Method and system for detecting and positioning steam leakage of corn oil deodorization equipment

Abstract

The invention relates to the technical field of leakage detection, and discloses a corn oil deodorization equipment steam leakage detection positioning method and system, and the method comprises the steps: constructing a three-dimensional digital space structure according to the spatial acoustic characteristics and structural information of deodorization equipment, and marking virtual monitoring nodes; performing calibration mapping on the steady-state sound field of the virtual monitoring node by combining steady-state background sound field data and a three-dimensional digital space structure during steady-state operation of the deodorization equipment to obtain baseline sound pressure representation; based on the three-dimensional digital space structure, carrying out reverse reconstruction on the real-time sound field data of the deodorization equipment to obtain real-time sound pressure representation of the virtual monitoring node; performing time sequence regular alignment on the real-time sound pressure characterization and the baseline sound pressure characterization, and performing energy difference comparison on an alignment result to obtain sound field residual energy; performing spatial clustering analysis on the residual energy of the sound field to obtain steam leakage information of the deodorization equipment; the detection and positioning efficiency of steam leakage of the corn oil deodorization equipment can be improved.

Description

Technical Field

[0001] This invention relates to the field of leakage detection technology, and in particular to a method and system for detecting and locating steam leaks in corn oil deodorization equipment. Background Technology

[0002] In the petrochemical and coal chemical industries, deodorization equipment is a core environmental protection device for treating odorous steam containing sulfur and ammonia. Its operational stability is directly related to environmental compliance and operational safety. This equipment operates under high temperature and high pressure conditions for a long time, and its internal steam pipelines are prone to leakage due to medium corrosion, stress fatigue, and seal aging, which can lead to energy waste, environmental pollution, and even safety accidents. Therefore, high precision and real-time requirements are placed on its steam leak detection technology. With the digital transformation of industry, acoustic detection technology and three-dimensional digital modeling technology have been gradually applied to equipment monitoring, and the industrial field has seen a monitoring approach that combines the two. However, in the specific application of steam leak detection in deodorization equipment, there are still key technical bottlenecks that limit the improvement of detection accuracy and practicality.

[0003] Existing monitoring solutions fail to combine the spatial acoustic characteristics and internal pipeline structure of deodorization equipment to accurately delineate acoustically sensitive areas and optimize the layout of virtual monitoring nodes. This results in blind spots or redundant node layouts in the monitoring range, making it impossible to specifically capture leakage sound source signals and affecting the accuracy and efficiency of leakage detection. Current technologies lack sound field baseline calibration and real-time sound field reverse reconstruction technology based on the equipment's three-dimensional digital spatial structure, making it difficult to eliminate the influence of sound field propagation attenuation, environmental interference, and other factors on monitoring signals. This leads to insufficient differentiation between leakage signals and normal operating background signals, especially in densely structured areas where positioning errors are prone to occur. Therefore, how to improve the detection and positioning efficiency of steam leaks in corn oil deodorization equipment has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for detecting and locating steam leaks in corn oil deodorization equipment, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for detecting and locating steam leaks in a corn oil deodorization equipment, comprising: S1. Based on the spatial acoustic characteristics and structural information of the deodorization equipment, construct a three-dimensional digital spatial structure of the deodorization equipment, and mark the virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital spatial structure; S2. Based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, perform sound field calibration mapping on the steady-state sound field of the virtual monitoring node to obtain the baseline sound pressure characterization of the virtual monitoring node. S3. Based on the three-dimensional digital space structure, the real-time sound field data of the deodorization device is reverse-reconstructed to obtain the real-time sound pressure characterization of the virtual monitoring node; S4. Perform time-series normalization and alignment on the real-time sound pressure characterization and the baseline sound pressure characterization, and compare the energy difference of the alignment results to obtain the sound field residual energy of the virtual monitoring node. S5. Perform spatial cluster analysis on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment.

[0006] In a preferred embodiment, the step of constructing a three-dimensional digital spatial structure of the deodorization device based on its spatial acoustic characteristics and structural information, and marking virtual monitoring nodes associated with the deodorization device in the three-dimensional digital spatial structure, includes: Collect multi-dimensional structural information of the deodorization equipment; Based on the multidimensional structural information, the deodorization equipment is digitally constructed to obtain the three-dimensional digital space structure of the deodorization equipment; Based on the layout and connection characteristics of the internal steam pipeline in the deodorization equipment, the acoustically sensitive monitoring area of ​​the deodorization equipment is delineated in the three-dimensional digital space structure. Based on the acoustic monitoring area and the preset node layout rules, virtual monitoring nodes associated with the deodorization device are deployed at key locations in the three-dimensional digital space structure.

[0007] In a preferred embodiment, the step of performing sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization device during steady-state operation and the three-dimensional digital spatial structure to obtain the baseline sound pressure characterization of the virtual monitoring node includes: The original acoustic signal of the deodorization equipment during steady-state operation is collected, and the original acoustic signal is analyzed in the time and frequency domain to obtain the steady-state background sound field data of the deodorization equipment. Wideband filtering is performed on the non-stationary noise in the steady-state background sound field data to obtain purified steady-state sound field data of the steady-state background sound field data. Based on the spatial coordinates of the virtual monitoring node, the sound field propagation path of the purified steady-state sound field data is subjected to reverse attenuation compensation to obtain the sound pressure data of the virtual monitoring node. Multidimensional feature fusion is performed on the principal component features of the time-frequency characteristics in the sound pressure data to obtain the baseline sound pressure characterization of the virtual monitoring node.

[0008] In a preferred embodiment, the step of performing reverse attenuation compensation on the sound field propagation path of the purified steady-state sound field data based on the spatial coordinates of the virtual monitoring node to obtain the sound pressure data of the virtual monitoring node includes: The spatial coordinates of the virtual monitoring node are analyzed to obtain the spatial location coordinates of the virtual monitoring node. Based on the spatial coordinates, the propagation direction of the purification steady-state sound field data is identified to obtain the propagation vector of the purification steady-state sound field data. Based on the propagation vector, the sound field propagation path of the purified steady-state sound field data is reconstructed in reverse to obtain the complete trajectory of the sound field propagation path; Based on the complete trajectory, sound pressure attenuation compensation is performed on the purified steady-state sound field data to obtain the sound pressure data of the virtual monitoring node.

[0009] In a preferred embodiment, the step of reversing the real-time sound field data of the deodorization device based on the three-dimensional digital spatial structure to obtain the real-time sound pressure characterization of the virtual monitoring node includes: Spatial sound field acquisition is performed on the deodorization equipment to obtain real-time sound field data of the deodorization equipment; Based on the three-dimensional digital space structure, multi-channel correlation analysis is performed on the real-time sound field data to obtain the sound wave propagation path parameters of the real-time sound field data. The sound field transfer matrix of the virtual monitoring node is obtained by weighted topology construction of the sound wave propagation path parameters; Based on the sound field transfer matrix, the sound source distribution of the real-time sound field data is reconstructed in reverse to obtain the real-time sound pressure characterization of the virtual monitoring node.

