Example: A 50cm long anchor rod (ferromagnetic) is mixed into the material in conveyor belt 2, and the anchor rod is located in the middle area of ​​the conveyor belt bearing surface. During the detection process, the voltage change of the coil in the 10th row and 8th column of the detection panel 1 is measured. =+30%, Phase shift =-8°, reactance change =+5Ω; The BP neural network model output is "anchor bolt, confidence level 95%"; Positioning calculation: The infrared sensor detects the speed of conveyor belt 2 as 2m / s, the peak signal occurrence time is t=7.65s, and the fixed distance between the starting end of detection panel 1 and the head of conveyor 5 is 5m. Therefore, the longitudinal position of the foreign object from the head is X=5+2×7.65=20.3m; the width direction is Y=8×6cm=0.48m; the voltage response of the upper and lower detection panels 1 are respectively V 上 =0.8V, V 下=0.2V, height direction Z=(0.2 / (0.8+0.2))×1=0.2m; explosion-proof display screen outputs: "Level 1 warning: anchor bolt, position X=20.3m, Y=0.48m, Z=0.2m", simultaneously triggering audible and visual alarms, and uploading the data to the monitoring system. Maintenance personnel found and removed the foreign object 10 minutes later, realizing foreign object detection, positioning and removal operations without stopping the machine.

[0131] This invention is based on the principle of transient electromagnetic induction, operating within a closed-loop logic of alternating magnetic field excitation, metal eddy current induction, differential signal acquisition, and intelligent algorithm analysis. The core of the invention, the synergistic action of excitation coil 4 and receiving coil 3, is used to achieve metal foreign object detection and location. First, excitation coil 4 outputs alternating currents at 4kHz, 8kHz, and 12kHz, forming a multi-frequency alternating magnetic field uniformly covering the entire width of conveyor belt 2, providing a stable excitation source for foreign object induction. Then, under multi-frequency excitation, the eddy current responses of different metal materials differ significantly. The array of receiving coils 3 captures the superimposed signal of the primary magnetic field (directly generated by excitation coil 4) and the secondary magnetic field (generated by metal eddy currents), providing key features for accurate identification of foreign object types. After processing with Kalman filtering and wavelet threshold fusion algorithms, environmental noise from underground dust and vibration is effectively removed, retaining only the core effective signal, thus better adapting to the complex environment of high dust and strong vibration underground. Subsequently, feature extraction is performed on the pre-processed signal, and the voltage change is calculated. Phase shift Reactance change Key parameters such as the foreign object are analyzed; further, a pre-trained BP neural network is used to intelligently analyze these differential parameters, thereby completing the determination of the presence of foreign objects and the identification of their specific types; at the same time, the centroid positioning and nonlinear least squares fitting algorithm are used to calculate the three-dimensional spatial coordinates of the foreign object, realizing the precise positioning of the foreign object and triggering risk warnings simultaneously, providing effective technical support for rapid disposal.

[0132] Compared with the prior art, the present invention has the following advantages:

[0133] 1. The detection capability has been comprehensively improved from missed detection to full coverage; the array coil layout combined with the dynamic calibration algorithm can adapt to the harsh working conditions of downhole with dust concentration of 100mg / m³ and vibration frequency of 5~10Hz. The detection rate of large metal foreign objects such as anchor bolts and scrapers is ≥99%, while the missed detection rate of traditional technology in this environment exceeds 30%, and the stability is significantly better than the traditional solution.

[0134] 2. The recognition accuracy has been significantly optimized from false alarms to accurate differentiation. Through the fusion of multi-frequency excitation and BP neural network, it can accurately distinguish between ferromagnetic and non-ferromagnetic foreign objects, as well as the specific type of foreign object. The accuracy of distinguishing between ferromagnetic and non-ferromagnetic foreign objects is ≥98%, and the accuracy of identifying specific types is ≥92%. The false alarm rate has been significantly reduced. Taking a coal mine application as an example, the false alarm rate was found to have dropped from 20% of the traditional technology to below 3%.

[0135] 3. Positioning efficiency has been significantly improved from blind to precise location; practical verification shows that the 3D positioning accuracy error is ≤10cm, allowing maintenance personnel to quickly and accurately locate foreign objects directly based on the positioning coordinates, reducing the traditional 2-3 hour investigation time to 10-15 minutes. In a coal mine application, the time for a single foreign object disposal was reduced from 120 minutes to 15 minutes, fully demonstrating the significant improvement in positioning efficiency.

