Predictive test system and method based on digital twinborn and environmental Monte Carlo simulation

Abstract

The invention discloses a predictive test system and method based on digital twinning and environmental Monte Carlo simulation, and relates to the technical field of equipment environmental adaptability tests.The system comprises a test cabin, an environmental parameter actuator, a climate Monte Carlo engine, an online sensing network and a health state evaluation and feedback control module; the method comprises the steps that an engine generates an initial multi-dimensional environment control parameter sequence according to a target environment model, an actuator applies environment stress according to the initial multi-dimensional environment control parameter sequence, a sensing network acquires internal state parameters of equipment in real time, an evaluation module judges whether the parameters reach a cascade failure threshold value or not, if yes, an engine algorithm is adjusted or the parameters are corrected, and if not, testing continues; and after finishing, constructing an aged digital twin model which can reproduce a multi-environment coupling effect and realize self-adaptive loading, thereby accurately predicting the failure and the service life of the equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of equipment environmental adaptability testing, in particular to a predictive testing system and method based on digital twinning and environmental Monte Carlo simulation. BACKGROUND

[0002] In the fields of power, communication, transportation and other critical infrastructures, a large number of equipment (such as optical cable splice closures, outdoor sensors, power transmission line accessories, etc.) are exposed to complex and variable natural environments for a long time, and their reliability directly affects the overall operation safety of the system. In order to evaluate the environmental adaptability of the equipment in the laboratory environment, the traditional method usually uses single or simply alternating environmental stresses such as temperature, humidity, vibration, salt spray for accelerated testing.

[0003] For example, patent CN114925916A discloses an important user power protection emergency auxiliary decision-making method based on digital twinning, which simulates power grid failure scenarios using Monte Carlo sampling to achieve risk assessment; another patent CN118228587A proposes a MCMC-based uncertainty analysis method for digital twinning of the reactor core, which quantifies the uncertainty of model parameters through the Bayesian framework. These existing technologies have improved the credibility and decision-making scientificity of system simulation to some extent.

[0004] However, the above methods focus on fault simulation or parameter optimization in virtual space and do not involve closed-loop testing of physical devices in real multi-physical field coupled environments. In the field of physical device testing, existing environmental testing equipment mostly applies stress based on pre-set static scripts or linear superposition models, lacking dynamic modeling capabilities for non-linear interactions between multiple environmental factors such as temperature, humidity, vibration, and chemical corrosion, and cannot reproduce cascading failure effects such as vibration-induced micro-cracks accelerating salt spray corrosion. In addition, the testing process is usually open-loop, and cannot dynamically adjust the testing strategy based on real-time state feedback of the tested device, resulting in low testing efficiency and poor correlation with actual aging process.

[0005] Therefore, one of the outstanding problems in the existing technology is: how to dynamically reproduce the non-linear coupling effects of multiple environmental factors in high-fidelity environmental simulation testing, and perform adaptive stress loading based on the real-time state of the device, so as to accurately predict the failure behavior and remaining life of the device in the real environment. This problem restricts the in-depth application of environmental testing technology in the field of predictive maintenance.

[0006] In view of the above shortcomings, there is an urgent need for a high-fidelity environmental simulation testing system and method that can integrate digital twinning technology. It not only has the ability to dynamically evolve multiple environmental parameters, but more importantly, it introduces a closed-loop feedback mechanism based on the perception of the internal state of the device, thereby realizing the transformation of the testing process from one-way loading to interactive adaptation, significantly improving the fidelity and prediction accuracy of the test. SUMMARY

[0007] The present application aims to make up for the deficiencies of the prior art, and provides a predictive testing system and method based on digital twinning and environmental Monte Carlo simulation, which can generate a dynamic evolving multi-dimensional environmental control parameter sequence, monitor the internal state of the device in real time and form a closed-loop feedback, reproduce the nonlinear coupling effect of multiple environmental factors, realize adaptive stress loading, and accurately predict the failure behavior and residual life of the device, by setting a climate Monte Carlo engine, an online sensing network and a health state evaluation and feedback control module, in combination with a test cabin and an environmental parameter executor.

