A method for identifying the operating state of a flour raw material feeding device

Abstract

This invention relates to the field of visual inspection technology, specifically to a method for identifying the operating status of a flour raw material feeding device. The method includes: acquiring infrared thermal imaging and acoustic emission signals through a non-contact sensor array arranged in the silo or discharge port; extracting the spatiotemporal characteristics of the temperature distribution field and texture variation characteristics of the material surface to identify initial signs of bridging or voids; analyzing the energy changes and frequency shifts of the acoustic emission signals to determine mouse holes or feeding interruptions; spatiotemporally fusing the infrared and acoustic emission characteristics to generate an arching risk index and a mouse hole probability index, and issuing an early warning when the threshold is exceeded. This invention uses long-wave infrared to penetrate dust and dry dust removal to achieve early fault identification in dark dust environments. It can also be extended to foreign object sorting, conveyor chain damage, bearing warning, etc., improving the operating safety and intelligence level of the feeding device.

Description

Technical Field

[0001] This invention relates to the field of visual inspection technology, and specifically to a method for identifying the operating status of a flour raw material feeding device. Background Technology

[0002] Flour raw materials are prone to arching, bridging, internal voids, or rat holes in silos and feed chutes due to factors such as humidity, compression, and static electricity. The surface may appear to have sufficient material, but in reality, there is no material below or the blockage occurs, causing downstream equipment to run dry due to material shortage. This is one of the main causes of unexpected shutdowns of flour production lines. Early and accurate identification of such internal flow faults in opaque, dusty, and completely dark sealed silos has always been a challenge in the industry. In existing technologies, infrared thermal imagers are used to acquire surface temperature images of grain inside grain silos. By analyzing areas of abnormal temperature distribution, it is determined whether the grain has become moldy, condensed, or compacted. This approach uses a non-contact method to acquire infrared radiation information from the material surface, providing a new approach for monitoring the internal condition of grain silos. However, the above approach mainly targets the detection of temperature anomalies caused by heat generated by mold during grain storage. When using the single infrared image analysis method to identify arching and rat holes in flour feeding devices, it has the following drawbacks: the temperature changes caused by friction and compression during material feeding are slight, and the temperature difference artifacts in the infrared image are difficult to distinguish from real bridging voids due to the influence of humidity and temperature differences between batches of incoming materials, resulting in a high false alarm rate. At the same time, for situations where a through-hole rat hole has formed inside but material is still slowly collapsing on the surface, there is no significant abnormality in the surface temperature distribution, and a single infrared method cannot achieve early detection of material feeding interruption. In view of this, in order to overcome the above-mentioned technical problems, the present invention proposes a method for identifying the operating status of a flour raw material feeding device. Summary of the Invention

[0003] The technical problem to be solved by this invention is to address the issue that existing technologies cannot achieve early identification of silo arching and rat holes under the requirements of high dust, darkness, and explosion protection in flour feeding devices. This invention provides a method for identifying operating status based on multimodal perception and resisting dust interference, and further solves at least one of the following: non-metallic foreign object sorting, early damage monitoring of conveyor belts / scraper chains, early fault warning of low-speed heavy-load bearings, continuous clean operation of vision systems, dust concentration field monitoring, and multi-fault root cause tracing.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for identifying the operating status of a flour raw material feeding device includes the following steps: S1, by using a non-contact sensor array arranged in the hopper and / or discharge port, multimodal sensing data of the target area is acquired in real time, wherein the multimodal sensing data includes at least an infrared thermal imaging image sequence and an acoustic emission signal; S2, Process the infrared thermal imaging image sequence to extract the spatiotemporal characteristics of the temperature distribution field and texture variation characteristics of the material surface, and identify the initial signs of material bridging or void formation based on the uniformity of temperature distribution and local temperature anomalies. S3, synchronously analyze the energy change, event count and frequency center shift of the acoustic emission signal under material flow and static states, and combine the material flow acoustic feature library to determine whether there is a rat hole or internal material feeding interruption. S4, the spatiotemporal characteristics of the temperature distribution field, texture variation characteristics and acoustic emission characteristics are spatiotemporally aligned and fused at the decision level to generate the arch risk index and the mouse hole probability index. S5. When the arch risk index or rat hole probability index exceeds the preset threshold, an early warning signal is output and the fault type is marked.

[0005] Preferably, the non-contact sensor array in S1 further includes a terahertz imaging module for acquiring a density distribution image below the surface of the material inside the silo; S2 further includes: registering and fusing the terahertz density distribution image with the infrared thermal imaging temperature distribution field, using the density sparse anomaly region to verify the location of the void inside the bridge, and correcting the arching risk index.

[0006] Preferably, the method further includes an online identification step for non-metallic foreign objects on the conveyor belt: acquiring hyperspectral images and infrared thermal images above the material flow on the conveyor belt; extracting the difference in reflectance of characteristic bands representing organic and inorganic matter in the hyperspectral image, and combining the specific heat capacity temperature difference between the foreign object and the flour in the infrared thermal image to identify stones, clods of soil, moldy grains and straw foreign objects, and outputting a rejection signal before the foreign object reaches the discharge port.