[0136] 4. The operation and maintenance costs have been significantly reduced from high consumption to high efficiency; no additional zero-adjustment coils and dense single-point sensors are required, and the hardware cost is reduced by about 40% compared with the traditional solution; at the same time, through automatic calibration and remote early warning functions, the frequency of manual inspection is reduced from 3 times a day to 1 time, reducing labor costs and workload.

[0137] This method is simple to implement, low in cost, and highly accurate in detection and positioning. Through the effective integration of array-type electromagnetic induction, multi-frequency excitation, and intelligent recognition algorithms, it constructs a complete solution for foreign object detection and positioning on conveyor belt 2. Ultimately, it achieves the goals of early warning, accurate positioning, and rapid processing, significantly reducing the probability of accidents such as conveyor belt 2 tearing, and greatly improving the production efficiency and safety factor of coal mines. It provides effective technical support for the safe and stable operation of belt conveyor 5 and can be effectively used in complex underground production environments.

[0138] like Figure 4 and Figure 5 As shown, the present invention also provides a conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics, used to implement a method for detecting and locating conveyor metal foreign objects using array-type transient electromagnetics, comprising:

[0139] An array-type detection module includes two symmetrically arranged detection panels 1, which are distributed opposite each other on the upper and lower sides of the conveyor belt 2 carrying section. Each detection panel 1 is provided with several receiving coils 3 arranged in a matrix to form a detection area that covers the entire area.

[0140] As a preferred option, during use, it is necessary to clean the dust on the detection panel 1 regularly (to avoid affecting the accuracy of electromagnetic signal acquisition), and at the same time, check the connection status of the cable regularly to prevent vibration from causing it to loosen.

[0141] As a preferred embodiment, the array-type detection module is located 5m away from the head of the conveyor 5 (while avoiding the area where the rollers rotate). The two detection panels 1 are set parallel to the conveyor belt 2, forming a detection area that clamps the conveyor belt 2 from both above and below. The distance between the edge of the detection panel 1 and the sides of the conveyor belt 2 is preferably 5cm. The spacing between the two detection panels 1 can be set as needed, for example, it can be set to 0.8m to 1.2m, preferably 1m. This can adapt to the thickness of the conveyor belt 2 and avoid affecting the coal and rock conveying. At the same time, it can match the height of the maximum foreign object. It can also ensure that when a foreign object enters the detection area, the coils on both sides can synchronously sense the signal change, so as to ensure real-time and effective detection of foreign objects on the conveyor belt 2.

[0142] The multi-frequency excitation module includes one main frequency converter, two slave frequency converters, and eight excitation coils 4. The eight excitation coils 4 are connected in parallel and evenly distributed around the perimeter of two detection panels 1, with a distance of 10 cm between them. The multi-frequency excitation module outputs alternating magnetic fields of 4kHz, 8kHz, and 12kHz. Figure 2 The diagram shows the arrangement of an excitation coil 4 and multiple receiving coils 3, where h = 10 cm;

[0143] The multi-frequency excitation module adopts a configuration of 1 master and 2 slave frequency converters, wherein the master frequency converter is 160kW and the slave frequency converter is 110kW. More preferably, the signal processing module can be connected to the multi-frequency excitation module via PROFINET bus to achieve synchronous control (frequency deviation <0.5Hz), and sequentially drive 8 parallel excitation coils 4 to output 3 kinds of fixed frequency alternating magnetic fields to generate an electromagnetic field with a certain power and meeting the requirements of the operating distance and frequency.