[0008] To solve the above technical problems, the present application provides the following technical solutions: on the one hand, a predictive testing system based on digital twinning and environmental Monte Carlo simulation, comprising a test cabin (200) for accommodating a device under test (300), and at least one environmental parameter executor (210) connected with the test cabin (200), characterized in that the system further comprises a climate Monte Carlo engine (100), an online sensing network (310) arranged inside the device under test (300), and a health state evaluation and feedback control module (400):

[0009] The climate Monte Carlo engine (100) is configured to generate a set of multi-dimensional environmental control parameter sequences dynamically evolving over time and correlated with each other according to a preset target environmental model;

[0010] The online sensing network (310) is used to monitor the internal state parameters of the device under test (300) in real time;

[0011] One end of the health state evaluation and feedback control module (400) is connected with the online sensing network (310), and the other end is connected with the climate Monte Carlo engine (100) and the environmental parameter executor (210);

[0012] The health state evaluation and feedback control module (400) is configured to receive the internal state parameters and evaluate the health state of the device under test (300) based on a preset degradation model, and dynamically adjust the evolution algorithm of the climate Monte Carlo engine (100) according to the health state to correct the multi-dimensional environmental control parameter sequence to form a closed-loop feedback control.

[0013] Further, the target environmental model includes meteorological statistical data of a target geographic location, historical extreme event data and a preset accelerated aging model;

[0014] The meteorological statistical data at least includes temperature, humidity, solar intensity, rainfall and probability distribution and correlation matrix of typhoon events;

[0015] The historical extreme event data refers to records of extreme climate conditions occurring in the target region within a preset historical period;

[0016] The accelerated aging model is a model framework that compresses actual environmental stress into laboratory accelerated test stress through mathematical transformation, which is constructed based on the principle of material degradation kinetics.

[0017] Further, the climate Monte Carlo engine (100) generates the multi-dimensional environmental control parameter sequence using an adaptive Monte Carlo evolutionary algorithm, which realizes the dynamic evolution of environmental parameters through the following mathematical formula:

[0018]

[0019] wherein, represents the multi-dimensional environmental control parameter vector at time step , the dimension of which is determined by the number of factors in the target environment model; represents the updated multi-dimensional environmental control parameter vector at time step ; and is an environmental evolution rate coefficient, which is a positive real number and is used to control the overall speed of parameter change; is an environmental factor coupling matrix, which is a symmetric matrix, and the matrix elements represent the correlation strength between different environmental parameters; is an external environmental disturbance vector, which is randomly generated based on historical data in the target environment model; is a state feedback adjustment coefficient, which is a real number and is dynamically adjusted by the output of the health state assessment and feedback control module (400); is a device state response function, the input of which is the internal state parameter vector monitored by the online sensing network (310), and the output is the normalized adjustment amount calculated based on the degradation model.

[0020] Further, the environmental parameter executor (210) includes at least two of a temperature controller, a humidity controller, a mechanical stress loading device, an electromagnetic interference generator, and a biochemical micro-injection system;

[0021] The temperature controller can realize refrigeration and heating functions and accurately adjust the temperature within a preset temperature range;

[0022] The humidity controller maintains the relative humidity in the test cabin (200) within a set interval through humidification and dehumidification modules;

[0023] The mechanical stress loading device is a multi-degree-of-freedom vibration table that can simulate various mechanical environments;

[0024] The electromagnetic interference generator generates a controllable electromagnetic field to simulate electromagnetic interference conditions.

[0025] The biochemical micro-injection system is composed of a liquid storage tank, a precision metering pump and a nozzle, and is used for injecting corrosive media into the test cabin (200).

[0026] Further, the biochemical micro-injection system is configured to inject acidic media, alkaline media and salt spray media into the test cabin (200) according to the instructions of the health state evaluation and feedback control module (400);

[0027] The acidic medium is a solution with a pH value less than 7, the alkaline medium is a solution with a pH value greater than 7, and the salt spray medium is a sodium chloride solution atomized;

[0028] The injection process is controlled by the precision metering pump with a flow rate of 0.1 milliliter per minute, and the injection timing and duration are dynamically triggered by the health state evaluation and feedback control module (400) based on the cascade failure threshold.

