Power station operation and maintenance management system and method based on Internet of Things and digital twinning

Abstract

The invention discloses a power station operation and maintenance management system and method based on the Internet of Things and digital twinning, and relates to the technical field of industrial automation control. The sensing mapping module is used for monitoring the current health state of the equipment in real time and determining the safety bearing performance limit of the equipment; the injection simulation module is used for comparing a plan instruction with the safety bearing performance limit, determining a final plan instruction according to a comparison result, and performing simulation deduction at the same time to obtain a fatigue damage increment prediction value; and the arbitration decision module is used for comparing the predicted value of the fatigue damage increment with a fatigue threshold interval and determining a final operation instruction. According to the method, the fatigue damage increment caused by the plan instruction to the equipment can be predicted through the simulation deduction of the digital twin, so that the high-risk instruction can be intercepted before the instruction is issued, the operation with diseases and overload operation can be fundamentally avoided, the active protection of equipment assets is realized, and the service life of the equipment assets is effectively prolonged.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, specifically to a power plant operation and maintenance management system and method based on the Internet of Things and digital twins. Background Technology

[0002] With the global energy structure transformation and the advancement of "dual carbon" goals, the power system is rapidly developing towards cleaner, smarter, and more efficient systems. As the core link in energy conversion and transmission, the safe and stable operation of power plants (including thermal power, hydropower, wind power, photovoltaic, and integrated energy plants) is crucial to ensuring the reliability of the power grid. However, traditional power plant operation and maintenance management models have long relied on manual inspections, periodic maintenance, and experience-based judgment, resulting in problems such as delayed response, weak fault early warning capabilities, high operation and maintenance costs, and severe data silos, making it difficult to meet the needs of modern power plants for high availability, refined management, and intelligent decision-making.

[0003] In recent years, the rapid development of Internet of Things (IoT) technology has provided a brand-new means for monitoring the status of power plant equipment. By deploying a large number of sensors, edge computing nodes, and communication networks, it is possible to collect and transmit in real time the operating parameters (temperature, vibration, current, voltage, and humidity, etc.) of key power plant equipment (such as transformers, circuit breakers, wind turbines, and inverters), thus building a comprehensive, continuous, and high-dimensional operational data foundation.

[0004] Meanwhile, digital twin technology, as a novel cyber-physical fusion method integrating multiphysics modeling, data-driven simulation, and artificial intelligence, can construct a dynamic digital mapping in virtual space that is highly consistent with the physical power plant. This digital twin can not only reflect the real-time operating status of the physical power plant, but also support predictive maintenance, intelligent diagnosis, and auxiliary decision-making through functions such as historical data backtracking, fault simulation, performance prediction, and optimized scheduling.

[0005] Chinese invention patent CN111596604A discloses an intelligent fault diagnosis and self-healing control system for engineering equipment based on digital twins. The system includes a physical entity module, a data acquisition module, an information processing module, a fault diagnosis module, a self-healing control module, and a digital twin module. The data acquisition module collects real-time operational information data from the physical entity module and transmits the data to the digital twin module for digital twin simulation of the equipment. Simultaneously, the data is processed by the information processing module and then intelligently diagnosed and analyzed in the fault diagnosis module. The self-healing control module performs self-healing control on any faults that occur. The digital twin module interacts and provides feedback to other modules, enabling information exchange and closed-loop optimization. This invention can improve the accuracy of fault prediction, reduce the fault incidence rate, lower equipment maintenance costs, and enhance the stability and robustness of equipment operation, showing broad prospects for engineering applications.

[0006] However, the above and similar technical solutions still have the following shortcomings: Current control systems usually only determine the corresponding operating instructions based on relevant process requirements, without fully considering the real-time physical state of the corresponding actuators, such as bearing temperature, vibration value and cumulative start-stop count. As a result, when the equipment already has potential failure risks, control instructions that exacerbate its damage will still be issued to the equipment, which will further shorten the service life of the equipment. Summary of the Invention

[0007] The purpose of this invention is to provide a power plant operation and maintenance management system and method based on the Internet of Things and digital twins to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a power plant operation and maintenance management system based on the Internet of Things and digital twins, comprising: Perception mapping module: Through a set high-frequency sensor array, it monitors the current health status of the device in real time and determines the safety tolerance limit of the device based on the current health status; The simulation module compares the planned instructions of the distributed control system with the safety tolerance performance limit, determines the final planned instructions based on the comparison result, and simulates the final planned instructions to obtain the predicted fatigue damage increment value, including: SB1: Instruction Judgment: Based on the security tolerance performance limit, set up the business rule engine, and take the planned instructions that pass the judgment of the business rule engine as the final planned instructions, and reconstruct the planned instructions that fail the judgment of the business rule engine; SB2: Instruction Reconstruction: Multiple backup plan instructions are obtained by randomly sampling and reconstructing the plan instructions through Gaussian distribution. The backup plan instructions are then filtered through the business rule engine. Based on the total evaluation data of the filtered backup plan instructions, the highest total evaluation data is determined. The backup plan instruction corresponding to the highest total evaluation data is the final plan instruction. SB3: Simulation Prediction: The final planned instructions are simulated and deduced using the constructed simulation model, and the total damage increment is determined based on the simulation results; Arbitration decision module: compares the predicted value of fatigue damage increment with the set fatigue threshold range, and determines the final running instruction based on the comparison result.

