Turbine heat preservation data monitoring method and system based on big data

Abstract

The application provides a steam turbine heat preservation data monitoring method and system based on big data, and belongs to the field of steam turbine heat preservation data monitoring. The method is used for obtaining distributed impedance spectrum data of a steam turbine body and a pipeline on which a self-limiting temperature electric heat tracing band is applied. The distributed impedance spectrum data is calculated into continuous temperature field data along the surface. Operation data of a steam turbine distributed control system is obtained, and is aligned with the continuous temperature field data to generate a multi-physical field correlation data set. Impedance drift characteristic information obtained by analyzing the multi-physical field correlation data set by a big data platform is received, and a mapping relationship between impedance and temperature is corrected based on the impedance drift characteristic information. The continuous temperature field data is updated according to the corrected mapping relationship, and a heat preservation state evaluation report and early warning information are generated. The self-limiting temperature electric heat tracing band is multiplexed into a distributed temperature sensing element, and through edge computing and cooperation with a big data platform, adaptive, multi-physical field correlation intelligent monitoring of the heat preservation state of the steam turbine is realized.

Description

Technical Field

[0001] This invention relates to the field of steam turbine insulation data monitoring, and in particular to a method and system for monitoring steam turbine insulation data based on big data. Background Technology

[0002] As a core power source in thermal power generation, nuclear power generation, and industrial drive, the safety and economy of steam turbine operation are highly dependent on the integrity of the insulation system. The steam turbine itself and its associated piping typically operate under high-temperature (250℃ to 600℃) and high-pressure steam conditions. To reduce heat loss, prevent burns to personnel, and improve the plant environment, high-performance insulation layers must be applied to the high- and medium-pressure cylinders of the steam turbine and all high-temperature piping. However, the insulation system is not a one-time solution. During long-term service, the insulation material will experience performance degradation due to factors such as thermal aging, mechanical vibration, and moisture erosion. Defects such as settlement, cracking, and detachment may occur in the insulation structure, and even corrosion under the insulation layer may develop, leading to potential equipment hazards.

[0003] Currently, the monitoring and management of steam turbine insulation status has long been in a relatively extensive stage, with the following three main problems.

[0004] First, the monitoring methods are outdated, resulting in numerous blind spots in perception.

[0005] Traditional insulation monitoring relies primarily on two methods. The first is manual inspection, where operation or maintenance personnel use portable infrared thermometers or thermal imagers to periodically sample the temperature of the outer surface of the insulation layer. This method yields discrete, discontinuous information and depends on the inspection cycle and subjective judgment, failing to capture the dynamic evolution of the temperature field under transient conditions such as unit start-up, shutdown, and load changes. The second method involves embedding a certain number of thermocouples or resistance temperature sensors on the outer surface or inside the insulation layer. However, a large steam turbine can have a surface area of ​​hundreds of square meters, and due to cost and construction feasibility limitations, sensors are typically installed at only a few points (e.g., a dozen or so). This results in over 90% of the turbine cylinder's surface area, complex and irregularly shaped components (such as flanges, valves, and the bolt area on the cylinder's split surface), and the stationary insulation jacket surrounding rotating components being in unmonitored "blind spots." If an insulation defect happens to occur within these blind spots, the system will be unable to detect it.

[0006] Secondly, the monitoring signal and the heat preservation function are independent of each other, resulting in a single information dimension.

[0007] In existing technical solutions, self-regulating heating cables and other heat tracing elements are merely considered functional devices providing antifreeze or heat compensation, their sole purpose being to regulate power on / off based on ambient or set temperatures. Temperature sensors, on the other hand, are independent measuring elements used solely for collecting temperature data. The two are completely separate in physical form and data link. This separation leads to the problem that when the system detects an abnormal temperature, it can only passively issue an alarm, unable to use the heat tracing actuators for reverse intervention or assisted location. Furthermore, existing systems collect data with a single dimension, focusing only on the "temperature value" itself, lacking the extraction of thermodynamic dynamic characteristics such as "temperature rise rate" and "temperature gradient," and failing to correlate temperature field data with mechanical operating parameters such as turbine vibration and differential expansion. This makes it difficult for operators to quickly distinguish whether the root cause of increased unit vibration or abnormal temperature rise stems from thermal deformation or pure mechanical failure, thus missing the optimal response time.

[0008] Third, it lacks the ability to assess the status of the entire life cycle, making predictive maintenance impossible.

[0009] Even if some units have deployed online temperature monitoring systems, their function is limited to uploading temperature data to the monitoring backend for threshold comparison and over-temperature alarms. This "post-event alarm" mode can only inform that "the temperature at a certain point is exceeding the standard," but cannot answer deeper questions such as "how much of the insulation performance is left," "whether the aging rate of the insulation material is normal," or "whether there are internal defects invisible to the naked eye." Especially for corrosion under the insulation layer, its process is extremely insidious: moisture seeps in from the damaged outer protective plate, accumulates inside the insulation cotton, causing the outer wall of the pipe or cylinder to remain wet for a long time and undergo electrochemical corrosion. Before the corrosion penetrates the pipe wall or causes large-scale delamination, the temperature of the outer surface of the insulation often does not show obvious abnormalities. Existing technologies struggle to accurately detect such hidden dangers in the early stages, leading to maintenance work often being "post-event repairs" or "over-maintenance," which compromises both economy and safety.

[0010] In summary, there is an urgent need in this field for an intelligent monitoring solution that can overcome the physical limitations of discrete point monitoring, achieve continuous, blind-spot-free sensing along the turbine surface, and deeply integrate and analyze insulation status data with unit operation data, thereby supporting predictive maintenance throughout the entire life cycle. Summary of the Invention

[0011] In view of the above situation, the main objective of this invention is to propose a method and system for monitoring steam turbine insulation data based on big data, so as to solve the above-mentioned technical problems.

[0012] This invention proposes a method for monitoring turbine insulation data based on big data, the method comprising the following steps: Step 1: For the turbine body and pipelines that have been equipped with self-regulating electric heating tape, obtain the distributed impedance spectrum data of the self-regulating electric heating tape. Step 2: Based on the impedance-temperature mapping relationship, the distributed impedance spectrum data is solved into continuous temperature field data along the turbine body and pipe surface; Step 3: Obtain the operating data from the distributed control system of the steam turbine, align the continuous temperature field data with the operating data using timestamps, and generate a multiphysics correlation dataset; the operating data includes vibration data, load data, and differential expansion data; Step 4: Receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field related datasets by the big data platform; correct the impedance-temperature mapping relationship based on the received impedance drift feature information; Step 5: Update the continuous temperature field data according to the corrected impedance-temperature mapping relationship, and generate a turbine insulation status assessment report and early warning information.

[0013] This invention also proposes a steam turbine insulation data monitoring system based on big data, wherein the system applies the aforementioned steam turbine insulation data monitoring method based on big data, and the system includes: The sensing layer module includes self-regulating electric heating tapes installed on the turbine body and pipelines, used for: Obtain distributed impedance spectrum data of self-limiting electric heating tape; Edge computing layer module, used for: Based on the impedance-temperature mapping relationship, the distributed impedance spectrum data is solved into continuous temperature field data along the turbine body and pipe surface; The system acquires operational data from the distributed control system of the steam turbine, aligns the continuous temperature field data with the operational data using timestamps, and generates a multiphysics correlation dataset. The operational data includes vibration data, load data, and differential expansion data. Receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field associated datasets by a big data platform; based on the received impedance drift feature information, correct the impedance-temperature mapping relationship; The continuous temperature field data is updated based on the corrected impedance-temperature mapping relationship, and a turbine insulation status assessment report and early warning information are generated.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention utilizes the positive temperature coefficient (PTC) characteristic of its conductive polymer composite material, reusing it as a distributed temperature sensor distributed along the entire length of the pipeline. By applying a high-frequency probe carrier signal to the metal conductor of the self-regulating heating cable and calculating its distributed impedance spectrum in real time, the system can acquire continuous temperature field data along the turbine body and pipeline surface, with a spatial resolution down to the centimeter level. Combined with the cross-linking and tightly wound characteristics of the self-regulating heating cable, this invention can achieve blind-spot-free coverage of monitoring dead zones such as the insulation jackets of rotating components, irregular flanges, and valves, which are insufficiently covered by traditional methods. Unlike the traditional monitoring paradigm that uses a dozen discrete points to characterize an area of ​​hundreds of square meters, this invention provides a more comprehensive temperature field monitoring solution.

[0015] 2. This invention reuses the parallel metal conductors of the self-regulating heating cable as a power line carrier communication bus. While acquiring temperature sensing data, it can interact with passive sensing tags (such as humidity tags) deployed within the insulation layer by superimposing a high-frequency communication modulated carrier signal. This design eliminates the need for separate power and signal lines for auxiliary sensors, significantly reducing the number of complex cables on the turbine body, lowering on-site construction difficulty, and improving the long-term reliability of the system under high-temperature and high-vibration environments.

[0016] 3. This invention further explores the piezoresistive effect of conductive polymer composite materials, enabling the self-regulating heating cable to function as an ultra-long vibration / noise sensing array when subjected to DC bias. By acquiring resistance fluctuation signals with high precision at both ends and performing cross-correlation time-difference analysis, the system can accurately calculate the physical location of abnormal sound or vibration sources inside the insulation layer caused by steam leakage, airflow erosion, structural loosening, etc. This eliminates the need for maintenance personnel to blindly dismantle large areas of the insulation layer for inspection; instead, they can perform precise excavation based on the acoustic positioning coordinates provided by the system, significantly shortening troubleshooting time and reducing maintenance costs.

[0017] 4. This invention performs timestamp alignment and edge fusion analysis on the calculated continuous temperature field data and the vibration data, load data, and differential expansion data in the turbine DCS system at the edge computing layer. The system can dynamically extract the temperature rise rate and temperature gradient characteristics of key areas from the temperature field and perform causal contribution quantification analysis with the vibration data. This enables the system to clearly distinguish whether the current vibration anomaly is caused by thermal deformation (such as thermal bending due to insufficient warm-up) or by pure mechanical failure (such as mass imbalance), providing operators with differentiated, data-supported handling suggestions and effectively avoiding misjudgment and blind operation.

[0018] 5. To address the inherent problem of impedance drift in PTC materials under long-term thermo-oxidative aging conditions, which affects the accuracy of temperature calculation, this invention employs a cold-state reference impedance time-series analysis method. Before each cold start-up of the unit or after a planned shutdown and cylinder temperature stabilization, reference impedance values ​​are collected and normalized to form a cold-state reference impedance time series spanning several years. Long-term trend analysis of this time series extracts the aging drift pattern of the reference impedance over service time, generating impedance drift characteristic information. Based on this impedance drift characteristic information, the impedance-temperature mapping relationship is corrected, automatically compensating for measurement deviations caused by material aging. Furthermore, this invention introduces a temperature calculation method based on a soft measurement model as an optional alternative. The soft measurement model is trained in the cloud and then deployed locally on an edge intelligent gateway for inference. This effectively handles the unique temperature hysteresis effect of PTC materials and the impact of operating condition fluctuations on temperature calculation accuracy, and can still operate with high accuracy offline. Through the synergy of aging drift compensation and the soft measurement model, the system maintains high-precision temperature field calculation capabilities even after long-term service.

[0019] 6. This invention not only performs longitudinal trend analysis on individual units, but also establishes a statistical fingerprint database of impedance characteristics for self-regulating heating cables of the same type and batch through a big data platform. When the impedance characteristics of a local area of ​​a unit significantly deviate from the statistical distribution threshold of the group, the system can output an early warning signal of corrosion under the insulation layer or settlement of the insulation cotton, even if the external surface temperature has not yet become abnormal. This truly realizes a leap from "passive alarm" to "active prediction" in maintenance mode, allowing power plants sufficient time to rationally arrange shutdown maintenance windows and procure spare parts and materials, effectively avoiding economic losses caused by unplanned shutdowns.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by means of embodiments of the invention. Attached Figure Description

[0021] Figure 1 This is a flowchart of the turbine insulation data monitoring method based on big data proposed in this invention; Figure 2 This is a schematic diagram of the structure of the self-regulating electric heating tape of the present invention; Figure 3 This is a schematic diagram of a traditional discrete-point temperature monitoring method; Figure 4 This is a schematic diagram of the self-regulating electric heating cable of the present invention wrapped around the cylinder body; Figure 5 This is an architecture diagram of the steam turbine insulation data monitoring system based on big data proposed in this invention.