[0007] Preferably, the method further includes a step for monitoring early damage to the conveyor belt and / or scraper chain: acquiring a sequence of linear laser profile scan images and acoustic emission signals on the back of the conveyor belt or along the scraper chain path; extracting longitudinal edge continuity features and pitch fluctuation curves from the linear laser profile scan image sequence to identify longitudinal microcracks, steel cord core slippage, and scraper microbending; simultaneously extracting the sudden event features of the acoustic emission signals, and assessing the damage development speed and issuing graded warnings after spatiotemporally correlating the two.

[0008] Preferably, it also includes an early warning step for early failure of low-speed heavy-load bearings: simultaneously collecting vibration signals and acoustic emission signals at the bearing housing; extracting the energy envelope of the resonant frequency band after high-pass filtering in the vibration signal to identify the periodic modulation characteristics caused by defect impact; simultaneously extracting the burst pulse interval and energy steepness of the acoustic emission signal, and mutually verifying the vibration modulation frequency and the acoustic emission pulse interval. When both match the characteristic frequency of the bearing element, it is determined that there is an early pitting or spalling fault and the bearing status index is output.

[0009] Preferably, the infrared thermal imaging uses an 8-14μm long-wave infrared band to penetrate dust and eliminate ambient light interference; periodic dry dust removal operations are performed at the optical windows of the infrared thermal imaging and terahertz imaging, including electrostatic neutralization and micromechanical vibration, to maintain light transmittance without the use of compressed gas and liquid cleaning agents.

[0010] Preferably, the method also includes a real-time dust concentration field monitoring step: using multiple infrared thermal imagers at different locations to acquire multi-view infrared images of the silo and conveying corridor; based on the difference in the degree of background infrared radiation attenuation caused by dust scattering in the images, the spatial dust concentration distribution field is inverted, and when the local concentration exceeds the safety threshold, the precise dust suppression device is activated.

[0011] Preferably, when at least two of the following symptoms are present simultaneously: increased current, increased vibration, and slower material discharge, a multi-fault root cause tracing step is performed: the feature quantities extracted from each monitoring step are input into a pre-constructed fault causal probability model, which includes a causal chain of arching, foreign object blockage, bearing damage, conveyor belt slippage, and material dampness; the posterior probability of each hypothetical root cause is calculated and sorted, and the root cause with the highest probability and the corresponding elimination guidance information are output.

[0012] Preferably, the acoustic emission signal is transmitted to the acoustic emission sensor through a waveguide rod fixed to the outside of the silo wall. The end of the waveguide rod that contacts the silo wall has a coupling surface that contacts the material, so as to realize solid-state acoustic conduction and isolate the interference of dust in the air to the sensor. The warning signal output by S5 includes highlighting the spatial location of arching or rat hole on the three-dimensional model of the silo in a pseudo-color cloud map, and displaying the severity level and recommended treatment measures corresponding to the arching risk index or rat hole probability index in numerical terms.

[0013] Preferably, the steps of the above-described method are implemented using an electronic device; the electronic device includes a processor and a memory. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method described above.

[0014] The beneficial effects of this invention are as follows: 1. This invention employs long-wave infrared thermal imaging, utilizing the differences in frictional heat generation and heat capacity during material flow and stagnation to obtain clear temperature distribution images in a completely dark and dust-filled enclosed silo, overcoming the problem of visible light vision systems failing due to dust coverage. Simultaneously, acoustic emission signals are acquired through solid-state conduction, unaffected by dust and light, reflecting the material flow state and internal voids from an acoustic perspective. The spatiotemporal fusion of these two methods provides complementary verification, issuing early warnings before arching or rodent holes cause significant material interruptions, greatly reducing the risk of unexpected downtime. By further expanding sensing methods such as terahertz, hyperspectral, and line laser profilometry, problems such as foreign object sorting, conveyor chain damage, and early bearing failures can be solved simultaneously within the same framework. Furthermore, all optical sensing employs dust-resistant wavelengths and dry cleaning strategies, enabling the system to operate long-term without maintenance in dust explosion-proof areas without complex air paths. The multi-fault root cause tracing model utilizes the causal logic of multi-source features to avoid misjudgment based on a single symptom, significantly improving the accuracy and intelligence level of the feeding device's operating status identification. Attached Figure Description

[0015] Figure 1 This is an overall flowchart of the flour raw material feeding device operation status identification method of the present invention; Figure 2 This is a schematic diagram illustrating the arrangement of monitoring sensors in the silo and the principle of arching / rat hole identification in this invention. Figure 3 This is a schematic diagram of the foreign object identification and conveyor belt damage detection method of the present invention. Figure 4 This is the multi-fault root cause tracing logic diagram of the present invention; Figure 5 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the following embodiments are merely examples illustrating the technical solution of the present invention and are not intended to limit the scope of protection.