[0010] In a preferred embodiment, the step of reconstructing the sound source distribution inversely from the real-time sound field data based on the sound field transfer matrix to obtain the real-time sound pressure characterization of the virtual monitoring node includes: By spatially focusing the real-time sound field data, the sound source location estimate of the real-time sound field data is obtained. Based on the sound source location estimation, the sound field propagation inverse simulation is performed on the sound field transmission matrix to obtain the sound pressure distribution of the virtual monitoring node; Based on the sound pressure distribution, the sound pressure data in the real-time sound field data is corrected for distribution consistency to obtain the real-time sound pressure characterization of the virtual monitoring node.

[0011] In a preferred embodiment, the real-time sound pressure characterization and the baseline sound pressure characterization are time-ordered and aligned, and the energy difference of the alignment results is compared to obtain the sound field residual energy of the virtual monitoring node, including: Temporal features are extracted from the real-time sound pressure characterization and the baseline sound pressure characterization to obtain the real-time sound pressure time series and the baseline sound pressure time series of the virtual monitoring node; The real-time sound pressure time series is aligned with the baseline sound pressure time series in the time domain to obtain the sound pressure time series pair of the virtual monitoring node; Energy difference comparison analysis is performed on the sound pressure time series pairs to obtain the time series energy difference distribution of the virtual monitoring node; The acoustic field residual energy of the virtual monitoring node is obtained by performing energy integration and aggregation on the time-series energy difference distribution.

[0012] In a preferred embodiment, the formula for calculating the residual energy of the sound field is as follows: ; In the formula, Indicates the first The aforementioned sound field residual energy This indicates the duration of the aligned sound pressure timing pair. This indicates time integration. This represents the weighting matrix of the sound field propagation path. Indicates that the virtual monitoring node is in time The relevant feature matrix of real-time sound pressure characterization This represents the relevant feature matrix characterizing the baseline sound pressure level. This represents the difference matrix between the real-time sound pressure level representation and the baseline sound pressure level representation. Indicates the relationship with the first Spatial location of the virtual monitoring nodes The relevant Dirac function.

[0013] In a preferred embodiment, spatial clustering analysis is performed on the residual energy of the sound field to obtain steam leakage information of the deodorization equipment, including: Based on the three-dimensional digital space structure, the residual energy of the sound field is correlated by coordinate binding to obtain the residual energy coordinate data of the virtual monitoring node; The residual energy coordinate data is mapped in three dimensions to obtain the spatial distribution of the residual energy of the deodorization device; Based on a preset residual energy threshold, the spatial distribution of the residual energy is regionalized by energy threshold screening to obtain the abnormal energy accumulation area of ​​the deodorization device; Based on the energy distribution gradient characteristics of the abnormal energy accumulation area, the energy core coordinates of the abnormal energy accumulation area are determined to obtain the suspected leakage points of the deodorization equipment. The leakage degree of the deodorization equipment is assessed based on the extreme energy intensity and spatial volume of the abnormal energy accumulation area. By integrating the suspected leak locations with the degree of leakage, steam leakage information of the deodorization equipment is obtained.

[0014] To address the above problems, the present invention also provides a steam leakage detection and location system for corn oil deodorization equipment, the system comprising: The three-dimensional space and node construction module is used to construct the three-dimensional digital space structure of the deodorization equipment based on the spatial acoustic characteristics and structural information of the deodorization equipment, and to mark the virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital space structure. The steady-state sound field baseline calibration module is used to perform sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, so as to obtain the baseline sound pressure characterization of the virtual monitoring node. The real-time sound field inverse reconstruction module is used to inversely reconstruct the real-time sound field data of the deodorization device based on the three-dimensional digital space structure, so as to obtain the real-time sound pressure characterization of the virtual monitoring node. The sound pressure temporal residual calculation module is used to perform temporal normalization and alignment of the real-time sound pressure representation and the baseline sound pressure representation, and to compare the energy difference of the alignment results to obtain the sound field residual energy of the virtual monitoring node. The residual clustering leakage identification module is used to perform spatial clustering analysis on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a three-dimensional digital space structure that precisely corresponds to the physical equipment and delineates acoustically sensitive monitoring areas based on the layout and connection characteristics of the internal steam pipelines, thereby achieving optimized deployment of virtual monitoring nodes. This method deploys nodes densely and precisely in high-leakage locations, eliminating monitoring blind spots and avoiding node redundancy. This enables the monitoring network to capture characteristic sound source signals generated by leaks in a targeted and efficient manner, fundamentally improving the targeting and efficiency of leak detection.

[0016] 2. This invention effectively overcomes the effects of sound field propagation attenuation and environmental interference by employing sound field calibration mapping based on a three-dimensional digital spatial structure and real-time sound field inverse reconstruction technology. Specifically, in the baseline establishment stage, inverse attenuation compensation is performed on the sound propagation path, and in the real-time monitoring stage, the sound source distribution is inversely reconstructed using the sound field transfer matrix, ensuring that the real-time sound pressure characterization and the baseline sound pressure characterization are on the same comparable benchmark. By performing time-series regularization and energy difference comparison on the two, the abnormal sound field residual energy caused by leakage can be clearly separated, significantly improving the distinguishability between the leakage signal and background noise. Combined with spatial clustering analysis, precise location of the leakage point is finally achieved in complex structural environments. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of a method for detecting and locating steam leaks in a corn oil deodorization device according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a steam leakage detection and location system for a corn oil deodorization equipment according to an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for detecting and locating steam leaks in a corn oil deodorization device. The executing entity of this method includes, but is not limited to, at least one electronic device configured to execute the method provided in this application, such as a server or a terminal. In other words, the method for detecting and locating steam leaks in a corn oil deodorization device can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a method for detecting and locating steam leaks in a corn oil deodorization equipment according to an embodiment of the present invention. In this embodiment, the method for detecting and locating steam leaks in a corn oil deodorization equipment includes: S1. Based on the spatial acoustic characteristics and structural information of the deodorization equipment, construct a three-dimensional digital spatial structure of the deodorization equipment, and mark the virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital spatial structure; In this embodiment of the invention, the step of constructing a three-dimensional digital spatial structure of the deodorization device based on its spatial acoustic characteristics and structural information, and marking virtual monitoring nodes associated with the deodorization device in the three-dimensional digital spatial structure, includes: Collect multi-dimensional structural information of the deodorization equipment; Based on the multidimensional structural information, the deodorization equipment is digitally constructed to obtain the three-dimensional digital space structure of the deodorization equipment; Based on the layout and connection characteristics of the internal steam pipeline in the deodorization equipment, the acoustically sensitive monitoring area of ​​the deodorization equipment is delineated in the three-dimensional digital space structure. Based on the acoustic monitoring area and the preset node layout rules, virtual monitoring nodes associated with the deodorization device are deployed at key locations in the three-dimensional digital space structure.