[0144] The signal processing module integrates a filtering submodule, a data storage submodule, and a foreign object intelligent identification module. It is connected to both an array-type detection module and a multi-frequency excitation module to perform signal denoising, difference parameter calculation, and foreign object intelligent identification, as well as hierarchical early warning based on the identification results. As the intelligent brain for data processing, the signal processing module preferably uses a Siemens S7-1200 series PLC (with an Ethernet interface, supporting high-speed data processing), equipped with an EEPROM data storage device (storing reference signals and pre-trained models) and a Kalman filter circuit (for signal denoising). The core functions of the signal processing module are as follows: 1. Reference calibration; automatically collecting data upon device startup. The induced voltage (e.g., 0.48V~0.52V), phase (e.g., -1°~+1°), and reactance component (e.g., 9.5Ω~10.5Ω) of all coils under physical conditions are stored as a reference signal. Preferably, it can be automatically recalibrated every 2 hours to effectively compensate for signal drift caused by temperature (downhole temperature fluctuations are usually 0~40℃) and vibration. 2. Signal processing: The voltage, phase, and reactance data collected by coil 3 are received in real time, and noise caused by dust and motor interference is removed by Kalman filtering algorithm (the signal-to-noise ratio is improved to more than 40dB after filtering). Then, the difference between the actual value (signal) and the reference value (signal) is calculated (such as voltage change, phase shift, and reactance change).

[0145] The positioning and display module includes an infrared sensor, a gravity sensor, and an explosion-proof display screen. The positioning and display module is connected to a signal processing module. The infrared sensor is set at the drive shaft end of the drive roller and is used to collect the running speed signal of the conveyor belt. The gravity sensor is installed at the bottom of the detection panel 1 on the lower side and is used to detect the weight signal of foreign objects. The explosion-proof display screen is used to display three-dimensional coordinates and warning information. It can be installed in the machine head control room, preferably at a distance of ≤10cm from the detection panel 1.

[0146] As a preferred option, a 10-inch touchscreen display is used. The positioning and display module, serving as the intuitive monitoring terminal, functions as follows: The infrared sensor calculates the travel distance based on the conveyor belt 2's rotational speed, and combines this with the matrix coordinates of the receiving coil 3 to determine the object's length direction position (e.g., when the object moves 15.3m within the detection area, plus the fixed 5m distance from the starting end of the detection panel 1 to the machine head, the display shows X=20.3m, meaning 20.3m from the machine head); the peak coordinates of the coil matrix determine the width direction position (e.g., Y=0.4m, meaning 0.4m from the left edge of the conveyor belt 2); the signal difference between the upper and lower detection panels 1 is used to calculate the height direction position (e.g., Z=0.3m, representing the normalized height coordinates of the object in the vertical direction); the display screen shows the object's three-dimensional coordinates, type, weight, detection time, and warning level (red / yellow / blue three-color warning) in real time. To improve detection accuracy, the receiving coils 3 in the two detection panels 1 of the array detection module are completely symmetrical, the distance between the two detection panels 1 is 0.8 to 1.2 m, the size of a single receiving coil 3 is 6 cm × 6 cm, the number of receiving coils 3 in each detection panel 1 is ≥ 500, and the sampling frequency is 100 Hz.

[0147] As a preferred embodiment, the size of the detection panel 1 is matched with the size of the conveyor belt 2, such as a 1.2m wide conveyor belt 2 with a 1.2m×0.6m panel. More preferably, each detection panel 1 includes 500 receiving coils 3 and is arranged in a 25×20 matrix. Even more preferably, the size of each receiving coil 3 is 6cm×6cm.

[0148] To ensure detection accuracy, the coils in the two detection panels 1 are arranged in a completely symmetrical manner, such that the coil in the 3rd row and 5th column above is directly opposite the coil in the 3rd row and 5th column below.

[0149] As a further preferred option, each receiving coil 3 is connected to an independent voltage acquisition circuit, and the voltage is acquired in parallel through 16 AD ports (each AD port can be connected to 31 coils through a multi-channel switching switch to achieve time-division acquisition of 1000 coils on the two detection panels 1, with an acquisition frequency of 1000Hz, to ensure that when the conveyor belt 2 is running at a speed of 2m / s, 10 to 15 sets of data can be collected when foreign objects pass through the detection area).

[0150] To effectively eliminate signal drift caused by temperature, vibration, and noise, the signal processing module triggers dynamic calibration every 2 hours, completing data acquisition of the foreign object-free area during a 0.5s excitation pause. The reference value is updated when the acquired data deviates from the reference value by more than ±3%. Preferably, the positioning and display module features an infrared sensor detection accuracy ≤0.01m / s, a gravity sensor weighing error ≤50g, and an explosion-proof display screen with an explosion-proof rating of Ex ib 1 Mb.