[0009] Furthermore, the safe tolerance limits of the equipment are determined, including: SA1: State perception: By setting up vibration acceleration sensors, infrared temperature sensors, acoustic emission sensors and current Hall sensors, a high-frequency sensor array is constructed to collect vibration waveforms, temperature data, acoustic emission signals and current data; SA2: Index Calculation: Based on the vibration waveform, temperature data, acoustic emission signal and current data, characteristic signals are obtained, and the degradation degree of each characteristic signal is combined with the set data weight to obtain the overall degradation degree. Based on the overall degradation degree, the real-time health index is determined. SA3: Index Calculation: Based on the comparison results between the real-time health index and the RHI-performance limit comparison table, determine the safe performance limits of the equipment under the current state, including but not limited to the maximum allowable load, the maximum allowable speed, and the maximum operating frequency.

[0010] Furthermore, the degradation degree of each feature signal is combined with the set data weights to obtain the overall degradation degree, including: SA2.1: Feature Extraction: The vibration waveform, temperature data, acoustic emission signal and current data are analyzed by the analyzer to obtain the vibration signal, temperature signal, acoustic emission signal and current signal; SA2.2: Data Mapping: Based on the vibration signal, temperature signal, acoustic emission signal, and current signal, determine the average signal value and standard deviation of each signal data, and determine the degree of degradation of each signal data based on the average signal value and standard deviation, specifically as follows: ; in: The degree of degradation corresponding to the signal data. These are the characteristic values ​​corresponding to the signal data. This is the baseline value for health. This is the threshold for health warnings; SA2.3: Fusion Determination: The overall degradation degree is obtained by combining the degradation degree of the signal data with the set data weights.

[0011] Furthermore, the average value of the signal is used as the health baseline value, and the standard deviation of the signal is used as the health fluctuation range. At the same time, the health warning threshold is determined based on the health baseline value and the health fluctuation range.

[0012] Furthermore, based on the overall evaluation data of the selected backup plan instructions, the highest overall evaluation data is determined. The backup plan instruction corresponding to the highest overall evaluation data is the final plan instruction, which includes: SB2.1: Population Formation: Multiple backup plan instructions are obtained by randomly sampling and reconstructing the plan instructions through Gaussian distribution. The backup plan instructions are then filtered through the business rule engine. Based on the retained backup plan instructions, an initial instruction population is formed. SB2.2: Scoring and Filtering: Based on the data size of each backup plan instruction in the initial instruction population, determine the total evaluation data of each backup plan instruction. At the same time, based on the total evaluation data, sort all the backup plan instructions in descending order and filter out the 20% of backup plan instructions with the highest total evaluation data scores. SB2.3: Cross-determination: The backup plan instructions with the highest total evaluation data scores in the 20% are matched in pairs to obtain matching data pairs. Based on the average value of each matching data pair, a cross-plan instruction is constructed. At the same time, steps SB2.1-SB2.3 are repeated according to the cross-plan instruction until no new cross-plan instruction can be obtained. The highest total evaluation data score is determined from the total evaluation data scores of the last set of cross-plan instructions. The cross-plan instruction corresponding to the highest total evaluation data score is the final plan instruction.

[0013] Furthermore, the magnitude of the total damage increment is determined, including: SB3.1: Digital Twin Construction: A digital twin simulation model is constructed using first-principles calculations to describe the inherent physical laws of the device; SB3.2: Simulation and Deduction: The final plan instructions are simulated and deduced using the digital twin simulation model, and the stress-time history of each component is recorded. At the same time, the stress-time history is analyzed using the rainflow counting method to determine the stress amplitude. Based on the stress amplitude, the number of damage-causing cycles is determined from the SN curve corresponding to the material of each component. The total damage increment is determined by the number of damage-causing cycles and the total number of stress cycles.

[0014] Furthermore, mechanical motion is simulated using a multibody dynamics model, fluid behavior is simulated using a fluid dynamics model, stress, strain, and heat distribution are simulated using a finite element analysis model, and the temperature field is simulated using a thermodynamic model.

