Intelligent state monitoring method and system based on petrochemical equipment

Abstract

The invention discloses an intelligent state monitoring method and system based on petrochemical equipment, and relates to the technical field of intelligent state monitoring, and the method comprises the steps: constructing a petrochemical equipment digital twin based on the parameters and physical characteristics of target petrochemical equipment, and collecting multi-source time sequence data; inputting the multi-source time sequence data into the digital twin of the petrochemical equipment, and performing real-time simulation to obtain theoretical operation state data; and based on the multi-source time sequence data and the theoretical operation state data, obtaining deviation data between the measured value and the theoretical value, and updating the parameters of the digital twin of the petrochemical equipment by using the deviation data to obtain equivalent real operation state data. Through deep fusion of a mechanism model, data driving and an intelligent algorithm, accurate prediction of equipment life, effective distinguishing of fault types and automatic generation of a diagnosis report are realized, and the industrial problems of high false alarm rate, poor interpretability and lack of predictability in the prior art are fundamentally solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent condition monitoring technology, and in particular to an intelligent condition monitoring method and system for petrochemical equipment. Background Technology

[0002] The condition monitoring technology for petrochemical equipment is gradually transforming from the traditional periodic maintenance and threshold alarm mode to a data-based predictive maintenance mode. By deploying vibration, temperature, and pressure sensors on key equipment (such as compressors and pump units) to collect high-frequency time-series data, and using deep neural networks to automatically extract features from the data, the technology can classify or regress and predict the health status of the equipment. This can detect abnormal patterns to a certain extent, reduce the risk of unplanned downtime, and has become a research hotspot in recent years. The advantage of this technology is that it avoids the complex physical modeling process and achieves end-to-end condition assessment through a data-driven approach. It is especially suitable for scenarios with sufficient and distinctive fault samples.

[0003] However, while existing data-driven methods perform well in specific scenarios, their core limitation lies in the lack of deep integration between the physical mechanisms of equipment and its operating environment. Existing methods typically treat equipment as a black box, relying on statistical learning of anomalous patterns in historical data to assess its health status. This fails to effectively couple the continuous physicochemical processes faced by equipment in real-world operating environments, such as time-varying corrosion and material degradation. This makes it difficult for the system to distinguish between normal performance degradation and abnormal failures, and in the complex, highly corrosive, and variable operating conditions of the petrochemical industry, it is prone to false alarms due to environmental disturbances or normal aging. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent status monitoring method for petrochemical equipment, which solves the problem of high false alarms caused by the lack of mechanism integration in existing data-driven methods, making it difficult to distinguish between normal performance degradation and abnormal faults.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an intelligent status monitoring method for petrochemical equipment, which includes constructing a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collecting multi-source time-series data. Multi-source time-series data is input into the digital twin of petrochemical equipment to perform real-time simulation and obtain theoretical operating status data; Based on multi-source time-series data and theoretical operating status data, the deviation data between measured values ​​and theoretical values ​​is obtained. The parameters of the digital twin of the petrochemical equipment are updated using the deviation data to obtain equivalent real operating status data. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, an equipment association network is constructed. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time series data to construct a corrosion kinetic model. Multi-source time-series data are input into the corrosion kinetics model to calculate the material performance degradation rate and generate a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, a real-time health index is calculated. The real-time health index and the dynamic health baseline are evaluated to generate an alarm signal. Based on the alarm signal, the vibration data sequence during the alarm period is input into the physical information neural network to obtain physical perturbation signals and generate a diagnostic report.

[0007] As a preferred embodiment of the intelligent status monitoring method for petrochemical equipment described in this invention, the method involves: constructing a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collecting multi-source time-series data, including the following steps: Obtain the design drawings, material properties, and physical characteristics of the target petrochemical equipment, and construct a digital twin of the petrochemical equipment in simulation software using the parameters and physical characteristics of the target petrochemical equipment; Vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors are deployed in the digital twin model of petrochemical equipment; Vibration data, temperature data, pressure data, and medium concentration data are collected synchronously using vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors. The collected vibration data, temperature data, pressure data, and medium concentration data are synchronized and aligned in time and packaged to form a multi-source time-series data set.

[0008] As a preferred embodiment of the intelligent status monitoring method for petrochemical equipment described in this invention, the method involves inputting multi-source time-series data into a digital twin of the petrochemical equipment and performing real-time simulation to obtain theoretical operating status data, including the following steps: The process parameter vector driving the simulation is parsed and extracted from the multi-source time series data set, and the process parameter vector is input into the digital twin of the petrochemical equipment; The digital twin of petrochemical equipment integrates a finite element solver to perform transient dynamic simulation calculations, solve for the dynamic response of the equipment under the current operating conditions, and obtain theoretical operating state data; The theoretical operating state data expression is: ; in, For digital twins of petrochemical equipment Theoretical operating state data at any given time. For the quality matrix, Here is the damping matrix. Here is the stiffness matrix. For digital twins of petrochemical equipment Load vector at time t, For time indexing.

[0009] As a preferred embodiment of the intelligent status monitoring method for petrochemical equipment described in this invention, the method includes the following steps: Based on multi-source time-series data and theoretical operating status data, deviation data between measured and theoretical values ​​is obtained; the parameters of the digital twin of the petrochemical equipment are updated using the deviation data to obtain equivalent real operating status data. Measured vibration data collected at the bearing housing measurement point is extracted from multi-source time series data, and theoretical vibration data at the bearing housing measurement point is extracted from theoretical operating state data. The measured vibration data and theoretical vibration data are compared in the frequency domain to generate the relative error and phase difference of the characteristic frequency amplitude, thus forming the deviation data between the measured value and the theoretical value. Based on the deviation data between measured and theoretical values, the extended Kalman filter algorithm is used to estimate and update the parameters to be corrected in the digital twin of the petrochemical equipment in reverse. Using the corrected parameters in the updated digital twin of the petrochemical equipment, the digital twin of the petrochemical equipment is driven to perform simulation to obtain equivalent real operating status data.