[0029] Further, the online sensing network (310) includes a fiber Bragg grating sensor array implanted inside the material of the device under test (300), a piezoelectric sensor pasted on the surface of the device under test (300), a chemical corrosion probe embedded on the device under test (300), and a miniature acoustic probe installed inside the device under test (300);

[0030] The fiber Bragg grating sensor array is used to monitor strain and temperature changes;

[0031] The piezoelectric sensor is used to detect vibration and stress signals;

[0032] The chemical corrosion probe is used to measure local chemical environmental parameters;

[0033] The miniature acoustic probe is used to capture acoustic emission events generated by material micro-cracks.

[0034] Further, the health state evaluation and feedback control module (400) is further configured to: based on the time series data of the internal state parameters, construct an aging digital twin model of the device under test (300), and further output a remaining useful life prediction value of the device under test;

[0035] The aging digital twin model is a dynamic mathematical model, the input of which is a sequence of external environmental parameters, and the output is an estimated value of a key internal state parameter of the device;

[0036] The remaining useful life prediction value is calculated by integrating the cumulative damage amount of the degradation model.

[0037] In another aspect, a predictive testing method based on digital twin and environmental Monte Carlo simulation, for testing a device under test (300) in a test chamber (200) comprising an environmental parameter executor (210), the method comprising the steps of:

[0038] S100, generating a set of initial multi-dimensional environmental control parameter sequences according to a target environmental model by a climate Monte Carlo engine (100);

[0039] S200, applying environmental stress in the test chamber (200) according to the parameter sequences by the environmental parameter executor (210);

[0040] S300, collecting internal state parameters of the device under test (300) in real time by an online sensing network (310) arranged inside the device under test (300);

[0041] S400, determining whether the internal state parameters reach a preset cascade failure threshold by a health state evaluation and feedback control module (400);

[0042] S500, if the cascade failure threshold is reached, the health state evaluation and feedback control module (400) dynamically adjusts the evolution algorithm of the climate Monte Carlo engine (100) or corrects the parameter sequences to simulate the coupling acceleration effect of environmental factors;

[0043] S600, if the threshold is not reached, continue to perform the test according to the original parameter sequences, if the test reaches a preset time length or triggers a termination condition, end the test process and enter S700.

[0044] Further, the cascade failure threshold in S400 is defined as the strain value when the first microcrack appears in the material, and the leakage rate when the sealing performance decreases by more than a predetermined percentage, the strain value is measured by a fiber Bragg grating sensor, the unit is micro-strain, and the leakage rate is detected by a pressure decay method.

[0045] Further, the method further comprises the step of integrating all collected internal state parameters by the health state evaluation and feedback control module (400) to generate an aging digital twin model of the device under test (300) after the test process is completed;

[0046] S700, the aging digital twin model is trained using a neural network structure, the input layer receives the environmental parameter sequences, the hidden layer includes multiple fully connected layers, and the output layer predicts the device state parameters, the training process uses time series data and optimizes the weights through a back propagation algorithm.

[0047] Compared with the prior art, the predictive test system and method based on digital twinning and environmental Monte Carlo simulation have the following beneficial effects:

[0048] Firstly, the climate Monte Carlo engine can generate a multi-dimensional environmental control parameter sequence dynamically evolving over time and interrelated according to the target environment model, the online perception network can monitor the internal state parameters of the device under test in real time, the health state evaluation and feedback control module evaluates the device health state based on the internal state parameters and dynamically adjusts the evolution algorithm of the climate Monte Carlo engine to correct the multi-dimensional environmental control parameter sequence, forming a closed-loop feedback control, thereby realizing dynamic reproduction of the nonlinear coupling effect of multiple environmental factors and adaptive stress loading based on the real-time state of the device, solving the problem that the prior art cannot reproduce the cascade failure effect caused by the coupling of multiple environmental factors, the test process is open-loop and it is difficult to accurately predict the failure behavior and residual life of the device in the real environment.

[0049] Secondly, the online perception network collects the internal state parameter time series data of the device under test during the whole test period, and the health state evaluation and feedback control module constructs an aging digital twinning model of the device under test, which can dynamically output the estimated value of the internal key state parameters of the device, and calculate the residual effective life prediction value of the device through the cumulative damage amount of the integral degradation model, so as to provide a high-fidelity digital model support for the whole life cycle reliability evaluation of the device under test, and facilitate technicians to intuitively master the device aging law.