[0015] Furthermore, the final execution instructions are determined, including: SC1: Threshold Determination: Using historical operation instructions from the digital twin simulation model, the total damage increment for each historical operation instruction is obtained. Based on this total damage increment, a baseline threshold for the fatigue threshold range is set. Simultaneously, the baseline threshold is adjusted based on the real-time health index to obtain the final fatigue threshold range. Specifically: ; in: The updated upper / lower fatigue thresholds, The upper / lower limit reference threshold, This is the current real-time health index. The real-time health index corresponding to the baseline threshold; SC2: Threshold Comparison: Based on the updated upper / lower fatigue thresholds, a final fatigue threshold range is set. Simultaneously, the total damage increment is compared with this final fatigue threshold range, and the corresponding decision range is determined based on the comparison result. Specifically: When the total damage increment is less than the lower limit of the final fatigue threshold interval, the corresponding decision interval is safe passage; when the total damage increment is within the range of the final fatigue threshold interval, the corresponding decision interval is instruction optimization, and steps SB1-SB3 are repeated according to the planned instruction corresponding to the total damage increment until the total damage increment is less than the lower limit of the final fatigue threshold interval; when the total damage increment is greater than the upper limit of the final fatigue threshold interval, the corresponding decision interval is interception alarm.

[0016] Furthermore, based on multiple total damage increments, a healthy operation total damage increment data set is constructed. The total damage increment corresponding to the 80th-90th percentile in the healthy operation total damage increment data set is used as the lower limit benchmark threshold of the fatigue threshold interval, and the total damage increment corresponding to the 95th-98th percentile is used as the upper limit benchmark threshold of the fatigue threshold interval.

[0017] The power plant operation and maintenance management method based on the Internet of Things and digital twins uses any one of the above-mentioned power plant operation and maintenance management systems based on the Internet of Things and digital twins.

[0018] Compared with the prior art, the beneficial effects of the present invention are: Firstly, this invention, through simulation and deduction using a digital twin, can predict the incremental fatigue damage to equipment caused by planned instructions. This allows for the interception of high-risk instructions that could exacerbate equipment wear or even cause malfunctions before the instructions are issued. Consequently, it can fundamentally avoid operating with defects or under overload, thus achieving proactive protection of equipment assets and effectively extending their service life. Secondly, this invention quantifies the real-time health index and damage increment of the equipment, providing managers with an intuitive and scientific basis for decision-making. At the same time, through a genetic algorithm-like optimization strategy, it can automatically generate and select the optimal instructions that meet the process requirements and minimize the damage to the equipment, thereby further improving the level of intelligent decision-making. Thirdly, this invention dynamically adjusts the alarm threshold for fatigue damage through the equipment health index, so that when the predicted damage exceeds the safe range, an alarm or automatic optimization command can be triggered, thereby reducing the risk of unplanned shutdowns caused by sudden equipment failures and ensuring the continuity of power plant operation and the stability of the power grid. Attached Figure Description

[0019] Figure 1 This is a system block diagram of the power plant operation and maintenance management system in this invention; Figure 2 This is a schematic diagram of the process for obtaining the overall degradation degree in this invention; Figure 3 This is a schematic diagram illustrating the process for determining the total damage increment in this invention; Figure 4 This is a schematic diagram of the instruction refactoring process in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 refer to Figure 1 and Figure 3This embodiment provides a power plant operation and maintenance management system based on the Internet of Things (IoT) and digital twins. The system includes a perception mapping module, an injection simulation module, and an arbitration decision module. The perception mapping module uses a set high-frequency sensor array to collect corresponding physical state signals and converts these signals into a quantitative understanding of the current health status of the equipment. The injection simulation module determines the performance limit that the equipment can safely withstand in the current state based on the quantitative understanding of the equipment's current health status, compares it with the issued control commands, determines the corresponding optimization commands based on the comparison results, and performs real-time simulation of the determined optimization commands in the digital twin model to obtain the predicted value of the fatigue damage increment. The arbitration decision module compares the predicted value of the fatigue damage increment with a set fatigue threshold and determines the final operating command based on the comparison results.

[0022] In this embodiment, the perception mapping module is used to construct a high-frequency sensor array using vibration acceleration sensors, infrared temperature sensors, acoustic emission sensors, and current Hall sensors. This high-frequency sensor array is then used to monitor the current health status of the device in real time and quantify its corresponding physical state signals. Simultaneously, based on the quantized physical state signals, the performance limits that the device can safely withstand under the current state are determined. Specifically: Step SA1: State Awareness. Based on the equipment's structure and fault mode analysis, the high-frequency sensor array is installed at the location closest to the fault source with the shortest signal transmission path, such as the radial and vertical directions of the bearing housing, the input and output ends of the gearbox, and the drive and non-drive ends of the motor. Simultaneously, the high-frequency sensor array is rigidly connected to its corresponding mounting position on the equipment, for example, using stainless steel studs to fix a vibration acceleration sensor to a machined smooth surface. It is worth noting that for mounting positions where drilling is not possible, this embodiment uses high-strength industrial adhesive to bond and fix the high-frequency sensor array to the corresponding mounting position.