[0021] Multidimensional structural information of the deodorization equipment is collected. This information refers to all information related to the spatial structure and internal components of the equipment, including its external dimensions, shell thickness, internal steam pipe routing, pipe diameter, connection point locations, and the distribution of mounting supports. A high-precision laser scanner is used to comprehensively scan the external contour of the deodorization equipment, acquiring its external spatial coordinate data. An industrial endoscope is inserted into the equipment to record the position, routing, and connection relationships of the pipes segment by segment. Simultaneously, calipers are used to measure the pipe diameter, wall thickness, and connection point spacing. All collected information is recorded and organized to form a complete set of multidimensional structural information for the deodorization equipment.

[0022] The deodorization equipment is digitally constructed based on multidimensional structural information. The organized multidimensional structural information is input into 3D modeling software. According to the actual size ratio and spatial position relationship, a 3D model of the shell of the deodorization equipment is first constructed. Then, based on the recorded internal steam pipeline information, the pipeline direction, connection points and pipe diameter are accurately restored inside the shell model. At the same time, 3D models of auxiliary structures such as equipment supports are added to ensure that each part of the model completely corresponds to the actual equipment. Finally, a 3D digital space structure that can completely reflect the spatial form and internal structure of the deodorization equipment is formed. This structure is a digital replica of the real state of the deodorization equipment.

[0023] Based on the layout and connection characteristics of the internal steam pipelines in the deodorization equipment, acoustically sensitive monitoring areas are delineated. The layout of the internal steam pipelines refers to the distribution and arrangement order of the pipelines within the equipment, while the connection characteristics refer to the features of the connections between pipelines via flanges, welding, etc. The specific distribution of all internal steam pipelines and the location of each connection point are clearly defined in a three-dimensional digital spatial structure. Since steam leakage is most likely to occur at pipeline connection points and stress concentration areas such as bends and diameter changes, these areas, along with the space extending outwards from them, are designated as acoustically sensitive monitoring areas. This area is the key focus for subsequently deploying virtual monitoring nodes.

[0024] Virtual monitoring nodes are deployed at key locations in the 3D digital space structure based on acoustically sensitive monitoring areas and pre-defined node layout rules. These pre-defined node layout rules are standards established to ensure that monitoring nodes can comprehensively capture the acoustic signals generated by leaking steam, based on the characteristics of acoustic propagation. Specifically, within the acoustically sensitive monitoring area, the node density is determined according to the principle of uniform distribution, taking into account the size and shape of the area, ensuring that the distance between adjacent nodes covers the effective range of acoustic signal propagation and eliminates monitoring blind spots. Key locations include connection points, bends, and diameter changes in steam pipelines, as well as areas on the casing where sealing may be weak. Virtual monitoring markers are added to the corresponding 3D digital space structure using 3D modeling software; each marker is a virtual monitoring node. These nodes accurately correspond to the key monitoring locations in the actual equipment.

[0025] The beneficial effects include ensuring the completeness and accuracy of the multi-dimensional structural information of the deodorization equipment, providing a reliable foundation for the subsequent construction of the three-dimensional digital space structure, ensuring the accuracy of the detection and positioning process, forming a virtual model corresponding to the actual equipment, realizing the visualized and precise operation of monitoring area delineation and node deployment, breaking the limitations of internal equipment observation, identifying high-risk leakage areas and key monitoring ranges, avoiding resource waste, comprehensively covering potential leakage points, improving the targeting and efficiency of monitoring, ensuring that the virtual monitoring nodes are distributed reasonably without blind spots, accurately capturing leakage acoustic signals, providing a stable monitoring benchmark, and ensuring the accuracy of leakage detection and positioning.

[0026] Based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, the steady-state sound field of the virtual monitoring node is calibrated and mapped to obtain the baseline sound pressure characterization of the virtual monitoring node. In this embodiment of the invention, the step of performing sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization device during steady-state operation and the three-dimensional digital spatial structure to obtain the baseline sound pressure characterization of the virtual monitoring node includes: The original acoustic signal of the deodorization equipment during steady-state operation is collected, and the original acoustic signal is analyzed in the time and frequency domain to obtain the steady-state background sound field data of the deodorization equipment. Wideband filtering is performed on the non-stationary noise in the steady-state background sound field data to obtain purified steady-state sound field data of the steady-state background sound field data. Based on the spatial coordinates of the virtual monitoring node, the sound field propagation path of the purified steady-state sound field data is subjected to reverse attenuation compensation to obtain the sound pressure data of the virtual monitoring node. Multidimensional feature fusion is performed on the principal component features of the time-frequency characteristics in the sound pressure data to obtain the baseline sound pressure characterization of the virtual monitoring node.

[0027] The process of performing inverse attenuation compensation on the sound field propagation path of the purified steady-state sound field data based on the spatial coordinates of the virtual monitoring node to obtain the sound pressure data of the virtual monitoring node includes: The spatial coordinates of the virtual monitoring node are analyzed to obtain the spatial location coordinates of the virtual monitoring node. Based on the spatial coordinates, the propagation direction of the purification steady-state sound field data is identified to obtain the propagation vector of the purification steady-state sound field data. Based on the propagation vector, the sound field propagation path of the purified steady-state sound field data is reconstructed in reverse to obtain the complete trajectory of the sound field propagation path; Based on the complete trajectory, sound pressure attenuation compensation is performed on the purified steady-state sound field data to obtain the sound pressure data of the virtual monitoring node.

[0028] The process involves collecting raw acoustic signals from the deodorization equipment during steady-state operation. These raw signals represent all sounds generated by the equipment's operation and internal steam flow during stable operation, including basic acoustic information for normal operation and any potential interference. Time-frequency domain analysis simultaneously observes the temporal variation and frequency distribution of the raw acoustic signals. Specifically, the raw acoustic signals are broken down chronologically, and the sound intensity at each time point is recorded. Then, the overall signal is decomposed into frequencies to determine the proportion and intensity of different frequency components. Through this analysis, a set of persistent acoustic information with minimal frequency and intensity fluctuations is selected. This set constitutes the steady-state background sound field data of the deodorization equipment, accurately reflecting the basic acoustic state during normal steady-state operation.

[0029] Wideband filtering is applied to non-stationary noise in steady-state background sound field data. Non-stationary noise refers to interfering acoustic signals in steady-state background sound field data whose frequency and intensity change irregularly over time, such as sudden vibrations in the external environment or sounds generated by transient electromagnetic interference. These noises are unrelated to the acoustic signals generated during normal equipment operation. Wideband filtering selects a filtering interval that covers all possible frequency ranges where non-stationary noise may exist, allowing the steady-state background sound field data to pass through this interval. Signals that conform to the acoustic characteristics of normal equipment operation and have stable frequencies can pass smoothly, while non-stationary noise signals, due to frequency fluctuations exceeding the normal acoustic signal range of the equipment, are filtered out. The result after this processing is a set containing only the acoustic signals of normal steady-state operation of the equipment; this is called purified steady-state sound field data. It eliminates irrelevant interference, making the acoustic signals more closely resemble the true steady-state operating state of the equipment.