[0050] Other advantages, objects, and features of the present application will be apparent to those skilled in the art from the following specification, in some degree of certainty, based on the study of the following, or can be taught from the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creating any creative labor.

[0052] Figure 1 The workflow diagram of the present application;

[0053] Figure 2 The system composition and data interaction diagram of the present application;

[0054] Figure 3 The closed-loop feedback control logic diagram of the present application. Detailed Implementation

[0055] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0056] Example 1

[0057] like Figure 2 and Figure 3 As shown, this embodiment uses the evaluation of the full lifecycle reliability of a smart optical cable splice box in the coastal area of ​​Sanya, Hainan as an application scenario, and elaborates on the specific implementation process of a predictive testing system and method based on digital twins and environmental Monte Carlo simulation. In this embodiment, the device under test is a smart optical cable splice box, the target environment is the natural environmental characteristics of the coastal area of ​​Sanya, Hainan over the past 50 years, and the test objective is to simulate the aging process of the device after 20 years of deployment in this area, while constructing its aging digital twin model and predicting its remaining effective lifespan. Dynamically evolving multi-dimensional environmental parameters are generated through a climate Monte Carlo engine, combined with real-time monitoring of the device's internal state by an online sensing network, and then closed-loop feedback adjustment is achieved by a health status assessment and feedback control module. Finally, the cascading failure modes caused by the nonlinear coupling of multiple environmental factors are reproduced, improving the correlation between the test results and the actual environment and the accuracy of the remaining effective lifespan prediction.

[0058] Overall system configuration:

[0059] The predictive testing system in this embodiment mainly consists of a test chamber 200, an environmental parameter actuator 210, a climate Monte Carlo engine 100, an online sensing network 310, and a health status assessment and feedback control module 400. The connection relationships and functional coordination of each component are as follows:

[0060] The device under test 300 is fixed in the center of the test chamber 200, ensuring that all its key parts are within the range of action of the environmental parameter actuator 210. The control end of the environmental parameter actuator 210 is electrically connected to the output end of the health status assessment and feedback control module 400, receiving control commands to apply environmental stress. The climate Monte Carlo engine 100 is integrated into the industrial control computer and communicates bidirectionally with the health status assessment and feedback control module 400 via Ethernet. On the one hand, it receives the target environmental model parameters, and on the other hand, it outputs a multi-dimensional environmental control parameter sequence. The signal output end of the online sensing network 310 is connected to the input end of the health status assessment and feedback control module 400 through a data acquisition card, transmitting the internal status parameters of the device in real time. Based on the preset degradation model and the collected internal status parameters, the health status assessment and feedback control module 400 completes the health status assessment, engine algorithm adjustment, and actuator command issuance, forming a closed-loop control.

[0061] Detailed structure and parameter configuration of each functional module:

[0062] Test cabin 200:

[0063] The test cabin 200 is a rectangular closed cavity made of 304 stainless steel material, with a high-temperature-resistant and corrosion-resistant polytetrafluoroethylene coating on the inner wall. The cavity volume meets the installation requirements of the measured device 300 and the online sensing network 310, while ensuring that the environmental stress generated by the environmental parameter actuator 210 can uniformly act on the measured device 300. The side wall of the test cabin 200 is provided with an observation window and a data interface. The observation window is used for real-time observation of the state of the measured device 300, and the data interface is used for signal extraction of the online sensing network 310 and pipeline connection of the environmental parameter actuator 210.

[0064] Environmental parameter actuator 210:

[0065] The environmental parameter actuator 210 includes five execution units, and the configuration and functions of each unit are as follows:

[0066] Temperature controller: uses a semiconductor refrigeration / heating integrated module, with a temperature regulation range of -70°C to +150°C, a temperature control accuracy of ±0.5°C, and copper heat dissipation / heating coils uniformly distributed in the inner wall of the test cabin 200. It can achieve rapid temperature rise and fall according to the control command, simulating the day-night temperature difference and seasonal temperature change in the coastal area of Sanya;

[0067] Humidity controller: uses a combination of ultrasonic humidification module and compressor refrigeration dehumidification module, with a humidity regulation range of 10%RH to 98%RH and a humidity control accuracy of ±3%RH. The humidification and dehumidification amount can be adjusted by pulse width modulation (PWM) signal, simulating the high humidity environment in the coastal area of Sanya and the humidity fluctuation before and after rainfall;