[0023] Specifically, the high-frequency sensor array in this embodiment includes a vibration acceleration sensor, an acoustic emission sensor, an infrared temperature sensor, and a current Hall sensor. The vibration acceleration sensor is an IEPE-type piezoelectric accelerometer; the acoustic emission sensor is an acoustic emission sensor with a resonant frequency in the range of 150kHz–300kHz; the infrared temperature sensor is a non-contact infrared thermometer or thermal imager; and the current Hall sensor is a clamp-on or through-hole Hall current sensor.

[0024] In other words, by setting up a high-frequency sensor array, the operating parameters of the corresponding location on the device are collected, including vibration waveforms, temperature data, acoustic emission signals and current data.

[0025] Step SA2: Exponent Calculation. This involves extracting the corresponding feature signals from the sensor data obtained in Step SA1. Simultaneously, based on the extracted feature signals, the degree of degradation corresponding to each feature signal is determined. This degree of degradation is then combined with the set data weights to determine the overall degree of degradation.

[0026] Furthermore, based on the obtained overall degradation level, the corresponding real-time health index is determined, specifically as follows:

[0027] in: For real-time health index, This represents the overall degradation level.

[0028] Step SA3: Index Calculation. This involves comparing the real-time health index determined in Step SA2 with the established RHI-Performance Limit Lookup Table to determine the performance limits that the corresponding equipment can safely withstand under the current conditions, including but not limited to maximum allowable load, maximum allowable speed, and maximum operating frequency. It is worth noting that the RHI-Performance Limit Lookup Table set in this embodiment can be specifically configured according to actual needs; therefore, it is not specifically described in this embodiment.

[0029] In this embodiment, the injection simulation module receives planning instructions from the distributed control system embedded in the power plant, compares the received planning instructions with the performance limits determined by the perception mapping module, and determines the final planning instructions based on the comparison results. Simultaneously, the determined final planning instructions are simulated in real time to obtain the corresponding predicted fatigue damage increment values. Specifically: Step SB1: Instruction Judgment. Based on the performance limits determined in Step SA3, the corresponding business rule engine is set. This means that the maximum allowable load, maximum allowable speed, and maximum action frequency are determined within the business rule engine. Simultaneously, the distributed control system embedded within the power plant receives and sends planning instructions in real time, and extracts the corresponding planned load, planned speed, and planned action frequency from these instructions.

[0030] Furthermore, the business rules engine compares and judges the received planning instructions. That is, it compares the extracted planned load with the maximum allowable load, the extracted planned speed with the maximum allowable speed, and the extracted planned action frequency with the maximum action frequency. Based on the comparison results, the corresponding final planning instruction is determined. Specifically, if the planned load is greater than the maximum allowable load and / or the planned speed is greater than the maximum allowable speed and / or the planned action frequency is greater than the maximum action frequency, then the next step SB2 is executed to reconstruct the received planning instruction. Otherwise, the received planning instruction is the final planning instruction.

[0031] Step SB2: Instruction Reconstruction. This involves performing Gaussian distribution random sampling on the received planning instructions to generate multiple backup planning instructions. Each backup planning instruction is then compared and evaluated using the business rule engine to select the best one. Simultaneously, based on the selected backup planning instructions, the overall evaluation data for each instruction is obtained, and the highest overall evaluation data is determined. The backup planning instruction corresponding to this highest overall evaluation data is the final planning instruction.

[0032] Step SB3: Simulation Prediction. This involves using the constructed simulation model to simulate and extrapolate the final planned instructions obtained in Step SB1 and / or Step SB2, and determining the corresponding predicted fatigue damage increment value, i.e., the corresponding total damage increment, based on the simulation results. Details are as follows: Step SB3.1: Digital Twin Construction. This involves constructing a corresponding digital twin simulation model using first principles of mechanics such as Newtonian mechanics and thermodynamics to describe the inherent physical laws of the equipment. Specifically, in this embodiment, a multibody dynamics model is used to simulate mechanical motion. For example, when constructing the pump rotor system, components such as the shaft, impeller, and bearings are modeled, and their corresponding parameters such as mass, stiffness, and damping are defined. Simultaneously, during simulation, the dynamic load on the bearings and the vibration response of the shaft can be obtained.