[0010] As a preferred embodiment of the intelligent status monitoring method for petrochemical equipment described in this invention, the method includes the following steps: constructing an equipment association network based on the equipment component structure and connection relationships of the petrochemical equipment digital twin; extracting process features associated with nodes and edges in the equipment association network from multi-source time-series data; and constructing a corrosion kinetic model. Based on the finite element mesh and assembly constraint information of the digital twin of petrochemical equipment, the structure and connection relationship of the equipment components of the digital twin of petrochemical equipment are extracted. By utilizing the extracted equipment component structure and connection relationships of the digital twin of petrochemical equipment, an equipment association network is automatically generated. Based on the physical meaning and spatial location of each node and edge in the device association network, local operating parameters are mapped and extracted from multi-source time series data; A corrosion kinetics model is constructed by utilizing the process characteristics associated with nodes and edges in the equipment-associated network.

[0011] As a preferred embodiment of the intelligent condition monitoring method for petrochemical equipment described in this invention, the method includes the following steps: inputting multi-source time-series data into a corrosion kinetic model, calculating the material performance degradation rate and generating a dynamic health baseline, calculating a real-time health index based on equivalent real-world operating state data and multi-source time-series data, evaluating the real-time health index against the dynamic health baseline, and generating an alarm signal. Medium concentration data, temperature data, pressure data, and flow rate data are extracted from a multi-source time-series dataset and input into a corrosion kinetics model. The corrosion kinetics model is calculated based on electrochemical principles and local environmental parameters to obtain the material performance degradation rate. The expression for the material property degradation rate is: ; in, For a moment The rate of degradation of material properties, Pre-exponential factor, For activation energy, Let be the ideal gas constant. For a moment The medium concentration below, For a moment The absolute temperature below, For a moment The flow rate below, The concentration is the reaction order. The flow rate is the reaction order; By introducing a decay factor to exponentially decay the cumulative erosion depth value at historical moments, and using a time-varying weight function to dynamically adjust the integral contribution based on the confidence index of the deviation data to suppress error accumulation and focus on recent information, an improved integral algorithm for decay memory is obtained. An improved integral algorithm with sliding window and decay memory is used to process the material performance degradation rate, and the local cumulative erosion depth is obtained from the current moment back to the finite time window. Based on the ratio of local cumulative erosion depth to critical corrosion depth, a dynamic health baseline is generated. The improved integral expression for decaying memory is: ; in, For a moment The local cumulative erosion depth below As the attenuation factor, The length of the sliding window. For a moment The time-varying weight function, For integration variables, This represents the cumulative erosion depth at a historical moment. By performing fast Fourier transform on theoretical operating state data and measured vibration data respectively, the theoretical amplitude and measured amplitude of characteristic frequency are extracted. Theoretical vibration features are extracted from equivalent real operating state data, and measured vibration features are extracted from multi-source time series data. The relative deviation between theoretical vibration features and measured vibration features is calculated, and the real-time health index is obtained by fusing the relative deviation. The real-time health index expression is: ; in, For real-time health index, The total number of characteristic frequencies. For characteristic frequency index, For the first The theoretical amplitude of the first-order characteristic frequency, For the first Measured amplitude of the first characteristic frequency; The dynamic health baseline serves as a reference value for whether to trigger an alarm based on the health index. An alarm signal is generated when the real-time health index is lower than the dynamic health baseline.

[0012] As a preferred embodiment of the intelligent status monitoring method for petrochemical equipment described in this invention, the method includes the following steps: based on alarm signals, the vibration data sequence during the alarm period is input into a physical information neural network to obtain physical perturbation signals and analyze them to generate a diagnostic report. Based on the time stamp of alarm signal generation, the vibration data sequence within the alarm signal period is extracted from the stored multi-source time-series data. The vibration data sequence within the time period corresponding to the alarm signal is input into the physical information neural network. Through the adversarial game between the generator and the discriminator, physical perturbation signals are synthesized. A knowledge base is established based on the historical fault case library and the equipment design failure mode library. Perform spectral analysis on physical perturbation signals to identify fault characteristics in the physical perturbation signals, and generate diagnostic reports by matching the fault characteristics with a knowledge base.

[0013] Secondly, the present invention provides an intelligent status monitoring system based on petrochemical equipment, including an intelligent status monitoring method based on petrochemical equipment, characterized in that: it includes a digital twin construction module, which constructs a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collects multi-source time-series data; The simulation module inputs multi-source time-series data into the digital twin of the petrochemical equipment to perform real-time simulation and obtain theoretical operating status data; The prediction module, based on multi-source time-series data and theoretical operating status data, obtains the deviation data between the measured values ​​and the theoretical values. The deviation data is used to update the parameters of the digital twin of the petrochemical equipment to obtain equivalent real operating status data. A corrosion kinetics model module is constructed. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, an equipment association network is built. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time series data to construct a corrosion kinetics model. The evaluation module inputs multi-source time-series data into the corrosion kinetics model, calculates the material performance degradation rate and generates a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, it calculates the real-time health index, evaluates the real-time health index and the dynamic health baseline, and generates an alarm signal. The visualization module, based on the alarm signal, inputs the vibration data sequence within the alarm period into the physical information neural network to obtain physical perturbation signals and analyze them to generate a diagnostic report.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the intelligent status monitoring method for petrochemical equipment as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent status monitoring method for petrochemical equipment as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By constructing an equipment association network and extracting local process features to build a corrosion kinetic model, the invention achieves accurate quantification and dynamic prediction of the corrosion environment at different locations of the equipment, laying the foundation for generating a high-precision health baseline; by generating a dynamic health baseline through the corrosion model and adaptively comparing it with the real-time health index, the invention achieves a scientific distinction between normal performance degradation and abnormal failures, significantly reducing the false alarm rate under varying operating conditions and high corrosion environments; and by deeply integrating the mechanism model, data-driven approach, and intelligent algorithms, the invention achieves accurate prediction of equipment lifespan, effective differentiation of fault types, and automatic generation of diagnostic reports, fundamentally solving the industry problems of high false alarm rate, poor interpretability, and lack of predictability in existing technologies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an intelligent condition monitoring method for petrochemical equipment.