[0022] In the diagram, 1. Self-regulating heating cable; 101. Copper core conductor; 102. Conductive plastic layer; 103. Insulation layer; 104. Shielding layer; 105. Sheath layer; 2. Steam turbine; 3. Temperature sensor; 4. Digital temperature display; 5. Signal line. Detailed Implementation

[0023] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0024] These and other aspects of the embodiments of the present invention will become clear from the following description and accompanying drawings. In these descriptions and drawings, some specific embodiments of the present invention are specifically disclosed to illustrate some ways of implementing the principles of the embodiments of the present invention; however, it should be understood that the scope of the embodiments of the present invention is not limited thereto.

[0025] Example 1

[0026] This embodiment discloses a method for monitoring turbine insulation data based on big data, as well as a monitoring system for executing the method. First, the five core steps of the method are explained step by step, then the additional technical features are supplemented, and finally the modular architecture of the system is described.

[0027] I. Execution process of monitoring methods; Please see Figure 1 This embodiment provides a method for monitoring turbine insulation data based on big data, which is executed collaboratively by an edge intelligent gateway deployed at the power plant site and a big data platform deployed in a remote data center. The method specifically includes the following steps: Step S11: For the turbine body and pipelines that have been covered with self-regulating electric heating tape, obtain the distributed impedance spectrum data of the self-regulating electric heating tape.

[0028] In this step, for the turbine 2 body and pipelines already equipped with self-regulating electric heating cable 1, the distributed impedance spectrum data of the self-regulating electric heating cable 1 is acquired through an edge smart gateway. The self-regulating electric heating cable 1 comprises a conductive polymer composite material with a positive temperature coefficient (PTC) and two parallel metal wires. In this invention, the function of the metal wires is greatly expanded: they not only serve as power supply channels for transmitting industrial frequency AC to generate Joule heat, but more importantly, they are multiplexed as excitation signal transmission channels and sensing signal receiving channels. The distributed impedance spectrum data is obtained by periodically applying a low-power high-frequency probe carrier signal (e.g., frequency sweeping in the range of 1kHz to 100kHz, with a voltage amplitude much lower than the rated operating voltage of the self-regulating electric heating cable) to the metal wires through a precision impedance analysis unit built into the edge smart gateway, and simultaneously measuring the response current signal flowing through the wires and the response voltage signal at both ends of the wires. The complex impedance frequency response data, i.e., the distributed impedance spectrum data, is then obtained through vector voltage-current ratio calculation or Fourier transform analysis. This data reflects the electrical impedance distribution of the self-regulating heating cable core material at various points along its laying path.

[0029] Please see Figure 2 The cross-sectional structure of the self-regulating heating cable, from the inside out, includes a copper core conductor 101, a conductive plastic layer 102 covering the copper core conductor, an insulating layer 103 covering the conductive plastic layer, a shielding layer 104 covering the insulating layer, and a sheath layer 105 covering the shielding layer. The copper core conductor consists of two parallel metal conductors (preferably tin-plated copper conductors with a stranded structure to improve flexibility), and a conductive polymer composite core material, i.e., the conductive plastic layer, is coated between the two metal conductors using a co-extrusion process. This conductive polymer composite material uses a semi-crystalline polyolefin or fluoropolymer as a matrix and is filled with highly conductive nano-sized carbon black particles, thus exhibiting a significant positive temperature coefficient (PTC) characteristic. Its microscopic mechanism is as follows: at room temperature, the carbon black particles form continuous conductive chains in the polymer matrix, resulting in low material resistivity; when the temperature rises to near the melting transition zone of the polymer matrix, the matrix volume expands, breaking the conductive chains and causing a sharp, non-linear increase in resistivity. The core material is sequentially encased in a high-performance fluoropolymer (such as perfluoroethylene propylene, FEP) inner insulation layer, a tin-plated copper wire braided shielding layer, and a high-temperature and oil-resistant polyolefin or fluororubber sheath layer. This multi-layered structure allows it to withstand the harsh environment of power plants, characterized by high temperatures and the potential presence of oil, moisture, and corrosive gases. In this invention, the function of the two parallel metal conductors is greatly expanded: they not only serve as power supply channels for transmitting industrial frequency AC power to generate Joule heating, but more importantly, they are multiplexed as transmission channels for excitation signals and reception channels for sensing signals.

[0030] Step S12: Based on the impedance-temperature mapping relationship, the distributed impedance spectrum data is solved into continuous temperature field data along the turbine body and pipe surface.

[0031] Figure 3 The diagram illustrates a traditional discrete-point temperature monitoring method. As can be seen from the diagram, a set of temperature sensors 3 are installed on the left and right sides of the high-pressure and intermediate-pressure cylinder shaft covers of the turbine 2 body, and a set of temperature sensors are installed on the left and right sides of the upper and lower cylinder bodies of the high-pressure and intermediate-pressure cylinders. The temperature sensors are connected to a temperature digital display instrument 4 via signal lines 5 for temperature monitoring.

[0032] In this step, the edge smart gateway invokes the impedance-temperature mapping relationship pre-stored in its solid-state storage unit. This mapping relationship characterizes the nonlinear correspondence between the resistivity of the conductive polymer composite material and temperature, typically existing in the form of a calibration curve (e.g., a lookup table recording the reference impedance values ​​at different temperature points). The central processing unit of the edge smart gateway maps the impedance characteristic values ​​(e.g., resistance components or impedance moduli at specific frequencies) in the distributed impedance spectrum data acquired in real time in step S11 to the corresponding temperature values. Since the self-regulating heating cable is continuously laid along the pipeline, its continuous impedance change along the line corresponds to the continuous temperature distribution. Therefore, the output of the calculation is a set of continuous temperature field data along the turbine body and pipeline surface, rather than discrete point temperature readings like those of traditional thermocouples. For example, based on the principle of frequency domain reflection, by applying a frequency sweep excitation signal to the metal wire of the heating cable, the change of the reflection coefficient with frequency is measured, and the impedance distribution curve along the line is obtained through inverse Fourier transform. This continuous temperature field data can present the fine distribution of the turbine surface temperature in the form of a two-dimensional curve or a three-dimensional surface plot.

[0033] Step S13: Obtain the operating data from the distributed control system of the steam turbine, align the continuous temperature field data with the operating data using timestamps, and generate a multiphysics correlation dataset.

[0034] In this step, the edge smart gateway establishes a data interface with the distributed control system (DCS) of the steam turbine through standard industrial communication protocols (such as Modbus TCP, OPC-UA, Profinet, etc.) to acquire in real time the steam turbine operating parameters closely related to the insulation status, including but not limited to: vibration data (shaft vibration or bearing vibration signals measured by eddy current sensors or acceleration sensors installed on the bearing housing), load data (generator active power or main steam flow), and expansion difference data (axial relative expansion difference between the cylinder and the rotor). The time synchronization module built into the edge smart gateway accurately timestamps all received data and the continuous temperature field data generated in step S12. Subsequently, the edge smart gateway performs data fusion processing: it timestamps the continuous temperature field data at the same moment or within the same preset short time window with the vibration data, load data, and expansion difference data, and performs data structured reorganization to generate a multi-dimensional multi-physics correlation dataset. The core value of this dataset lies in characterizing the temporal correlation between the thermodynamic state (reflected by the temperature field) and the mechanical state (reflected by vibration and expansion difference) of the steam turbine, providing a data foundation for subsequent root cause analysis of failures.

[0035] Step S14: Receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field associated datasets by a big data platform; correct the impedance-temperature mapping relationship based on the received impedance drift feature information.

[0036] In this step, the impedance drift characteristic information includes a set of aging drift coefficients, which are used to characterize the offset of the reference impedance value caused by material aging at different temperature points in the impedance-temperature mapping relationship. Based on the aging drift coefficients, the edge smart gateway uses a preset compensation algorithm to shift or scale the locally stored initial impedance-temperature mapping relationship to obtain the corrected impedance-temperature mapping relationship.

[0037] The edge smart gateway first securely uploads the multiphysics correlation dataset generated in step S13 to a big data platform located in a remote data center or cloud server via a pre-established quantum-encrypted communication channel. The quantum-encrypted channel ensures the absolute security of critical infrastructure operation data during transmission, preventing data from being eavesdropped on or tampered with.

[0038] After receiving massive amounts of historical and real-time data, the big data platform will initiate in-depth analysis tasks. Specifically, the platform will extract and analyze features from the received distributed impedance spectrum data and continuous temperature field data. A core task is to identify the impedance drift characteristics of the conductive polymer composite material at different aging stages. This identification process is based on long-term trend analysis of the cold-state reference impedance time series. Impedance drift characteristics refer to the slow, irreversible shift in the reference impedance value (e.g., resistance at 25°C) of PTC materials over time under long-term thermo-oxidative aging. This characteristic reflects the degree of overall translation or shape distortion of the PTC characteristic curve. By performing long-term trend analysis on the cold-state reference impedance values ​​of all self-regulating heating cables of the same model and batch in the database at different service years, the impedance drift characteristics of the material at the current stage can be accurately predicted. This impedance drift characteristic information is encapsulated as the aforementioned aging drift coefficient and fed back to the edge intelligent gateway for active querying and downloading, or directly pushed to the edge intelligent gateway after calculation on the platform side. After receiving this information, the edge smart gateway autonomously and dynamically corrects the impedance-temperature mapping relationship used in step S12, compensates for measurement errors caused by material aging, and ensures that the long-term accuracy of its temperature sensing does not completely depend on the continuous network connection with the big data platform.

[0039] It is important to clarify that the "impedance-temperature mapping relationship" in this invention is a higher-level concept, and its specific physical implementation is a data structure or model parameter that can be stored in the solid-state storage unit of the edge smart gateway. When using the traditional lookup table method for temperature calculation, the mapping relationship is specifically reflected in the initial impedance-temperature calibration curve, i.e., a data table recording discrete temperature points and their corresponding reference impedance values. When using a soft-sensor model, the mapping relationship is specifically reflected in the internal weights and bias parameters of the LSTM model. Under this architecture, "correcting the impedance-temperature mapping relationship" in step S14 means that the edge smart gateway compensates and adjusts the currently used mapping relationship based on the received aging drift coefficient—if it is a lookup table method, the reference impedance values ​​in the data table are subjected to overall translation or scaling correction; if it is a soft-sensor model, the aging drift coefficient is used as a bias compensation term for the model input layer or the first hidden layer to perform translation correction on the model output, or the model is periodically fine-tuned and updated using the corrected temperature data. Regardless of the form used, the corrected mapping relationship is used for the continuous temperature field data update in step S15.

[0040] Step S15: Update the continuous temperature field data according to the corrected impedance-temperature mapping relationship, and generate a turbine insulation status assessment report and early warning information.

[0041] In this step, the edge smart gateway applies the revised and more accurate impedance-temperature mapping relationship from step S14 to recalculate or calibrate the original continuous temperature field data, thereby obtaining updated continuous temperature field data. This updated data eliminates the influence of material aging drift and better reflects the true thermodynamic state of the turbine insulation layer.

[0042] Finally, by comprehensively utilizing the updated continuous temperature field data, the analysis results of the multiphysics correlation dataset, and other auxiliary information, a comprehensive turbine insulation condition assessment report and corresponding early warning information are generated. The assessment report may include quantitative scores for insulation performance, a visualization map of heat loss, identification of suspected corrosion areas under the insulation layer, acoustic location point identification, and an assessment of the aging status of the PTC material. The early warning information is used to indicate areas where insulation performance has significantly deteriorated, predict the location of potential corrosion under the insulation layer, or alert to potential fault points where the insulation structure is loose, thereby guiding operation and maintenance personnel to take targeted preventative or corrective measures.