[0017] Existing technologies primarily target temperature anomalies caused by mold growth during grain storage. However, when using a single infrared image analysis method to identify bridging and rat holes in flour feeding devices, the following drawbacks exist: the temperature changes caused by friction and compression during material feeding are slight, and the temperature difference artifacts in the infrared image are difficult to distinguish from actual bridging voids due to variations in humidity and temperature between batches of incoming materials, resulting in a high false alarm rate. Furthermore, in cases where interconnected rat holes have formed internally but material is still slowly collapsing on the surface, the surface temperature distribution shows no significant abnormalities, and a single infrared method cannot achieve early detection of feeding interruptions. Example

[0018] Reference Figure 1 and Figure 2 A non-contact sensor array is installed at the conical section and / or discharge chute of the flour raw material silo via a flange window or embedded installation. The sensor array includes at least one long-wave infrared thermal imager and one acoustic emission sensor. The long-wave infrared thermal imager operates in the 8-14μm band, effectively penetrating flour dust and unaffected by the dark environment inside the silo. The acoustic emission sensor is coupled to the outside of the silo wall via a stainless steel waveguide rod. The front end of the waveguide rod has a coupling surface that contacts the material inside the silo, transmitting the acoustic emission signal generated by the material flow and silo wall friction to the sensor via a solid path. This isolates the interference of airborne dust on the sensor's sensitive element and avoids opening large holes in the silo wall that could affect the explosion-proof performance.

[0019] S1, using a non-contact sensor array arranged in the hopper and / or discharge port, multimodal sensing data of the target area is acquired in real time. The multimodal sensing data includes at least an infrared thermal imaging image sequence and an acoustic emission signal. The infrared thermal image sequence and acoustic emission signal are acquired. The infrared thermal imager continuously captures images of the material surface and discharge port area at a frame rate of 5-10 frames per second to obtain a grayscale sequence of spatial temperature distribution. The acoustic emission system continuously records waveform streams and event characteristic parameters at a sampling frequency of not less than 1MHz. S2, the infrared thermal imaging image sequence is processed to extract the spatiotemporal characteristics of the temperature distribution field and texture variation characteristics of the material surface. Based on the temperature distribution uniformity and local temperature anomalies, initial signs of material bridging or void formation are identified. First, non-uniformity correction and bad pixel replacement are performed on each frame of the image. Then, the spatiotemporal characteristics of the temperature distribution field on the material surface are extracted. These spatiotemporal characteristics include, but are not limited to, the variance, skewness, and radial temperature gradient modulus of the temperature spatial distribution. When the material flows normally and continuously, due to continuous friction between particles and scraping against the silo wall, the material surface exhibits… The temperature rise is relatively uniform, and the overall temperature variance is small. Once a bridging occurs, the material below the bridging stops flowing, and a local low-temperature zone appears in this area due to reduced frictional heat generation. When a mouse hole forms, the heat capacity of the cavity area is extremely low, and its temperature change rate is significantly different from that of the surrounding dense material under slight environmental temperature fluctuations. At the same time, texture variation features are extracted, and by comparing the changes in local texture patterns in consecutive frame images, texture stagnation caused by the cessation of material movement can be detected. Based on the degree of decrease in temperature distribution uniformity and the spatial connectivity of local low-temperature / high-temperature anomaly areas, suspected areas of bridging or cavities are initially marked. S3. Simultaneously analyze the energy changes, event counts, and frequency centroid shifts of the acoustic emission signal under material flow and static states. Combine this with the material flow acoustic feature library to determine whether there are rat holes or internal material feeding interruptions. During normal material feeding, the material flow generates continuous, broadband acoustic emission signals with relatively stable root mean square energy values ​​and event count rates, and the frequency centroid is located in the mid-to-high frequency range. When rat holes or bridging cause material feeding interruptions, the acoustic emission energy drops sharply, the event count rate drops drastically, and the frequency centroid shifts to lower frequencies. Using a pre-established material flow acoustic feature library, which contains acoustic emission templates of wheat with different humidity and particle sizes under various states such as normal flow, bridging, rat holes, and material feeding interruptions, the real-time extracted energy change sequence, event counts, and frequency centroid shifts are matched with the templates to determine whether there are rat holes or internal material feeding interruptions, and a qualitative confidence level is given. S4. The spatiotemporal characteristics of the temperature distribution field, texture variation characteristics, and acoustic emission characteristics are spatiotemporally aligned and fused at the decision level to generate an arching risk index and a mouse hole probability index. The timestamps of the infrared image sequence and acoustic emission characteristics are aligned, and the suspected areas marked in the infrared images are associated with the spatial orientation information during the period of abnormal acoustic emission. Decision-level fusion is performed using weighted evidence theory or fuzzy logic: if the infrared temperature anomaly area and the acoustic emission discontinuity event are highly consistent in spatiotemporal terms, the confidence of arching or mouse hole formation is greatly increased; if there is only a local infrared temperature anomaly and no obvious discontinuity is found in the acoustic emission, it may be a transient temperature difference caused by uneven material humidity, and the risk index is lowered accordingly. Finally, a quantitative arching risk index and mouse hole probability index are generated, for example, a value of 0-100. S5, when the arching risk index or the rat hole probability index exceeds the preset threshold, an early warning signal is output and the fault type is marked; when the arching risk index exceeds 60 or the rat hole probability index exceeds 0.7, the system outputs an early warning signal; the early warning signal not only includes a text prompt of the fault type, but also highlights the spatial location of the arching or rat hole on the three-dimensional silo model of the operator station with a pseudo-color cloud map, and marks the severity level in numerical form, while giving suggested handling measures, such as suggesting to start the silo wall vibrator, reduce the discharge rate, etc. Example