[0033] Furthermore, fluid behavior is simulated using fluid dynamics models, such as simulating pressure, flow rate, and cavitation within a pump. Simultaneously, stress, strain, and heat distribution are simulated using finite element analysis models. For instance, based on loads obtained from the fluid dynamics model, a finite element model of the bearing is run to determine the microscopic stress distribution at the bearing raceway contact points. Additionally, the temperature field is simulated using thermodynamic models; for example, the bearing's temperature rise is determined based on frictional power dissipation and cooling conditions.

[0034] Step SB3.2: Simulation and Deduction. Based on the digital twin simulation model constructed in Step SB3.1, the final plan instructions are simulated and deduced. Simultaneously, during the simulation and deduction process, the stress-time history of each key component, such as the bearing, is recorded to obtain the stress spectrum of each key component, such as the bearing.

[0035] Furthermore, the stress amplitude is extracted from the obtained stress-time history using the rainflow counting method. Simultaneously, the number of damage-causing cycles is determined from the SN curve corresponding to each component material based on the extracted stress amplitude. In other words, the corresponding damage increment is obtained based on the determined number of damage-causing cycles. Specifically:

[0036] in: Let j be the damage increment during the j-th stress cycle. The number of damage-causing cycles corresponds to the j-th stress cycle.

[0037] Furthermore, based on all the stress cycle counts determined during the simulation, these are combined with the determined damage increment to determine the corresponding total damage increment, specifically:

[0038] in: This represents the total increase in damage. This represents the damage increment during the j-th stress cycle.

[0039] In this embodiment, the arbitration decision module compares the total damage increment obtained by the injection simulation module with the set fatigue threshold range, and determines the final running instruction based on the comparison result. Specifically: Step SC1: Threshold Determination. This involves obtaining the total damage increment corresponding to each historical operation command from the digital twin simulation model. Simultaneously, based on the obtained total damage increments, a corresponding healthy operation total damage increment data set is constructed. Further, based on the healthy operation total damage increment data set, the total damage increment corresponding to the 80th-90th percentile is used as the lower limit benchmark threshold of the fatigue threshold interval, and the total damage increment corresponding to the 95th-98th percentile is used as the upper limit benchmark threshold of the fatigue threshold interval.

[0040] Furthermore, based on the real-time health index obtained from the perception mapping module, the set upper and lower baseline thresholds are adjusted in real time to determine the corresponding final fatigue threshold range, specifically:

[0041] in: The updated upper / lower fatigue thresholds, The upper / lower limit reference threshold, This is the current real-time health index. This is the real-time health index corresponding to the baseline threshold.

[0042] Step SC2: Threshold Comparison. Based on the updated upper / lower fatigue thresholds obtained in Step SC1, the corresponding final fatigue threshold range is determined. Simultaneously, the total damage increment obtained from the simulation module is compared with the final fatigue threshold range, and based on the comparison result, the corresponding decision range is determined, specifically: When the obtained total damage increment is less than the lower limit of the final fatigue threshold interval, the corresponding decision interval is safe release. That is, an operating instruction is sent to the field actuator according to the planned instruction corresponding to the total damage increment. When the obtained total damage increment is within the range of the final fatigue threshold interval, the corresponding decision interval is instruction optimization. That is, steps SB1-SB3 are repeated according to the planned instruction corresponding to the total damage increment until the obtained total damage increment is less than the lower limit of the final fatigue threshold interval. When the obtained total damage increment is greater than the upper limit of the final fatigue threshold interval, the corresponding decision interval is interception alarm. That is, the planned instruction corresponding to the total damage increment is intercepted and an early warning signal is issued.

[0043] This embodiment also provides a power plant operation and maintenance management method based on the Internet of Things and digital twins, which uses the aforementioned power plant operation and maintenance management system based on the Internet of Things and digital twins.

[0044] Example 2 This embodiment provides a power plant operation and maintenance management system based on the Internet of Things and digital twins. Its specific implementation method is the same as in Embodiment 1, except that, based on the extracted feature signals, the degree of degradation corresponding to each feature signal is determined, and the degree of degradation corresponding to each feature signal is combined with the set data weights to determine the corresponding overall degree of degradation. The invention will now be illustrated with specific examples of this embodiment. Figure 2 The details are as follows: Step SA2.1: Feature Extraction. This involves determining the corresponding vibration, temperature, acoustic emission, and current signals based on the acquired vibration waveform, temperature data, acoustic emission signal, and current data. Specifically, a Fast Fourier Transform (FFT) analyzer is used to analyze the acquired vibration waveform to determine the overall intensity, peak factor, and characteristic frequency peak value. Temperature trend analysis software is used to analyze the acquired temperature data to determine the instantaneous values, temperature difference values, and temperature rise rate. An acoustic emission parameter analyzer is used to analyze the acquired acoustic emission signal to determine the corresponding ring count, energy, and amplitude. A power quality analyzer is used to analyze the acquired current data to determine the total harmonic distortion (THD) and amplitude variations of specific harmonics.