[0043] II. Additional technical features of the monitoring methods; The following describes the additional technical features that can be further optimized or included in the above basic method flow.

[0044] (a) Densely wrapped sensing for special areas; As a preferred embodiment of step S11, this method adopts a differentiated data acquisition strategy for special areas on the steam turbine that are difficult to cover using traditional monitoring methods. Specifically, for the rotating component insulation jackets of the steam turbine 2 (e.g., the static insulation sleeves at the protruding parts of the high- and medium-pressure rotor journals) and complex irregular components (e.g., the split flanges in the cylinder, steam inlet valves, drain valve bodies, etc.), in the early laying stage, the self-regulating heating cable 1 is laid in a tight-wound manner, taking advantage of its softness, ability to overlap without overheating risk. The so-called tight-wound manner refers to tightly winding or attaching the self-regulating heating cable to the surface of the aforementioned special areas in a spiral, "S"-shaped, or conformal bending manner to achieve gapless coverage. Correspondingly, when acquiring distributed impedance spectrum data in step S11, the system will specifically collect and analyze the impedance spectrum data generated by these tightly-wound self-regulating heating cable sections. Thanks to the physical coverage created by the densely wrapped insulation, the calculated continuous temperature field data can extend to monitoring blind spots that traditional discrete sensors simply cannot reach, thus forming a distributed temperature sensing network without blind spots, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of a self-regulating electric heating cable wrapped around the cylinder body.

[0045] (ii) Power line carrier communication multiplexing of metal conductors for self-limiting heating cables; As an extended implementation of step S11, this method also simultaneously performs data communication multiplexing during the acquisition of distributed impedance spectrum data. Specifically, the power line carrier communication (PLC) modulation and demodulation module built into the edge smart gateway applies a high-frequency probe carrier signal to the self-regulating heating cable while simultaneously superimposing a high-frequency communication modulated carrier signal onto the self-regulating heating cable. The frequency band of this communication carrier signal (e.g., 2MHz to 30MHz) is fully separated from the frequency band of the probe carrier signal in the frequency domain to avoid mutual interference. In this way, the two parallel metal wires of the self-regulating heating cable are successfully multiplexed into a power line carrier communication bus.

[0046] Multiple passive sensing tags (such as surface acoustic wave (SAW) temperature / humidity sensors) are pre-deployed at key locations within the insulation layer (e.g., deep within flange bolt holes, areas prone to water accumulation between insulation layers, and the outer chamber of the shaft seal labyrinth seal). These tags are battery-free and operate by capturing energy from an external electromagnetic field. When a high-frequency communication modulated carrier signal transmitted on the self-regulating heating cable passes the location where the tag is attached, the tag captures the electromagnetic energy in the signal through its antenna sensing, powering its brief measurement and data transmission cycle. After the tag's built-in sensitive element detects the humidity or local temperature data at that location, it encodes the measured value and modulates the encoded information onto the metal wire of the self-regulating heating cable by backscattering it through its antenna reflection impedance, forming auxiliary sensing data that is transmitted back.

[0047] The PLC module of the edge smart gateway receives and demodulates these weak backscattered signals to extract auxiliary sensing data. This auxiliary sensing data (especially the humidity inside the insulation layer) will be fused with the distributed impedance spectrum data obtained in step S11 and used together for the calculation and correction of the continuous temperature field data in step S12.

[0048] (iii) Humidity correction for continuous temperature field data; As a preferred embodiment of the continuous temperature field data calculation in step S12, this method also includes a step of humidity correction for the calculated continuous temperature field data. Specifically, the edge smart gateway maintains a "humidity-thermal conductivity correction coefficient" lookup table, which is obtained through experimental calibration.

[0049] For the fusion correction of humidity data and distributed impedance spectrum data, this embodiment provides the following specific implementation method. When the auxiliary sensing data returned by the passive humidity sensing tag indicates that the relative humidity of a certain area within the insulation layer exceeds a preset threshold (e.g., RH>40%), the edge smart gateway retrieves the thermal conductivity correction coefficient corresponding to the current humidity value from the stored "humidity-thermal conductivity correction lookup table". Since the thermal conductivity of water is much higher than that of dry insulation cotton, the heat dissipation rate of the damp area is accelerated, resulting in the self-regulating heating cable core material being over-cooled, and its original temperature calculated directly from the impedance is too low. The edge smart gateway calculates the temperature measurement deviation ΔT caused by humidity based on the thermal conductivity correction coefficient and the heat flux density of the pipe wall under the current unit load conditions. wet and all the original temperature values ​​T in that area measured The temperature T is uniformly corrected to humidity compensation. corrected = T measured + ΔT wet This eliminates the interference of moisture intrusion on the accuracy of temperature field calculation. Furthermore, the edge smart gateway cross-validates the timing and location of humidity data triggers with the abnormal frequency band shift characteristics in the impedance spectrum caused by abrupt changes in dielectric constant. When both spatially and temporally match, the system sets the confidence level for "water ingress into the insulation layer" to the highest level, directly triggering an early warning for maintenance personnel. When only humidity data is abnormal without significant changes in the impedance spectrum, the system lowers the confidence level and continues monitoring to avoid false alarms caused by brief condensation. When the humidity label indicates a dry state (RH≤40%), no correction is triggered, and the original temperature value calculated from the impedance spectrum is directly used as the continuous temperature field data.

[0050] It should be noted that the humidity correction and the aging drift correction in step S14 work together in this embodiment, and their execution logic has a clear priority order. The aging drift correction completes the calibration of the "sensor" measurement reference by adjusting the impedance-temperature mapping relationship itself; the humidity correction, based on the updated mapping relationship after the aging correction, further deducts the abnormal heat dissipation deviation caused by water ingress into the insulation layer from the calculated temperature value. When updating the continuous temperature field data in step S15, the intermediate temperature field is first calculated using the impedance-temperature mapping relationship corrected based on the aging drift coefficient, and then the humidity correction module is called to compensate for the damp areas in the intermediate temperature field to obtain the final continuous temperature field data.

[0051] (iv) Two implementation forms of impedance-temperature mapping relationship; It should be noted that the "impedance-temperature mapping relationship" in step S12 is a higher-level concept, and its specific implementation includes two forms: impedance-temperature calibration curves and soft-sensor models. During normal operation, the edge smart gateway defaults to using a lookup table method based on the impedance-temperature calibration curve to perform daily temperature calculations. When deployment conditions are met and higher accuracy is required, the edge smart gateway can switch to a soft-sensor model-based method for temperature calculation. The acquisition and usage of these two implementations are explained below.

[0052] Obtaining the impedance-temperature calibration curve (basics of the table lookup method); The process of obtaining the impedance-temperature calibration curve. The calibration curve used in this embodiment is derived from the full-temperature-range calibration process that the self-regulating heating cable undergoes before leaving the factory. This full-temperature-range calibration refers to placing a sample of a certain length of self-regulating heating cable in a high-precision temperature-controlled chamber within the design operating temperature range of the self-regulating heating cable (e.g., from the lower limit of ambient temperature -20°C to the highest temperature that the outer surface of the turbine insulation may reach, 120°C), and maintaining a constant temperature at preset temperature intervals (e.g., every 5°C). After stabilization at each temperature point, an AC excitation signal of a specific frequency and amplitude is applied to the two parallel metal wires of the self-regulating heating cable using a precision LCR meter, and its complex impedance value at that temperature is accurately measured. The series of temperature-impedance data obtained is then subjected to curve fitting or interpolation processing to form a unique, initial impedance-temperature calibration curve data file for this batch of self-regulating heating cables. This data file is pre-written and stored in the solid-state storage unit of the edge smart gateway executing this method during the power plant deployment phase. During unit operation, the pre-stored calibration curve is invoked, and the real-time impedance value is mapped to the corresponding temperature value through lookup table interpolation, thereby completing the calculation of continuous temperature field data.

[0053] Acquisition and application of soft measurement models (LSTM method)

[0054] This implementation is an alternative to the method described above for calculating continuous temperature field data based on impedance-temperature mapping in step S12. It can be used to replace the table lookup method to address the temperature hysteresis effect unique to PTC materials and the impact of operating condition fluctuations on the accuracy of temperature calculation.

[0055] Cloud-based training of soft measurement models; A long short-term memory (LSTM) neural network model was constructed to extract historical impedance spectrum data of self-limiting electric heating cables under different operating conditions and corresponding continuous temperature field data from a multiphysics correlation dataset. The historical impedance spectrum data was used as the training set input, and the corresponding continuous temperature field data was used as the supervision label. The continuous temperature field data constituting this supervision label had undergone the aforementioned humidity correction process before being added to the training set to eliminate temperature deviations caused by non-material aging factors, ensuring that the mapping relationship learned by the model accurately reflects the impedance-temperature characteristics of the PTC material itself and its aging evolution. The model was then trained under supervision. The LSTM model, due to its unique gating mechanism (forget gate, input gate, output gate), can effectively capture the nonlinear mapping relationship between impedance and temperature, as well as the hysteresis effect unique to PTC materials. After training convergence, the big data platform distributed the model file (including network structure and weight parameters) of the soft measurement model to the edge intelligent gateway via a quantum-encrypted channel and deployed it in the inference engine of the edge intelligent gateway.

[0056] Edge-local inference in soft measurement models; During the real-time monitoring phase, the edge smart gateway inputs the currently acquired distributed impedance spectrum data and load data into the locally deployed soft measurement model to obtain a high-precision real-time temperature prediction value for the current moment, which serves as continuous temperature field data without relying on a cloud network connection. Aging compensation under soft measurement model; When the edge smart gateway uses a soft sensor model for daily temperature calculation, after receiving the aging drift coefficient set from the big data platform, it applies a translation correction to the output layer of the soft sensor model locally. Specifically, it calculates the temperature deviation compensation ΔT caused by material aging under the current service condition based on the aging drift coefficient set. drift The temperature prediction value T from the soft sensor model predicted With compensation amount ΔT drift Superimposed (i.e., temperature T after aging compensation) calibrated =T predicted + ΔT drift This allows for automatic compensation of temperature measurement deviations caused by material aging.

[0057] Aging drift modeling based on cold-state reference impedance timing; To address the slow, year-long aging process of PTC materials, this embodiment innovatively employs a cold-state reference impedance analysis method. Specifically, before each cold start of the unit (i.e., when the turbine is shut down for maintenance or in standby mode, and the self-regulating heating cable is not energized), the system automatically collects a segment of distributed impedance spectrum data and simultaneously collects the temperature of one or more reference points (e.g., the surface temperature of the main cylinder) at the location of the self-regulating heating cable. Based on the short-segment temperature coefficient of the factory calibration curve, the measured impedance value is normalized to an equivalent value at a uniform reference temperature (e.g., 25°C), thus obtaining a "normalized cold-state reference impedance value" that eliminates the influence of short-term temperature fluctuations. Arranging these reference impedance values, collected and normalized before each cold start along a time axis, forms a pure aging effect time series spanning several years.

[0058] Given that a true cold start is a relatively rare event, this embodiment introduces a supplementary sampling strategy to increase effective sampling points and improve the stability and statistical significance of long-term trend analysis. In addition to cold starts, if the system determines that the conditions for a cold start are not met, but a planned shutdown is detected, and the turbine block temperature has dropped to a preset threshold (e.g., below 80°C) and remained stable for more than 24 hours, it automatically triggers a reference impedance acquisition and normalization process—the "quasi-cold reference impedance acquisition." The normalization method for these quasi-cold sampling points is completely consistent with that for cold sampling; they only need to be marked as different sampling sources in the time series for weighted processing in subsequent analysis. By merging cold sampling points (high weight) and quasi-cold sampling points (slightly lower weight), a normalized reference impedance time series with a significantly increased sampling frequency and a more uniform time distribution can be formed, sufficient to support reliable long-term trend analysis and prediction function modeling.

