Automatic analysis system and method for abnormal state of pressure gauge

Abstract

The present application discloses a system and method for automatically analyzing the abnormal state of a pressure gauge. The system uses a collaborative model based on an autoencoder to learn the collaborative pattern between the real-time flow rate value sequence and the real-time pressure value sequence under normal working conditions, and infers the expected real-time pressure value sequence from the real-time flow rate value sequence. By calculating the residual between the actual real-time pressure value sequence and the inferred real-time pressure value sequence, the degree of deviation from the collaborative relationship between pressure and flow rate can be effectively quantified. The residual sequence and its time series characteristics, combined with the pressure time series characteristics of the target pressure gauge itself, are jointly input into the trained abnormal state diagnosis multi-classifier model, thereby achieving accurate and robust diagnosis of the abnormal state of the pressure gauge, and effectively distinguishing whether the pressure change is caused by the pressure gauge itself or the system process abnormality.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to a system and method for automatically analyzing abnormal conditions of a pressure gauge. Background Art

[0002] In modern industrial production and pipeline network operations, pressure gauges, as key process monitoring instruments, bear the important responsibility of ensuring the safe and stable operation of the system. Because abnormal pressure often indicates equipment failure, pipeline leakage, or other potential risks, timely detection and accurate diagnosis of abnormal pressure gauge conditions are crucial for preventing accidents, reducing losses, and improving operational efficiency. With the continuous advancement of automation and intelligence, relying on manual inspections or single threshold alarms can no longer meet the actual needs for efficient and accurate anomaly detection. Therefore, there is an urgent need to develop a new technical solution that can automatically analyze abnormal pressure gauge conditions.

[0003] Currently, most existing automatic analysis solutions for abnormal pressure gauge conditions are based on single parameter trend judgment, such as implementing simple alarms by setting upper and lower pressure thresholds. However, these existing methods generally have limitations. They often rely solely on single pressure data for analysis and fail to fully consider other relevant parameters in the system environment in which the pressure gauge is located, such as flow rate. This single-dimensional analysis method makes it difficult to effectively distinguish whether the pressure changes are caused by pressure gauge failures themselves (such as sensor drift and blockage) or system process abnormalities (such as pipeline leakage and pump failure) when faced with complex and changing industrial scenarios. This leads to insufficient accuracy, robustness and specificity in abnormal diagnosis, especially in distinguishing between sensor failures and process failures. It is prone to false alarms or missed alarms and cannot meet the needs of high-precision and high-reliability industrial diagnosis.

[0004] Therefore, an optimized automatic analysis solution for abnormal status of pressure gauges is expected. Summary of the Invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a system and method for automatic analysis of the abnormal state of a pressure gauge, which learns the collaborative pattern between the real-time flow rate value sequence and the real-time pressure value sequence under normal working conditions by utilizing a collaborative model based on an autoencoder, and infers the expected real-time pressure value sequence from the real-time flow rate value sequence. By calculating the residual between the actual real-time pressure value sequence and the inferred real-time pressure value sequence, the degree of deviation of the collaborative relationship between pressure and flow rate can be effectively quantified. The residual sequence and its time series characteristics, combined with the pressure time series characteristics of the target pressure gauge itself, are jointly input into the trained abnormal state diagnosis multi-classifier model, thereby achieving accurate and robust diagnosis of the abnormal state of the pressure gauge, and effectively distinguishing whether the pressure change is caused by the pressure gauge itself or the system process abnormality.

[0006] According to one aspect of the present application, a method for automatically analyzing an abnormal state of a pressure gauge is provided, comprising:

[0007] Obtain the real-time pressure value sequence of the target pressure gauge;

[0008] obtaining a sequence of real-time flow rate values ​​collected by a flow meter associated with a target pressure gauge;

[0009] Performing time stamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence;

[0010] Extract individual signal features from the aligned real-time pressure value sequence to obtain the target pressure gauge pressure time series feature vector;

[0011] Input the aligned real-time flow velocity value sequence into the autoencoder-based collaborative model to obtain the inferred real-time pressure value sequence;

[0012] The pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence is calculated, and the pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature are input into the trained abnormal state diagnosis multi-classifier model to obtain the abnormal diagnosis result.

[0013] According to another aspect of the present application, a pressure gauge abnormal state automatic analysis system is provided, comprising:

[0014] A pressure value acquisition module is used to obtain the real-time pressure value sequence of the target pressure gauge;

[0015] a flow rate value acquisition module, for acquiring a real-time flow rate value sequence collected by a flow meter associated with a target pressure gauge;

[0016] A timestamp alignment module is used to perform timestamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence;

[0017] An individual signal feature extraction module is used to extract individual signal features from the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector;

[0018] A pressure value inference module is used to input the aligned real-time flow velocity value sequence into the autoencoder-based collaborative model to obtain an inferred real-time pressure value sequence;

[0019] The abnormal state diagnosis module is used to calculate the pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, and input the pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature into the trained abnormal state diagnosis multi-classifier model to obtain the abnormal diagnosis result.