[0030] The spatial coordinates of the virtual monitoring nodes are analyzed. The spatial coordinates of the virtual monitoring nodes are information set for each virtual monitoring node to identify its position in three-dimensional space when constructing the three-dimensional digital space structure of the deodorization equipment. Coordinate analysis reads the coordinate records of each virtual monitoring node stored in the three-dimensional digital space structure and clarifies the specific values ​​of the length, width and height of each node in three-dimensional space. These specific values ​​are combined to form the spatial position coordinates of the virtual monitoring node. It can accurately locate the specific orientation of each virtual monitoring node in three-dimensional digital space and provide a clear position reference for subsequent sound field propagation analysis.

[0031] Based on spatial coordinates, the propagation direction of the sound field is identified in the purified steady-state sound field data. This identification involves observing the direction of sound signal propagation from inside the device to the virtual monitoring nodes, based on the spatial coordinates of each virtual monitoring node and the source of the purified steady-state sound field data. Specifically, it involves comparing the timing and intensity differences of the purified steady-state sound field data received by different virtual monitoring nodes. Nodes closer to the sound source receive the signal earlier and with stronger intensity, while nodes farther away receive it later and with weaker intensity. Through this comparison, the specific direction of the sound signal from the sound source to each virtual monitoring node is determined. This acoustic signal representation with a clear propagation direction is the propagation vector of the purified steady-state sound field data, which clearly reflects the direction of sound field propagation.

[0032] Based on the propagation vector, the propagation path of the sound field data of the purification steady-state sound field is reconstructed in reverse. The reverse path reconstruction is to trace the propagation path of the purification steady-state sound field data backward from the location of the virtual monitoring node along the direction indicated by the propagation vector. Combined with the three-dimensional digital space structure of the deodorization equipment, considering the internal pipeline layout, structural obstructions, etc., the various positions, path turns, and structural components that the sound signal passes through during its propagation from the sound source to the virtual monitoring node are restored. The final path record that can completely show the entire process of the sound signal propagating from the sound source to the virtual monitoring node is the complete trajectory of the sound field propagation path, which presents the specific route of the sound field propagation in detail.

[0033] Based on the complete trajectory, sound pressure attenuation compensation is performed on the purified steady-state sound field data. Sound pressure attenuation refers to the phenomenon that the pressure of the sound signal gradually weakens as the purified steady-state sound field data propagates along the complete trajectory due to factors such as increased propagation distance and encounters with equipment structures. Sound pressure attenuation compensation calculates the amount of sound pressure attenuation caused by these factors during propagation based on information such as the propagation distance recorded in the complete trajectory and the type and number of obstacles encountered during propagation. This attenuation is then added to the current sound pressure value of the purified steady-state sound field data. The resulting value, which accurately reflects the actual sound pressure level at the virtual monitoring node, is the sound pressure data of the virtual monitoring node. It eliminates the influence of attenuation during propagation on sound pressure measurement, ensuring that the sound pressure data accurately corresponds to the actual acoustic state of the virtual monitoring node.

[0034] Multidimensional feature fusion is performed on the principal component features of the time-frequency characteristics in the sound pressure data. Time-frequency features are the attributes of sound pressure data in the time and frequency dimensions. Time-dimensional features reflect the variation of sound pressure data over time, such as the change in sound pressure intensity at different time points. Frequency-dimensional features reflect the frequency components contained in the sound pressure data and the intensity proportion of each frequency. Principal component features refer to the most representative features that play a major role in reflecting the acoustic state of the equipment during normal operation, such as the sound pressure variation pattern corresponding to the dominant frequency during normal operation and the stable range of sound pressure within a fixed time interval. Multidimensional feature fusion integrates these principal component features according to their importance to form a unified, comprehensive, and accurate integrated feature information that reflects the acoustic characteristics of the equipment during normal steady-state operation at the virtual monitoring node. This integrated feature information is the baseline sound pressure characterization of the virtual monitoring node, which provides a standard acoustic reference for subsequent judgment of whether there is steam leakage in the equipment.

[0035] The beneficial effects include: comprehensively capturing acoustic information during normal equipment operation, providing complete and realistic basic data, providing reliable support for establishing baseline sound pressure characterization, eliminating non-stationary noise interference, making the acoustic signal more purely reflect the normal acoustic characteristics of the equipment, improving the accuracy of subsequent analysis, clarifying the specific location of virtual monitoring nodes, providing accurate location basis for subsequent acoustic analysis, avoiding deviations caused by ambiguous locations, determining the direction of sound field propagation, providing clear guidance for reverse path reconstruction, improving the pertinence and accuracy of path restoration, restoring the entire sound field propagation process, providing specific path reference for sound pressure attenuation compensation, ensuring the authenticity and reliability of the compensated sound pressure data, eliminating the influence of sound pressure attenuation during propagation, making the sound pressure data truly reflect the sound pressure level during normal equipment operation, providing an accurate basis for feature extraction and fusion, integrating the core acoustic characteristics of normal equipment operation, forming a unified and comprehensive standard acoustic reference, and providing a clear and reliable benchmark for leak identification.

[0036] S3. Based on the three-dimensional digital space structure, the real-time sound field data of the deodorization device is reverse-reconstructed to obtain the real-time sound pressure characterization of the virtual monitoring node; In this embodiment of the invention, the step of reversing the real-time sound field data of the deodorization device based on the three-dimensional digital spatial structure to obtain the real-time sound pressure characterization of the virtual monitoring node includes: Spatial sound field acquisition is performed on the deodorization equipment to obtain real-time sound field data of the deodorization equipment; Based on the three-dimensional digital space structure, multi-channel correlation analysis is performed on the real-time sound field data to obtain the sound wave propagation path parameters of the real-time sound field data. The sound field transfer matrix of the virtual monitoring node is obtained by weighted topology construction of the sound wave propagation path parameters; Based on the sound field transfer matrix, the sound source distribution of the real-time sound field data is reconstructed in reverse to obtain the real-time sound pressure characterization of the virtual monitoring node.

[0037] The step of reconstructing the sound source distribution inversely from the real-time sound field data based on the sound field transfer matrix to obtain the real-time sound pressure characterization of the virtual monitoring node includes: By spatially focusing the real-time sound field data, the sound source location estimate of the real-time sound field data is obtained. Based on the sound source location estimation, the sound field propagation inverse simulation is performed on the sound field transmission matrix to obtain the sound pressure distribution of the virtual monitoring node; Based on the sound pressure distribution, the sound pressure data in the real-time sound field data is corrected for distribution consistency to obtain the real-time sound pressure characterization of the virtual monitoring node.

[0038] Multiple acoustic sensors are arranged in the shell, pipe interfaces, valves and other parts of the deodorization equipment that are prone to steam leakage, as well as in the surrounding space, according to the principle of uniform distribution and coverage of the entire equipment range. The spacing between the sensors is set according to the size of the equipment to ensure that there are no signal acquisition blind spots. All sensors are started synchronously and continuously capture the sound signals of the surrounding space during the operation of the equipment. These signals include the background sound generated by the normal operation of the equipment and the characteristic sound generated when steam leaks. The real-time sound field data is the set of raw signals collected by these sensors in the same time period, which can reflect the sound intensity and frequency characteristics of different locations.