[0059] The big data platform performs long-term trend analysis on this time series, extracting the aging drift pattern of the reference impedance over service time through methods such as data fitting, regression analysis, or dedicated lightweight time-series neural networks, and generating impedance drift characteristic information that quantitatively characterizes this pattern. The impedance drift characteristic information includes a set of aging drift coefficients and a prediction function describing the future drift trend. Its graphical representation is the predicted curve of the impedance drift characteristics. The set of aging drift coefficients includes the current cumulative drift amount describing the current aging state. The predicted curve reflects the irreversible resistance change law caused by microscopic mechanisms such as conductive chain breakage and polymer matrix relaxation under long-term thermo-oxidative aging of PTC materials.

[0060] Automatic compensation for impedance-temperature mapping; During normal operation, the edge smart gateway defaults to using a lookup table method for daily temperature calculation, employing the factory-calibrated impedance-temperature mapping relationship. When deployment conditions are met and higher accuracy is required, the edge smart gateway can switch to a soft-sensor model for temperature calculation. Regardless of the method used, the system periodically (e.g., after each unit startup from a cold state or monthly) triggers a second-stage aging analysis: the big data platform retrieves all normalized reference impedance values ​​up to the most recent cold state from the database, re-executes long-term trend analysis based on this updated time-series data, generates the latest impedance drift characteristic information, and sends the corresponding aging drift coefficient set to the edge smart gateway via a quantum-encrypted channel. After receiving the data, the edge smart gateway performs online compensation on the currently used impedance-temperature mapping relationship based on the coefficient set. If a lookup table method is currently used, the locally stored impedance-temperature calibration curve is shifted or its gain is adjusted, and the corrected mapping table is continuously used for daily temperature calculation in step S12. If a soft measurement model is currently used, the edge smart gateway applies a shift correction to the output layer of the soft measurement model locally. That is, it calculates the temperature deviation compensation amount based on the received aging drift coefficient set, and superimposes the temperature prediction value output by the soft measurement model with the compensation amount to obtain the final temperature value that has been automatically compensated for the effects of material aging. This automatically compensates for temperature measurement deviations caused by material aging.

[0061] In addition, the system also has an anomaly self-check function: if the latest collected normalized cold-state reference impedance value deviates from the corresponding value on the predicted curve beyond the expected threshold, it may indicate abnormal accelerated deterioration or local failure of the self-regulating heating cable, and the system will trigger a corresponding "self-regulating heating cable health status abnormal" warning. This two-stage closed-loop mechanism ensures that even if the self-regulating heating cable has been in service for many years and the material has aged significantly, the system can still maintain high-precision temperature field calculation capability, realizing the long-term reliability of the impedance-temperature mapping relationship throughout the entire life cycle of the self-regulating heating cable.

[0062] (v) The specific content of the assessment report and early warning information; As a specific implementation of step S15, the generated turbine insulation status assessment report and early warning information include a number of visualized contents and diagnostic conclusions generated by the aforementioned steps that have clear engineering guidance significance.

[0063] First, based on the updated continuous temperature field data from step S15, the edge intelligent gateway calls the visualization engine to generate a temperature distribution map of the turbine body surface. This map, in the form of a 3D model or a 2D planar unfolded diagram, visually displays the degree of heat loss on the outer surface of the high and medium pressure cylinders and main pipes of the turbine using different color levels (e.g., blue represents low-temperature areas and red represents high-temperature areas), making weak points in insulation and local overheating points immediately apparent.

[0064] Secondly, the edge smart gateway performs multi-dimensional analysis on the raw distributed impedance spectrum data obtained in step S11, extracting the impedance spectrum at each spatial location point. It then identifies anomalous characteristic frequency components caused by a sudden change in the local dielectric constant due to water ingress into the insulation layer. Combining this with the spatial location information of these anomalous components, it determines and marks suspected areas of corrosion under the insulation layer (CUI). Specifically, the edge smart gateway can send the distributed impedance spectrum data to a big data platform, which uses its spectrum analysis algorithm to detect anomalous characteristic frequency components and determine suspected CUI areas, returning the results to the edge smart gateway. The edge smart gateway highlights these areas in the 3D model of the evaluation report using a specific color (e.g., yellow). When water enters the insulation layer, the dielectric constant and thermal conductivity of the water differ significantly from those of dry insulation cotton and air. This sudden change in local physical properties significantly alters the response characteristics of the PTC material at that location to specific frequency excitation signals, manifesting as unexpected distortions or new resonance peaks in the amplitude of certain frequency bands of the impedance spectrum. The platform's algorithm detects these anomalous components using pattern recognition technology, thereby identifying and marking suspected areas of corrosion under the insulation layer (CUI), and highlighting them with a specific color (such as yellow) in the 3D model of the evaluation report.

[0065] Furthermore, the edge intelligent gateway extracts temperature change characteristics of key areas from continuous temperature field data. These key areas include, but are not limited to, the high-pressure cylinder shaft seal area and the upper and lower cylinder flange areas of the turbine. The extracted features include temperature rise rate characteristics and temperature gradient characteristics. The edge intelligent gateway performs time-series correlation analysis on the extracted temperature rise rate and temperature gradient characteristics with vibration data acquired from the DCS system during the same period, quantifying the impact of temperature changes on vibration amplitude and obtaining a quantitative indicator—causal contribution. Simultaneously, the edge intelligent gateway jointly evaluates the temperature gradient characteristics with expansion difference data acquired from the DCS system. When the temperature gradient of a certain area exceeds a preset threshold, and the corresponding expansion difference data shows an abnormal trend, the system raises the confidence level of the thermal deformation risk assessment to the highest level and issues a "high risk of cylinder thermal deformation" warning, recommending that operators reduce the load rate or strengthen warm-up. This causal contribution is used to effectively distinguish the root cause of vibration anomalies in the warning information. In cases with a high causal contribution, the warning information will prompt operators to pay attention to the warm-up rate or insulation status, that is, to distinguish between thermally induced vibration faults and purely mechanical vibration faults, thereby providing operators with differentiated and targeted handling suggestions to avoid blind operation.

[0066] To facilitate understanding, a specific calculation example for quantifying causal contribution is provided below. The high-pressure cylinder shaft seal region is selected as the analysis object, and a sliding time window of length T is set (e.g., T = 30 minutes). Within this window, the following steps are performed: The first step is to extract the temperature rise rate and temperature gradient features. The maximum temperature rise rate ΔT of the surface temperature field in the shaft seal region within the window is calculated. rate (Unit: °C / min) and maximum temperature gradient ΔT grad (That is, the difference between the highest and lowest surface temperatures in the region at the same time, in °C).

[0067] The second step is to extract the vibration variation. Within the same time window, obtain the peak-to-peak value V of the high-voltage rotor front bearing in the X direction. pp The change ΔV (the difference between the final value of the window and the initial value of the window, in μm).

[0068] The third step is to perform time-lag cross-correlation analysis. This involves analyzing the temperature gradient time series ΔT. grad We perform cross-correlation calculations between the cross-correlation function (t) and the vibration signal ΔV(t), and find the lag time τ corresponding to the maximum value of the cross-correlation function. peak If τ peak If the value is positive and the temperature change leads the vibration change, it indicates a physical causal temporal relationship between the two. Record the normalized cross-correlation peak value R at this point. max ∈[0,1].

[0069] The fourth step is to quantify the causal contribution. This involves defining the thermally induced vibration contribution index HI. thermal : HI thermal = α × (ΔT grad / ΔT ref ) + β × R max +γ × f(τ peak ); Where f() is a linear decay function, f(τ) peak ) = 1 - (τ peak / T), where T is the time window length, ΔT ref The reference temperature difference (which can be set according to the unit design data, for example, the maximum allowable temperature difference between the upper and lower cylinders during normal operation of the high-pressure cylinder is 50℃), ΔT grad = max( ΔT grad (t) ), where α, β, and γ are weighting coefficients (in this embodiment, α = 0.5, β = 0.4, and γ = 0.1). HI thermal The value range is approximately [0, 1].

[0070] Step 5: Judgment and Early Warning. When HI thermal When the value is ≥ 0.75, it is determined to be a "high causal contribution - thermally induced vibration fault," and the warning message prompts operators to pay attention to the warm-up rate or check the cylinder block insulation status; when HI thermalWhen ≤ 0.30, it is judged as "low causal contribution - pure mechanical vibration fault", and the warning message suggests checking the rotor balance or bearing condition; when 0.30 < HI thermal If the value is less than 0.75, it is considered a "mixed fault" and a comprehensive inspection is recommended.

[0071] (vi) Acoustic vibration localization function; As an extended implementation of generating early warning information in step S15, this method can be automatically triggered when an anomaly is detected. Simultaneously, this function can also be executed independently, for example, upon receiving a system-preset periodic inspection command, an externally triggered diagnostic request, or during continuous monitoring when the system identifies abnormal fluctuations in continuous temperature field data or vibration data, it can utilize the parasitic physical effects of the self-regulating heating cable to achieve acoustic vibration localization. Specifically, during the generation of the turbine insulation status assessment report and early warning information, if abnormal fluctuations in continuous temperature field data or vibration data are detected, the acoustic localization mode is triggered. The edge intelligent gateway controls its precision impedance analysis unit to switch to a high-frequency sampling mode. In this mode, the system temporarily stops applying the sweep frequency probe carrier and instead supplies a weak DC bias current to the self-regulating heating cable, and activates a high-precision, high-sampling-rate analog-to-digital converter (ADC) to continuously acquire the microvolt-level resistance fluctuation signal superimposed on the DC component at an extremely high sampling frequency (e.g., 1 million sampling points per second).

[0072] The generation mechanism of this microvolt-level resistance fluctuation signal is as follows: When there is micro-vibration or aerodynamic noise inside the insulation layer caused by damage, loosening, or airflow erosion, the core material of the self-regulating heating cable tightly attached to the pipe wall will be subjected to this mechanical vibration. Due to the piezoresistive effect of the conductive polymer composite material, when its micro-conductive chains are subjected to alternating stress, the resistance value will fluctuate slightly in sync with the stress change, and this fluctuation manifests as a microvolt-level resistance fluctuation signal.

[0073] The digital signal processor (DSP) built into the edge smart gateway performs cross-correlation analysis on two wave signals simultaneously acquired from the power supply end and the tail end of the self-regulating heating cable, accurately calculating the time difference of arrival (TDOA) of the same vibration characteristic waveform at both ends. Combining this with the acoustic wave propagation characteristic parameters of the core material of this type of self-regulating heating cable (i.e., the phase velocity of stress wave propagation in PTC material) measured and stored in advance through experiments, the distance of the vibration or noise source generating the resistance wave signal relative to the end (power supply end or tail end) of the self-regulating heating cable can be calculated using a simple time difference distance formula (distance equals wave velocity multiplied by half the propagation time difference).

[0074] The system reads the calculated distance to the location and maps the distance data onto the turbine's 3D digital model. A flashing warning icon is generated at the corresponding coordinate point on the model, thus identifying the acoustic location of loose or damaged insulation. This function greatly simplifies maintenance personnel: they no longer need to blindly remove large areas of expensive insulation for inspection, but can directly target the problem, precisely excavating near the marked point, significantly shortening maintenance time and reducing maintenance costs.

[0075] (vii) Early warning based on a group feature fingerprint database; As an advanced application of generating early warning information in step S15, this method introduces a cross-unit data comparison mechanism. The edge intelligent gateway can send the key impedance characteristic values ​​from the insulation status assessment report of this turbine to the big data platform, which then calls its maintained impedance characteristic fingerprint database for comparative analysis. This fingerprint database does not store raw impedance data, but rather stores the statistical distribution range of impedance characteristics (e.g., the average impedance modulus μ and standard deviation σ) at certain characteristic frequencies for multiple turbines of the same model, under similar operating conditions (e.g., the same load rate, similar ambient temperature), and similar service years, using the same batch of self-regulating heating cables. The big data platform returns the comparison results to the edge intelligent gateway.