[0020] Compared with the prior art, the present application provides a system and method for automatically analyzing the abnormal state of a pressure gauge. By utilizing a collaborative model based on an autoencoder, it learns the collaborative pattern between the real-time flow rate value sequence and the real-time pressure value sequence under normal working conditions, and based on this, infers the expected real-time pressure value sequence from the real-time flow rate value sequence. By calculating the residual between the actual real-time pressure value sequence and the inferred real-time pressure value sequence, the degree of deviation from the collaborative relationship between pressure and flow rate can be effectively quantified. The residual sequence and its time series characteristics, combined with the pressure time series characteristics of the target pressure gauge itself, are jointly input into the trained abnormal state diagnosis multi-classifier model, thereby achieving accurate and robust diagnosis of the abnormal state of the pressure gauge, and effectively distinguishing whether the pressure change is caused by the pressure gauge itself or the system process abnormality. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0022] Figure 1 Flowchart of a method for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application;

[0023] Figure 2 Schematic diagram of data flow of the method for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application;

[0024] Figure 3 Flowchart of sub-step S6 of the method for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application;

[0025] Figure 4 4 is a block diagram of a system for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0027] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0028] Although the present application makes various references to certain modules in the system according to embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0029] Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0030] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0031] In the technical solution of the present application, a method for automatically analyzing abnormal conditions of a pressure gauge is proposed. Figure 1 Flowchart of a method for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the method for automatically analyzing abnormal state of a pressure gauge according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the method for automatically analyzing the abnormal state of a pressure gauge according to an embodiment of the present application includes the following steps: S1, obtaining a real-time pressure value sequence of a target pressure gauge; S2, obtaining a real-time flow rate value sequence collected by a flow meter associated with the target pressure gauge; S3, performing timestamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence; S4, performing individual signal feature extraction on the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector; S5, inputting the aligned real-time flow rate value sequence into a collaborative model based on an autoencoder to obtain an inferred real-time pressure value sequence; S6, calculating the pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, and inputting the pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature into the trained abnormal state diagnosis multi-classifier model to obtain an abnormal diagnosis result.

[0032] In particular, the S1 obtains the real-time pressure value sequence of the target pressure gauge; and obtains the real-time flow rate value sequence collected by the flow meter associated with the target pressure gauge. It should be understood that the data of a single parameter is often difficult to accurately reflect the actual state of equipment operation under complex working conditions, and multi-source data fusion can reveal a deeper synergistic relationship between parameters. For example, under normal circumstances, there should be a certain physical correlation between flow rate and pressure. When the pipeline leaks, is blocked, or the instrument itself fails, this correlation will be broken, causing the data between the two to show abnormal characteristics. Therefore, by simultaneously acquiring and aligning these two types of time series data, a complete and accurate information basis can be provided for subsequent model analysis.

[0033] In particular, in step S3, the real-time pressure value sequence and the real-time flow rate value sequence are timestamp aligned to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence. Due to the influence of factors such as sampling frequency, communication delay, and device clock error in actual operating environments, the pressure value sequence and the flow rate value sequence are often time-asynchronous. If the original data is directly input into the subsequent analysis model, the correlation between parameters is easily weakened or even lost, thereby affecting the accuracy of anomaly detection. Therefore, in the technical solution of the present application, the real-time pressure value sequence and the real-time flow rate value sequence are timestamp aligned to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence. This not only helps the model accurately explore the dynamic correlation between parameters, but also provides a prerequisite for intelligent algorithms such as residual analysis and autoencoder inference. Through the aligned data, complex anomalies that are difficult to detect with a single variable can be more effectively identified, achieving efficient fault diagnosis driven by multi-parameter collaboration. In a specific example of the present application, a target timeline is constructed with a step size of seconds. For each target moment, the data point closest to that moment is found from the original sequence and numerical completion is performed. For data with significant missing data or extremely uneven sampling, smoothing can be performed using methods such as sliding window weighted averaging. This step significantly improves the sensitivity and robustness of the anomaly detection system to dynamic changes in the process, laying a solid foundation for intelligent and automated operation and maintenance management.

[0034] In particular, S4 extracts individual signal features from the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector. It should be understood that the pressure data collected by the pressure gauge is often affected by a variety of operating conditions and exhibits complex time series dynamic characteristics. Therefore, in order to achieve efficient and accurate identification of abnormal conditions of the pressure gauge, in the technical solution of the present application, after completing the timestamp alignment of the real-time pressure value sequence and the flow rate value sequence, individual signal features are further extracted from the aligned real-time pressure value sequence to fully explore and express the change patterns, fluctuation patterns, and potential abnormal trends of the target pressure gauge itself within the historical period, thereby obtaining a target pressure gauge pressure time series feature vector. The target pressure gauge pressure time series feature vector not only contains the trend information of the pressure signal changing over time, but also captures key dynamic attributes such as periodic fluctuations, mutation points, and abnormal persistence, thereby effectively improving the subsequent multi-classifier model's ability to distinguish different types of abnormal conditions (such as leakage, blockage, or instrument failure). In a specific example of this application, LSTM-based individual signal feature extraction is performed on the aligned real-time pressure value sequence to obtain the target pressure gauge pressure time series feature vector. Here, LSTM, as a recurrent neural network specifically designed to handle time series data dependencies, can effectively model the complex relationship between the current pressure value and historical multi-step data. By extracting individual signal features from the aligned real-time pressure value sequence based on the LSTM model, not only is the information expression capability of the original data greatly enriched, but it also provides solid data support for the entire intelligent diagnostic system.

[0035] In particular, in S5, the aligned real-time flow rate value sequence is input into a collaborative model based on an autoencoder to obtain an inferred real-time pressure value sequence. It should be understood that there is a close physical coupling relationship between pressure and flow rate, and traditional methods often ignore the complex and implicit coupling mechanism between different parameters. When the system is in a normal state, flow rate changes usually affect pressure changes in a certain pattern, but when abnormalities such as leakage, blockage or instrument failure occur, this pattern will be broken. Therefore, in the technical solution of the present application, the aligned real-time flow rate value sequence is input into a collaborative model based on an autoencoder to obtain an inferred real-time pressure value sequence. Here, the autoencoder is an unsupervised learning neural network structure, which is mainly used to automatically learn effective features in the input data. It consists of two parts: an encoder and a decoder: the encoder compresses the input signal into a low-dimensional latent space representation, and the decoder attempts to reconstruct the original input from the latent representation. In the specific implementation process, the autoencoder is used to capture the complex mapping collaborative relationship between flow rate and pressure, and the decoding end outputs the theoretical pressure value sequence inferred at the corresponding moment.