[0151] Working principle: such as Figure 1 As shown, during the detection process, a multi-frequency alternating magnetic field is generated by the excitation coil 4. When a metallic foreign object in the conveyor belt 2 enters the magnetic field range, induced eddy currents are generated inside it under the action of the applied alternating magnetic field, which in turn changes the direction of the magnetic field. This eddy current generates a secondary induced magnetic field opposite to the original excitation magnetic field. This secondary magnetic field changes the magnetic flux of the surrounding receiving coil, thereby causing a change in the induced electromotive force of the receiving coil 3, achieving the initial detection of the presence of the metallic foreign object. At the same time, the eddy current further generates a diffuse induced magnetic field during the decay process. Due to the differences in core physical parameters such as conductivity of different media, the response characteristics of the induced magnetic field generated by the metal positioning technology based on transient electromagnetic array will also show significant differences. Based on this, the induced magnetic field signal can be collected by the receiving coil 3, and combined with the pattern of the response signal of different media under multi-frequency magnetic excitation, the physical parameters of the medium can be identified in reverse, initially achieving accurate detection and positioning of metallic foreign objects in the conveyor belt 2.

[0152] In this invention, two detection panels, one above the other, form a pincer detection layout on the conveyor belt 2. Each detection panel 1 is equipped with several receiving coils 3 arranged in a matrix, achieving full-width and full-length detection of the conveyor belt 2 without blind spots, effectively eliminating missed detection zones. Eight excitation coils 4 are evenly arranged around the two detection panels 1, forming a double-layered, surrounding magnetic field structure. This ensures that the magnetic field strength distribution deviation across the cross-section of the conveyor belt 2 is ≤5%, further eliminating the detection blind spot in the central area. The annular arrangement of four coils in each group creates a region with a nearly uniform magnetic field within the detection panel 1, making the eddy current induction intensity more stable regardless of the width of the metal foreign object on the conveyor belt 2, avoiding missed detections due to magnetic field gradients. The signal processing module facilitates not only noise reduction preprocessing and difference parameter calculation of the acquired signals, but also allows for accurate identification of the foreign object material and specific type based on the difference parameters using the intelligent foreign object identification module. Furthermore, it enables tiered early warning based on the identification results. The positioning and display module is designed to facilitate the acquisition of basic parameters required for positioning using its infrared and gravity sensors, thereby enabling the signal processing module to quickly obtain positioning results. The explosion-proof display feature allows for the easy display of the three-dimensional coordinates of foreign objects and early warning information, enabling relevant personnel to intuitively observe positioning and warning information.

[0153] The system has a simple structure, high level of intelligence, and low manufacturing cost. It can quickly and accurately detect and locate foreign objects on conveyor belt 2, which can significantly reduce the incidence of accidents such as conveyor belt 2 tearing and improve coal mine production efficiency.

Claims

1. A method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array, characterized in that, Including the following methods; Step 1: System initialization and calibration; Collect the voltage, phase, and reactance reference values ​​of the array coil under foreign object-free conditions, and periodically and dynamically update the reference values ​​during the operation of the detection and positioning system; Step 2: Multi-frequency magnetic field excitation; output an alternating magnetic field at a preset frequency cycle to generate eddy currents and secondary induced magnetic fields in the metal foreign objects on the conveyor belt; Step 3: Signal Acquisition and Preprocessing; The induced signal is acquired through an array coil and preprocessed for noise reduction using a wavelet thresholding and Kalman filtering fusion algorithm; Step 4: Anomaly detection and ROI extraction; Calculate the difference parameters between the real-time signal and the reference signal, identify suspected foreign object regions based on the difference parameters, perform cluster analysis on the suspected foreign object regions, and extract the region of interest corresponding to each suspected foreign object region; The foreign object identification process is as follows: S51: Rapid material screening based on physical rules; Using the change in reactance With phase shift The preliminary material determination is based on the following rules: when and satisfy At that time, it was initially determined to be a ferromagnetic foreign object; when and satisfy At that time, it was initially determined to be a non-ferromagnetic foreign object; S52: Accurate type identification based on BP neural network; voltage change Phase shift Reactance change The components are combined into a feature vector, which is then input into a pre-trained BP neural network model. The model then performs computations to identify and output the specific type of the foreign object and its corresponding recognition confidence level. S53: Recognition result fusion verification; If the material properties corresponding to the specific type output by the BP neural network are consistent with the results of the rapid material screening based on physical rules, then the specific type output by the neural network shall prevail; if the two results contradict each other, it is determined that the current signal is affected by environmental interference or encounters an unknown new foreign object. The system automatically increases the sampling frequency for secondary confirmation and adopts the safety-oriented conclusion of the physical rules to ensure the safe operation of the conveyor belt. Step 5: Foreign object identification; Based on physical rules, the material of the foreign object is quickly initially determined, and then the specific type is accurately identified through a BP neural network, forming a dual identification mechanism that complements rapid screening and precise judgment; Step Six: 3D positioning of the foreign object; the planar coordinates are solved using a planar positioning algorithm, the height coordinates are solved based on the signal strength ratio of the upper and lower coils, and the planar coordinates and height coordinates are fused to obtain the 3D position; Step 7: Early warning output; tiered early warnings are issued based on foreign object parameters, and external devices are activated in conjunction with these warnings.