[0019] Figure 2 This is a schematic diagram of an intelligent status monitoring system based on petrochemical equipment. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figure 1 and Figure 2 This is one embodiment of the present invention, which provides an intelligent status monitoring method for petrochemical equipment, comprising the following steps: S1. Based on the parameters and physical characteristics of the target petrochemical equipment, construct a digital twin of the petrochemical equipment and collect multi-source time-series data.

[0024] S1.1 Obtain the design drawings, material properties and physical characteristics of the target petrochemical equipment, and construct a digital twin of the petrochemical equipment in the simulation software using the parameters and physical characteristics of the target petrochemical equipment.

[0025] Furthermore, based on the design drawings of the target petrochemical equipment, the geometric dimensions and assembly relationships of the equipment are obtained; based on the material certificate, material property parameters such as elastic modulus and density are obtained; and based on the process operation procedures, the physical properties of the medium such as operating density and corrosion characteristics are obtained. Using the obtained parameters and physical properties of the target petrochemical equipment, a geometric model of the petrochemical equipment digital twin is established in ANSYS finite element analysis software, material property parameters are defined, boundary constraints are applied, mesh generation rules are set, and control equations are integrated to form the petrochemical equipment digital twin.

[0026] S1.2 Deploy vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors according to the digital twin model of the petrochemical equipment.

[0027] Furthermore, based on the key components and simulation requirements identified in the digital twin model of the petrochemical equipment, ICP-type vibration acceleration sensors, PT100 temperature sensors, pressure sensors, and online electrochemical medium concentration sensors are deployed at corresponding physical locations on the bearing housings, pump casing surfaces, process pipelines, and media flow paths of the target petrochemical equipment. These sensors together constitute a sensor group for data acquisition.

[0028] S1.3. Vibration data, temperature data, pressure data and medium concentration data are collected synchronously through vibration acceleration sensor, temperature sensor, pressure sensor and medium concentration sensor.

[0029] Furthermore, vibration data is collected through deployed vibration acceleration sensors, temperature data through temperature sensors, pressure data through pressure sensors, and medium concentration data through medium concentration sensors. The vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors are synchronously triggered using a unified clock source based on the IEEE 1588 precision time protocol, and continuously collect vibration data, temperature data, pressure data, and medium concentration data at a preset sampling rate.

[0030] S1.4. Synchronize and align the collected vibration data, temperature data, pressure data, and medium concentration data in time and package them to form a multi-source time-series data set.

[0031] Furthermore, the vibration data collected by the vibration acceleration sensor, the temperature data collected by the temperature sensor, the pressure data collected by the pressure sensor, and the medium concentration data collected by the medium concentration sensor are received through a data acquisition card. The vibration data, temperature data, pressure data, and medium concentration data are timestamped using a precise network time protocol. The time-synchronized vibration data, temperature data, pressure data, and medium concentration data are then packaged into a structured multi-source time-series data set.

[0032] S2. Input multi-source time-series data into the digital twin of the petrochemical equipment and perform real-time simulation to obtain theoretical operating status data.

[0033] S2.1. Parse and extract the process parameter vectors driving the simulation from the multi-source time series data set, and input the process parameter vectors into the digital twin of the petrochemical equipment.

[0034] Furthermore, the multi-source time-series data set is analyzed to extract data on rotational speed, flow rate, inlet pressure, inlet temperature, and medium concentration. These data are then combined into a process parameter vector that drives the simulation. The process parameter vector is a time-varying array whose elements represent rotational speed, flow rate, inlet pressure, inlet temperature, and medium concentration, respectively. The process parameter vector is used as boundary conditions and load inputs and is input into the digital twin of the petrochemical equipment through the application programming interface.

[0035] By directly extracting and structuring driving parameters from multi-source time-series data sets, the simulation conditions and real-time operating conditions are accurately matched, overcoming the lag and distortion problems caused by manually setting operating condition parameters in traditional methods, and ensuring the timeliness and accuracy of simulation input.

[0036] S2.2 The petrochemical equipment digital twin integrated finite element solver performs transient dynamic simulation calculations to solve the equipment dynamic response under the current operating conditions and obtain theoretical operating state data.

[0037] The theoretical operating state data expression is: ; in, For digital twins of petrochemical equipment Theoretical operating state data at any given time. For the quality matrix, Here is the damping matrix. Here is the stiffness matrix. For digital twins of petrochemical equipment Load vector at time t, For time indexing.

[0038] Furthermore, the digital twin of the petrochemical equipment calls upon the integrated ANSYS transient dynamics solver, taking the operating conditions defined by the process parameter vector as input, to perform transient dynamics simulation calculations. The solution process is based on the finite element method, and the solver directly solves the motion differential equations through numerical integration, calculating the displacement, velocity, and acceleration responses of the digital twin of the petrochemical equipment at each node. All node response data output by the simulation calculation constitute the theoretical operating state data, which fully describes the theoretical dynamic behavior of each part of the target petrochemical equipment under the current input operating conditions.

[0039] By integrating a commercial finite element solver and directly solving the equipment dynamics equations, the simulation results strictly follow the laws of physics, resulting in high-fidelity theoretical operating state data, which provides a reliable benchmark reference for subsequent accurate state assessment.

[0040] It should be noted that the quality matrix Damping matrix and stiffness matrix These are the core parameters describing the inherent physical characteristics of the equipment, respectively characterizing the system's inertia, energy dissipation, and elastic recovery characteristics, and the mass matrix. Damping matrix and stiffness matrix Together they define the device's DNA and determine its inherent dynamic response characteristics.

[0041] Load vector It is the mathematical expression of external excitation, with elements calculated from the input process parameter vector (such as rotational speed and pressure) according to mechanical principles, representing... The real external forces that act on the equipment at all times, time It is a variable, indicating that the response is a dynamic process that changes over time.

[0042] Theoretical operating status data It is the result of the combined action of the above physical quantities, including physical responses such as displacement, velocity, and acceleration, and fully describes the dynamic behavior of the device under a given excitation.