[0036] In particular, the step S6 calculates the pressure-flow rate synergistic deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, and inputs the pressure-flow rate synergistic deviation feature and the target pressure gauge pressure time series feature into the trained abnormal state diagnosis multi-classifier model to obtain an abnormal diagnosis result. In a specific example of the present application, Figure 3 As shown, the S6 includes: S61, calculating the residual between the inferred real-time pressure value sequence and the real-time pressure value sequence to obtain a real-time pressure value residual sequence; S62, performing residual time series feature extraction on the real-time pressure value residual sequence to obtain a pressure-flow rate collaborative deviation feature vector as a pressure-flow rate collaborative deviation feature; S63, performing multi-scale joint progressive perception on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain a target pressure gauge main behavior-collaborative behavior time series joint coding vector; S64, obtaining an abnormal diagnosis result based on the target pressure gauge main behavior-collaborative behavior time series joint coding vector.

[0037] Specifically, the S61 calculates the residual between the inferred real-time pressure value sequence and the real-time pressure value sequence to obtain a real-time pressure value residual sequence. It should be understood that residual analysis can effectively capture those complex anomalies that are difficult to identify or easily overlooked by relying solely on a single parameter trend. Even if certain anomalies have not yet caused absolute numerical values ​​to exceed the limit, as long as they destroy the normal synergistic relationship between flow rate and pressure, they will appear as significant fluctuations or continuous deviations in the residuals. For example, when a small leak occurs in the pipeline, the actual pressure will be lower than the model inferred value, and instrument failure may cause a continuous deviation between the two. Therefore, in the technical solution of the present application, the inferred real-time pressure value sequence is compared with the actual observed real-time pressure value sequence, and the residual between the two is calculated to effectively capture those complex anomalies that are difficult to identify by relying solely on a single parameter trend.

[0038] In specific implementation, for each moment, the actual collected pressure value is subtracted from the theoretical pressure value obtained by the autoencoder model based on flow rate prediction to obtain the residual at that moment; similarly, after traversing each time window, a real-time pressure value residual sequence is obtained. Among them, the real-time pressure value residual sequence not only intuitively reflects the operating status and health level of the system at each stage of the process, but also provides a rich information basis for subsequent feature extraction and multi-classification diagnosis. By analyzing this residual sequence, not only can various small but critical data deviations be keenly captured, and higher precision and stronger robustness of automated anomaly detection effects be achieved, but also the traditional monitoring methods can be greatly improved for fine-grained anomaly recognition capabilities under complex working conditions, providing solid data support and technical guarantees for intelligent operation and maintenance and risk prevention and control in industrial sites.

[0039] Specifically, the S62 performs residual time series feature extraction on the residual sequence of real-time pressure values ​​to obtain a pressure-flow rate coordination deviation feature vector as a pressure-flow rate coordination deviation feature. It should be understood that a simple residual value can only reveal the deviation at a certain moment, but cannot fully reflect the evolution law and persistence of the abnormal state in the time dimension. Industrial working conditions are often complex and changeable, and some abnormalities (such as leakage, blockage or sensor drift) have obvious time series characteristics and dynamic evolution processes. Therefore, in the technical solution of the present application, residual time series feature extraction is performed on the residual sequence of real-time pressure values ​​to convert the original high-dimensional, redundant and noisy residual signal into a highly generalized low-dimensional vector, so that it can fully express the key information such as the dynamic change trend, periodic fluctuations, mutation points and continuous offsets of the flow rate-pressure coordination disorder during the operation of the system, and obtain the pressure-flow rate coordination deviation feature vector.

[0040] In a specific example of this application, a gated recurrent unit (GRU) network can be used as the primary residual time series feature extraction model. GRU is a recurrent neural network structure that efficiently handles long-term dependencies and effectively suppresses the gradient vanishing problem. It is well suited for modeling nonlinear dynamic processes in industrial data. In this process, the fully aligned real-time pressure value residual sequence is input into the GRU network. The network automatically learns and filters important historical information related to anomalies through a gating mechanism, gradually encodes the input signal, and generates a fixed-length, highly expressive, and discriminative pressure-flow rate collaborative deviation feature vector at the output.