[0020] As one embodiment of the present invention, the non-contact sensor array in S1 further includes a terahertz imaging module for acquiring density distribution images below the surface of the material inside the silo; S2 further includes: registering and fusing the terahertz density distribution image with the infrared thermal imaging temperature distribution field, using density sparse anomaly regions to verify the location of voids inside the bridging, and correcting the arching risk index; the non-contact sensor array in S1 also includes a terahertz imaging module; terahertz waves have a certain penetrating ability to non-polar materials such as flour and wheat, and can acquire density distribution images several centimeters to tens of centimeters below the material surface; in S2, the terahertz density distribution image is registered with the infrared thermal imaging temperature distribution field in three-dimensional space; the infrared image shows surface temperature difference characteristics, while the terahertz image reveals the density of the material below the surface; when the infrared image shows local temperature anomalies, and the terahertz image at the same location shows obvious low-density areas, i.e., density sparse anomaly areas, it can be confirmed as voids generated by internal bridging, thereby correcting the arching risk index and improving the identification accuracy to over 95%. Example

[0021] As one embodiment of the present invention, it further includes an online identification step for non-metallic foreign objects on the conveyor belt: acquiring hyperspectral images and infrared thermal images above the material flow on the conveyor belt; extracting the characteristic band reflectance differences representing organic and inorganic matter in the hyperspectral images, and combining the specific heat capacity temperature difference between the foreign objects and flour in the infrared thermal images to identify stones, clods of earth, moldy grains, and straw foreign objects, and outputting a rejection signal before the foreign objects reach the discharge port; refer to Figure 3 A multispectral imaging unit, including a visible-near-infrared hyperspectral camera and an infrared thermal imager, is added above the conveyor belt between the feed inlet and the bucket elevator. The hyperspectral camera acquires spectral images of the material flow in the 900-1700nm band, while the infrared thermal imager simultaneously acquires thermal images of the same area. The process is expanded to extract characteristic bands from the hyperspectral images. Organic matter such as straw and moldy particles have characteristic absorption peaks near 1450nm, while inorganic matter such as stones and clods of soil do not. At the same time, foreign objects of different materials have different specific heat capacities and thermal conductivity than flour. After friction during the conveying process, their surface temperatures differ from those of flour, and this difference is identifiable in the infrared thermal images. By fusing the two types of information, light-colored stones, clods of soil, white moldy particles, and small pieces of paper that are very similar in color to flour can be accurately identified. When a foreign object is identified, the system activates a downstream rapid pneumatic rejection device to remove it before it reaches the feed inlet, without requiring a complete shutdown of the entire line. Example

[0022] As one embodiment of the present invention, it further includes a step of monitoring early damage to the conveyor belt and / or scraper chain: acquiring a linear laser profile scanning image sequence and acoustic emission signal on the back of the conveyor belt or along the scraper chain path; extracting longitudinal edge continuity features and pitch fluctuation curves from the linear laser profile scanning image sequence to identify longitudinal microcracks, steel cord core slippage, and scraper microbending; simultaneously extracting the sudden event characteristics of the acoustic emission signal, assessing the damage development speed and issuing a graded warning after spatiotemporal correlation of the two; installing a linear laser profiler and a high-sensitivity acoustic emission sensor on the bottom surface of the return section of the conveyor belt or at the straight section of the scraper chain; the linear laser profiler emits fan-shaped laser stripes, and a high-speed camera captures the profile deformation of the lower surface of the conveyor belt or scraper; Extended to: extracting longitudinal edge continuity features from the linear laser profile image sequence; when longitudinal microcracks appear on the conveyor belt, the edge profile will show local discontinuities and abrupt changes in depth; steel wire rope core twitching will manifest as periodic changes in the local thickness of the conveyor belt; for scraper chains, extracting the pitch fluctuation curves of adjacent chain links; when the chain elongates due to wear or the scraper is slightly bent, the pitch fluctuation amplitude increases; acoustic emission sensors synchronously collect sudden acoustic emission events caused by microcrack propagation, steel rope twitching, or scraper bending, and extract the event amplitude, duration, and event interval; spatiotemporally correlating the abnormal laser profile location with the acoustic emission event source location to confirm the damage location and severity; issuing warnings at three levels—attention, warning, and shutdown—based on the damage development speed. Example