[0043] Specifically, traditional data-driven status monitoring methods typically bypass physical modeling and use purely statistical or machine learning models (such as LSTM and CNN) to directly learn the mapping relationship between input and output (device status) from historical data.

[0044] S3. Based on multi-source time-series data and theoretical operating status data, the deviation data between measured values ​​and theoretical values ​​is obtained. The parameters of the digital twin of the petrochemical equipment are updated using the deviation data to obtain equivalent real operating status data.

[0045] S3.1 Extract the measured vibration data collected at the bearing housing measurement point from the multi-source time series data, and extract the theoretical vibration data at the bearing housing measurement point from the theoretical operating state data. Compare the measured vibration data and the theoretical vibration data in the frequency domain to generate the relative error and phase difference of the characteristic frequency amplitude, thus forming the deviation data between the measured value and the theoretical value.

[0046] Furthermore, measured vibration data collected by vibration acceleration sensors at the bearing housing measurement points are extracted from the multi-source time-series data set, and theoretical vibration data at the bearing housing measurement points are extracted from the theoretical operating state data. Fast Fourier transforms are performed on both the measured and theoretical vibration data to convert them to the frequency domain. In the frequency domain, key characteristic frequency components such as power frequency, harmonics, and bearing characteristic frequency are selected to obtain the relative amplitude error and phase difference of the measured spectrum and theoretical spectrum at these characteristic frequency points. These relative amplitude error and phase difference data are combined to form the deviation data between the measured value and the theoretical value.

[0047] By accurately stripping away environmental noise and focusing on key fault characteristic frequencies through frequency domain analysis, the essential deviation information characterizing abnormal equipment status is effectively extracted. This overcomes the drawback of time domain signals being susceptible to random interference, providing a high signal-to-noise ratio input for subsequent parameter updates.

[0048] S3.2 Based on the deviation data between measured values ​​and theoretical values, the extended Kalman filter algorithm is used to estimate and update the parameters to be corrected in the digital twin of the petrochemical equipment.

[0049] Furthermore, the deviation data between the measured values ​​and the theoretical values ​​is used as the observation error input to the extended Kalman filter algorithm. The extended Kalman filter algorithm is based on a simplified state-space model of the petrochemical equipment digital twin. It compares the deviation data with the model prediction values ​​and estimates the optimal values ​​of the parameters to be corrected in the petrochemical equipment digital twin through a recursive algorithm. The parameters to be corrected include the bearing support stiffness coefficient and damping coefficient. The estimated optimal values ​​are then used to update the parameters to be corrected in the petrochemical equipment digital twin.

[0050] By employing state estimation theory, closed-loop online identification and self-correction of model parameters are achieved. By dynamically adjusting the internal parameters of the model in real time by absorbing observation biases, the digital twin continuously approximates the real state of the physical entity, thus solving the problem of long-term accuracy decay caused by the solidification of model parameters in traditional open-loop simulation.

[0051] S3.3 Using the corrected parameters in the updated petrochemical equipment digital twin, drive the petrochemical equipment digital twin to perform simulation and obtain equivalent real operating status data.

[0052] Furthermore, the corrected parameters in the petrochemical equipment digital twin, updated using the extended Kalman filter algorithm, including the updated support stiffness coefficient and damping coefficient, are used to drive the petrochemical equipment digital twin to perform transient dynamic simulation calculations again. The data output by the simulation calculations is the equivalent real operating state data.

[0053] By using a high-fidelity model that has been calibrated online for simulation, the simulation output data can be equivalently represented to the actual operating status of the physical entity of the equipment at the current moment. This provides a highly reliable benchmark that is synchronized with the physical equipment status for subsequent health assessments, laying the foundation for accurate status monitoring.

[0054] S4. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, construct the equipment association network, extract the process features associated with the nodes and edges in the equipment association network from multi-source time series data, and construct the corrosion kinetic model.

[0055] S4.1 Based on the finite element mesh and assembly constraint information of the petrochemical equipment digital twin, extract the equipment component structure and connection relationship of the petrochemical equipment digital twin.

[0056] Furthermore, based on the node element topology relationships in the finite element mesh model of the petrochemical equipment digital twin and the contact pair definitions in the assembly constraint information, the equipment component structure and connection relationships of the petrochemical equipment digital twin are extracted. The equipment component structure and connection relationships of the petrochemical equipment digital twin include the geometry of each component, spatial position, and assembly relationships between components, such as flange connections, welding, and interaction relationships.

[0057] By deeply analyzing the underlying mesh topology and assembly constraints of the digital twin model, the component and connection information describing the physical structure of the equipment is accurately extracted. This provides a complete structured data foundation for the subsequent construction of an associated network that reflects the real equipment topology, overcoming the limitations of traditional methods that can only handle simple equipment or rely on manually defined component relationships.

[0058] S4.2 Utilize the extracted equipment component structure and connection relationships of the petrochemical equipment digital twin to automatically generate an equipment association network.

[0059] Furthermore, by utilizing the extracted equipment component structure and connection relationships of the digital twin of the petrochemical equipment, an equipment association network is automatically generated. The equipment association network is a graph structure, where nodes represent specific components such as impellers and pump casings, and edges represent physical connection relationships between components such as bolt connections or functional dependencies such as fluid transmission paths.

[0060] By automatically transforming the structural relationships of physical devices into network models in graph theory, the mathematical abstraction of device topology is realized, creating conditions for subsequent accurate data mapping and feature extraction based on network nodes and edges. This avoids the drawback of traditional monitoring methods that treat devices as a whole and cannot perform fine-grained local analysis.

[0061] S4.3. Based on the physical meaning and spatial location of each node and edge in the device association network, local operating parameters are mapped and extracted from multi-source time series data.

[0062] Furthermore, based on the physical meaning and spatial location of each node and edge in the equipment association network, local operating parameters directly associated with them are mapped and extracted from the multi-source time series data set; these local operating parameters constitute the process features associated with the nodes and edges in the equipment association network.