[0076] During the generation of the assessment report in step S15, the system automatically compares and analyzes the key impedance characteristic values ​​in the insulation status assessment report of this steam turbine with the statistical data of the corresponding group in the impedance characteristic fingerprint database. Key impedance characteristic values ​​include long-term drift characteristics extracted from the cold-state reference impedance time series of this unit, such as the annual average drift rate or cumulative drift percentage. When the system determines that the impedance characteristics of a local area on the steam turbine (e.g., the normalized impedance value of a section of self-regulating heating cable at a reference temperature of 25°C) significantly deviate from the group statistical threshold (e.g., deviating from ±3 standard deviations of the average of similar units within the same service life), even if the external surface temperature data calculated by step S12 or step S15 for that area is still within the normal range specified by national standards, the system will determine that there is a potential, nascent anomaly in that area. At this time, the edge intelligent gateway outputs an early warning signal for corrosion under the insulation layer or settlement of the insulation cotton. The core value of this early warning mechanism lies in its leap from "passive response alarm" to "proactive predictive early warning." Its early warning signal is issued before the temperature of the outer surface of the insulation layer becomes significantly abnormal, thus providing the power plant with sufficient time to formulate maintenance plans and purchase spare parts. It is a key technical support for achieving predictive maintenance.

[0077] III. Modular architecture of the monitoring system; Please see Figure 5This embodiment also discloses a steam turbine insulation data monitoring system based on big data. This system is specifically designed to execute any of the monitoring methods described in the first and second parts of this embodiment. From the perspective of physical deployment and functional logic, the system includes a perception layer module, an edge computing layer module, and a big data platform layer module.

[0078] The sensing layer module is deployed beneath the insulation layer at the turbine and pipeline site. The core of this module is a self-regulating heating cable. This cable comprises a conductive polymer composite material with a positive temperature coefficient (PTC) and two parallel metal wires. In this system, the metal wires are designed to serve as both excitation signal transmission channels and sensing signal receiving channels. The sensing layer module is in direct contact with the surface of the monitored turbine equipment, responsible for sensing changes in the physical state along its entire length. When the edge computing layer module applies an excitation signal, the self-regulating heating cable in the sensing layer module generates a corresponding electrical response, thereby sensing the changes in distributed impedance along the entire length of the cable, providing the upper-layer modules with the most basic sensing data. In addition, the sensing layer module may also include auxiliary sensing elements such as passive sensing tags deployed within the insulation layer.

[0079] The edge computing layer module, deployed near the turbine at the power plant site (e.g., in the turbine platform's electronics room or field cabinet), is essentially one or more edge intelligent gateways. It plays a crucial bridging role. First, it acquires the distributed impedance spectrum data of the self-limiting heating cable from the sensing layer module and uses the preset impedance-temperature mapping relationship stored in its internal storage unit to perform real-time calculations, converting the raw impedance spectrum data into continuous temperature field data along the turbine body and pipe surfaces. Second, it acquires vibration, load, and differential expansion data from the turbine's distributed control system via a communication interface with the power plant's DCS system. Third, it performs data fusion, aligning the continuous temperature field data with the vibration, load, and differential expansion data using timestamps, and then using edge computing algorithms to generate a multiphysics-related dataset. The deployment location of the edge computing layer module enables it to perform real-time calculations and preprocessing of local data with extremely low latency, effectively reducing the burden on the core network and central platform. The edge computing layer module is also used to receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field associated datasets by the big data platform, correct the impedance-temperature mapping relationship based on the received impedance drift feature information, update the continuous temperature field data according to the corrected impedance-temperature mapping relationship, and generate a turbine insulation status assessment report and early warning information.

[0080] The big data platform layer module is deployed in a remote data center far from the power plant or in a server cluster using a cloud service architecture. This module is the brain of the entire system, responsible for processing massive amounts of data and executing complex intelligent algorithms. First, it receives multi-physics correlation datasets uploaded by the edge computing layer module through a quantum-encrypted channel via a network interface. Second, it utilizes its powerful computing resources to extract features from the received historical and real-time distributed impedance spectrum data and continuous temperature field data. One of its core objectives is to identify the impedance drift characteristics of conductive polymer composite materials at different aging stages. Based on the identified drift characteristics, the big data platform layer module can calculate correction parameters, and then the edge computing layer module corrects the impedance-temperature mapping relationship based on the impedance drift characteristic information. The big data platform layer module is also responsible for performing cross-unit data comparison analysis and model training tasks to achieve advanced functions such as early warning.

[0081] Example 2

[0082] This embodiment, based on the framework scheme described in Embodiment 1, takes a complete industrial site deployment and operation scenario as its background, and provides a more detailed description of the technical details, preferred parameters, alternative solutions, process requirements, and application effects of the present invention. The description in this embodiment completely corresponds to the corresponding steps and modules in Embodiment 1, supplementing its depth and breadth. The combination of the two enables those skilled in the art to implement the present invention without any doubt.

[0083] The core concept of this invention lies in breaking through the traditional technical bias of regarding self-regulating electric heating cables as merely single-function heating devices, and creatively using them as multi-physics fusion devices that integrate heating, distributed temperature sensing, data communication bus, and acoustic vibration sensing functions. Through the collaboration of edge computing and big data platforms, a comprehensive, full-lifecycle intelligent monitoring and early warning system for the thermal insulation status of steam turbines is constructed, from "point" to "line" and then to "surface".

[0084] Please refer to it again. Figure 5 This embodiment provides a preferred system deployment scheme and its corresponding workflow. The method in this embodiment can be executed by an integrated intelligent monitoring system.

[0085] I. System Hardware Architecture and Perception Layer Deployment; The system in this embodiment can be divided into three layers in terms of physical architecture: the perception layer, the edge computing layer, and the big data platform layer, which correspond to the system modules in Embodiment 1: the perception layer module, the edge computing layer module, and the big data platform layer module, respectively.

[0086] (a) Deployment of the perception layer module; The core of the sensing layer module is the self-regulating electric heating cable laid on the turbine body and related pipelines. The self-regulating electric heating cable used in this embodiment is not an ordinary self-regulating electric heating cable product on the market, but one that has been specifically designed and calibrated.

[0087] 1. Structure and characteristics of self-regulating electric heating tape: The self-regulating heating cable comprises two parallel metal conductors (typically tin-plated copper conductors) and a conductive polymer composite core sandwiched between the two conductors. This conductive polymer composite exhibits a significant positive temperature coefficient (PTC), meaning its resistivity increases with temperature and displays a "switching" characteristic of rapidly and non-linearly increasing resistivity within a specific temperature range. The core is externally coated with a high-performance fluoropolymer or polyolefin insulating sheath to withstand the harsh environment of power plants, characterized by high temperatures and the potential presence of oil and corrosive gases. In this invention, the function of the two parallel metal conductors is greatly expanded: they not only serve as power supply channels for transmitting AC power to generate Joule heating, but more importantly, they are multiplexed as both excitation signal transmission channels and sensing signal receiving channels.

[0088] 2. Impedance-temperature calibration before shipment: To ensure the accuracy of temperature sensing, each batch of self-regulating heating cables used in this embodiment undergoes a complete full-temperature-range calibration process in a controlled laboratory environment before leaving the factory. Specifically, this process involves placing a sample of self-regulating heating cable of a certain length (e.g., 100 meters) in a high-precision temperature-controlled chamber and maintaining a constant temperature at preset intervals (e.g., every 5°C) within the designed operating temperature range of the self-regulating heating cable (e.g., from -20°C to the highest temperature that the turbine surface may reach, 120°C). After stabilization at each temperature point, a specific frequency and amplitude AC excitation signal is applied to the two parallel metal wires of the self-regulating heating cable using a precision LCR meter (inductance, capacitance, and resistance meter), and the complex impedance value (including resistive and reactive components) at that temperature is accurately measured. The obtained series of temperature-impedance data are then curve-fitted to form a unique, initial impedance-temperature calibration curve for that batch of self-regulating heating cables. The impedance-temperature calibration curve data file will be encrypted and pre-stored in the solid-state storage unit of the edge smart gateway that will be deployed at the power plant site, serving as a benchmark database for subsequent temperature calculations.

[0089] 3. Differentiated installation processes for turbine components: In this embodiment, the deployment of the sensing layer is not uniform, but rather a differentiated laying strategy is adopted according to the structural characteristics and monitoring needs of different parts of the steam turbine.

[0090] A. For areas with regular pipes and large flat cylinder blocks: For the relatively flat outer surfaces of the main steam pipes, reheat steam pipes, and the upper and lower cylinders of the high-pressure and intermediate-pressure cylinders of the steam turbine, a parallel straight-line laying method is adopted. Specifically, using specialized pressure-sensitive aluminum foil tape, the self-regulating heating cable is firmly pressed onto the surface of the pipe or cylinder after the anti-corrosion coating has been applied, every 30 to 50 centimeters. A tight contact between the self-regulating heating cable and the equipment surface must be ensured, with no air gaps, to maximize heat transfer efficiency and sensing sensitivity. By applying a high-frequency detection carrier signal to the metal wire of the self-regulating heating cable and measuring its response signal, the distributed impedance spectrum data along that section of the self-regulating heating cable can be obtained in real time. Since the impedance spectrum data is a continuously changing function along the line, after calculation, continuous temperature field data along the surface of the pipe or cylinder can be obtained, rather than the discrete point temperature of a traditional thermocouple.

[0091] B. For areas with thermal insulation jackets for rotating components and complex, irregularly shaped components: This is the key technological step in solving the monitoring "blind spot" problem of this invention. For rotating components of a steam turbine, such as the journal extension of the high-pressure rotor, although the rotor itself is rotating at high speed, its external insulation jacket is stationary. Traditional temperature sensing elements are prone to failure due to difficult wiring and vibration environments, creating monitoring blind spots. This embodiment utilizes the characteristics of self-regulating heating tape—soft, flexible, and capable of overlapping without the risk of overheating—to tightly wrap it around the inner wall of the stationary insulation jacket. Specifically, the self-regulating heating tape is tightly coiled in a spiral or "S" shape around the inner wall of the jacket, ensuring that the entire circumference of the shaft seal area is covered by the self-regulating heating tape.

[0092] Similarly, for complex and irregularly shaped components, such as cylinder split flanges, steam inlet valves, and drain valve bodies, where there are many protrusions, depressions, and bolts, traditional rigid or large-bending-radius temperature sensing elements cannot fit properly. In this embodiment, the construction personnel will bend and wrap the self-regulating electric heating cable along the complex contours of the flange edge and valve neck, and use high-temperature resistant cable ties or metal wires for auxiliary fixation.

[0093] Through the aforementioned close-wrap and conformal laying methods, the self-regulating heating cable achieves seamless wrapping or close following of the measured surface in physical space, thereby constructing a distributed temperature sensing network without blind spots. This enables the subsequently calculated continuous temperature field data to cover monitoring blind spots and irregular curved surfaces that traditional discrete sensors simply cannot reach, greatly improving the spatial resolution and completeness of temperature monitoring.

[0094] 4. The perception layer serves as a multiplexed structure for power line carrier communication buses: To address the challenge of laying numerous signal cables in the high-temperature, high-electromagnetic-interference environment of steam turbines, this invention further explores the communication potential of self-regulating heating cables. In the sensing layer deployment of this embodiment, the two parallel metal wires of the self-regulating heating cable are designed to be multiplexed as a power line carrier communication bus.

[0095] The specific implementation is as follows: At the power input terminal of the self-regulating heating cable (usually located inside the junction box), in addition to connecting the 220V or 380V AC power supply line, a high-frequency coupler / signal isolator couples the high-frequency communication signal line from the edge smart gateway to the same pair of metal conductors. This high-frequency communication modulation carrier signal, the power supply signal, and the high-frequency sensing carrier signal used for impedance measurement coexist on the same pair of conductors using frequency division multiplexing technology, without interference. For example, the power supply frequency is 50Hz, the impedance sensing carrier can be selected from the 10kHz to 100kHz frequency band, while the communication modulation carrier can be selected from a higher frequency band of 2MHz to 30MHz.