[0041] Specifically, S63 performs multi-scale joint progressive perception on the pressure-flow velocity collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain a joint encoding vector for the target pressure gauge's main behavior and collaborative behavior time series. It should be understood that the dynamic collaborative relationship between pressure and flow velocity exhibits complex coupling characteristics at multiple levels and granularities. For example, when a pipeline leak occurs, the linkage between pressure and flow velocity may manifest as a combination of short-term, sudden local anomalies (e.g., a sudden drop in pressure accompanied by a sharp increase in flow velocity) and long-term, trend-dependent deviations (e.g., pressure continuously falling below the normal range). Conversely, when a pressure gauge experiences a drift fault, its pressure time series may exhibit a steady-state offset, while the collaborative deviation with flow velocity remains normal. Traditional single-layer fusion methods are unable to distinguish these abnormal patterns at different abstraction levels. Therefore, in the technical solution of the present application, multi-scale joint progressive perception is performed on the pressure-flow velocity collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain a joint encoding vector for the target pressure gauge's main behavior and collaborative behavior time series. Specifically, by hierarchically decoupling the fine-grained differences between the pressure gauge's primary behavior (its own temporal characteristics) and its coordinated behavior (the dynamic relationship between pressure and flow rate), the essential characteristics of the fault can be more accurately captured. Specifically, through deep nonlinear transformations and cross-level feature interactions, a joint semantic space is constructed that reflects both the pressure gauge's own operating state (such as individual behaviors like periodic fluctuations and trend drift) and its dynamic coordinated relationship with flow rate parameters (such as the breakdown of pressure-flow rate correlation due to leakage and the delay in coordinated response due to blockage). For example, in low-level feature fusion, the system may capture the correspondence between instantaneous pressure jitter and high-frequency flow rate noise to eliminate environmental interference. In mid-level fusion, the coordinated response patterns of pressure and flow rate at specific process stages (such as pump start-up and shutdown, valve adjustment) can be identified to determine whether there are any coordinated anomalies. In deep-level fusion, global reasoning at the semantic level (such as the contradiction between long-term pressure trends and accumulated flow rates) is used to identify instrument failures or systemic faults. This hierarchical and progressive perception mechanism enables the system to provide comprehensive coverage from local details of the time series to global semantics, avoiding pattern omissions caused by single-scale analysis, thereby achieving high-precision fault type discrimination in the classifier (such as distinguishing between leaks, blockages, or instrument failures), and reducing the risk of misjudgment caused by single-dimensional analysis or simple feature splicing in traditional methods.

[0042] Specifically, first, multi-level implicit feature extraction is performed on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, the target pressure gauge pressure middle-layer time series implicit feature coding vector, the pressure-flow rate collaborative deviation deep-layer implicit feature coding vector, and the target pressure gauge pressure deep-layer time series implicit feature coding vector. It should be understood that there are multi-dimensional abnormal representations between the dynamic relationship between pressure and flow rate, and traditional single-layer feature extraction cannot capture such cross-level contradictory signals. Therefore, in the technical solution of the present application, multi-level implicit feature extraction is performed on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, the target pressure gauge pressure middle-layer time series implicit feature coding vector, the pressure-flow rate collaborative deviation deep-layer implicit feature coding vector, and the target pressure gauge pressure deep-layer time series implicit feature coding vector.

[0043] Specifically, the mid-level implicit feature encoding vector focuses on the structural characteristics of coordinated deviations between pressure and flow velocity (such as local waveform distortion and phase lag within a specific time window), while the deep-level implicit feature encoding vector is dedicated to capturing the global semantics of the individual and coordinated behaviors of pressure gauges (such as the contradiction between the long-term deviation of pressure from the normal range and the accumulated flow velocity). This layered abstraction mechanism enables the system to focus on both local details (such as the matching degree between a specific pressure pulse and the flow velocity response) and overall trends (such as the relationship between the mean pressure shift and the flow velocity distribution). Ultimately, it forms a multi-scale fault recognition framework, significantly improving the separability of abnormal patterns.

[0044] In a specific example of the present application, multi-level implicit feature extraction is performed on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector using the following formula to obtain a pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, a target pressure gauge pressure middle-layer time series implicit feature coding vector, a pressure-flow rate collaborative deviation deep-layer implicit feature coding vector, and a target pressure gauge pressure deep-layer time series implicit feature coding vector; wherein, the formula is:

[0045]

[0046] in, is the pressure-flow rate cooperative deviation eigenvector, is the target pressure gauge pressure time series feature vector, represents the weight matrix of mid-level feature extraction, represents the bias term for mid-level feature extraction, express function, is the implicit feature encoding vector of the pressure-velocity coordinated deviation, is the target pressure gauge pressure mid-layer temporal implicit feature coding vector, represents the weight matrix for deep feature extraction, represents the bias term for deep feature extraction, is the deep implicit feature encoding vector of the pressure-flow rate collaborative deviation, is the deep temporal implicit feature encoding vector of the target pressure gauge pressure.

[0047] Next, low-level feature fusion is performed on the pressure-flow rate coordination deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the target pressure gauge main behavior-cooperative behavior low-level time series fusion feature encoding vector. It should be understood that the time series data collected by pressure gauges and flow meters are highly dynamically coupled. For example, when a small crack causes an initial leak in a pipeline network, the coordinated deviation of pressure and flow rate may manifest as a misalignment of the precise timestamps of the sudden pressure drop and the sudden flow rate increase on the millisecond time scale. Directly interacting with mid- and high-level abstract features can easily lose the fine temporal correlation of the original time series signals due to the layer-by-layer abstraction of deep neural networks. However, by directly operating on the fine structure of the original time series data, low-level feature fusion can preserve the millisecond-level synchronization characteristics of the pressure pulse waveform and flow rate response, providing a physically clear data foundation for identifying such transient anomalies. By building this raw signal-level fusion foundation, the system not only enhances the early identification of complex anomalies such as micro-leaks and intermittent blockages, but also ensures the interpretability of the physical mechanisms during the multi-scale feature evolution.

[0048] In a specific example of the present application, the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector are subjected to low-level feature fusion using the following formula to obtain the target pressure gauge main behavior-collaborative behavior low-level time series fusion feature encoding vector; wherein, the formula is:

[0049]

[0050] in, It means adding by position. represents a multilayer perceptron, A low-level temporal fusion feature encoding vector of the target pressure gauge main behavior-cooperative behavior is obtained.