[0063] This step achieves a leap from macroscopic system monitoring to precise perception of the microscopic local environment by intelligently distributing global process data to specific components and connections based on the equipment topology network. It solves the pain point that traditional methods cannot quantify the differentiated corrosion environment at different locations inside the equipment.

[0064] S4.4 Construct a corrosion kinetics model by utilizing the process characteristics associated with nodes and edges in the equipment-related network.

[0065] Furthermore, a corrosion kinetics model is constructed by using the process features associated with nodes and edges in the equipment network as input variables. The corrosion kinetics model is a mathematical equation based on electrochemical principles and local environmental parameters, used to quantify the material corrosion rate under the coupled effects of specific local process features such as temperature, concentration, and flow rate.

[0066] This step utilizes precisely mapped local process features to drive the corrosion model, enabling dynamic and precise simulation of the differentiated corrosion behavior of different parts of the equipment. This provides a scientific basis for generating predictive dynamic health baselines and fundamentally improves the foresight and accuracy of condition monitoring.

[0067] S5. Input multi-source time-series data into the corrosion kinetic model, calculate the material performance degradation rate and generate a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, calculate the real-time health index, evaluate the real-time health index and the dynamic health baseline, and generate an alarm signal.

[0068] S5.1 Extract medium concentration data, temperature data, pressure data, and flow rate data from the multi-source time series data set and input them into the corrosion kinetic model. The corrosion kinetic model calculates the material performance degradation rate based on electrochemical principles and local environmental parameters.

[0069] The expression for the material property degradation rate is: ; in, For a moment The rate of degradation of material properties, Pre-exponential factor, For activation energy, Let be the ideal gas constant. For a moment The medium concentration below, For a moment The absolute temperature below, For a moment The flow rate below, The concentration is the reaction order. denoted as the flow rate reaction order.

[0070] Furthermore, medium concentration data, temperature data, pressure data, and flow rate data are extracted from the multi-source time-series data set and input into the corrosion kinetic model. The corrosion kinetic model is calculated based on electrochemical principles and local environmental parameters, and the material performance degradation rate is obtained by solving the Arrhenius form reaction rate equation.

[0071] By inputting local environmental parameters into a mechanism-driven corrosion kinetics model, the instantaneous degradation rate of materials under specific corrosion conditions is accurately quantified, providing a scientific basis for the generation of predictive health baselines and overcoming the limitations of traditional empirical models in adapting to varying operating conditions.

[0072] It should be noted that, It is the core part of the Arrhenius equation, which describes the exponential relationship between the chemical reaction rate constant and temperature.

[0073] activation energy It is the energy barrier that needs to be overcome to initiate a corrosion reaction; its value is determined by the nature of the material and the medium, and is the ideal gas constant. It is a physical constant used to unify the unit of energy.

[0074] absolute temperature It directly affects the kinetic energy of the reaction particles. As the temperature increases, the kinetic energy of the particles increases, making it easier for them to overcome the activation energy barrier, resulting in an exponential increase in the reaction rate.

[0075] Pre-exponential factors It represents the frequency factor of the reaction, which is related to the collision frequency and spatial configuration of the reacting particles.

[0076] yes The concentration of the corrosive medium at any given time, and the concentration-reaction order. It is an index determined by the reaction mechanism, reflecting the dependence of the corrosion rate on the concentration of the medium. yes The flow rate of the fluid at any given time, and the flow rate reaction order. It is an empirical index that reflects the complex influence of flow rate on the corrosion process.

[0077] Yes, the present invention adopts This hybrid model, which integrates mechanisms and data, is based on the following core considerations: Ensuring physical extrapolation: The core framework of the model is based on robust physical and chemical laws (Arrhenius equation, law of mass action), which means that even under completely new working conditions that have not been experienced before, as long as the input parameters are physically reasonable, a mechanistically reliable prediction result can be given, which greatly overcomes the fundamental defects of poor extrapolation ability of purely empirical models and data-driven models.

[0078] Achieving locally accurate predictions: the key inputs to this model , , It refers to local process characteristics associated with specific nodes and edges in the equipment's network, rather than macroscopic averages of the entire equipment. This allows the model to calculate different corrosion rates for different locations on the equipment, accurately quantifying the differences in local corrosion environments, and achieving a leap from overall averaging to precise local analysis.

[0079] S5.2 By introducing a decay factor to exponentially decay the cumulative erosion depth value at historical moments, and using a time-varying weight function to dynamically adjust the integral contribution based on the confidence index of the deviation data to suppress error accumulation and focus on recent information, an improved integral algorithm for decay memory is obtained.

[0080] Furthermore, by introducing a decay factor to exponentially decay the cumulative erosion depth value at historical moments, and using a time-varying weight function to dynamically adjust the integral contribution based on the confidence index of the deviation data to suppress error accumulation and focus on recent information, an improved integral algorithm for decay memory is obtained.

[0081] The over-attenuation factor reduces the weight of historical errors and the time-varying weight function focuses on new data with high confidence, effectively suppressing the error accumulation and lag problems in traditional full-time domain integration, and improving the real-time performance and reliability of cumulative erosion depth calculation.

[0082] S5.3. An improved integral algorithm with sliding window and decay memory is used to process the material performance degradation rate, and the local cumulative erosion depth is obtained from the current moment back to the finite time window. Based on the ratio of the local cumulative erosion depth to the critical corrosion depth, a dynamic health baseline is generated.

[0083] The improved integral expression for decaying memory is: ; in, For a moment The local cumulative erosion depth below As the attenuation factor, The length of the sliding window. For a moment The time-varying weight function, For integration variables, This represents the cumulative erosion depth at a historical moment.

[0084] Furthermore, an improved integral algorithm with sliding window and decay memory is used to process the material performance degradation rate, and the local cumulative erosion depth is obtained from the current moment back to the finite time window. Based on the ratio of the local cumulative erosion depth to the critical corrosion depth, a dynamic health baseline is generated.

[0085] By limiting the integration range through a sliding window and combining it with an improved integration algorithm, a health prediction trajectory that dynamically adjusts with changes in the corrosive environment is generated. This provides an intelligent threshold that can adapt to changes in operating conditions for equipment health status assessment, solving the fundamental problem that fixed thresholds cannot distinguish between normal aging and abnormal faults.