[0096] Multiple passive sensor tags are pre-embedded or affixed to key locations within the insulation layer, such as deep within the bolt holes of the cylinder split flange, areas prone to water accumulation between insulation layers, and the outer cavity of the shaft seal labyrinth seal—locations difficult to monitor externally. These passive sensor tags can be, for example, surface acoustic wave (SAW) temperature / humidity sensors, which do not contain batteries and operate using piezoelectric substrates and interdigital transducers. When the high-frequency communication modulated carrier signal transmitted on the self-regulating heating cable passes the location where the tag is attached, the tag's antenna captures the energy in the signal through electromagnetic induction, providing power for its brief measurement and data transmission cycle. After the tag's built-in sensitive element senses the humidity or local temperature data at that location, it encodes the measured value and backscatters the encoded information onto the metal wire of the self-regulating heating cable by changing the antenna's reflection impedance. This weak backscattered signal returns along the original path of the self-regulating heating cable and is demodulated and extracted by the high-sensitivity receiving circuit of the edge smart gateway.

[0097] In this embodiment, the passive sensing tag includes a miniature antenna or coupling coil capable of sensing specific high-frequency signals. The passive sensing tag is attached to the interior of the insulation layer using a high-temperature resistant adhesive, with its antenna positioned parallel to the self-regulating heating cable. The PLC module of the edge smart gateway emits a high-frequency communication modulation carrier wave, which is transmitted along the metal wire of the self-regulating heating cable and received by the antenna of the passive tag via near-field coupling, providing operating power to the passive sensing tag. After the humidity sensor inside the passive sensing tag completes its measurement, it modulates the received high-frequency communication modulation carrier wave by changing its internal load impedance. The weak modulated signal is then transmitted back through the metal wire or spatial coupling and demodulated by the high-sensitivity receiving circuit of the edge smart gateway to obtain humidity data. This communication process is completely isolated from the low-frequency impedance detection signal in frequency and separated at the edge smart gateway side by a filter bank.

[0098] The advantage of this innovative design lies in the fact that acquiring auxiliary sensing data completely eliminates the need for separate power and signal lines for these tags, thus completely solving the pain points of cumbersome wiring and low reliability in traditional solutions. This auxiliary sensing data, especially the humidity data inside the insulation layer, will be fused with the surface temperature data calculated through impedance at the edge computing layer, and used together for humidity correction of the continuous temperature field data in step S12. When calculating the temperature deviation caused by humidity, load data obtained from the multiphysics correlation dataset will also be introduced to estimate the pipe wall heat flux density in real time, thereby more accurately determining the compensation amount and improving the accuracy of predicting corrosion risk under the insulation layer.

[0099] (II) Configuration and functions of the edge computing layer module; The edge computing layer module is centered around one or more edge intelligent gateways deployed near the steam turbine. An edge intelligent gateway is an industrial-grade embedded computer with powerful local data processing, protocol conversion, and intelligent analysis capabilities.

[0100] 1. Acquisition of distributed impedance spectrum data and calculation of continuous temperature field: The edge smart gateway incorporates a precision impedance analysis unit. This unit, via the aforementioned high-frequency coupler, periodically applies a low-power, high-frequency probe carrier signal (e.g., a frequency sweep from 1kHz to 100kHz, with a voltage amplitude far lower than the rated operating voltage of the self-regulating heating cable to avoid affecting the heating function and temperature measurement accuracy) to the self-regulating heating cable circuit of the sensing layer. The impedance analysis unit simultaneously acquires the response current signal flowing through the self-regulating heating cable conductor and the response voltage signal at both ends of the conductor. Through Fourier transform or vector voltage-current ratio calculation, it calculates the distributed impedance spectrum data of the self-regulating heating cable circuit at multiple frequency points in real time.

[0101] After acquiring real-time impedance spectrum data, the edge smart gateway's central processing unit (CPU) retrieves the impedance-temperature calibration curve pre-stored in solid-state storage. Using a lookup table or interpolation algorithm, the measured impedance values ​​(typically the resistance component or impedance magnitude at a specific frequency) are mapped to the corresponding temperature values. Since the impedance-temperature mapping relationship of PTC materials characterizes the nonlinear relationship between its resistivity and temperature, the calculation process can accurately invert the temperature at various points along the self-limiting heating cable. By analyzing the continuous impedance changes along the route, continuous temperature field data covering the entire laying area is ultimately generated.

[0102] 2. Generation of multiphysics correlation datasets: Another key function of the edge smart gateway is to achieve the fusion of multi-source heterogeneous data. The edge smart gateway establishes a data interface with the distributed control system (DCS) of the steam turbine through standard industrial communication protocols (such as Modbus TCP, OPC-UA, Profinet, etc.) to acquire steam turbine operating parameters closely related to the insulation status in real time, including but not limited to: vibration data (taken from eddy current sensors or acceleration sensors installed on the bearing housing), load data (power generation or main steam flow), and expansion difference data (relative expansion difference between the cylinder and the rotor).

[0103] The edge smart gateway's built-in time synchronization module (synchronizing with the plant clock via NTP or PTP protocol) accurately timestamps all received data. Subsequently, the edge smart gateway executes an edge computing fusion algorithm to align the timestamps of continuous temperature field data from the same moment or within the same short time window with vibration, load, and expansion difference data acquired by the DCS, and restructure the data to generate a multi-dimensional multiphysics correlation dataset. The core value of this dataset lies in characterizing the temporal correlation between the turbine's thermodynamic state (reflected by the temperature field) and mechanical state (reflected by vibration and expansion difference). Simultaneously, the edge computing module jointly evaluates temperature gradient characteristics and expansion difference data. When synchronization between the two is abnormal, a "high-confidence thermal deformation risk" marker is added to the generated multiphysics correlation dataset, providing a basis for early warning generation in subsequent step 5.

[0104] 3. Edge processing for accessibility functions: For multiplexing power line carrier communication functions, the edge smart gateway integrates a power line carrier communication (PLC) modulation and demodulation module. This module is responsible for generating a high-frequency communication modulation carrier signal and superimposing it onto the self-limiting heating tape, while simultaneously receiving and demodulating auxiliary sensing data backscattered from passive sensing tags.

[0105] For acoustic positioning, the impedance analysis unit of the edge smart gateway features a high-frequency sampling mode. During the generation of evaluation reports and early warning information, if abnormal fluctuations are detected in continuous temperature field data or vibration data, the acoustic positioning mode is triggered. In this mode, the system temporarily stops applying the sweep frequency probe carrier and instead introduces a weak DC bias current into the self-regulating heating cable, activating a high-precision analog-to-digital converter (ADC) to continuously acquire the microvolt-level AC resistance fluctuation signal superimposed on the DC component at an extremely high sampling rate (e.g., 1 million sampling points per second). The generation mechanism of this microvolt-level fluctuation signal is as follows: when there is micro-vibration or aerodynamic noise inside the insulation layer caused by damage, loosening, or airflow erosion, the core material of the self-regulating heating cable, which is in close contact with the pipe wall, will be subjected to this mechanical vibration. Due to the piezoresistive effect of the conductive polymer composite material, its microscopic conductive chains will undergo a small change in resistance value when subjected to stress; this change manifests as a resistance fluctuation signal. The digital signal processor (DSP) of the edge smart gateway performs cross-correlation analysis on the two acquired signals (one from the power supply end of the self-regulating heating cable and the other from the tail end) to accurately calculate the time difference of arrival (TDOA) of the same vibration characteristic waveform at both ends. Combined with the pre-determined acoustic wave propagation characteristic parameters of the core material of this type of self-regulating heating cable (i.e., the phase velocity of the stress wave propagating in the PTC material), the distance from the vibration or noise source generating the resistance fluctuation signal to the power supply end or tail end of the self-regulating heating cable can be calculated using the simple time difference distance formula: distance = (propagation time difference × wave velocity) / 2. This location distance information will be timestamped and used to identify the acoustic location point when generating the evaluation report in step S15.

[0106] (III) Configuration and functions of big data platform layer modules; The big data platform layer module is the "brain" of the system, typically deployed in a data center or a reliable cloud server cluster. It is responsible for receiving data uploaded from edge intelligent gateways of various power plants and multiple generating units, and performing in-depth data mining and model training on massive amounts of data.

[0107] 1. Secure data transmission: The data communication channel between the edge computing layer and the big data platform layer employs a quantum-encrypted communication channel. Quantum Key Distribution (QKD) terminal devices are deployed at both the power plant and central computer room sides. Before each data transmission, both parties negotiate a completely random, one-time pad, and theoretically uneavesdroppable symmetric key via the quantum channel. The multiphysics correlation dataset generated by the edge intelligent gateway is encrypted at high speed using this quantum key before uploading. The encrypted data is then transmitted to the big data platform via traditional classical networks (such as fiber optic leased lines or Virtual Private Networks, VPNs). Even if the classical network channel is intercepted, the data content cannot be deciphered due to the absolute security of the encryption key. This ensures the security of turbine operating status data, especially sensitive information involving insulation defects and vibration characteristics, which is critical infrastructure.

[0108] 2. Acquisition and mapping correction of impedance drift characteristic information: After receiving massive amounts of historical data, the big data platform will initiate an offline training task. It should be noted that in this embodiment, the temperature calculation in step S12 defaults to a lookup table method based on the impedance-temperature calibration curve; when deployment conditions are met and high accuracy is required, the edge smart gateway can switch to a soft measurement model for temperature calculation. The following sections explain the training and edge deployment of the soft measurement model, the acquisition of impedance drift characteristic information, and the collaborative method between the two in aging compensation.

[0109] (1) Cloud training and edge deployment of soft measurement models (optional high-precision temperature calculation method); First, the big data platform constructs a Long Short-Term Memory (LSTM) neural network model to achieve higher-precision real-time mapping from impedance and load data to temperature. The training dataset is extracted from a multi-physics correlation dataset: the input consists of historical impedance spectrum snapshots of self-regulating electric heating cables under typical operating conditions such as unit start-up and shutdown, full-load steady-state operation, and peak-shaving load shifting, along with synchronously collected load data; the supervision label is continuous temperature field data for the corresponding time period, calculated by the edge smart gateway using factory calibration curves and with known errors eliminated. Due to its unique gating mechanism (forget gate, input gate, output gate), the LSTM model can effectively capture the nonlinear mapping relationship between impedance, load, and temperature, as well as the temperature hysteresis effect unique to PTC materials. After training convergence, the big data platform distributes the model file (including network structure and weight parameters) of the soft measurement model to the edge smart gateway via a quantum-encrypted channel and deploys it in the local inference engine of the edge smart gateway. In daily monitoring, the edge smart gateway inputs the distributed impedance spectrum data and load data acquired in real time into the locally deployed soft measurement model to obtain a high-precision real-time temperature prediction value for the current moment. As continuous temperature field data, it does not rely on cloud network connection.

[0110] (2) Acquisition of impedance drift characteristic information - Aging drift modeling based on cold reference impedance timing; To address the slow, year-long aging process of PTC materials, this embodiment employs a cold-state reference impedance analysis method. Specifically, before each cold start of the unit (i.e., when the turbine is shut down for maintenance or in standby mode, and the self-regulating heating cable is not energized), the system automatically collects a segment of distributed impedance spectrum data and simultaneously collects the temperature of one or more reference points (e.g., the surface temperature of the main cylinder) at the location of the self-regulating heating cable. Based on the short-segment temperature coefficient of the factory calibration curve, the measured impedance value is normalized to an equivalent value at a uniform reference temperature (e.g., 25°C), thus obtaining a "normalized cold-state reference impedance value" that eliminates the influence of short-term temperature fluctuations. Arranging these reference impedance values, collected and normalized before each cold start along a time axis, forms a pure aging effect time series spanning several years.