[0039] The three-dimensional digital spatial structure is a virtual spatial model constructed in a computer that perfectly corresponds to the actual equipment. This model is created by comprehensively scanning the physical structure of the deodorization equipment using 3D scanning technology to obtain the size, shape, and spatial location information of each component. The model includes precise spatial data for all components such as the equipment casing, pipes, and valves. Based on this three-dimensional digital spatial structure, real-time sound field data collected by different sound sensors are used as signals from independent channels. The temporal sequence, signal intensity changes, and frequency characteristic matching of each channel's signals are compared one by one. Because sound waves generated at the same steam leak point will propagate to different sensors through different paths, the signals from each channel exhibit specific correlations. Through this comprehensive comparative analysis, the specific route, propagation distance, and attenuation of sound intensity during propagation of the sound waves from the possible leak point to each sensor are determined. This information collectively constitutes the sound wave propagation path parameters.

[0040] Virtual monitoring nodes are virtual locations uniformly arranged at a preset density in a three-dimensional digital space structure. These locations cover all areas around the equipment where steam leakage may occur, simulating the monitoring effect of actual monitoring points. Using the three-dimensional digital space structure as a framework, each virtual monitoring node is treated as an independent node. Based on the sound wave propagation distance and attenuation in the sound wave propagation path parameters, corresponding weight values ​​are assigned to the sound wave propagation correlation between each virtual monitoring node and each sound sensor, as well as between virtual monitoring nodes themselves. The closer the propagation distance and the smaller the attenuation, the larger the weight value, and vice versa. Based on these weighted correlations, a network structure that can clearly reflect the transmission efficiency and influence of sound waves in virtual space is constructed. The sound field transmission matrix is ​​the digital representation of this network structure, containing the weight information and correlation rules of sound wave transmission between all virtual monitoring nodes and sensors, and between virtual monitoring nodes.

[0041] By utilizing the sound wave transmission laws contained in the sound field transmission matrix, real-time sound field data is filtered to remove background noise signals generated during normal equipment operation, retaining sound signals with steam leakage characteristics. These characteristic signals are then summarized and focused according to signal strength and frequency characteristics to pinpoint the source direction of the strongest signal with a frequency consistent with leakage characteristics. Sound source location estimation is the approximate spatial range where steam leakage may occur, obtained through this focusing analysis. This includes the coordinate range of this range in the three-dimensional digital space structure and its positional relationship relative to specific components of the equipment.

[0042] Starting from the approximate spatial range determined by the sound source location estimation, and combining the spatial position information of the equipment components in the three-dimensional digital space structure, the process of sound waves propagating from this area to each virtual monitoring node is reversed. During the deduction process, the sound wave transmission weight and correlation law reflected in the sound field transmission matrix are strictly followed. The sound intensity that each virtual monitoring node will receive during this propagation process is calculated. The sound pressure distribution is the overall distribution of the sound intensity values ​​corresponding to all virtual monitoring nodes, which can intuitively reflect the strength of the sound pressure generated by the leakage sound source that may be felt at different virtual monitoring nodes.

[0043] The sound pressure distribution obtained from the inverse simulation of sound field propagation is compared point by point with the actual sound pressure values ​​in the real-time sound field data collected by the sound sensor. The differences in the magnitude and trend of change between the two are examined. For virtual monitoring nodes with large differences, the sound pressure values ​​of the nodes are adjusted according to the change law of the actual sound pressure values ​​and the environmental characteristics of the three-dimensional digital space structure. This ensures that the sound pressure distribution of all virtual monitoring nodes is consistent with the actual collected sound pressure data, eliminating deviations caused by minor differences between the virtual model and the actual environment or sensor errors. The real-time sound pressure representation is the specific value corresponding to each virtual monitoring node after adjustment, which can accurately reflect the real sound pressure situation at that location around the device.

[0044] The beneficial effects include: comprehensive coverage of potential leakage areas, avoiding signal omissions, providing complete and authentic raw data support for subsequent detection, accurately mining signal correlation patterns, clarifying sound wave propagation paths, providing a reliable basis for constructing sound field transmission matrices, clearly presenting the sound wave virtual space transmission patterns, clarifying node correlation strength, providing scientific model support for reverse simulation and localization, eliminating background noise interference, quickly locating the source of leakage characteristic signals, narrowing the search range, improving detection targeting and efficiency, simulating the sound wave propagation process, accurately calculating node sound pressure distribution, providing a clear reference standard for sound pressure data correction, eliminating the influence of various errors, ensuring the accuracy of real-time sound pressure characterization, and providing a reliable judgment basis for precise location of leakage points.

[0045] S4. Perform time-series normalization and alignment on the real-time sound pressure characterization and the baseline sound pressure characterization, and compare the energy difference of the alignment results to obtain the sound field residual energy of the virtual monitoring node. In this embodiment of the invention, the real-time sound pressure characterization and the baseline sound pressure characterization are time-ordered and aligned, and the energy difference of the alignment results is compared to obtain the sound field residual energy of the virtual monitoring node, including: Temporal features are extracted from the real-time sound pressure characterization and the baseline sound pressure characterization to obtain the real-time sound pressure time series and the baseline sound pressure time series of the virtual monitoring node; The real-time sound pressure time series is aligned with the baseline sound pressure time series in the time domain to obtain the sound pressure time series pair of the virtual monitoring node; Energy difference comparison analysis is performed on the sound pressure time series pairs to obtain the time series energy difference distribution of the virtual monitoring node; The acoustic field residual energy of the virtual monitoring node is obtained by performing energy integration and aggregation on the time-series energy difference distribution.

[0046] The specific formula for calculating the residual energy of the sound field is as follows: ; In the formula, Indicates the first The aforementioned sound field residual energy This indicates the duration of the aligned sound pressure timing pair. This indicates time integration. This represents the weighting matrix of the sound field propagation path. Indicates that the virtual monitoring node is in time The relevant feature matrix of real-time sound pressure characterization This represents the relevant feature matrix characterizing the baseline sound pressure level. This represents the difference matrix between the real-time sound pressure level representation and the baseline sound pressure level representation. Indicates the relationship with the first Spatial location of the virtual monitoring nodes The relevant Dirac function.

[0047] For both real-time and baseline sound pressure levels (SPL) representations, the time-dependent characteristics of SPL variations are captured. These characteristics include time-related attributes such as changes in SPL intensity and frequency. The real-time SPL representation is a comprehensive feature representation of the real-time SPL of the virtual monitoring node, while the baseline SPL representation is a benchmark feature representation of the SPL of the virtual monitoring node during steady-state operation. During the extraction process, the feature data of both types of representations are recorded sequentially in chronological order, forming a continuous data sequence. The sequence corresponding to the real-time SPL representation is the real-time SPL time series, and the sequence corresponding to the baseline SPL representation is the baseline SPL time series. Each sequence contains multiple feature data points arranged chronologically.