2. The method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array according to claim 1, characterized in that, In step two, the preset frequencies are 4kHz, 8kHz, and 12kHz, and the frequencies are output alternately in the order of 4kHz, 8kHz, and 12kHz, with a continuous excitation time of 0.1s for each frequency.

3. A method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array, as described in claim 1 or 2, characterized in that... In step four, the anomaly detection and ROI extraction process is as follows: S41: Set voltage change threshold Voltage change for all coils Perform a comprehensive detection, first filtering out voltage changes. Exceeding the voltage change threshold The candidate abnormal coils are then checked for their adjacency. When there are no less than 3 adjacent candidate abnormal coils that are triggered simultaneously, the group of adjacent coils is determined to be the trigger coil. S42: A distance-based clustering algorithm is used to cluster the selected coil groups, and coil groups in the same connected domain are identified as signal regions triggered by the same foreign object, thus realizing the preliminary division of suspected foreign object regions. S43: For each suspected foreign object region obtained by division, extract the coil signal data of the region and its surrounding preset range to form a ROI data set; the preset range is a 5×5 coil centered on the suspected foreign object region.

4. The method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array according to claim 3, characterized in that, In step six, the process of fusing to obtain the three-dimensional position is as follows: S61: Planar coordinate positioning is performed using the centroid positioning method or the nonlinear least squares fitting method; S61-1: Planar coordinate positioning using the centroid positioning method: Obtain the relative coordinates of the foreign object in the width and thickness directions of the conveyor belt according to formulas (1) and (2) respectively. , ; （1）； （2）； In the formula, , The first The physical coordinates of each detection coil in the width and thickness directions; For the first The amount of voltage change sensed by each detection coil; The total number of valid detection coils involved in the calculation; S61-2: Planar coordinate positioning using nonlinear least squares fitting method: First, establish the location of the foreign object based on the electromagnetic field propagation theory. The mathematical mapping relationship between it and the response signals of each coil Then, with the goal of minimizing the sum of squares of the residuals between the measured voltage change and the theoretical prediction, an optimization function is constructed according to formula (3); subsequently, a nonlinear optimization algorithm is used to gradually approach the optimal solution from the initial value. （3）； S62: Absolute position fusion in the length direction; first, infrared sensors are used for speed monitoring and time recording to obtain the conveyor belt running speed. The time point at which a foreign object triggers a specific reference coil Then, through spatiotemporal information fusion and coordinate correction, the global absolute coordinates are obtained, as shown in formula (4). （4）； In the formula, This refers to the longitudinal position of the foreign object in the conveyor belt. To detect the fixed distance from the starting end of the panel to the head of the conveyor; S63: Determine the position in the width direction; combine the peak coil coordinates to perform an intuitive conversion of the position in the width direction, and obtain the absolute position of the foreign object in the conveying width direction according to formula (5). ; （5）； In the formula, These are the column coordinates of the peak coil; The width of a single coil; S64: Height coordinate positioning; first, target the already determined horizontal projection area. Based on formulas (6) and (7), summarize all effective voltage responses of the upper and lower detection panels respectively. , Then, the signal amplitude is standardized according to formula (8) to obtain the normalized intensity scaling factor. Then, the vertical distance between the foreign object and the conveyor belt surface, i.e., the height position, is obtained according to formula (9). ; （6）； （7）； （8）； （9）； In the formula, This refers to the installation interval between the upper and lower detection panels.