[0111] Given that a true cold start is a relatively rare event, this embodiment introduces a supplementary sampling strategy to increase effective sampling points and improve the stability and statistical significance of long-term trend analysis. In addition to cold starts, if the system determines that the conditions for a cold start are not met, but a planned shutdown is detected, and the turbine block temperature has dropped to a preset threshold (e.g., below 80°C) and remained stable for more than 24 hours, it automatically triggers a reference impedance acquisition and normalization process—the "quasi-cold reference impedance acquisition." The normalization method for these quasi-cold sampling points is completely consistent with that for cold sampling; they only need to be marked as different sampling sources in the time series for weighted processing in subsequent analysis. By merging cold sampling points (high weight) and quasi-cold sampling points (slightly lower weight), a normalized reference impedance time series with a significantly increased sampling frequency and a more uniform time distribution can be formed, sufficient to support reliable long-term trend analysis and prediction function modeling.

[0112] The big data platform performs long-term trend analysis on this time series, extracting the aging drift pattern of the reference impedance over service time through methods such as data fitting, regression analysis, or dedicated lightweight time-series neural networks, and generating impedance drift characteristic information that quantitatively characterizes this pattern. The impedance drift characteristic information includes a set of aging drift coefficients and a prediction function describing the future drift trend. Its graphical representation is the predicted curve of the impedance drift characteristics. The set of aging drift coefficients includes the current cumulative drift amount describing the current aging state. The predicted curve reflects the irreversible resistance change law caused by microscopic mechanisms such as conductive chain breakage and polymer matrix relaxation under long-term thermo-oxidative aging of PTC materials.

[0113] (3) Local implementation of aging compensation; During normal operation, the edge smart gateway defaults to using a lookup table method for daily temperature calculations, employing the factory-calibrated impedance-temperature mapping. When deployment conditions are met and higher accuracy is required, the edge smart gateway can switch to a soft-sensor model for temperature calculations. Regardless of the method used, the system periodically (e.g., after each unit startup from a cold state or monthly) triggers aging analysis: the big data platform retrieves all normalized reference impedance values ​​up to the most recent cold state from the database, re-executes long-term trend analysis based on this updated time-series data, generates the latest impedance drift characteristic information, and sends the corresponding aging drift coefficient set to the edge smart gateway via a quantum-encrypted channel. Upon receiving this data, the edge smart gateway performs online compensation on the currently used impedance-temperature mapping based on the aging drift coefficient set.

[0114] When the edge smart gateway uses the lookup table method, it directly performs an overall shift or gain adjustment on the locally stored impedance-temperature calibration curve. The corrected mapping table is continuously used for daily temperature calculation, thereby automatically compensating for temperature measurement deviations caused by material aging.

[0115] When the edge smart gateway uses a soft measurement model, this embodiment specifically adopts a "result correction" strategy: that is, keeping the internal structure and parameters of the soft measurement model deployed locally on the edge smart gateway unchanged, the edge smart gateway calculates the temperature deviation compensation amount ΔT caused by material aging under the current service state based on the received aging drift coefficient set. drift The temperature prediction value T from the soft sensor model predicted With compensation amount ΔT drift Superimposed (i.e., temperature T after aging compensation) calibrated = T predicted + ΔT drift This allows for automatic compensation of temperature measurement deviations caused by material aging.

[0116] In addition, the system also has an anomaly self-check function: if the latest collected normalized cold-state reference impedance value deviates from the corresponding value on the predicted curve beyond the expected threshold, it may indicate abnormal accelerated deterioration or local failure of the self-regulating heating cable, and the system will trigger a corresponding "self-regulating heating cable health status abnormal" warning. This closed-loop mechanism ensures that even if the self-regulating heating cable has been in service for many years and the material has aged significantly, the system can still maintain high-precision temperature field calculation capability, realizing the long-term reliability of the impedance-temperature mapping relationship throughout the entire life cycle of the self-regulating heating cable.

[0117] II. System workflow and typical application scenarios; The following detailed steps illustrate the workflow of the system in this embodiment during a complete monitoring cycle.

[0118] (I) Real-time monitoring and data acquisition phase; After the power plant's steam turbine is put into operation, the sensing layer and edge computing layer operate continuously. The edge smart gateway periodically (e.g., every 10 seconds) applies a high-frequency probe carrier signal to the self-regulating heating cable to acquire real-time distributed impedance spectrum data. The moment the data is acquired, the edge smart gateway CPU immediately calls the initial impedance-temperature calibration curve in the storage unit and quickly converts the impedance data into continuous temperature field data.

[0119] Meanwhile, accessibility functions are executed in parallel: The PLC communication module maintains periodic polling communication with the passive sensor tags inside the insulation layer to acquire auxiliary sensing data such as humidity and local temperature.

[0120] When the system detects abnormal fluctuations in continuous temperature field data or DCS vibration data, it triggers an acoustic localization mode. The edge intelligent gateway controls the self-limiting heating tape to briefly switch to DC bias mode, initiates high-frequency sampling to capture microvolt-level resistance fluctuation signals, and immediately calculates the location distance of the sound source / vibration source.

[0121] (II) Edge Data Fusion and Analysis Stage; The edge intelligent gateway continuously acquires vibration, load, and differential expansion data of the steam turbine from the DCS system. Upon receiving a new frame of temperature field data, the data fusion module immediately performs timestamp alignment to generate a data packet of a multiphysics-related dataset.

[0122] Next, the edge computing module performs the following analysis tasks: Threshold alarm: Determines whether the temperature at any location in the continuous temperature field data exceeds a preset safety threshold (e.g., the maximum allowable temperature of the outer surface of the insulation as specified in national standards). If so, a local audible and visual alarm is immediately generated.

[0123] Key area feature extraction: Continuously calculate the temperature rise rate and temperature gradient characteristics of key areas such as high-pressure cylinder shaft seals and flanges.

[0124] Causal contribution calculation: The above characteristics are correlated with vibration data to dynamically calculate the causal contribution of thermal deformation to vibration.

[0125] (III) The stage of collaborative cloud-based in-depth analysis and edge-based local diagnostics on big data platforms; The edge intelligent gateway packages all data, including multiphysics correlation datasets, auxiliary sensor data, and acoustic positioning distance information, and uploads them to the big data platform via a quantum-encrypted channel. This stage adopts a collaborative architecture of "cloud analysis, edge execution," with the edge intelligent gateway acting as the execution entity, responsible for initiating all analysis tasks and generating the final report. For computationally intensive sub-tasks, the edge intelligent gateway obtains the results by calling the cloud analysis service of the big data platform.

[0126] First, the edge intelligent gateway initiates a cloud analytics service request to the big data platform via a quantum-encrypted channel, uploading the multiphysics correlation dataset. The big data platform responds to the request and executes the following cloud analytics tasks: Data parsing and storage: Decrypt data packets and store the data in a distributed database according to dimensions such as unit, time, and measurement point.

[0127] Aging drift coefficient set query and distribution: The big data platform queries the latest generated "aging drift coefficient set", obtains the cumulative offset of the reference impedance corresponding to the current service time, and returns the aging drift coefficient set to the edge smart gateway.

[0128] Distributed impedance spectrum anomaly frequency band analysis and CUI identification: The big data platform performs spectrum analysis on the uploaded distributed impedance spectrum data to find abnormal characteristic frequency components caused by water ingress into the insulation layer, determines and identifies suspected areas of corrosion under the insulation layer (CUI), and returns the CUI risk area information to the edge smart gateway.

[0129] Cross-unit fingerprint comparison and early warning: The big data platform maintains an impedance characteristic fingerprint database for similar units, storing the statistical distribution range of impedance characteristics for steam turbines of the same model and self-regulating heating cables of the same batch under similar operating conditions and service years. The big data platform automatically performs comparative analysis on the impedance characteristics of the unit. When it determines that the impedance characteristics of a local area deviate from the group statistical threshold, it generates an early warning trigger signal and returns the comparison results and the early warning trigger signal to the edge intelligent gateway.

[0130] Subsequently, after receiving the aging drift coefficient set, CUI risk area information, fingerprint comparison results, and early warning trigger signals returned from the cloud, the edge intelligent gateway continues to perform the following real-time diagnostic and comprehensive summary tasks locally: Temperature Field Correction and Update: The edge smart gateway generates and compensates for continuous temperature field data locally. If a lookup table method is currently used, the edge smart gateway first uses the locally stored impedance-temperature calibration curve to calculate the distributed impedance spectrum data into the original temperature field. Then, it applies the received aging drift coefficient set to perform an overall shift or gain adjustment on the calibration curve data table and updates the continuous temperature field data using the corrected mapping table. If a soft sensor model has been deployed, the edge smart gateway inputs the real-time distributed impedance spectrum data and load data into the locally deployed soft sensor model to obtain the temperature prediction value. Then, it calculates the temperature deviation compensation amount based on the received aging drift coefficient set, performs a shift correction on the temperature prediction value, and obtains an accurate temperature field that has automatically compensated for the effects of material aging.

[0131] Comprehensive condition assessment and map generation: Based on the updated continuous temperature field data, the local visualization engine is invoked to generate a surface temperature distribution map of the turbine body, which visually displays the degree of heat loss in different regions in the form of a three-dimensional model or a two-dimensional unfolded diagram.

[0132] Based on the CUI risk area information returned from the cloud, it is highlighted in the 3D model of the assessment report with a specific color (such as yellow).

[0133] The temperature rise rate and temperature gradient characteristics of key areas in continuous temperature field data are extracted and correlated with vibration data of the same period to quantify the causal contribution, which is used to distinguish between thermally induced vibration faults and purely mechanical vibration faults.

[0134] At the same time, combining expansion difference data for joint assessment improves the accuracy of thermal deformation risk assessment.

[0135] Read the locally calculated acoustic positioning distance results, call the three-dimensional digital model of the steam turbine, mark the acoustic positioning points where the insulation layer is loose or damaged at the corresponding coordinate points on the model, and highlight them by flashing.

[0136] Summary of final assessment report and early warning information: The edge intelligent gateway integrates local real-time diagnostic results (temperature spectrum, causal contribution, acoustic location points) with cloud-returned analysis results (CUI risk areas, cross-unit comparison results) to generate a unified "Comprehensive Assessment Report on Turbine Insulation Status." Based on early warning trigger signals returned from the cloud, it outputs early warnings of corrosion under the insulation layer or settlement of the insulation cotton. The report is pushed to operation and maintenance personnel via local HMI, web interface, or mobile application. The report content includes: Overall thermal insulation performance rating and trend chart of the unit.

[0137] High-resolution heat loss map and list of key overheated areas.

[0138] Corrosion Under Insulation (CUI) Risk Area Identification Map and Risk Level.

[0139] Precise location indication of insulation loosening / damage points based on acoustic positioning.

[0140] Recommendations for operators (e.g., warm-up curve optimization recommendations based on causal contribution).

[0141] Based on the maintenance personnel's suggestions (e.g., suggesting that the insulation layer at flange number XX be thoroughly excavated and inspected during the next shutdown).

[0142] The warning information list distinguishes between real-time over-temperature alarms and early risk warnings based on big data comparison.

[0143] (iv) Examples of specific application scenarios; Scenario 1: Monitoring of thermal stress during the cold start-up of a steam turbine; During the cold start-up of a steam turbine, controlling the warm-up rate is crucial. Traditional operation relies solely on the temperature measurement point of the inner cylinder. With this invention, the edge intelligent gateway calculates the temperature rise rate and temperature gradient characteristics of the high-pressure cylinder flange area in real time. When the system detects an excessive temperature difference between the inner and outer walls of the flange, and the causal contribution analysis shows a strong correlation between increased unit vibration and thermal deformation, the edge intelligent gateway generates an early warning and sends it to the operators via the local HMI: "At the current warm-up rate, the thermal stress level of the high-pressure cylinder flange is high, contributing 85% to shaft vibration. Simultaneously, the expansion difference data shows an abnormal trend, indicating a high confidence level of thermal deformation risk. It is recommended to appropriately extend the medium-speed warm-up time and closely monitor changes in the expansion difference." Based on this, operators can make precise operational adjustments to avoid thermal bending accidents.