[0048] Using the time scale of the baseline sound pressure time series as a standard reference, the time marker of the real-time sound pressure time series is adjusted so that the time position of each feature data point in the real-time sound pressure time series is completely consistent with the time position of the corresponding feature data point in the baseline sound pressure time series. During the adjustment process, if there are slight differences in the sampling interval between the real-time sound pressure time series and the baseline sound pressure time series, the data is supplemented or normalized by uniform interpolation to ensure that both time series have corresponding feature data at the same time point. Finally, a pair of sound pressure time series is formed in which each time point contains both real-time sound pressure feature data and baseline sound pressure feature data.

[0049] The energy of sound pressure data refers to the acoustic energy carried by the sound pressure. Based on the real-time sound pressure characteristic data and baseline sound pressure characteristic data corresponding to each time point in the sound pressure time series, the acoustic energy corresponding to the two types of data is calculated separately. Then, the difference between the real-time sound pressure energy and the baseline sound pressure energy at each time point is calculated. When the real-time sound pressure energy is higher than the baseline sound pressure energy, the difference is positive; when the real-time sound pressure energy is lower than the baseline sound pressure energy, the difference is negative; when the two energies are equal, the difference is zero. The energy differences of all time points are arranged sequentially according to their corresponding time order to form a continuous difference sequence, which is the time series energy difference distribution.

[0050] First, determine the duration of the sound pressure timing pair. This duration is the total time from the first time point to the last time point of the sound pressure timing pair. Within this duration, the energy difference at each time point in the timing energy difference distribution is accumulated. The accumulation process is to add the energy differences at all time points in sequence. The final sum is the sound field residual energy. This value comprehensively reflects the total energy difference between the real-time sound pressure and the baseline sound pressure over the entire duration.

[0051] In the formula, The residual energy of the sound field quantifies the degree of energy deviation between the real-time sound pressure of the virtual monitoring node and the baseline sound pressure, and is a core quantitative indicator for determining whether there is a steam leak in the area corresponding to that node. The time window for time series analysis is determined based on the acquisition cycle of the acoustic sensor and the steady-state operating characteristics of the deodorization equipment. This ensures that a stable sound pressure time series can be captured completely, avoiding misjudgments due to excessively short time segments. Implement a time window as a time integral operator. The accumulation of instantaneous energy differences corresponds to the energy integration aggregation step, which transforms dynamic temporal differences into accumulated energy values. This is a time-regulation correction operator used to correct the time deviation between real-time sound pressure level and baseline sound pressure level, ensuring precise alignment between the two in the time domain and guaranteeing the effectiveness of difference comparison. The time-domain covariance matrix represents the real-time sound pressure level, comprehensively reflecting the dynamic fluctuation pattern of sound pressure under real-time operating conditions. The time-domain covariance matrix characterizing the baseline sound pressure reflects the inherent properties of sound pressure during normal steady-state operation of the equipment and serves as a benchmark for judging whether the sound pressure is abnormal. The sound pressure covariance difference term is quantized at time [time]. The deviation of real-time sound pressure fluctuations from the baseline is the core indicator for capturing the acoustic characteristics of steam leaks; leaks will significantly increase this difference. Spatial weighting factor, based on the coordinates of virtual monitoring nodes. In conjunction with acoustic sensitivity level settings, the differential contribution of high-risk leakage areas is enhanced, improving the targeting of leakage identification. The L2 norm squared operator transforms the vector form of the sound pressure covariance difference term into a scalar form of energy intensity, following the acoustic energy quantization logic to ensure the difference results conform to physical calculation rules. The time averaging factor is used to average the cumulative energy after integration, eliminating energy fluctuations caused by transient disturbances during equipment operation, and ensuring... It more closely matches the deviation trend of the real sound field.

[0052] When steam leakage causes sudden changes in real-time sound pressure, frequency shifts, or other anomalies... and The differences will increase significantly, directly driving The higher the temperature, and the greater the difference, especially with more severe leaks. The higher the value, the more nodes in the acoustically sensitive area. With higher weighting, under the same sound pressure difference, these nodes... It will be larger, enabling sensitive identification of areas prone to leaks, avoiding missed detections in critical areas, and measuring the duration of the difference; if the sound pressure difference is within the time window As it persists within the time frame, the time integral result will accumulate and increase. This increases the accuracy of the detection, effectively distinguishing between transient interference and continuous leakage, thus reducing the false positive rate. The accuracy of the calibration directly affects and The more precise the alignment, the more accurately the difference between the two will reflect the acoustic anomalies caused by the leak. The more reliable the quantification results, the better to avoid misjudgments or omissions caused by timing deviations.

[0053] The beneficial effects are: preserving the changing patterns of sound pressure characteristics over time, providing a unified and continuous basic data, avoiding comparison errors caused by fragmented feature data, achieving accurate time alignment with baseline sound pressure data in real time, eliminating time deviation interference, improving the accuracy of difference analysis, intuitively presenting the degree and direction of sound energy deviation at different time points, providing detailed and comprehensive difference data, avoiding omissions of details in the overall comparison, transforming the differences at scattered time points into unified quantitative indicators, clearly reflecting the overall degree of sound energy deviation, and providing core and reliable data for leak identification.

[0054] S5. Perform spatial clustering analysis on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment; In this embodiment of the invention, spatial clustering analysis is performed on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment, including: Based on the three-dimensional digital space structure, the residual energy of the sound field is correlated by coordinate binding to obtain the residual energy coordinate data of the virtual monitoring node; The residual energy coordinate data is mapped in three dimensions to obtain the spatial distribution of the residual energy of the deodorization device; Based on a preset residual energy threshold, the spatial distribution of the residual energy is regionalized by energy threshold screening to obtain the abnormal energy accumulation area of ​​the deodorization device; Based on the energy distribution gradient characteristics of the abnormal energy accumulation area, the energy core coordinates of the abnormal energy accumulation area are determined to obtain the suspected leakage points of the deodorization equipment. The leakage degree of the deodorization equipment is assessed based on the extreme energy intensity and spatial volume of the abnormal energy accumulation area. By integrating the suspected leak locations with the degree of leakage, steam leakage information of the deodorization equipment is obtained.

[0055] The three-dimensional digital spatial structure based on the deodorization equipment is digitally constructed by collecting multi-dimensional structural information of the equipment, including the spatial location information of each component and labeled virtual monitoring nodes. The acoustic field residual energy is the energy difference result obtained after time-series normalization and alignment and energy difference comparison between the real-time sound pressure characterization and the baseline sound pressure characterization. Coordinate binding and association involves associating the acoustic field residual energy corresponding to each virtual monitoring node with its specific spatial coordinates in the three-dimensional digital spatial structure, ensuring that each energy data point accurately matches a specific location on the equipment. The resulting residual energy coordinate data is a complete data set composed of the spatial coordinates of each virtual monitoring node and its corresponding acoustic field residual energy. Each data item clearly points to the location of a virtual monitoring node and its energy difference.