[0045] Furthermore, before performing a Fourier transform on the vibration waveform, a hardware anti-aliasing filter is used to filter the original vibration waveform, removing frequency components higher than half the sampling frequency to prevent spectral distortion. Simultaneously, based on a preset data length, the filtered original vibration waveform is segmented into multiple data blocks of fixed length, and each data block is multiplied by a preset window function (such as a Hanning window) to obtain the preprocessed vibration waveform. In this embodiment, a Fourier transform is performed on the preprocessed vibration waveform to obtain the corresponding spectrum, and the overall intensity, peak value factor, and characteristic frequency peak value are extracted from the obtained spectrum. Specifically, in this embodiment, the overall intensity is calculated by squaring, averaging, and then taking the square root of the instantaneous values ​​of the vibration waveform within a preset time period; the peak value factor is the ratio between the peak value and the effective value of the vibration waveform within the preset time period; and the characteristic frequency peak value is the amplitude of the characteristic frequency points located on the spectrum corresponding to the equipment components, including bearing fault frequencies and gear meshing frequencies.

[0046] Furthermore, based on the acquired temperature data, the instantaneous temperature values ​​at each moment under the current operating condition are determined. Simultaneously, the acquired temperature data is compared with historical temperature data under the same / similar operating conditions to obtain the temperature difference between the current and historical data. Furthermore, based on the temperature data's time-varying curve, the slope of the curve is determined, thereby identifying the corresponding temperature rise rate.

[0047] Furthermore, the waveform of the acoustic emission signal is counted and analyzed using a preset voltage threshold. Specifically, the preset voltage threshold is compared with the waveform of the acoustic emission signal. When the waveform of the acoustic emission signal crosses the preset voltage threshold once, the corresponding ringing count is incremented by one. At the same time, the corresponding energy and amplitude are determined based on the area under the envelope of the acoustic emission signal waveform.

[0048] Furthermore, Fourier transform analysis is performed on the acquired current data to obtain the amplitude of the corresponding fundamental frequency (i.e., 50Hz) and various harmonics (i.e., including but not limited to 100Hz and 150Hz). Simultaneously, harmonics of specific orders are monitored to obtain their amplitude variations. For example, a rotor bar breakage fault generates a special (1±2ks)f1 frequency band in the spectrum. Here, k is the harmonic order index, s is the slip, and f1 is the fundamental frequency.

[0049] Step SA2.2: Data Mapping. Based on the vibration signal, temperature signal, acoustic emission signal, and current signal obtained in step SA2.1, determine the average value and standard deviation of each signal data. Simultaneously, use the determined average value as the corresponding health baseline value and the determined standard deviation value as the corresponding health fluctuation range.

[0050] Furthermore, based on the set health baseline value and health fluctuation range, the corresponding health warning threshold is determined using the health baseline value as a benchmark and in conjunction with the health fluctuation range. That is, the health baseline value and the upper limit of the health fluctuation range are added together to obtain the corresponding health warning threshold. Simultaneously, each acquired signal data point is compared with the corresponding set health baseline value and health warning threshold, and based on the comparison result, the degree of degradation corresponding to each signal data point is determined, specifically:

[0051] in: The degree of degradation corresponding to the signal data. These are the characteristic values ​​corresponding to the signal data. This is the baseline value for health. This is the threshold for health warnings.

[0052] In the specific implementation process, the healthy baseline value of bearing temperature in this embodiment, that is, the average temperature of stable operation, is 65℃, while the corresponding health alarm threshold, that is, the highest temperature value, is 85℃, and the current detected temperature value is 75℃. Therefore, the deterioration degree corresponding to the current temperature value is 0.5.

[0053] Step SA2.3: Fusion Determination. This involves combining the degradation levels of each signal data point determined in Step SA2.2 with the set data weights, and determining the overall degradation level based on the combination result. Specifically:

[0054] in: For overall degradation, Let i be the degradation degree corresponding to the i-th signal data. Let be the data weight corresponding to the i-th signal data.

[0055] It is worth noting that the data weights set in this embodiment can be determined according to a pre-set weight configuration table. Each different device (such as a pump, fan, and compressor) has a corresponding weight configuration table, and each weight configuration table is jointly set and constructed by equipment engineers, maintenance experts, and data scientists. In other words, it can be specifically set according to actual data needs; therefore, it is not specifically described in this embodiment.