[0023] As one embodiment of the present invention, it also includes an early warning step for early failure of low-speed heavy-load bearings: simultaneously acquiring vibration signals and acoustic emission signals at the bearing housing; extracting the energy envelope of the resonant frequency band after high-pass filtering from the vibration signal to identify the periodic modulation characteristics caused by defect impact; simultaneously extracting the burst pulse interval and energy steepness of the acoustic emission signal, cross-verifying the vibration modulation frequency and the acoustic emission pulse interval, and when both match the characteristic frequency of the bearing element, determining that there is an early pitting or spalling fault and outputting bearing status indicators; installing acceleration sensors and acoustic emission sensors at the bearing housings of each shaft of the feeding device; when the bearing speed is below 200 rpm... The impact energy caused by the fault is weak; the steps are expanded to: the acceleration signal is filtered by high-pass to retain the high-frequency resonant frequency band, and its envelope signal is extracted to identify whether there is periodic modulation related to the characteristic frequencies of the bearing inner ring, outer ring, rolling elements or cage; at the same time, the rise time, energy steepness and pulse interval of the burst pulse are extracted from the acoustic emission signal; when the vibration envelope modulation frequency and the acoustic emission pulse repetition interval are both consistent with the theoretical fault characteristic frequency of a certain bearing element, and the two occur in the same time period, it can be determined that the element has early pitting or spalling, and the bearing condition index is output; this method effectively suppresses the misjudgment of background vibration and random noise through dual physical field mutual verification. Example

[0024] In one embodiment of the present invention, the infrared thermal imaging employs an 8-14μm long-wave infrared band to penetrate dust and eliminate ambient light interference. Periodic dry dust removal is performed at the optical windows of the infrared thermal imaging and terahertz imaging modules, including electrostatic neutralization and micromechanical vibration, maintaining transmittance without the use of compressed gas or liquid cleaning agents. Instead of traditional air curtain blowing and electric wipers, the optical windows of the infrared thermal imager and terahertz imaging module utilize the excellent dust penetration characteristics of the 8-14μm long-wave infrared band, minimizing the impact of slight dust accumulation on image quality. Simultaneously, the window glass is made of germanium or zinc selenide, with a hard carbon antistatic film coated on the surface. A piezoelectric ceramic micro-vibrator is installed at the outer edge of the window, performing a high-frequency micro-vibration for 5 seconds every 30 minutes to shake off loosely attached flour. An ion wind generator is also configured to periodically release positive and negative ions in front of the window to neutralize the static electricity of the flour and further reduce adsorption. Thus, long-term clean imaging is achieved without the consumption of compressed gas or any moving or wearing parts.

[0025] Furthermore, the system includes a real-time dust concentration field monitoring step: using multiple infrared thermal imagers at different locations to acquire multi-view infrared images of the silo and conveyor corridor; based on the differences in the attenuation of background infrared radiation caused by dust scattering in the images, the spatial dust concentration distribution field is inverted, and when the local concentration exceeds the safety threshold, the precision dust suppression device is activated; using multiple infrared thermal imagers already installed on the upper part of the silo and conveyor corridor, the infrared radiation is scattered and absorbed when it propagates in dusty air, resulting in attenuation of background infrared radiation in the images; by comparing the degree of infrared intensity attenuation of the same static background target by thermal imagers at different locations, combined with a pre-calibrated mass concentration-attenuation model, the integral concentration on each optical path is inverted, and the spatial dust concentration distribution field is obtained using a tomographic reconstruction algorithm; when the concentration in a certain area exceeds the safety threshold, the system automatically activates the precision spray dust suppression nozzles above that area to achieve on-demand dust suppression and avoid material dampness caused by continuous large-area spraying. Example

[0026] As one embodiment of the present invention, when at least two of the following symptoms—increased current, increased vibration, and slower material discharge—are present simultaneously, a multi-fault root cause tracing step is executed: the feature quantities extracted from each monitoring step are input into a pre-constructed fault causal probability model, which includes a causal chain of arching, foreign object blockage, bearing damage, conveyor belt slippage, and material dampness; the posterior probability of each hypothesized root cause is calculated and ranked, and the root cause with the highest probability and corresponding elimination guidance information are output; refer to Figure 4When the online monitoring system simultaneously detects multiple symptoms such as increased drive motor current, increased bearing housing vibration acceleration, and slowed discharge flow sensor readings, all the feature quantities extracted in Examples 1 to 6 are sent to the root cause tracing module. This module pre-constructs a fault causal probability model based on historical fault cases and expert knowledge. The model uses a directed acyclic graph to describe the conditional probability relationship between faults such as arching, foreign object blockage, bearing damage, conveyor belt slippage, and material dampness and intermediate symptoms. After inputting real-time features, the posterior probability of each hypothetical root cause is calculated and sorted according to probability. For example, when the arching risk index is low and the bearing condition index deteriorates, although vibration and current increase, the root cause is likely to point to bearing damage rather than arching. If the slowed discharge and increased current are accompanied by a sudden drop in the acoustic emission energy of the silo, the arching probability will rise to the top. The system sends the most probable root cause and recommended inspection and maintenance steps to the operator, greatly shortening the fault diagnosis time. Example