[0051] Next, a mid-level feature fusion is performed on the mid-level implicit feature encoding vector of the pressure-flow rate collaborative deviation and the mid-level temporal implicit feature encoding vector of the target pressure gauge pressure to obtain the mid-level temporal fusion feature encoding vector of the target pressure gauge's main behavior-collaborative behavior. It should be understood that when a regulating valve in a pipe network section becomes stuck, the mid-level temporal features of the pressure gauge may manifest as a phase shift of periodic pressure fluctuations, while the corresponding mid-level features of the flow rate collaborative deviation may capture the step-like mutation of the flow rate regulation curve. Such cross-modal structural anomalies may be masked by noise at the raw signal level, but after mid-level feature extraction, the component-level behavioral patterns of both (such as the correlation between valve opening and flow resistance) are made explicit.

[0052] Compared to low-level fusion, which focuses on transient signal correspondences, mid-level fusion focuses on structured pattern matching within the equipment's operating cycle. For example, in the case of early-stage centrifugal pump bearing wear, the mid-level pressure gauge feature might extract harmonic distortion of the impeller's pass frequency, while the mid-level flow velocity synergy deviation feature might identify abnormal modulation of the flow pulsation amplitude. Mid-level feature fusion can establish a causal relationship model between the two within the vibration transmission path of the rotating machinery. This fusion mechanism enables the system to transcend the limitations of a single sensor perspective. By analyzing the physical interactions between equipment components, it integrates the response patterns of pressure gauge measurements and flow velocity parameters into a unified mechanical system model, thereby distinguishing transient anomalies caused by external disturbances from structural failures due to actual equipment degradation. This structured feature fusion, based on equipment operating mechanisms, not only enhances the system's ability to analyze multiple concurrent fault scenarios, but more importantly, by capturing the stable correlation patterns between pressure and flow velocity at the component level, it significantly reduces the probability of misjudgment caused by random noise or transient disturbances, providing a more interpretable diagnostic basis for operational decisions.

[0053] In a specific example of the present application, the mid-level feature fusion is performed on the pressure-flow rate collaborative deviation mid-level implicit feature coding vector and the target pressure gauge pressure mid-level temporal implicit feature coding vector using the following formula to obtain the target pressure gauge main behavior-collaborative behavior mid-level temporal fusion feature coding vector; wherein, the formula is:

[0054]

[0055] in, Indicates point multiplication by position, represents the attention mechanism, express function, represents the characteristic dimension of the vector, A hierarchical temporal fusion feature encoding vector for the target pressure gauge main behavior-cooperative behavior.

[0056] Then, deep feature fusion is performed on the deep implicit feature encoding vector of the pressure-flow rate synergistic deviation and the deep temporal implicit feature encoding vector of the target pressure gauge pressure to obtain the deep-level temporal fusion feature encoding vector of the target pressure gauge's main behavior and synergistic behavior. It is understandable that when a pipeline system experiences structural aging or regional corrosion, the pressure gauge's deep temporal features may implicitly encode the entropy increase of long-term pressure decay, while the deep features of the synergistic deviation may reflect systematic deviations from the pressure-flow rate energy conservation relationship. Such global anomalies are difficult to detect in local time series segments, but through deep feature extraction, the semantic-level representations of both (such as degradation of system energy transfer efficiency and qualitative changes in fluid dynamics) are made explicit. Here, deep fusion combines the long-term evolution of individual pressure gauge behaviors (such as the cumulative effect of zero-point drift) with the essential changes in the pressure-flow rate synergistic relationship (such as the anomaly of the friction coefficient in Darcy's equation) within a unified system-level health assessment framework, thereby constructing a causal relationship model between equipment failure and fluid state changes at a higher level of abstraction.

[0057] Compared to the component-level interaction patterns focused on by mid-level fusion, deep feature fusion focuses on the global operational nature of industrial systems and the collaborative mechanisms constrained by physical laws. For example, in a scenario of overall elastic degradation of a pipeline network, deep pressure features may capture changes in pressure wave propagation characteristics caused by the decay of Young's modulus, while deep collaborative deviation features may reflect anomalies in the dependence of the velocity-pressure drop relationship on pipeline stiffness. Deep fusion can establish cross-scale correlations between material performance degradation and fluid dynamic response. This fusion mechanism enables the system to transcend the reliance on explicit parameters of traditional monitoring methods, linking pressure gauge measurement anomalies with deep semantic changes in velocity parameters as system-level state transitions, thereby identifying progressive failure modes that are invisible to traditional methods. This fusion approach, based on the inherent laws of the system, not only enhances the interpretability of diagnostic results but, more importantly, by capturing the deep coupling relationship between pressure and velocity at the level of physical constraints, establishes an intelligent judgment benchmark that transcends empirical thresholds, providing a new technical path for resilience assessment and life prediction of industrial systems.

[0058] In a specific example of the present application, the deep implicit feature coding vector of the pressure-flow rate collaborative deviation and the deep temporal implicit feature coding vector of the target pressure gauge pressure are deeply fused using the following formula to obtain the target pressure gauge main behavior-collaborative behavior deep-level temporal fusion feature coding vector; wherein, the formula is:

[0059]

[0060] in, represents the weight matrix of deep fusion, Representation layer normalization, express function, and is the low-rank projection matrix, A deep-level temporal fusion feature encoding vector of the target pressure gauge main behavior-cooperative behavior is provided.