[0068] Mechanical stress loading device: uses a six-degree-of-freedom vibration table with a vibration frequency range of 5Hz to 2000Hz, a maximum acceleration of 50g, and a displacement range of ±50mm. It can simulate the vibration caused by typhoon in the coastal area of Sanya and the mechanical stress in the equipment transportation process by presetting the vibration spectrum;

[0069] Electromagnetic interference generator: uses a Helmholtz coil structure to generate a uniform electromagnetic field of 0 to 100V / m, with a frequency range of 1kHz to 1GHz, simulating the electromagnetic interference environment generated by power transmission lines and communication base stations in the coastal area of Sanya;

[0070] Biochemical micro-injection system: composed of 3 liquid storage tanks, 3 precision metering pumps and 3 atomizing nozzles, the liquid storage tanks store acid medium, alkaline medium and salt fog medium respectively, the flow control precision of the precision metering pump is 0.1 milliliter / minute, the atomized particle diameter of the atomizing nozzle is 5 to 10 μm, the nozzles are uniformly distributed on the top of the test cabin 200, and the accurate injection of different corrosive media can be realized according to the control instruction to simulate the marine atmospheric corrosion environment in the coastal area of Sanya.

[0071] The climate Monte Carlo engine 100:

[0072] The climate Monte Carlo engine 100 is a software module running in an industrial computer, developed in C++ language, integrating Markov Chain Monte Carlo (MCMC) algorithm. The engine preloads the target environment model, which includes meteorological statistical data, historical extreme event data and accelerated aging model of the coastal area of Sanya in the past 50 years:

[0073] The meteorological statistical data includes the probability distribution of temperature, humidity, solar intensity, rainfall and typhoon events (temperature conforms to normal distribution N(25,8²)℃, humidity conforms to Beta distribution Beta(2.5,1.2), solar intensity conforms to Weibull distribution Weibull(3.0,200)W / m², rainfall conforms to Poisson distribution P(1500)mm / year, and typhoon events conform to Poisson distribution P(2)times / year) and the correlation matrix (the correlation coefficient of temperature and humidity is-0.6, the correlation coefficient of temperature and solar intensity is 0.8, and the correlation coefficient of rainfall and typhoon events is 0.7);

[0074] The historical extreme event data includes the records of extreme high temperature, extreme low temperature, extreme rainfall and strong typhoon in the coastal area of Sanya in the past 50 years;

[0075] The accelerated aging model is constructed based on the principle of material degradation kinetics, and the Arrhenius equation is used to convert the actual environmental temperature stress into laboratory accelerated temperature stress, and the accelerated factor calculation formula is:

[0076]

[0077] Wherein is the activation energy of the material, is the gas constant, is the average temperature of the actual environment, is the laboratory accelerated temperature, and the calculation result is .

[0078] The climate Monte Carlo engine 100 generates a multi-dimensional environment control parameter sequence through an adaptive Monte Carlo evolution algorithm, and the mathematical expression of the algorithm is as follows:

[0079]

[0080] Online sensing network 310:

[0081] Online sensing network 310 is composed of 5 fiber Bragg grating (FBG) sensors, 2 micro-humidity sensors and 1 MEMS acoustic probe. The installation position and function of each sensor are as follows:

[0082] The 5 FBG sensors are pasted on the stress concentration points of the shell of the measured device 300, near the seal ring and the surface of the internal circuit board by epoxy resin. The center wavelength of the sensor is 1550 nm, the wavelength resolution is 1 pm, the strain measurement range is -1500 to +1500 με, the temperature measurement range is -50 to +150℃, and it is used to monitor the internal strain and temperature changes of the device in real time;

[0083] The 2 micro-humidity sensors are installed in the form of a patch inside the sealed cavity and outside the shell of the measured device 300. The measurement range is 0%RH to 100%RH, and the measurement accuracy is ±2%RH. It is used to monitor the humidity difference inside and outside the device in real time and evaluate the sealing performance;

[0084] The 1 MEMS acoustic probe is fixed by screw inside the shell of the measured device 300. The frequency response range is 20 kHz to 1 MHz, and the sensitivity is -70 dBV / Pa. It is used to capture the acoustic emission signals generated by material micro-cracks and identify early signs of failure.