5. The method for detecting and locating metallic foreign objects in a conveyor using an array-type transient electromagnetic array according to claim 3, characterized in that, In step seven, the process of issuing graded early warnings based on foreign object parameters and linking external devices is as follows: S71: Based on the type, size, and location of large ferromagnetic foreign objects, the warning level is determined according to the following priority: Level 1 warning corresponds to: any foreign object ≥ 50cm; or a ferromagnetic foreign object > 30cm located in the middle; Level 2 warning corresponds to: ferromagnetic foreign objects > 30cm and < 50cm located at the edge; ferromagnetic foreign objects ≥ 10cm and ≤ 30cm; non-ferromagnetic foreign objects ≥ 30cm and < 50cm; unknown foreign objects > 30cm; Level 3 warning corresponds to: ferromagnetic foreign objects < 10cm; non-ferromagnetic foreign objects < 30cm; unknown foreign objects ≤ 30cm. If a foreign object meets multiple conditions simultaneously, the highest level shall be applied. Unknown foreign objects are defined when the identification confidence level is below 70%, and the warning level is no longer downgraded solely based on the confidence level. When location information is missing, the warning level is determined according to the case that the foreign object is located in the middle area, so as to ensure that large ferromagnetic foreign objects are not missed due to misjudgment of location. S72: The warning level results are displayed in real time on the screen. At the same time, the warning level is uploaded to the coal mine production monitoring system, and the on-site audible and visual alarms are triggered simultaneously. The first-level warning is indicated by a red light, the second-level warning by a yellow light, and the third-level warning by a blue light. In addition, in conjunction with the foreign object location linkage equipment control, if the foreign object is located near the machine head, the conveyor will be automatically slowed down to buy time for on-site handling.

6. A conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics, used to implement the conveyor metal foreign object detection and positioning method based on array-type transient electromagnetics as described in any one of claims 1 to 5, characterized in that, include: An array-type detection module includes two symmetrically arranged detection panels, which are distributed opposite each other on the upper and lower sides of the conveyor belt carrying section. Each detection panel is equipped with several receiving coils arranged in a matrix to form a detection area that covers the entire area. The multi-frequency excitation module includes one main frequency converter, two slave frequency converters, and eight excitation coils. The eight excitation coils are connected in parallel and evenly distributed around the four edges of the two detection panels, with a distance of 10cm between them and the detection panels. The multi-frequency excitation module is used to output 4kHz, 8kHz, and 12kHz alternating magnetic fields. The signal processing module integrates a filtering submodule, a data storage submodule, and a foreign object intelligent identification module. The signal processing module is connected to the array detection module and the multi-frequency excitation module, respectively, and is used to realize signal denoising, difference parameter calculation and foreign object intelligent identification, as well as hierarchical early warning based on the identification results. The positioning and display module includes an infrared sensor, a gravity sensor, and an explosion-proof display screen. The positioning and display module is connected to a signal processing module. The infrared sensor is installed at the drive shaft end of the drive roller to collect the running speed signal of the conveyor belt. The gravity sensor is installed at the bottom of the detection panel on the lower side to detect the weight signal of foreign objects. The explosion-proof display screen is used to display three-dimensional coordinates and warning information.

7. The conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics according to claim 6, characterized in that, The receiving coils in the two detection panels of the array-type detection module are completely symmetrical, the distance between the two detection panels is 0.8 to 1.2 m, the size of a single receiving coil is 6 cm × 6 cm, the number of receiving coils in each detection panel is ≥500, and the sampling frequency is 100 Hz.

8. The conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics according to claim 7, characterized in that, The signal processing module triggers dynamic calibration every 2 hours, completes data acquisition of the foreign object-free area during the 0.5s excitation pause period, and updates the reference value when the acquired data deviates from the reference value by more than ±3%.

9. A conveyor metal foreign object detection and positioning system based on array-type transient electromagnetics according to claim 8, characterized in that, The positioning and display module has an infrared sensor detection accuracy of ≤0.01m / s, a gravity sensor weighing error of ≤50g, and an explosion-proof display screen with an explosion-proof rating of Ex ib I Mb.

Images