[0056] Example 3 This embodiment provides a power plant operation and maintenance management system based on the Internet of Things and digital twins. Its specific implementation method is the same as in Embodiment 1, except that, based on the selected backup plan instructions, the total evaluation data corresponding to each backup plan instruction is obtained, and the highest total evaluation data is determined. The backup plan instruction corresponding to the highest total evaluation data is the corresponding final plan instruction. The invention will now be illustrated with specific examples of this embodiment. (Refer to...) Figure 4 The details are as follows: Step SB2.1: Population Building. This involves performing Gaussian distribution random sampling based on the received planning instructions to generate multiple backup planning instructions. Simultaneously, the business rule engine compares and judges each backup planning instruction, and filters the generated backup planning instructions based on the comparison results. In other words, backup planning instructions that pass the business rule engine's judgment are saved, while those that do not are deleted. Based on the saved backup planning instructions, an initial instruction population is built.

[0057] Step SB2.2: Scoring and Filtering. This involves summing the data for each backup plan instruction in the initial instruction population obtained from the initial instruction group to obtain the total evaluation data for each backup plan instruction. Simultaneously, based on the total evaluation data, the backup plan instructions in the initial instruction population are sorted in descending order. Furthermore, based on the sorted backup plan instructions in the initial instruction population, the top 20% of backup plan instructions with the highest total evaluation data scores are selected.

[0058] Step SB2.3: Cross-matching. This involves pairwise matching of the 20% of standby plan instructions selected in Step SB2.2 to obtain matched data pairs. Simultaneously, based on the magnitude of the two corresponding data values ​​for each matched data pair, the average value is determined. Based on this average value, the cross-matching plan instruction corresponding to each matched data pair is constructed.

[0059] Furthermore, based on the multiple cross-planning instructions obtained, steps SB2.1 to SB2.3 are repeated until no new cross-planning instructions can be obtained. Simultaneously, based on the last group of multiple cross-planning instructions obtained, the total evaluation data score corresponding to each cross-planning instruction is determined, and the highest total evaluation data score is identified. In other words, the cross-planning instruction corresponding to this highest total evaluation data score is the corresponding final planning instruction.

[0060] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A power plant operation and maintenance management system based on the Internet of Things and digital twins, characterized in that: Including: Perception mapping module: Through a set high-frequency sensor array, it monitors the current health status of the device in real time and determines the safety tolerance limit of the device based on the current health status; The simulation module compares the planned instructions of the distributed control system with the safety tolerance performance limit, determines the final planned instructions based on the comparison result, and simulates the final planned instructions to obtain the predicted fatigue damage increment value, including: SB1: Instruction Judgment: Based on the security tolerance performance limit, set up the business rule engine, and take the planned instructions that pass the judgment of the business rule engine as the final planned instructions, and reconstruct the planned instructions that fail the judgment of the business rule engine; SB2: Instruction Reconstruction: Multiple backup plan instructions are obtained by randomly sampling and reconstructing the plan instructions through Gaussian distribution. The backup plan instructions are then filtered through the business rule engine. Based on the total evaluation data of the filtered backup plan instructions, the highest total evaluation data is determined. The backup plan instruction corresponding to the highest total evaluation data is the final plan instruction. SB3: Simulation Prediction: The final planned instructions are simulated and deduced using the constructed simulation model, and the total damage increment is determined based on the simulation results; Arbitration decision module: compares the predicted value of fatigue damage increment with the set fatigue threshold range, and determines the final running instruction based on the comparison result.

2. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 1, characterized in that, Determine the safe tolerance limits of the equipment, including: SA1: State perception: By setting up vibration acceleration sensors, infrared temperature sensors, acoustic emission sensors and current Hall sensors, a high-frequency sensor array is constructed to collect vibration waveforms, temperature data, acoustic emission signals and current data; SA2: Index Calculation: Based on the vibration waveform, temperature data, acoustic emission signal and current data, characteristic signals are obtained, and the degradation degree of each characteristic signal is combined with the set data weight to obtain the overall degradation degree. Based on the overall degradation degree, the real-time health index is determined. SA3: Index Calculation: Based on the comparison results between the real-time health index and the RHI-performance limit comparison table, determine the safe performance limits of the equipment under the current state, including but not limited to the maximum allowable load, the maximum allowable speed, and the maximum operating frequency.

3. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 2, characterized in that, The degradation degree of each feature signal is combined with the set data weights to obtain the overall degradation degree, which includes: SA2.1: Feature Extraction: The vibration waveform, temperature data, acoustic emission signal and current data are analyzed by the analyzer to obtain the vibration signal, temperature signal, acoustic emission signal and current signal; SA2.2: Data Mapping: Based on the vibration signal, temperature signal, acoustic emission signal, and current signal, determine the average signal value and standard deviation of each signal data, and determine the degree of degradation of each signal data based on the average signal value and standard deviation, specifically as follows: in: The degree of degradation corresponding to the signal data. These are the characteristic values ​​corresponding to the signal data. This is the baseline value for health. This is the threshold for health warnings; SA2.3: Fusion Determination: The overall degradation degree is obtained by combining the degradation degree of the signal data with the set data weights.

4. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 3, characterized in that, The average value of the signal is used as the health baseline value, and the standard deviation of the signal is used as the health fluctuation range. At the same time, the health warning threshold is determined based on the health baseline value and the health fluctuation range.

5. The power plant operation and maintenance management system based on the Internet of Things and digital twin as described in claim 1, characterized in that, Based on the overall evaluation data of the selected backup plan instructions, the highest overall evaluation data is determined. The backup plan instruction corresponding to the highest overall evaluation data is the final plan instruction, including: SB2.1: Population Formation: Multiple backup plan instructions are obtained by randomly sampling and reconstructing the plan instructions through Gaussian distribution. The backup plan instructions are then filtered through the business rule engine. Based on the retained backup plan instructions, an initial instruction population is formed. SB2.2: Scoring and Filtering: Based on the data size of each backup plan instruction in the initial instruction population, determine the total evaluation data of each backup plan instruction. At the same time, based on the total evaluation data, sort all the backup plan instructions in descending order and filter out the 20% of backup plan instructions with the highest total evaluation data scores. SB2.3: Cross-determination: The backup plan instructions with the highest total evaluation data scores in the 20% are matched in pairs to obtain matching data pairs. Based on the average value of each matching data pair, a cross-plan instruction is constructed. At the same time, steps SB2.1-SB2.3 are repeated according to the cross-plan instruction until no new cross-plan instruction can be obtained. The highest total evaluation data score is determined from the total evaluation data scores of the last set of cross-plan instructions. The cross-plan instruction corresponding to the highest total evaluation data score is the final plan instruction.

6. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 1, characterized in that, The total incremental damage was determined, including: SB3.1: Digital Twin Construction: A digital twin simulation model is constructed using first-principles calculations to describe the inherent physical laws of the device; SB3.2: Simulation and Deduction: The final plan instructions are simulated and deduced using the digital twin simulation model, and the stress-time history of each component is recorded. At the same time, the stress-time history is analyzed using the rainflow counting method to determine the stress amplitude. Based on the stress amplitude, the number of damage-causing cycles is determined from the SN curve corresponding to the material of each component. The total damage increment is determined by the number of damage-causing cycles and the total number of stress cycles.

7. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 6, characterized in that, Mechanical motion is simulated using a multibody dynamics model, fluid behavior is simulated using a fluid dynamics model, stress, strain, and heat distribution are simulated using a finite element analysis model, and the temperature field is simulated using a thermodynamic model.

8. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 1, characterized in that, The final execution instructions are determined, including: SC1: Threshold Determination: Using historical operation instructions from the digital twin simulation model, the total damage increment for each historical operation instruction is obtained. Based on this total damage increment, a baseline threshold for the fatigue threshold range is set. Simultaneously, the baseline threshold is adjusted based on the real-time health index to obtain the final fatigue threshold range. Specifically: ； in: The updated upper / lower fatigue thresholds. The upper / lower limit is the baseline threshold. This is the current real-time health index. The real-time health index corresponding to the baseline threshold; SC2: Threshold Comparison: Based on the updated upper / lower fatigue thresholds, a final fatigue threshold range is set. Simultaneously, the total damage increment is compared with this final fatigue threshold range, and the corresponding decision range is determined based on the comparison result. Specifically: When the total damage increment is less than the lower limit of the final fatigue threshold interval, the corresponding decision interval is safe passage; when the total damage increment is within the range of the final fatigue threshold interval, the corresponding decision interval is instruction optimization, and steps SB1-SB3 are repeated according to the planned instruction corresponding to the total damage increment until the total damage increment is less than the lower limit of the final fatigue threshold interval; when the total damage increment is greater than the upper limit of the final fatigue threshold interval, the corresponding decision interval is interception alarm.

9. The power plant operation and maintenance management system based on the Internet of Things and digital twins according to claim 8, characterized in that, Based on multiple total damage increments, a healthy operation total damage increment data set is constructed. The total damage increment corresponding to the 80th-90th percentile of the healthy operation total damage increment data set is used as the lower limit benchmark threshold of the fatigue threshold interval, and the total damage increment corresponding to the 95th-98th percentile is used as the upper limit benchmark threshold of the fatigue threshold interval.

10. A power plant operation and maintenance management method based on the Internet of Things and digital twins, characterized in that: The power plant operation and maintenance management system based on the Internet of Things and digital twins as described in any one of claims 1-9 was used.

Images