[0027] In one embodiment of the present invention, the acoustic emission signal is transmitted to the acoustic emission sensor via a waveguide rod fixed to the outside of the silo wall. The end of the waveguide rod in contact with the silo wall has a coupling surface that contacts the material, thereby achieving solid-state acoustic conduction and isolating the sensor from interference by dust in the air. The warning signal output by S5 includes highlighting the spatial location of arching or rat holes on the silo's 3D model using a pseudo-color cloud map, and displaying the severity level corresponding to the arching risk index or rat hole probability index and suggested treatment measures numerically. The acoustic emission signal is preferably transmitted from the silo wall solid through the waveguide rod to isolate dust interference. The warning signal can be fused with the 3D model for pseudo-color cloud map display.

[0028] Furthermore, the steps of the above-described method are implemented using an electronic device; the electronic device includes a processor and a memory. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method described above.

[0029] like Figure 5 As shown, the electronic device is represented in the form of a general-purpose computing device; the components of the electronic device may include, but are not limited to: one or more processors 410, memory 430, and communication bus 440 connecting different system components (including memory 430 and processing unit 410); Communication bus 440 represents one or more of several bus architectures, including memory bus or memory controller, peripheral bus, graphics acceleration port, processor, or local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, Industry Standard Architecture Bus (ISA), Microchannel Architecture Bus (MAC), Enhanced ISA Bus, Video Electronics Standards Association Local Bus (VESA), and Peripheral Component Interconnect Bus (PCI).

[0030] Comparative Experiment: Performance Comparison of Arch / Mouse Hole Recognition between the Invention Method and Single Infrared Method The performance advantages of the method of the present invention, such as infrared thermal imaging + acoustic emission fusion, compared with the existing technology, such as single infrared thermal imaging, in the early identification of arching in flour raw material silos and rat holes were verified. The evaluation focused on three indicators: identification accuracy, false alarm rate, and early warning lead time. Experimental silo system: One cone-bottom silo with a volume of 3m³, the silo body is made of Q235 carbon steel with a wall thickness of 4mm; the cone section of the silo wall has an inclination angle of 60° and a discharge port diameter of 300mm; equipped with a variable frequency vibrating feeder to adjust the discharge rate; the silo body has an observation window for experimental observation purposes, which should be closed during actual testing; Sensor setup; Method group of this invention: One long-wave infrared thermal imager with a wavelength of 8-14μm, a resolution of 640×480, and a frame rate of 9fps, installed at the center of the silo top, vertically photographing the material surface; Two acoustic emission sensors with a resonant frequency of 150kHz and a sensitivity of 65dB, coupled to the outer sides of the silo walls on both sides of the conical section via waveguide rods; One terahertz imaging module with a wavelength of 0.1-0.3THz, installed at the flange window of the conical section; Prior art control group: Only one long-wave infrared thermal imager of the same model is installed at the center of the silo top, with the same position and parameters as this method group; Experimental materials: Wheat flour with a moisture content of 12.5% ​​± 0.5%, a bulk density of 780 kg / m³, and an angle of repose of 32°; 200 kg each of high-moisture wheat (e.g., 15.5% moisture content) and low-moisture wheat (e.g., 10.0% moisture content) were also prepared to simulate batch differences in incoming materials. Data acquisition and processing system; Infrared thermal image: stored in real time as a 16-bit grayscale sequence; Acoustic emission signal: continuously acquired at a sampling frequency of 2MHz, real-time extraction of RMS value, event count, and frequency centroid; All data is timestamped and sent to a unified processing platform; Experimental sample preparation Four types of operating conditions were prepared, with each type repeated 10 times, for a total of 40 experimental samples: A Normal flow Normal feeding, discharge rate 2t / h 10 groups B The arched structure resembles a bridge. A removable arc-shaped baffle is inserted into the conical section to induce the formation of a stable bridging structure. 10 groups C The rat hole is an internal cavity. A dissolvable paper tube is pre-embedded in the center of the warehouse; it dissolves during the feeding process, forming a through-hole. 10 groups D Batch temperature difference interference Normal flow, but alternating the introduction of room temperature wheat (e.g., 25°C) and refrigerated wheat (e.g., 5°C) to create a non-faulty temperature difference. 10 groups Each experiment lasted 15 minutes, during which all sensor data were recorded. The experiment was conducted according to the following steps: Step 1: System calibration and baseline acquisition; Under empty conditions, acquire background noise and acoustic emission background noise of the infrared thermal imager for 5 minutes each; establish baseline data of acoustic emission feature library under normal flow conditions; Step 2: Normal flow test is condition A; add room temperature wheat (25℃) into the silo to 2 / 3 capacity; turn on the feeder and continuously discharge at a rate of 2t / h; collect data simultaneously for 15 minutes using both methods; record whether false alarms occur in either method. Step 3: The bridging failure test is under condition B; fill the bin with room temperature wheat to 2 / 3 capacity and pre-set the arc-shaped baffle; turn on the feeder to discharge normally, and after 2 minutes, remove the baffle to form a bridge; continue to collect data for 15 minutes; record the time when the first warning is issued for both methods, starting from the moment the bridge is formed; Step 4: Mouse hole malfunction test is condition C; fill the bin with room temperature wheat to 2 / 3 capacity, and pre-embed a paper tube in the center; let it stand for 20 minutes to allow the paper tube to dissolve; turn on the feeder to discharge the material; collect data for 15 minutes and record the time it takes for the two methods to identify the mouse hole; Step 5: Batch temperature difference interference test under condition D; alternately feed 100kg of wheat at 25℃ and 5℃ per layer, for a total of 4 alternations; discharge at normal feeding rate, no faults; collect data for 15 minutes, and record the number of false alarms for both methods. Step 6: Data collection and analysis; statistical analysis of detection results under various operating conditions; calculation of identification accuracy, false alarm rate, false negative rate, and early warning lead time. Experimental data Operating Condition A: False Alarm Rate During Normal Flow 0%（0 / 10） 0%（0 / 10） Condition B: Accuracy of Arch Recognition 100%（10 / 10） 90%（9 / 10） Operating Condition B: Average Early Warning Time for Arch Formation 38 seconds 52 seconds Condition C: Accuracy of mouse hole identification 100%（10 / 10） 30%（3 / 10） Condition C: Average lead time for rat burrow warning 25 seconds 67 seconds (only 3 groups were recognized) Operating Condition D: Batch Temperature Difference Interference False Alarm Rate 0%（0 / 10） 60%（6 / 10） Overall false alarm rate 0%（0 / 20） 30%（6 / 20） Analysis of Working Condition B: Both methods can identify bridging. The existing single infrared method has an accuracy of 90%, with one group not identified. The bridging thickness of this group is relatively thin, and the infrared temperature difference is only 0.8℃, which is lower than the fixed threshold of the existing method. The method of this invention complements the method by reducing the acoustic emission energy from 2.3mV to 0.4mV during normal flow, and all bridging was successfully identified. In terms of early warning time, the method of this invention issues an early warning on average 38 seconds after bridging is formed, which is earlier than the 52 seconds of the single infrared method. This is because the response speed of the acoustic emission signal to the interruption of material flow can be detected within about 5 seconds. The energy drop is faster than the time required for the infrared temperature field to establish a temperature difference. Due to the influence of the thermal conductivity of the material, it usually takes 30-50 seconds for the temperature difference to become obvious.