[0085] Health status evaluation and feedback control module 400:

[0086] The health status evaluation and feedback control module 400 is a hardware module based on an embedded system, which uses an ARM Cortex-A9 processor and integrates data acquisition interfaces, Ethernet communication interfaces and control signal output interfaces. The degradation model of the measured device 300 is embedded in the module. The model takes the strain value monitored by the FBG sensor, the humidity difference value monitored by the micro-humidity sensor and the acoustic emission signal amplitude monitored by the MEMS acoustic probe as input, and takes the device health index (0 to 1, 1 represents complete health, 0 represents failure) as output. A health status evaluation model is constructed by a multivariate linear regression algorithm:

[0087]

[0088] Wherein is the health index, is the actual strain value, is the cascade failure threshold strain, is the actual humidity difference value, is the cascade failure threshold humidity difference value, is the actual acoustic emission signal amplitude, Cascade failure threshold acoustic emission signal amplitude.

[0089] The module simultaneously integrates an aging digital twin model training function, the model adopts a neural network structure, an input layer contains 5 neurons, a hidden layer contains 3 fully connected layers, an output layer contains 3 neurons, a training process adopts time sequence data, network weights are optimized through a back propagation algorithm, and a convergence condition is that a loss function is less than 0.001.

[0090] Test method specific implementation steps:

[0091] Step one: target environment model configuration and initial parameter sequence generation (corresponding to method step S201)

[0092] An operator inputs target environment model parameters into the climate Monte Carlo engine 100 through a man-machine interface of an industrial computer, including meteorological statistical data in the coastal area of San Ya in the past 50 years, historical extreme event data and accelerated aging model parameters, and sets a test target as simulating a 20-year service life.

[0093] The climate Monte Carlo engine 100 starts an adaptive Monte Carlo evolution algorithm to generate an initial multi-dimensional environment control parameter sequence, a time resolution of the sequence is 1 second, a total time length is 2 years, the sequence contains parameter values in five dimensions of temperature, humidity, vibration acceleration, salt fog concentration and electromagnetic field intensity, for example, parameter values at a moment are: temperature 35 DEG C, humidity 85% RH, vibration acceleration 5g, salt fog concentration 0.5mg / m³, and electromagnetic field intensity 50V / m.

[0094] Step two: environmental stress application (corresponding to method step S202)

[0095] The health state evaluation and feedback control module 400 receives the initial multi-dimensional environment control parameter sequence output by the climate Monte Carlo engine 100 and converts it into control instructions of each environment parameter actuator 210:

[0096] A PWM signal is sent to a temperature controller to control a heating module of the temperature controller to work, so that the temperature in the test cabin 200 rises to 35 DEG C and the temperature is maintained;

[0097] A PWM signal is sent to a humidity controller to control a humidification module of the humidity controller to work, so that the humidity in the test cabin 200 rises to 85% RH and the humidity is maintained;

[0098] A digital signal is sent to a six-degree-of-freedom vibration table to control the vibration table to output a 5g, 100Hz sine vibration;

[0099] A pulse signal is sent to a biochemical micro-injection system to control a precision metering pump to inject a salt fog medium at a flow rate of 0.3 milliliter / minute, so that a 0.5mg / m³ salt fog concentration is formed through an atomizing nozzle;

[0100] The current signal is sent to the Helmholtz coil to control it to generate an electromagnetic field of 50V / m.

[0101] The environmental parameter actuators 210 work synchronously according to the control instructions to build an environmental stress field matching the initial parameter sequence in the test cabin 200.

[0102] Step three: device internal state parameter acquisition (corresponding to method step S203)

[0103] The online sensing network 310 acquires the internal state parameters of the device 300 in real time at a sampling frequency of 10Hz:

[0104] The FBG sensor converts the monitored strain and temperature signals into wavelength shift signals, which are transmitted to the data acquisition card through the optical fiber, and the data acquisition card converts the wavelength signals into digital signals;

[0105] The micro humidity sensor converts the monitored humidity signal into a voltage signal, which is transmitted to the data acquisition card through the wire, and the data acquisition card converts the voltage signal into a digital signal and calculates the internal and external humidity difference of the device;

[0106] The MEMS acoustic probe converts the monitored acoustic emission signal into a voltage signal, which is transmitted to the data acquisition card through the wire, and the data acquisition card converts the voltage signal into a digital signal.