[0086] It should be noted that, It is a historic moment The cumulative erosion depth value represents the total material loss of the equipment up to that point in time.

[0087] (attenuation factor, 0 < <1) is a physical forgetting factor that simulates the gradual and non-permanent memory characteristics of material corrosion processes. The actual impact of early corrosion history on the current health status of equipment gradually weakens over time, rather than always having a 100% weighting on the current state. <1 is precisely the mathematical characterization of this physical reality.

[0088] yes The material property degradation rate at a given time is calculated by the corrosion kinetics model based on the actual local environment at that time.

[0089] The time-varying weighting function is a data reliability factor, based on... The confidence level of the data at each moment dynamically adjusts the contribution weight of the corrosion rate at that moment to the cumulative result. High confidence data has a high weight, while low confidence data or data with noise interference has a low weight or even close to zero weight. This simulates the physical fact that in actual monitoring, the degree of trust in measurement data at different times is different.

[0090] Yes, the present invention adopts This improved integral algorithm with sliding window and decay memory is based on the following core physical and engineering considerations.

[0091] Simulating the forgetting effect of the physical world: By introducing a decay factor, the weight of premature historical data is actively reduced. This allows the algorithm to simulate the physical reality that the state of a device is not entirely determined by history, thus more accurately tracking the current true state of the device and suppressing the infinite accumulation of historical errors.

[0092] Focusing on recent dominant factors and responding quickly to changes: By using a sliding window to limit the integration to a recent period, the cumulative depth calculation primarily reflects the impact of the recent corrosive environment. This enables the algorithm to respond quickly to drastic changes in operating conditions, promptly capturing accelerated degradation of equipment health, and greatly improving the model's real-time performance and sensitivity.

[0093] Trusting reliable data and suppressing noise interference: By using a time-varying weighting function, the contribution of the data is dynamically adjusted according to the confidence level of the data. This allows the algorithm to prioritize trusting high-quality data during integration and automatically suppress interference from low-quality or unreliable data. As a result, even in real industrial environments with measurement noise and sensor anomalies, it can still calculate a more robust and reliable cumulative erosion depth.

[0094] S5.4. By performing Fast Fourier Transform on the theoretical operating state data and the measured vibration data respectively, the theoretical amplitude and the measured amplitude of the characteristic frequency are extracted. The theoretical vibration characteristics are extracted from the equivalent real operating state data, and the measured vibration characteristics are extracted from the multi-source time series data. The relative deviation between the theoretical vibration characteristics and the measured vibration characteristics is calculated, and the real-time health index is obtained by fusing the relative deviation.

[0095] The real-time health index expression is: ; in, For real-time health index, The total number of characteristic frequencies. For characteristic frequency index, For the first The theoretical amplitude of the first-order characteristic frequency, For the first The measured amplitude of the first characteristic frequency.

[0096] Furthermore, by performing Fast Fourier Transform on theoretical operating state data and measured vibration data respectively, the theoretical amplitude and measured amplitude of characteristic frequencies are extracted. Theoretical vibration characteristics are extracted from equivalent real operating state data, and measured vibration characteristics are extracted from multi-source time series data. The relative deviation between theoretical vibration characteristics and measured vibration characteristics is calculated, and the real-time health index is obtained by fusing the relative deviation.

[0097] By fusing and calculating the characteristic deviations of multiple frequency domains, a quantitative index that comprehensively represents the overall health status of the equipment was obtained, providing a data foundation for accurate status assessment.

[0098] It should be noted that, For the first The theoretical amplitude of the first-order characteristic frequency represents the theoretical amplitude of the vibration response at that characteristic frequency, which is obtained by the simulation of the equipment's digital twin based on its updated physical parameters under the current equivalent real operating state. It is a dynamically updated theoretical expected value based on physical laws.

[0099] For the first The measured amplitude of the first-order characteristic frequency represents the true amplitude of the vibration response at that characteristic frequency, which is actually measured by the sensor.

[0100] S5.5 The dynamic health baseline serves as a reference value for whether to trigger an alarm for the health index. When the real-time health index is less than the dynamic health baseline, an alarm signal is generated.

[0101] Furthermore, the dynamic health baseline serves as a reference value for whether to trigger an alarm based on the health index. When the real-time health index is lower than the dynamic health baseline, it indicates that the decline in the equipment's health status has exceeded the reasonable expectations under the current corrosive environment, thus generating an alarm signal.

[0102] By intelligently comparing real-time health indices with dynamic prediction baselines, early and accurate warnings of abnormal faults are achieved, significantly reducing the false alarm rate under varying operating conditions.

[0103] S6. Based on the alarm signal, the vibration data sequence during the alarm period is input into the physical information neural network to obtain the physical perturbation signal and generate a diagnostic report.

[0104] S6.1 Based on the time stamp generated by the alarm signal, extract the vibration data sequence within the alarm signal period from the stored multi-source time-series data.

[0105] Furthermore, based on the time stamp generated by the alarm signal, the vibration data sequence within the corresponding time period of the alarm signal is extracted from the stored multi-source time-series data. The vibration data sequence within the corresponding time period of the alarm signal contains complete high-frequency vibration waveform data within a period of time before and after the alarm is triggered, ensuring that the complete development process of potential faults can be covered.

[0106] S6.2. Input the vibration data sequence within the time period corresponding to the alarm signal into the physical information neural network. Through the adversarial game between the generator and the discriminator, the physical perturbation signal is synthesized. A knowledge base is established based on the historical fault case library and the equipment design failure mode library.

[0107] Furthermore, the vibration data sequence within the time period corresponding to the alarm signal is input into the physical information neural network. The physical information neural network is trained through an adversarial game between the generator and the discriminator. The generator attempts to synthesize a reference vibration signal that conforms to the constraints of the equipment dynamics equation, while the discriminator tries to distinguish between the input measured vibration data sequence and the reference vibration signal synthesized by the generator. Through adversarial training, the generator finally outputs a reference vibration signal that is most similar to the measured vibration data sequence in statistical characteristics. The residual between the measured vibration data sequence and the reference vibration signal is the physical perturbation signal. A knowledge base is constructed based on the fault spectrum template in the historical fault case library and the failure mechanism description in the equipment design failure mode library.