[0061] Furthermore, the low-level temporal fusion feature encoding vectors of the target pressure gauge's main behavior and collaborative behavior, the mid-level temporal fusion feature encoding vectors of the target pressure gauge's main behavior and collaborative behavior, and the deep-level temporal fusion feature encoding vectors of the target pressure gauge's main behavior and collaborative behavior are subjected to multi-scale progressive complementary perception fusion to obtain the target pressure gauge's main behavior and collaborative behavior temporal joint encoding vector. It should be understood that the low-level temporal fusion feature encoding vector carries the millisecond-level correlation details of the original pressure fluctuation and the flow rate coordinated deviation, while the mid-level feature encoding vector contains the system behavior pattern after preliminary abstraction. When performing cross-level interactive fusion, direct feature splicing or weighted superposition can easily lead to phase conflicts between hierarchical features. For example, the abstracted pressure trend feature at the mid-level may obscure the precise time-locked relationship between the sudden pressure pulse and flow rate oscillation recorded in the low-level features. Therefore, in a preferred example of the present application, first, the target pressure gauge main behavior-collaborative behavior low-level temporal fusion feature coding vector and the target pressure gauge main behavior-collaborative behavior middle-level temporal fusion feature coding vector are subjected to cross-level preliminary interactive fusion based on order parameter phase complementarity to obtain the target pressure gauge main behavior-collaborative behavior cross-level preliminary fusion semantic feature coding vector. That is, by establishing a dynamic phase compensation channel between the high-frequency details of the low-level features and the macro trends of the middle-level features, key cross-level associations such as the flow rate response delay corresponding to the pressure drop can maintain spatiotemporal consistency. In the oil pipeline leakage detection scenario, the local pressure drop recorded by the low-level features and the abnormal pressure gradient of the entire pipeline network revealed by the middle-level features form a synergistically enhanced signal expression through the phase-complementary order parameter field constraints, which can effectively suppress the interference of transient noise caused by the start-stop operation of pumps and valves on feature fusion. This cross-level preliminary semantic fusion not only provides fusion basis vectors with physical field synchronization characteristics for subsequent deep feature interaction, but also maintains the energy conservation characteristics in the process of multi-scale feature evolution through the constraints of path integral optimization, ensuring the integrity of dynamic correlation of leakage signals during the propagation process at different abstract levels, and laying the foundation for multi-scale joint perception for accurately distinguishing complex abnormal patterns such as instrument drift and pipe wall microcracks.

[0062] Furthermore, it should be understood that the cross-level preliminary fusion semantic feature coding vector carries the system behavior pattern formed by the interaction of low- and medium-level features, while the deep-level temporal fusion feature coding vector contains the dynamic evolution law of the system after high abstraction. When performing cross-level semantic interaction analysis, direct feature association is prone to phase conflict due to the transition of the degree of abstraction between levels. For example, the overall pressure attenuation trend of the pipeline network represented in the deep features may be misaligned with the local pressure pulse response recorded in the cross-level preliminary fusion features. Therefore, in the preferred example of the present application, the target pressure gauge main behavior-collaborative behavior cross-level preliminary fusion semantic feature coding vector and the target pressure gauge main behavior-collaborative behavior deep-level temporal fusion feature coding vector are further subjected to target pressure gauge main behavior-collaborative behavior temporal semantic interaction analysis based on order parameter field constraints and path integral optimization to obtain the target pressure gauge main behavior-collaborative behavior temporal joint coding vector. That is, by constructing a weight matrix field with self-organizing characteristics, the energy transfer path of pressure wave propagation and flow velocity response at different time and space scales is simulated. The interactive analysis based on order parameter field constraints and path integral optimization significantly improves the system's ability to decode hidden correlated faults. When the dynamic balance of multi-level features under physical constraints is accurately modeled, the system can extract systematically explanatory diagnostic evidence from seemingly discrete abnormal signals. For example, in the case of a compound fault where the pressure gauge's pressure pipe is blocked and the pipe network topology changes, path integral optimization identifies the coupling effect of the sudden increase in local flow resistance and the distortion of the global pressure distribution by analyzing the curvature change of the characteristic manifold. The order parameter field constraint ensures that this analysis conforms to the flow velocity-pressure drop relationship of Darcy's law, thereby effectively distinguishing between real blockage and sensor drift.

[0063] In this example, the target pressure gauge main behavior-cooperative behavior low-level temporal fusion feature coding vector, the target pressure gauge main behavior-cooperative behavior middle-level temporal fusion feature coding vector, and the target pressure gauge main behavior-cooperative behavior deep-level temporal fusion feature coding vector are subjected to multi-scale progressive complementary perception fusion of the target pressure gauge main behavior-cooperative behavior to obtain the target pressure gauge main behavior-cooperative behavior temporal joint coding vector; wherein, the formula is:

[0064]

[0065] in, represents the gating weight matrix, express function, represents the gating order parameter, A cross-level preliminary fusion semantic feature encoding vector for the target pressure gauge main behavior-cooperative behavior, represents the query weight matrix, represents the key weight matrix, represents the value weight matrix, represents the nuclear norm of the matrix, that is, the sum of the eigenvalues ​​of the matrix, To scale hyperparameters, is the field uniform alignment loss function, Indicates a query, represents the key vector, represents a value vector, It is the joint encoding vector of the target pressure gauge main behavior-cooperative behavior time series.

[0066] Specifically, the S64 obtains an abnormality diagnosis result based on the target pressure gauge main behavior-cooperative behavior time series joint coding vector. That is, in the technical solution of the present application, the target pressure gauge main behavior-cooperative behavior time series joint coding vector is input into the trained abnormal state diagnosis multi-classifier model to obtain the abnormality diagnosis result. It should be understood that the abnormal state of the pressure gauge may be caused by a variety of complex factors, including equipment aging, pipeline leakage, blockage or sensor failure. It is difficult to achieve accurate identification of multiple types of abnormalities by relying solely on a single parameter or simple rule. Therefore, in order to achieve efficient differentiation and accurate positioning of abnormal states of different types of pressure gauges (such as leakage, blockage, sensor drift, etc.), thereby improving the overall intelligent level of operation and maintenance and reducing the false alarm rate and the risk of missed detection, in the technical solution of the present application, the deep temporal features reflecting the main behavior and cooperative behavior during the system operation are jointly encoded and further input into a multi-classifier model with strong discrimination ability for intelligent diagnosis.