[0144] Scenario 2: Early detection of corrosion under insulation (CUI); Three years after a certain generating unit began operation, the edge intelligent gateway, during its routine assessment report generation, invoked the cross-unit fingerprint comparison service of the big data platform. The comparison results returned by the big data platform showed that the impedance characteristics of the self-regulating heating cable in a section of pipeline below the intermediate-pressure cylinder of the unit, at a reference temperature of 25°C, drifted downwards by 12% compared to the statistical average of units in the same batch, accompanied by an early warning trigger signal. However, the surface temperature spectrum of this area showed no obvious abnormalities. Based on the returned results, the edge intelligent gateway identified this as a suspected CUI (Constant Intake) area in the assessment report. Combined with auxiliary sensing data from passive humidity sensor tags deployed within the insulation layer of this area, the humidity reached 80% RH (relative humidity) at three different times. During the subsequent overhaul, maintenance personnel precisely removed this section of insulation based on the acoustic positioning points provided by the platform's 3D model. Inspection revealed that a minor tear in the external aluminum cladding allowed rainwater to seep in, causing the insulation cotton to become damp, and slight pitting corrosion had appeared on the outer wall of the pipeline. Because it was discovered in time, only surface rust removal was required, thus avoiding a potential pipeline leak and unplanned downtime.

[0145] Scenario 3: Acoustic location of steam leakage in the shaft seal area; The on-duty personnel heard a slight abnormal noise at the front shaft seal of the high-pressure cylinder of the steam turbine, suspecting a steam leak, but were unable to pinpoint the exact location. Because the area was covered by a thick insulation layer, traditional methods were ineffective for inspection. Engineers remotely activated the acoustic positioning function of the unit's edge intelligent gateway. The system activated high-frequency sampling mode, capturing a microvolt-level resistance fluctuation signal at a specific frequency band generated by steam eroding the insulation cotton, and quickly calculated the vibration source's location as "23.5 meters from the power supply end of the self-regulating heating cable." Mapping this distance onto the steam turbine's 3D model, the system marked the acoustic location point of loose or damaged insulation at approximately the 2 o'clock position on the shaft seal. Maintenance personnel precisely removed the small piece of insulation, discovering that a fixing hook had fallen off, causing a cavity to be blown out of the insulation cotton by steam. The defect was quickly located and repaired.

[0146] As can be seen from the detailed description of the above embodiments, the present invention has revolutionized the function of the self-regulating electric heating tape and deeply integrated it with technologies such as edge computing, big data analysis, and quantum communication. This effectively overcomes the shortcomings of existing technologies, such as numerous monitoring blind spots, weak data correlation, and poor early warning capabilities, and significantly improves the safety, reliability, and economy of power plant operation.

[0147] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for monitoring steam turbine insulation data based on big data, characterized in that, The method includes the following steps: Step 1: For the turbine body and pipelines that have been equipped with self-regulating electric heating tape, obtain the distributed impedance spectrum data of the self-regulating electric heating tape. Step 2: Based on the impedance-temperature mapping relationship, the distributed impedance spectrum data is solved into continuous temperature field data along the turbine body and pipe surface; Step 3: Obtain the operating data from the distributed control system of the steam turbine, align the continuous temperature field data with the operating data using timestamps, and generate a multiphysics correlation dataset; the operating data includes vibration data, load data, and differential expansion data; Step 4: Receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field correlation datasets by the big data platform; correct the impedance-temperature mapping relationship based on the received impedance drift feature information; Step 5: Update the continuous temperature field data according to the corrected impedance-temperature mapping relationship, and generate a turbine insulation status assessment report and early warning information.

2. The method for monitoring steam turbine insulation data based on big data according to claim 1, characterized in that, In step 1, the distributed impedance spectrum data of the self-regulating electric heating tape is obtained, which specifically includes the following steps: For the thermal insulation jacket of the rotating components and the irregularly shaped component areas of the steam turbine, the distributed impedance spectrum data generated by the self-regulating electric heating tape laid in a tightly wound manner is obtained to form a distributed temperature sensing network without blind spots.

3. The method for monitoring steam turbine insulation data based on big data according to claim 2, characterized in that, In step 1, the distributed impedance spectrum data of the self-regulating electric heating cable is obtained by applying a high-frequency probe carrier signal to the self-regulating electric heating cable and obtaining the distributed impedance spectrum data by detecting the response signal. The self-regulating heating cable is installed in the insulation layer of the turbine body and pipeline. The self-regulating heating cable includes a conductive polymer composite material with positive temperature coefficient characteristics and two parallel metal wires. The metal wires are used as the excitation signal transmission channel and the sensing signal receiving channel. During the acquisition of distributed impedance spectrum data, a high-frequency communication modulation carrier signal is superimposed on the self-regulating heating cable to multiplex the self-regulating heating cable into a power line carrier communication bus; the high-frequency communication modulation carrier signal and the high-frequency detection carrier signal are separated in the frequency domain to avoid mutual interference.

4. The method for monitoring steam turbine insulation data based on big data according to claim 3, characterized in that, Step 2 also includes a continuous temperature field data correction step, which is detailed below: It receives auxiliary sensing data transmitted back by passive sensing tags deployed at preset key locations within the insulation layer via backscattering. The passive sensing tags acquire working energy by sensing high-frequency communication modulated carrier signals, and the auxiliary sensing data includes humidity data or local temperature data inside the insulation layer. When the auxiliary sensing data returned by the passive humidity sensor tag indicates that the relative humidity in a certain area of ​​the insulation layer exceeds the preset threshold, the temperature measurement deviation caused by the current humidity is calculated using the auxiliary sensing data, and then the temperature measurement deviation caused by humidity is used to correct each data point in the continuous temperature field data.

5. The method for monitoring steam turbine insulation data based on big data according to claim 4, characterized in that, In step 4, the big data platform extracts and analyzes features from the multiphysics correlation dataset to obtain impedance drift feature information. Based on the received impedance drift feature information, the impedance-temperature mapping relationship is corrected, specifically including the following steps: Before each cold start of the unit, the distributed impedance spectrum data are collected under stable ambient temperature and when the self-limiting heating cable is not heated. Based on the synchronously collected reference temperature, the impedance value is normalized to a unified reference temperature and used as the reference impedance value. Arrange the normalized reference impedance values ​​along the time axis to form a cold reference impedance time series. Long-term trend analysis was performed on the cold-state reference impedance time series to extract the aging drift law of the reference impedance of conductive polymer composite material with service time, and impedance drift characteristic information was obtained. Based on impedance drift characteristics, the correction amount of the impedance-temperature mapping relationship under the current aging state is calculated, and the correction amount is applied to the impedance-temperature mapping relationship to automatically compensate for the temperature measurement deviation caused by material aging, so as to realize the correction of the impedance-temperature mapping relationship.

6. The method for monitoring steam turbine insulation data based on big data according to claim 5, characterized in that, In step 5, a turbine insulation status assessment report and early warning information are generated, which specifically includes the following steps: A temperature distribution map of the turbine body surface is generated based on continuous temperature field data; the temperature distribution map presents the degree of heat loss in different regions in a visual manner. Spectral analysis was performed on the distributed impedance spectrum data to obtain abnormal characteristic frequency components, and suspected areas of corrosion under the insulation layer were identified and marked. The abnormal characteristic frequency components are caused by the sudden change in local thermal conductivity due to water ingress into the insulation layer, which manifests as amplitude distortion in a specific frequency band in the impedance spectrum. Extract the temperature rise rate and temperature gradient characteristics of key regions from the continuous temperature field data; the key regions include the high-pressure cylinder shaft seal area and the upper and lower cylinder flange areas of the steam turbine; perform correlation analysis between the temperature rise rate characteristics, temperature gradient characteristics and vibration data of the same period to quantify the causal contribution of abnormal mechanical vibration of the steam turbine caused by thermal deformation. Based on the causal contribution, the current fault is determined to be either a thermally induced vibration fault or a purely mechanical vibration fault, and differentiated handling suggestions are provided to the operators.

7. The method for monitoring steam turbine insulation data based on big data according to claim 6, characterized in that, In step 5, generating the turbine insulation status assessment report and early warning information also includes: During the generation of turbine insulation status assessment report and early warning information, if abnormal fluctuations are detected in continuous temperature field data or vibration data, the acoustic positioning mode is triggered, DC bias current is supplied to the self-regulating heating cable, and high-frequency sampling is started to capture microvolt-level resistance fluctuation signals. The microvolt-level resistance fluctuation signals are generated by the piezoresistive effect of conductive polymer composite materials and are used to characterize the micro-vibration state of the pipe wall or equipment surface to which the self-regulating heating cable is attached. Based on the arrival time difference of the microvolt-level resistance fluctuation signal to the power supply end and tail end of the self-regulating heating cable, and combined with the acoustic wave propagation characteristic parameters of the conductive polymer composite material, the position distance of the vibration or noise source that generates the microvolt-level resistance fluctuation signal relative to the end of the self-regulating heating cable is calculated. Based on the calculated location distance, acoustic positioning points for loose or damaged insulation layers are marked on the three-dimensional model of the steam turbine. These acoustic positioning points are used to guide maintenance personnel to accurately excavate the insulation layer for repair work.

8. The method for monitoring steam turbine insulation data based on big data according to claim 7, characterized in that, Step 5 further includes: The turbine insulation status assessment report was compared and analyzed with the impedance characteristic fingerprint database of similar units; the impedance characteristic fingerprint database stores the statistical distribution range of impedance characteristics of self-regulating electric heating cables of multiple turbines of the same model under similar operating conditions. When the impedance characteristics of a local area deviate from the group statistical threshold, an early warning signal for corrosion under the insulation layer or settlement of the insulation cotton is output in advance. The early warning signal is issued before the temperature of the outer surface of the insulation layer becomes obviously abnormal, so as to achieve predictive maintenance.

9. The method for monitoring steam turbine insulation data based on big data according to claim 8, characterized in that, Impedance-temperature mapping relationship includes impedance-temperature calibration curve or soft measurement model; the process of obtaining impedance-temperature calibration curve is as follows: within the design operating temperature range of self-limiting electric heating cable, its impedance value is measured and recorded point by point at preset temperature intervals to form an initial impedance-temperature calibration curve. The process of acquiring the soft measurement model is as follows: a long short-term memory neural network model is constructed using a big data platform, and historical impedance spectrum data of the self-limiting electric heating cable under different operating conditions, as well as continuous temperature field data and load data for the corresponding time period, are extracted from the multi-physics field correlation dataset. The historical impedance spectrum data and load data are used as the training set input, and the continuous temperature field data for the corresponding time period is used as the supervision label to conduct supervised training on the long short-term memory neural network model, thereby obtaining a soft measurement model for real-time prediction of temperature values ​​from impedance spectrum data and load data.

10. A steam turbine insulation data monitoring system based on big data, characterized in that, The system employs the big data-based turbine insulation data monitoring method as described in any one of claims 1 to 9, and the system comprises: The sensing layer module includes self-regulating electric heating tapes installed on the turbine body and pipelines, used for: Obtain distributed impedance spectrum data of self-limiting electric heating tape; Edge computing layer module, used for: Based on the impedance-temperature mapping relationship, the distributed impedance spectrum data is solved into continuous temperature field data along the turbine body and pipe surface; The system acquires operational data from the distributed control system of the steam turbine, aligns the continuous temperature field data with the operational data using timestamps, and generates a multiphysics correlation dataset. The operational data includes vibration data, load data, and differential expansion data. Receive impedance drift feature information obtained by feature extraction and analysis of multi-physics field associated datasets by a big data platform; based on the received impedance drift feature information, correct the impedance-temperature mapping relationship; The continuous temperature field data is updated based on the corrected impedance-temperature mapping relationship, and a turbine insulation status assessment report and early warning information are generated.

Images