[0108] S6.3 Perform spectral analysis on the physical perturbation signal to identify fault characteristics in the physical perturbation signal, match the fault characteristics with the knowledge base, and generate a diagnostic report.

[0109] Furthermore, spectral analysis is performed on the physical perturbation signal output by the physical information neural network. The physical perturbation signal is converted to the frequency domain using a fast Fourier transform to identify the fault characteristic frequency components and modulation sidebands contained in the physical perturbation signal. The identified fault characteristic frequency and modulation sideband information is matched with the fault spectrum templates and failure modes stored in the knowledge base to determine the fault type, location, and severity, and generate a diagnostic report.

[0110] This embodiment also provides an intelligent status monitoring system based on petrochemical equipment, including: a digital twin construction module, which constructs a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collects multi-source time-series data; The simulation module inputs multi-source time-series data into the digital twin of the petrochemical equipment to perform real-time simulation and obtain theoretical operating status data; The prediction module, based on multi-source time-series data and theoretical operating status data, obtains the deviation data between the measured values ​​and the theoretical values. The deviation data is used to update the parameters of the digital twin of the petrochemical equipment to obtain equivalent real operating status data. A corrosion kinetics model module is constructed. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, an equipment association network is built. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time series data to construct a corrosion kinetics model. The evaluation module inputs multi-source time-series data into the corrosion kinetics model, calculates the material performance degradation rate and generates a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, it calculates the real-time health index, evaluates the real-time health index and the dynamic health baseline, and generates an alarm signal. The visualization module, based on the alarm signal, inputs the vibration data sequence within the alarm period into the physical information neural network to obtain physical perturbation signals and analyze them to generate a diagnostic report.

[0111] This embodiment also provides a computer device applicable to the intelligent status monitoring method for petrochemical equipment, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the intelligent status monitoring method for petrochemical equipment as proposed in the above embodiment.

[0112] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0113] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the intelligent status monitoring method for petrochemical equipment as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0114] In summary, this invention constructs a corrosion kinetic model by building an equipment association network and extracting local process features, achieving accurate quantification and dynamic prediction of the corrosion environment at different locations on the equipment, laying the foundation for generating a high-precision health baseline. By generating a dynamic health baseline through the corrosion model and adaptively comparing it with real-time health indices, it achieves a scientific distinction between normal performance degradation and abnormal faults, significantly reducing the false alarm rate under varying operating conditions and high corrosion environments. Through the deep integration of mechanistic models, data-driven approaches, and intelligent algorithms, it achieves accurate prediction of equipment lifespan, effective differentiation of fault types, and automatic generation of diagnostic reports, fundamentally solving the industry problems of high false alarm rates, poor interpretability, and lack of predictability in existing technologies.

[0115] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent status monitoring of petrochemical equipment, characterized in that: This includes constructing a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collecting multi-source time-series data; Multi-source time-series data is input into the digital twin of petrochemical equipment to perform real-time simulation and obtain theoretical operating status data; Based on multi-source time-series data and theoretical operating status data, the deviation data between measured values ​​and theoretical values ​​is obtained. The parameters of the digital twin of the petrochemical equipment are updated using the deviation data to obtain equivalent real operating status data. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, an equipment association network is constructed. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time series data to construct a corrosion kinetic model. Multi-source time-series data are input into the corrosion kinetics model to calculate the material performance degradation rate and generate a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, a real-time health index is calculated. The real-time health index and the dynamic health baseline are evaluated to generate an alarm signal. Based on the alarm signal, the vibration data sequence during the alarm period is input into the physical information neural network to obtain physical perturbation signals and generate a diagnostic report.

2. The intelligent status monitoring method for petrochemical equipment as described in claim 1, characterized in that: Based on the parameters and physical characteristics of the target petrochemical equipment, a digital twin of the petrochemical equipment is constructed, and multi-source time-series data is collected, including the following steps: Obtain the design drawings, material properties, and physical characteristics of the target petrochemical equipment, and construct a digital twin of the petrochemical equipment in simulation software using the parameters and physical characteristics of the target petrochemical equipment; Vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors are deployed in the digital twin model of petrochemical equipment; Vibration data, temperature data, pressure data, and medium concentration data are collected synchronously using vibration acceleration sensors, temperature sensors, pressure sensors, and medium concentration sensors. The collected vibration data, temperature data, pressure data, and medium concentration data are synchronized and aligned in time and packaged to form a multi-source time-series data set.

3. The intelligent status monitoring method for petrochemical equipment as described in claim 2, characterized in that: Multi-source time-series data is input into the digital twin of the petrochemical equipment for real-time simulation to obtain theoretical operating state data, including the following steps: The process parameter vector driving the simulation is parsed and extracted from the multi-source time series data set, and the process parameter vector is input into the digital twin of the petrochemical equipment; The digital twin of petrochemical equipment integrates a finite element solver to perform transient dynamic simulation calculations, solve for the dynamic response of the equipment under the current operating conditions, and obtain theoretical operating state data; The theoretical operating state data expression is: ； in, For digital twins of petrochemical equipment Theoretical operating state data at any given time. For the quality matrix, Here is the damping matrix. Here is the stiffness matrix. For digital twins of petrochemical equipment Load vector at time t, For time indexing.

4. The intelligent status monitoring method for petrochemical equipment as described in claim 3, characterized in that: Based on multi-source time-series data and theoretical operating status data, the deviation data between measured values ​​and theoretical values ​​is obtained. The parameters of the digital twin of the petrochemical equipment are updated using the deviation data to obtain equivalent real operating status data, including the following steps: Measured vibration data collected at the bearing housing measurement point is extracted from multi-source time series data, and theoretical vibration data at the bearing housing measurement point is extracted from theoretical operating state data. The measured vibration data and theoretical vibration data are compared in the frequency domain to generate the relative error and phase difference of the characteristic frequency amplitude, thus forming the deviation data between the measured value and the theoretical value. Based on the deviation data between measured and theoretical values, the extended Kalman filter algorithm is used to estimate and update the parameters to be corrected in the digital twin of the petrochemical equipment in reverse. Using the corrected parameters in the updated digital twin of the petrochemical equipment, the digital twin of the petrochemical equipment is driven to perform simulation to obtain equivalent real operating status data.