[0067] Multi-classifier models typically employ classic machine learning or deep learning structures such as softmax regression, multi-layer perceptrons (MLPs), support vector machines (SVMs), or ensemble learning methods. Their core working principle is to first model the characteristic distributions corresponding to various abnormal conditions using labeled historical sample data. Through repeated training and parameter optimization, the model automatically captures the implicit but critical distinction boundaries between different categories. Next, during the inference phase, the target pressure gauge's main behavior and collaborative behavior temporal joint encoding vector, obtained through multi-scale fusion at the current moment, is input. Based on the learned knowledge, the model makes a probabilistic prediction of the category to which it belongs and outputs the most likely corresponding abnormality type or types, enabling one-click automatic diagnosis of equipment health under complex operating conditions. This approach not only greatly improves diagnostic efficiency and accuracy, but also provides solid technical support for real-time, intelligent operation and maintenance management in industrial sites.

[0068] In summary, the method for automatic analysis of the abnormal state of a pressure gauge according to the embodiment of the present application is explained, which learns the collaborative pattern between the real-time flow rate value sequence and the real-time pressure value sequence under normal working conditions by utilizing a collaborative model based on an autoencoder, and infers the expected real-time pressure value sequence from the real-time flow rate value sequence. By calculating the residual between the actual real-time pressure value sequence and the inferred real-time pressure value sequence, the degree of deviation from the collaborative relationship between pressure and flow rate can be effectively quantified. The residual sequence and its time series characteristics, combined with the pressure time series characteristics of the target pressure gauge itself, are jointly input into the trained abnormal state diagnosis multi-classifier model, thereby achieving accurate and robust diagnosis of the abnormal state of the pressure gauge, and effectively distinguishing whether the pressure change is caused by the pressure gauge itself or the system process abnormality.

[0069] Furthermore, a pressure gauge abnormal state automatic analysis system is also provided.

[0070] Figure 4 FIG. 1 is a block diagram of a system for automatically analyzing abnormal conditions of a pressure gauge according to an embodiment of the present application. Figure 4 As shown, the pressure gauge abnormal state automatic analysis system 300 according to the embodiment of the present application includes: a pressure value acquisition module 310, which is used to obtain a real-time pressure value sequence of a target pressure gauge; a flow rate value acquisition module 320, which is used to obtain a real-time flow rate value sequence collected by a flow meter associated with the target pressure gauge; a timestamp alignment module 330, which is used to perform timestamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence; an individual signal feature extraction module 340, which is used to perform individual signal feature extraction on the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector; a pressure value inference module 350, which is used to input the aligned real-time flow rate value sequence into a collaborative model based on an autoencoder to obtain an inferred real-time pressure value sequence; an abnormal state diagnosis module 360, which is used to calculate the pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, and input the pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature into the trained abnormal state diagnosis multi-classifier model to obtain an abnormal diagnosis result.

[0071] As described above, the pressure gauge abnormal state automatic analysis system 300 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server equipped with an automatic pressure gauge abnormal state analysis algorithm. In one possible implementation, the pressure gauge abnormal state automatic analysis system 300 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the pressure gauge abnormal state automatic analysis system 300 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the pressure gauge abnormal state automatic analysis system 300 can also be one of the many hardware modules of the wireless terminal.

[0072] Alternatively, in another example, the pressure gauge abnormal state automatic analysis system 300 and the wireless terminal may also be separate devices, and the pressure gauge abnormal state automatic analysis system 300 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0073] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for automatically analyzing abnormal conditions of a pressure gauge, characterized in that: include: Obtain the real-time pressure value sequence of the target pressure gauge; obtaining a sequence of real-time flow rate values ​​collected by a flow meter associated with a target pressure gauge; Performing time stamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence; Extracting individual signal features from the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector, including: extracting individual signal features from the aligned real-time pressure value sequence based on LSTM to obtain a target pressure gauge pressure time series feature vector; Input the aligned real-time flow velocity value sequence into the autoencoder-based collaborative model to obtain the inferred real-time pressure value sequence; Calculating a pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, including: calculating a residual between the inferred real-time pressure value sequence and the real-time pressure value sequence to obtain a real-time pressure value residual sequence; performing residual time series feature extraction on the real-time pressure value residual sequence to obtain a pressure-flow rate collaborative deviation feature vector as the pressure-flow rate collaborative deviation feature; The pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature are input into the trained abnormal state diagnosis multi-classifier model to obtain abnormal diagnosis results, including: The pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector are subjected to multi-scale joint progressive perception to obtain the target pressure gauge main behavior-collaborative behavior time series joint coding vector, including: performing deep nonlinear transformation on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector respectively to obtain the pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, the target pressure gauge pressure middle-layer time series implicit feature coding vector, the pressure-flow rate collaborative deviation deep-layer implicit feature coding vector and the target pressure gauge pressure deep-layer time series implicit feature coding vector; performing multi-level cross-modal time series on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector The target pressure gauge main behavior-cooperative behavior low-level temporal fusion feature coding vector, the target pressure gauge main behavior-cooperative behavior middle-level temporal fusion feature coding vector and the target pressure gauge main behavior-cooperative behavior deep-level temporal fusion feature coding vector are sequentially fused to obtain the target pressure gauge main behavior-cooperative behavior low-level temporal fusion feature coding vector, the target pressure gauge main behavior-cooperative behavior middle-level temporal fusion feature coding vector and the target pressure gauge main behavior-cooperative behavior deep-level temporal fusion feature coding vector; the target pressure gauge main behavior-cooperative behavior low-level temporal fusion feature coding vector, the target pressure gauge main behavior-cooperative behavior middle-level temporal fusion feature coding vector and the target pressure gauge main behavior-cooperative behavior deep-level temporal fusion feature coding vector are subjected to multi-scale progressive complementary perception fusion of the target pressure gauge main behavior-cooperative behavior to obtain the target pressure gauge main behavior-cooperative behavior temporal joint coding vector; Based on the joint encoding vector of the target pressure gauge's main behavior and collaborative behavior time series, the abnormal diagnosis result is obtained.