[0107] The collected internal state parameters are transmitted to the health state evaluation and feedback control module 400 in real time through Ethernet, and the module pre-processes the data to remove high-frequency interference signals and outliers.

[0108] Step four: cascade failure threshold judgment (corresponding to method step S204)

[0109] The health state evaluation and feedback control module 400 inputs the pre-processed internal state parameters into the preset degradation model to calculate the device health index and judge whether each internal state parameter reaches the cascade failure threshold:

[0110] If the strain value monitored by the FBG sensor is ≤500με, the humidity difference value monitored by the micro humidity sensor is ≤20%RH, and the acoustic emission signal amplitude monitored by the MEMS acoustic probe is ≤50dB, i.e. HI≥0.8, it is determined that the cascade failure threshold is not reached, and step S202 is returned to continue applying the environmental stress according to the original parameter sequence.

[0111] If any internal state parameter exceeds the cascade failure threshold (for example, the strain value monitored by the FBG sensor reaches 550με, that is:

[0112] HI = 0.4(1-550 / 500) + 0.3(1-15 / 20) + 0.3(1-40 / 50) = 0.4(-0.1) + 0.30.25 + 0.30.2 = -0.04 + 0.075 + 0.06 = 0.095 < 0.8), indicating that the cascading failure threshold has been reached, and step S205 is executed.

[0113] In this embodiment, when the test ran for 0.6 years of accelerated laboratory time, the FBG sensor detected that the strain value at the stress concentration point of the device under test 300 shell reached 550με, and determined that the cascade failure threshold had been reached.

[0114] Step 5: Adjust the environmental parameter sequence (corresponding to step S205)

[0115] The health status assessment and feedback control module 400 dynamically adjusts the evolution algorithm parameters of the climate Monte Carlo engine 100 based on the internal state parameters that have reached the cascading failure threshold, and corrects the multidimensional environmental control parameter sequence.

[0116] In the evolutionary algorithm of the climate Monte Carlo Engine 100 The value was adjusted from the initial 0.2 to -0.3, which reduces the degree of drastic change in environmental parameters through negative adjustment and avoids rapid equipment failure;

[0117] The salt spray concentration parameter in the multidimensional environmental control parameter sequence was modified by increasing the salt spray concentration in the original sequence by 50% (e.g., from 0.5 mg / m³ to 0.75 mg / m³) to simulate the cascade failure effect of microcracks accelerating salt spray corrosion.

[0118] A control command is sent to the biochemical micro-injection system to increase the injection flow rate of the salt spray medium from 0.3 mL / min to 0.45 mL / min, maintaining the corrected salt spray concentration.

[0119] The adjusted multidimensional environmental control parameter sequence is fed back to the climate Monte Carlo engine 100 via Ethernet. The engine continues to generate subsequent parameter sequences based on the adjusted algorithm parameters, returns to step S202, and applies environmental stress according to the corrected parameter sequence.

[0120] Step Six: Construction of Aging Digital Twin Model and Prediction of Remaining Effective Life (Corresponding Method Step S206)

[0121] When the test reaches the second year of the accelerated laboratory duration, the test termination condition is triggered. The health status assessment and feedback control module 400 integrates the time series data of internal status parameters collected throughout the test and inputs it into the aging digital twin model for training.

[0122] The neural network is trained using the backpropagation algorithm with the environmental parameter sequence of the entire test as input data and the corresponding internal state parameter sequence as output data.

[0123] During training, the loss function is calculated every 100 iterations. When the iteration reaches 5000, the loss function drops to 0.0008, satisfying the convergence condition. Training is then stopped, and the final aging digital twin model is obtained.

[0124] Based on the trained aging digital twin model, the current internal state parameters of the device under test 300 are input, and the model outputs the future evolution trend of the internal state parameters. By integrating the cumulative damage of the degradation model, the remaining effective lifespan of the device under test 300 in the Sanya coastal area is calculated to be 2 years.