5. The intelligent status monitoring method for petrochemical equipment as described in claim 4, characterized in that: Based on the equipment component structure and connection relationships of the digital twin of petrochemical equipment, an equipment association network is constructed. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time-series data to construct a corrosion kinetic model, including the following steps: Based on the finite element mesh and assembly constraint information of the digital twin of petrochemical equipment, the structure and connection relationship of the equipment components of the digital twin of petrochemical equipment are extracted. By utilizing the extracted equipment component structure and connection relationships of the digital twin of petrochemical equipment, an equipment association network is automatically generated. Based on the physical meaning and spatial location of each node and edge in the device association network, local operating parameters are mapped and extracted from multi-source time series data; A corrosion kinetics model is constructed by utilizing the process characteristics associated with nodes and edges in the equipment-associated network.

6. The intelligent status monitoring method for petrochemical equipment as described in claim 5, characterized in that: Multi-source time-series data is input into the corrosion kinetics model to calculate the material performance degradation rate and generate a dynamic health baseline. Based on equivalent real-world operating data and multi-source time-series data, a real-time health index is calculated. The real-time health index and the dynamic health baseline are then evaluated to generate an alarm signal, including the following steps: Medium concentration data, temperature data, pressure data, and flow rate data are extracted from a multi-source time-series dataset and input into a corrosion kinetics model. The corrosion kinetics model is calculated based on electrochemical principles and local environmental parameters to obtain the material performance degradation rate. The expression for the material property degradation rate is: ； in, For a moment The rate of degradation of material properties, Pre-exponential factor, For activation energy, Let be the ideal gas constant. For a moment The medium concentration below, For a moment The absolute temperature below, For a moment The flow rate below, The concentration is the reaction order. The flow rate is the reaction order; By introducing a decay factor to exponentially decay the cumulative erosion depth value at historical moments, and using a time-varying weight function to dynamically adjust the integral contribution based on the confidence index of the deviation data to suppress error accumulation and focus on recent information, an improved integral algorithm for decay memory is obtained. An improved integral algorithm with sliding window and decay memory is used to process the material performance degradation rate, and the local cumulative erosion depth is obtained from the current moment back to the finite time window. Based on the ratio of local cumulative erosion depth to critical corrosion depth, a dynamic health baseline is generated. The improved integral expression for decaying memory is: ； in, For a moment The local cumulative erosion depth below As the attenuation factor, The length of the sliding window. For a moment The time-varying weight function, For integration variables, This represents the cumulative erosion depth at a historical moment. By performing fast Fourier transform on theoretical operating state data and measured vibration data respectively, the theoretical amplitude and measured amplitude of characteristic frequency are extracted. Theoretical vibration features are extracted from equivalent real operating state data, and measured vibration features are extracted from multi-source time series data. The relative deviation between theoretical vibration features and measured vibration features is calculated, and the real-time health index is obtained by fusing the relative deviation. The real-time health index expression is: ； in, For real-time health index, The total number of characteristic frequencies. For characteristic frequency index, For the first The theoretical amplitude of the first-order characteristic frequency, For the first Measured amplitude of the first characteristic frequency; The dynamic health baseline serves as a reference value for whether to trigger an alarm based on the health index. An alarm signal is generated when the real-time health index is lower than the dynamic health baseline.

7. The intelligent status monitoring method for petrochemical equipment as described in claim 6, characterized in that: Based on the alarm signal, the vibration data sequence during the alarm period is input into the physical information neural network to obtain physical perturbation signals and generate a diagnostic report, including the following steps: Based on the time stamp of alarm signal generation, the vibration data sequence within the alarm signal period is extracted from the stored multi-source time-series data. The vibration data sequence within the time period corresponding to the alarm signal is input into the physical information neural network. Through the adversarial game between the generator and the discriminator, physical perturbation signals are synthesized. A knowledge base is established based on the historical fault case library and the equipment design failure mode library. Perform spectral analysis on physical perturbation signals to identify fault characteristics in the physical perturbation signals, and generate diagnostic reports by matching the fault characteristics with a knowledge base.

8. An intelligent status monitoring system for petrochemical equipment, based on the intelligent status monitoring method for petrochemical equipment according to any one of claims 1 to 7, characterized in that: This includes a digital twin construction module, which constructs a digital twin of the petrochemical equipment based on the parameters and physical characteristics of the target petrochemical equipment, and collects multi-source time-series data; The simulation module inputs multi-source time-series data into the digital twin of the petrochemical equipment to perform real-time simulation and obtain theoretical operating status data; The prediction module, based on multi-source time-series data and theoretical operating status data, obtains the deviation data between the measured values ​​and the theoretical values. The deviation data is used to update the parameters of the digital twin of the petrochemical equipment to obtain equivalent real operating status data. A corrosion kinetics model module is constructed. Based on the equipment component structure and connection relationship of the digital twin of petrochemical equipment, an equipment association network is built. Process features associated with nodes and edges in the equipment association network are extracted from multi-source time series data to construct a corrosion kinetics model. The evaluation module inputs multi-source time-series data into the corrosion kinetics model, calculates the material performance degradation rate and generates a dynamic health baseline. Based on equivalent real operating state data and multi-source time-series data, it calculates the real-time health index, evaluates the real-time health index and the dynamic health baseline, and generates an alarm signal. The visualization module, based on the alarm signal, inputs the vibration data sequence within the alarm period into the physical information neural network to obtain physical perturbation signals and analyze them to generate a diagnostic report.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent status monitoring method based on petrochemical equipment as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent status monitoring method based on petrochemical equipment as described in any one of claims 1 to 7.

Images