2. The method for automatically analyzing abnormal conditions of a pressure gauge according to claim 1, characterized in that: The pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector are respectively subjected to deep nonlinear transformation to obtain the pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, the target pressure gauge pressure middle-layer time series implicit feature coding vector, the pressure-flow rate collaborative deviation deep-layer implicit feature coding vector and the target pressure gauge pressure deep-layer time series implicit feature coding vector, including: Multi-level implicit feature extraction is performed on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the pressure-flow rate collaborative deviation middle-layer implicit feature coding vector, the target pressure gauge pressure middle-layer time series implicit feature coding vector, the pressure-flow rate collaborative deviation deep-layer implicit feature coding vector and the target pressure gauge pressure deep-layer time series implicit feature coding vector.

3. The method for automatically analyzing abnormal conditions of a pressure gauge according to claim 2, characterized in that: Multi-level cross-modal temporal fusion is performed on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the target pressure gauge main behavior-collaborative behavior low-level temporal fusion feature encoding vector, the target pressure gauge main behavior-collaborative behavior middle-level temporal fusion feature encoding vector, and the target pressure gauge main behavior-collaborative behavior deep-level temporal fusion feature encoding vector, including: Perform low-level feature fusion on the pressure-flow rate collaborative deviation feature vector and the target pressure gauge pressure time series feature vector to obtain the target pressure gauge main behavior-collaborative behavior low-level time series fusion feature encoding vector; Perform mid-level feature fusion on the mid-level implicit feature coding vector of the pressure-flow rate collaborative deviation and the mid-level temporal implicit feature coding vector of the target pressure gauge pressure to obtain the mid-level temporal fusion feature coding vector of the target pressure gauge main behavior-collaborative behavior; The deep implicit feature coding vector of the pressure-flow rate collaborative deviation and the deep temporal implicit feature coding vector of the target pressure gauge pressure are deeply fused to obtain the deep-level temporal fusion feature coding vector of the target pressure gauge main behavior-collaborative behavior.

4. The method for automatically analyzing abnormal conditions of a pressure gauge according to claim 3, characterized in that: The target pressure gauge main behavior-cooperative behavior low-level temporal fusion feature encoding vector, the target pressure gauge main behavior-cooperative behavior middle-level temporal fusion feature encoding vector and the target pressure gauge main behavior-cooperative behavior deep-level temporal fusion feature encoding vector are subjected to multi-scale progressive complementary perception fusion of the target pressure gauge main behavior-cooperative behavior to obtain the target pressure gauge main behavior-cooperative behavior temporal joint encoding vector, including: The low-level temporal fusion feature coding vector of the target pressure gauge main behavior-cooperative behavior and the mid-level temporal fusion feature coding vector of the target pressure gauge main behavior-cooperative behavior are subjected to cross-level preliminary interactive fusion based on the order parameter phase complementarity to obtain the cross-level preliminary fusion semantic feature coding vector of the target pressure gauge main behavior-cooperative behavior; The target pressure gauge main behavior-collaborative behavior cross-level preliminary fusion semantic feature coding vector and the target pressure gauge main behavior-collaborative behavior deep-level temporal fusion feature coding vector are subjected to the target pressure gauge main behavior-collaborative behavior temporal semantic interaction analysis based on order parameter field constraints and path integral optimization to obtain the target pressure gauge main behavior-collaborative behavior temporal joint coding vector.

5. The method for automatically analyzing abnormal conditions of a pressure gauge according to claim 4, characterized in that: Based on the target pressure gauge main behavior and collaborative behavior time series joint coding vector, the abnormal diagnosis results are obtained, including: The target pressure gauge main behavior-cooperative behavior time series joint encoding vector is input into the trained abnormal state diagnosis multi-classifier model to obtain the abnormal diagnosis result.

6. A pressure gauge abnormal state automatic analysis system, used to execute the pressure gauge abnormal state automatic analysis method according to any one of claims 1 to 5, characterized in that: include: A pressure value acquisition module is used to obtain the real-time pressure value sequence of the target pressure gauge; a flow rate value acquisition module, for acquiring a real-time flow rate value sequence collected by a flow meter associated with a target pressure gauge; A timestamp alignment module is used to perform timestamp alignment on the real-time pressure value sequence and the real-time flow rate value sequence to obtain an aligned real-time pressure value sequence and an aligned real-time flow rate value sequence; An individual signal feature extraction module is used to extract individual signal features from the aligned real-time pressure value sequence to obtain a target pressure gauge pressure time series feature vector; A pressure value inference module is used to input the aligned real-time flow velocity value sequence into the autoencoder-based collaborative model to obtain an inferred real-time pressure value sequence; The abnormal state diagnosis module is used to calculate the pressure-flow rate collaborative deviation feature between the inferred real-time pressure value sequence and the real-time pressure value sequence, and input the pressure-flow rate collaborative deviation feature and the target pressure gauge pressure time series feature into the trained abnormal state diagnosis multi-classifier model to obtain the abnormal diagnosis result.

Images