[0056] The residual energy coordinate data includes the spatial coordinates of the virtual monitoring nodes and the corresponding acoustic field residual energy. The three-dimensional spatial mapping is to accurately project these data items into the previously constructed three-dimensional digital space structure according to their respective spatial coordinates. By marking the energy value corresponding to each coordinate point in the three-dimensional space, the discrete energy data is transformed into a continuous spatial distribution picture. The final residual energy spatial distribution is the spatial distribution of the acoustic field residual energy corresponding to all virtual monitoring nodes in the three-dimensional digital space structure. It can intuitively present the strength distribution of energy in different areas of the device and clearly show which areas have more obvious energy differences.

[0057] The preset residual energy threshold is a standard value determined by statistically analyzing the energy fluctuation range of the deodorization equipment during normal operation, based on long-term steady-state energy data. It is used to clearly distinguish between normal and abnormal energy differences. Regionalized energy threshold screening involves examining the energy value at each location in the spatial distribution of residual energy, identifying all locations where energy values ​​exceed the preset threshold, and then dividing these locations into regions based on their spatial continuity. Spatially connected locations exceeding the threshold are integrated into a complete region. The resulting abnormal energy aggregation region is a collection of all spatially continuous areas in the residual energy spatial distribution where energy values ​​exceed the preset threshold. Each region represents a potential area on the equipment where steam leakage may occur.

[0058] Energy distribution gradient characteristics refer to the pattern of energy change from strong to weak within an abnormal energy accumulation area. By comparing the energy values ​​at each location within this area, the trend of energy change can be clearly identified. Calibrating the energy core coordinates involves finding the point with the highest energy value within the abnormal energy accumulation area, following the direction of energy change from strong to weak. Then, the specific coordinates of this point in the three-dimensional digital space structure are determined. The resulting suspected leak point is the three-dimensional coordinates of this point with the highest energy value. These coordinates perfectly correspond to the specific location in the actual physical structure of the deodorization equipment, indicating the most likely location for steam leakage.

[0059] The extreme energy intensity value is the highest energy value among all locations within the abnormal energy accumulation area, which can be determined by traversing the energy data of all locations within the area. The spatial volume is the size of the space occupied by the abnormal energy accumulation area in the three-dimensional digital space structure. It is obtained by measuring the length, width, and height dimensions of the area in three-dimensional space and then calculating their product. When assessing the degree of leakage, a higher extreme energy intensity value indicates a stronger energy disturbance caused by the leakage, and a larger spatial volume indicates a wider area of ​​equipment affected by the leakage. Combining the actual situation of these two indicators, the degree of leakage is clearly divided into three levels: minor leakage, moderate leakage, and severe leakage, clearly defining the severity of the current leakage situation of the deodorization equipment.

[0060] Suspected leak locations are the three-dimensional coordinates of identified potential leak sites, while the leak severity level is the severity level of the leak determined through assessment. The integration process involves systematically summarizing these two pieces of information, matching each suspected leak location with its corresponding leak severity level to form a complete information set. The resulting steam leak information is this complete set containing specific leak locations and corresponding leak severity levels, which can comprehensively and accurately reflect the steam leak situation of the deodorization equipment.

[0061] The beneficial effects include: accurately binding energy and node coordinates to ensure that spatial analysis corresponds to the actual location of the equipment, avoiding positioning deviations and laying the foundation for accurate identification of leak locations; three-dimensional spatial mapping transforms discrete data into intuitive energy distribution maps, facilitating rapid observation of the morphology and extent of abnormal areas, providing an intuitive basis for locating potential leak areas; preset threshold filtering distinguishes between normal fluctuations and abnormal clusters, eliminating interference from slight energy differences and improving the accuracy of leak detection; calibrating core coordinates based on energy distribution gradients to accurately locate the leak source from abnormal areas, achieving precise leak point identification; assessing the degree of leakage by combining energy intensity and spatial range, comprehensively and objectively reflecting the severity of the leak, providing a reliable basis for developing maintenance plans; and integrating leak locations and leakage levels to avoid maintenance confusion, clarifying maintenance priorities, and improving maintenance efficiency.

[0062] like Figure 2 The diagram shown is a functional block diagram of a steam leakage detection and location system for a corn oil deodorization equipment according to an embodiment of the present invention.

[0063] The steam leak detection and location system 100 for corn oil deodorization equipment described in this invention can be installed in an electronic device. Depending on the functions implemented, the steam leak detection and location system 100 may include a three-dimensional space and node construction module 101, a steady-state sound field baseline calibration module 102, a real-time sound field inverse reconstruction module 103, a sound pressure temporal residual calculation module 104, and a residual clustering leak identification module 105. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0064] In this embodiment, the functions of each module / unit are as follows: The three-dimensional space and node construction module 101 is used to construct a three-dimensional digital space structure of the deodorization equipment based on the spatial acoustic characteristics and structural information of the deodorization equipment, and to mark virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital space structure. The steady-state sound field baseline calibration module 102 is used to perform sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, so as to obtain the baseline sound pressure characterization of the virtual monitoring node. The real-time sound field reverse reconstruction module 103 is used to reverse reconstruct the real-time sound field data of the deodorization device based on the three-dimensional digital space structure, so as to obtain the real-time sound pressure characterization of the virtual monitoring node. The sound pressure temporal residual calculation module 104 is used to perform temporal regularization and alignment of the real-time sound pressure representation and the baseline sound pressure representation, and to compare the energy difference of the alignment result to obtain the sound field residual energy of the virtual monitoring node. The residual clustering leakage identification module 105 is used to perform spatial clustering analysis on the acoustic field residual energy to obtain the steam leakage information of the deodorization equipment.

[0065] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0066] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0067] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0068] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0069] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting and locating steam leaks in a corn oil deodorization equipment, characterized in that, The method includes: S1. Based on the spatial acoustic characteristics and structural information of the deodorization equipment, construct a three-dimensional digital spatial structure of the deodorization equipment, and mark the virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital spatial structure; S2. Based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, perform sound field calibration mapping on the steady-state sound field of the virtual monitoring node to obtain the baseline sound pressure characterization of the virtual monitoring node. S3. Based on the three-dimensional digital space structure, the real-time sound field data of the deodorization device is reverse-reconstructed to obtain the real-time sound pressure characterization of the virtual monitoring node; S4. Perform time-series normalization and alignment on the real-time sound pressure characterization and the baseline sound pressure characterization, and compare the energy difference of the alignment results to obtain the sound field residual energy of the virtual monitoring node. S5. Perform spatial cluster analysis on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment.

2. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 1, characterized in that, The step of constructing a three-dimensional digital spatial structure of the deodorization equipment based on its spatial acoustic characteristics and structural information, and marking virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital spatial structure, includes: Collect multi-dimensional structural information of the deodorization equipment; Based on the multidimensional structural information, the deodorization equipment is digitally constructed to obtain the three-dimensional digital space structure of the deodorization equipment; Based on the layout and connection characteristics of the internal steam pipeline in the deodorization equipment, the acoustically sensitive monitoring area of ​​the deodorization equipment is delineated in the three-dimensional digital space structure. Based on the acoustic monitoring area and the preset node layout rules, virtual monitoring nodes associated with the deodorization device are deployed at key locations in the three-dimensional digital space structure.

3. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 1, characterized in that, The method of performing sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital spatial structure to obtain the baseline sound pressure characterization of the virtual monitoring node includes: The original acoustic signal of the deodorization equipment during steady-state operation is collected, and the original acoustic signal is analyzed in the time and frequency domain to obtain the steady-state background sound field data of the deodorization equipment. Wideband filtering is performed on the non-stationary noise in the steady-state background sound field data to obtain purified steady-state sound field data of the steady-state background sound field data. Based on the spatial coordinates of the virtual monitoring node, the sound field propagation path of the purified steady-state sound field data is subjected to reverse attenuation compensation to obtain the sound pressure data of the virtual monitoring node. Multidimensional feature fusion is performed on the principal component features of the time-frequency characteristics in the sound pressure data to obtain the baseline sound pressure characterization of the virtual monitoring node.

4. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 3, characterized in that, The process of performing inverse attenuation compensation on the sound field propagation path of the purified steady-state sound field data based on the spatial coordinates of the virtual monitoring node to obtain the sound pressure data of the virtual monitoring node includes: The spatial coordinates of the virtual monitoring node are analyzed to obtain the spatial location coordinates of the virtual monitoring node. Based on the spatial coordinates, the propagation direction of the purification steady-state sound field data is identified to obtain the propagation vector of the purification steady-state sound field data. Based on the propagation vector, the sound field propagation path of the purified steady-state sound field data is reconstructed in reverse to obtain the complete trajectory of the sound field propagation path; Based on the complete trajectory, sound pressure attenuation compensation is performed on the purified steady-state sound field data to obtain the sound pressure data of the virtual monitoring node.

5. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 1, characterized in that, The process of reconstructing the real-time sound field data of the deodorization device based on the three-dimensional digital spatial structure to obtain the real-time sound pressure characterization of the virtual monitoring node includes: Spatial sound field acquisition is performed on the deodorization equipment to obtain real-time sound field data of the deodorization equipment; Based on the three-dimensional digital space structure, multi-channel correlation analysis is performed on the real-time sound field data to obtain the sound wave propagation path parameters of the real-time sound field data. The sound field transfer matrix of the virtual monitoring node is obtained by weighted topology construction of the sound wave propagation path parameters; Based on the sound field transfer matrix, the sound source distribution of the real-time sound field data is reconstructed in reverse to obtain the real-time sound pressure characterization of the virtual monitoring node.

6. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 5, characterized in that, The step of reconstructing the sound source distribution inversely from the real-time sound field data based on the sound field transfer matrix to obtain the real-time sound pressure characterization of the virtual monitoring node includes: By spatially focusing the real-time sound field data, the sound source location estimate of the real-time sound field data is obtained. Based on the sound source location estimation, the sound field propagation inverse simulation is performed on the sound field transmission matrix to obtain the sound pressure distribution of the virtual monitoring node; Based on the sound pressure distribution, the sound pressure data in the real-time sound field data is corrected for distribution consistency to obtain the real-time sound pressure characterization of the virtual monitoring node.

7. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 1, characterized in that, The real-time sound pressure characterization and the baseline sound pressure characterization are time-ordered and aligned, and the energy difference of the alignment results is compared to obtain the sound field residual energy of the virtual monitoring node, including: Temporal features are extracted from the real-time sound pressure characterization and the baseline sound pressure characterization to obtain the real-time sound pressure time series and the baseline sound pressure time series of the virtual monitoring node; The real-time sound pressure time series is aligned with the baseline sound pressure time series in the time domain to obtain the sound pressure time series pair of the virtual monitoring node; Energy difference comparison analysis is performed on the sound pressure time series pairs to obtain the time series energy difference distribution of the virtual monitoring node; The acoustic field residual energy of the virtual monitoring node is obtained by performing energy integration and aggregation on the time-series energy difference distribution.

8. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 7, characterized in that, The specific formula for calculating the residual energy of the sound field is as follows: ； In the formula, Indicates the first The aforementioned sound field residual energy This indicates the duration of the aligned sound pressure timing pair. This indicates time integration. This represents the weighting matrix of the sound field propagation path. Indicates that the virtual monitoring node is in time The relevant feature matrix of real-time sound pressure characterization This represents the relevant feature matrix characterizing the baseline sound pressure level. This represents the difference matrix between the real-time sound pressure level representation and the baseline sound pressure level representation. Indicates the relationship with the first Spatial location of the virtual monitoring nodes The relevant Dirac function.

9. The method for detecting and locating steam leaks in a corn oil deodorization equipment as described in claim 1, characterized in that, Spatial clustering analysis is performed on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment, including: Based on the three-dimensional digital space structure, the residual energy of the sound field is correlated by coordinate binding to obtain the residual energy coordinate data of the virtual monitoring node; The residual energy coordinate data is mapped in three dimensions to obtain the spatial distribution of the residual energy of the deodorization device; Based on a preset residual energy threshold, the spatial distribution of the residual energy is regionalized by energy threshold screening to obtain the abnormal energy accumulation area of ​​the deodorization device; Based on the energy distribution gradient characteristics of the abnormal energy accumulation area, the energy core coordinates of the abnormal energy accumulation area are determined to obtain the suspected leakage points of the deodorization equipment. The leakage degree of the deodorization equipment is assessed based on the extreme energy intensity and spatial volume of the abnormal energy accumulation area. By integrating the suspected leak locations with the degree of leakage, steam leakage information of the deodorization equipment is obtained.

10. A steam leak detection and location system for corn oil deodorization equipment, characterized in that, The system for implementing the steam leak detection and location method for a corn oil deodorization equipment as described in claim 1 includes: The three-dimensional space and node construction module is used to construct the three-dimensional digital space structure of the deodorization equipment based on the spatial acoustic characteristics and structural information of the deodorization equipment, and to mark the virtual monitoring nodes associated with the deodorization equipment in the three-dimensional digital space structure. The steady-state sound field baseline calibration module is used to perform sound field calibration mapping on the steady-state sound field of the virtual monitoring node based on the steady-state background sound field data of the deodorization equipment during steady-state operation and the three-dimensional digital space structure, so as to obtain the baseline sound pressure characterization of the virtual monitoring node. The real-time sound field inverse reconstruction module is used to inversely reconstruct the real-time sound field data of the deodorization device based on the three-dimensional digital space structure, so as to obtain the real-time sound pressure characterization of the virtual monitoring node. The sound pressure temporal residual calculation module is used to perform temporal normalization and alignment of the real-time sound pressure representation and the baseline sound pressure representation, and to compare the energy difference of the alignment results to obtain the sound field residual energy of the virtual monitoring node. The residual clustering leakage identification module is used to perform spatial clustering analysis on the residual energy of the sound field to obtain the steam leakage information of the deodorization equipment.

Images