[0031] Analysis of Working Condition C: This is the working condition where the method of this invention has the most significant advantages. The single infrared method only identified 3 groups, with a false negative rate as high as 70%. The reason is that after the formation of the rat hole, the surface material still has intermittent collapse movement, and the surface temperature field does not form a significant temperature difference. The temperature distribution in the infrared image is not significantly different from that in the normal flow state. However, the method of this invention detects the intermittent characteristics of the flow through acoustic emission signals. The count of acoustic emission events in normal flow is stable at 120-150 times per second, while in the rat hole state, it drops sharply to 20-40 times per second and fluctuates periodically. The center of gravity of the frequency shifts from 45kHz to 12kHz, successfully identifying all 10 groups of rat hole working conditions. The early warning time is 25 seconds, which is much earlier than the 67 seconds that the single infrared method only detects the anomaly after the surface collapses.

[0032] Analysis of operating condition D: This is the main defect of the existing method; after wheat is refrigerated at 5℃, local low-temperature areas appear in the infrared image, which are similar in shape to the static low-temperature areas caused by arching. As a result, the single infrared method will have 6 false alarms, resulting in a false alarm rate of 60%; when the infrared image shows a temperature difference, the method of this invention will simultaneously detect the acoustic emission signal. Since the material is still in normal flow, the acoustic emission characteristics remain at a normal level, including RMS value, event count, and frequency centroid, all of which are within the baseline range. After fusion decision, it is determined to be a difference in the material's own properties, and no false alarms are generated.

[0033] Experimental data show that the method of the present invention, through decision-level fusion of infrared thermal imaging and acoustic emission, improves the accuracy of rat hole identification from 30% to 100% while maintaining 100% accuracy in knot identification, reduces the false alarm rate caused by batch temperature difference from 60% to 0%, and reduces the overall false alarm rate from 30% to 0%. In terms of early warning time, the method of the present invention provides an average of about 16 seconds (knots) and 42 seconds (rat holes) earlier than the single infrared method, giving operators more time to process the warnings. The experiment fully verifies the significant technical advantages of the method of the present invention under special working conditions of flour raw material feeding devices.

[0034] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying the operating status of a flour raw material feeding device, characterized in that, Includes the following steps: S1, by using a non-contact sensor array arranged in the hopper and / or discharge port, multimodal sensing data of the target area is acquired in real time, wherein the multimodal sensing data includes at least an infrared thermal imaging image sequence and an acoustic emission signal; S2, Process the infrared thermal imaging image sequence to extract the spatiotemporal characteristics of the temperature distribution field and texture variation characteristics of the material surface, and identify the initial signs of material bridging or void formation based on the uniformity of temperature distribution and local temperature anomalies. S3, synchronously analyze the energy change, event count and frequency center shift of the acoustic emission signal under material flow and static states, and combine the material flow acoustic feature library to determine whether there is a rat hole or internal material feeding interruption. S4, the spatiotemporal characteristics of the temperature distribution field, texture variation characteristics and acoustic emission characteristics are spatiotemporally aligned and fused at the decision level to generate the arch risk index and the mouse hole probability index. S5. When the arch risk index or rat hole probability index exceeds the preset threshold, an early warning signal is output and the fault type is marked.

2. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, The non-contact sensor array described in S1 also includes a terahertz imaging module for acquiring density distribution images below the surface of the material inside the silo; S2 also includes: registering and fusing the terahertz density distribution image with the infrared thermal imaging temperature distribution field, using the density sparse anomaly region to verify the location of the void inside the bridge, and correcting the arching risk index.

3. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, It also includes an online identification step for non-metallic foreign objects on the conveyor belt: acquiring hyperspectral images and infrared thermal images above the material flow on the conveyor belt; extracting the difference in reflectance of characteristic bands representing organic and inorganic matter in the hyperspectral image, and combining the specific heat capacity temperature difference between foreign objects and flour in the infrared thermal image to identify stones, clods of soil, moldy grains and straw foreign objects, and outputting a rejection signal before the foreign object reaches the discharge port.

4. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, It also includes a monitoring step for early damage to the conveyor belt and / or scraper chain: acquiring a sequence of linear laser profile scanning images and acoustic emission signals on the back of the conveyor belt or along the scraper chain path; extracting longitudinal edge continuity features and pitch fluctuation curves from the linear laser profile scanning image sequence to identify longitudinal microcracks, steel cord core slippage, and scraper microbending; simultaneously extracting the sudden event features of the acoustic emission signals, and assessing the damage development speed and issuing graded warnings after spatiotemporally correlating the two.

5. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, It also includes an early warning step for early failures of low-speed heavy-load bearings: simultaneously collecting vibration signals and acoustic emission signals at the bearing housing; extracting the energy envelope of the resonant frequency band after high-pass filtering from the vibration signal to identify the periodic modulation characteristics caused by defect impact; simultaneously extracting the burst pulse interval and energy steepness of the acoustic emission signal, and cross-verifying the vibration modulation frequency and the acoustic emission pulse interval. When both match the characteristic frequency of the bearing element, it is determined that there is an early pitting or spalling fault and the bearing status index is output.

6. A method for identifying the operating status of a flour raw material feeding device according to claim 1 or 2, characterized in that, The infrared thermal imaging uses the 8-14μm long-wave infrared band to penetrate dust and eliminate ambient light interference; the optical windows of the infrared thermal imaging and terahertz imaging perform periodic dry dust removal operations, including electrostatic neutralization and micromechanical vibration, to maintain light transmittance without the use of compressed gas and liquid cleaning agents.

7. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, It also includes a real-time dust concentration field monitoring step: using multiple infrared thermal imagers at different locations to acquire multi-view infrared images of the silo and conveyor corridor; based on the difference in the degree of background infrared radiation attenuation caused by dust scattering in the images, the spatial dust concentration distribution field is inverted, and when the local concentration exceeds the safety threshold, the precise dust suppression device is activated.

8. A method for identifying the operating status of a flour raw material feeding device according to any one of claims 1 to 7, characterized in that, When at least two of the following symptoms are present simultaneously: increased current, increased vibration, and slower material discharge, a multi-fault root cause tracing step is executed: the feature quantities extracted from each monitoring step are input into a pre-constructed fault causal probability model, which includes a causal chain of arching, foreign object blockage, bearing damage, conveyor belt slippage, and material dampness; the posterior probability of each hypothetical root cause is calculated and sorted, and the root cause with the highest probability and the corresponding elimination guidance information are output.

9. The method for identifying the operating status of a flour raw material feeding device according to claim 1, characterized in that, The acoustic emission signal is transmitted to the acoustic emission sensor through a waveguide rod fixed to the outside of the silo wall. The end of the waveguide rod that contacts the silo wall has a coupling surface that contacts the material, so as to realize solid-borne sound conduction and isolate the interference of dust in the air to the sensor. The warning signal output by S5 includes highlighting the spatial location of arching or rat hole on the silo 3D model with a pseudo-color cloud map, and displaying the severity level and recommended treatment measures corresponding to the arching risk index or rat hole probability index in numerical terms.

10. A method for identifying the operating status of a flour raw material feeding device according to any one of claims 1 to 8, characterized in that, The steps of the method are implemented using an electronic device; the electronic device includes a processor and a memory. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 8.

Images