[0125] Detailed explanation of the parameters of the climate Monte Carlo engine evolution algorithm:

[0126] The meaning, value basis, and physical significance of each parameter in the mathematical expression of the adaptive Monte Carlo evolution algorithm used by the Climate Monte Carlo Engine 100 are as follows:

[0127] Indicates time step The multidimensional environmental control parameter vector is determined by the number of environmental factors in the target environment model. In this embodiment, the target environment model includes five environmental factors: temperature, humidity, vibration acceleration, salt spray concentration, and electromagnetic field strength. For a 5-dimensional vector, the expression is: ,in for Temperature at all times for Humidity at all times for Constant vibration acceleration, for Constant salt spray concentration, for Electromagnetic field strength at any given time The initial value is determined by the mean of the meteorological statistics data of the target environment model. In this embodiment, the initial value is...

[0128] .

[0129] : Indicates the time step The updated multidimensional environmental control parameter vector is the algorithm's output, used to guide the operation of the environmental parameter actuator 210. Its dimensions are... Consistent, passed The calculation, which is superimposed with subsequent adjustment terms, reflects the dynamic evolution trend of environmental parameters over time.

[0130] : represents the environmental evolution rate coefficient, which is a positive real number, and its value range is 0.1 to 1.0, used to control the overall speed of the change of the environmental parameters, The greater the value is, the greater the change range of the environmental parameters in the adjacent time step is, The smaller the value is, the smaller the change range is; in this embodiment, based on the flatness of the environmental change in the coastal area of Sanya, The initial value is 0.8, which ensures that the environmental parameter evolution conforms to the actual environmental change rhythm.

[0131] : represents the environmental factor coupling matrix, which is a symmetric matrix, and its dimension is consistent with The matrix element represents the correlation strength between the th environmental factor and the th environmental factor, and its value range is -1 to 1, a positive value represents a positive correlation, a negative value represents a negative correlation, and the greater the absolute value is, the stronger the correlation is; in this embodiment, the value of is:

[0132]

[0133] Among them, represents that the temperature and humidity are negatively correlated, represents that the humidity and salt mist concentration are positively correlated, which conforms to the environmental characteristics of the coastal area of Sanya.

[0134] : represents the external environmental disturbance vector, and its dimension is consistent with Its element value is randomly generated based on the historical extreme event data in the target environmental model, and the generation probability is determined by the occurrence frequency of the historical extreme event; for example, when simulating a typhoon event, the disturbance value of the vibration acceleration is 5g, the disturbance value of the temperature is -2°C, the disturbance value of the humidity is 15%RH, the disturbance value of the salt mist concentration is 0.3mg / m3, and the disturbance value of the electromagnetic field intensity is 10V / m, The value of ensures that the simulated extreme environment conforms to the actual situation of the coastal area of Sanya.

[0135] : represents the state feedback adjustment coefficient, which is a real number, and its value range is -0.5 to 0.5, and it is dynamically adjusted by the health state evaluation and feedback control module 400 according to the device health index ; when ≥0.8, take a positive value, increase the change range of the environmental parameters through positive adjustment, and speed up the aging process; when <0.8, Taking negative values, reducing the amplitude of environmental parameter changes through negative regulation, avoiding rapid failure of equipment, and ensuring accurate capture of cascading failure process; in this embodiment, β = 0.2 before 0.6 years before testing; after testing for 0.6 years, = -0.3.

[0136] : represents the time step of the device internal state parameter vector, the dimension is determined by the number of monitoring parameters of the online perception network 310, in this embodiment, the online perception network 310 monitors three parameters of strain, humidity difference and acoustic emission signal amplitude, so is a 3-dimensional vector, the expression is , wherein is the strain value at , is the humidity difference value at , is the acoustic emission signal amplitude at , the value of is collected by the online perception network 310 in real time and transmitted to the health state evaluation and feedback control module 400. : represents the device state response function, which is a function of

[0137] , the input is , and the output is the standardized adjustment amount calculated based on the degradation model, the dimension is consistent with ; in this embodiment, the calculation expression of is:

[0138]

[0139] , wherein to are weight coefficients, the change of internal state parameters is converted into the adjustment amount of environmental parameters through standardization processing, for example, when , ;

[0140] , the adjustment amount of the corresponding temperature parameter in is , realizing fine adjustment of environmental parameters based on the state of the equipment.

[0141] ​​In summary, the present embodiment takes the reliability evaluation of intelligent optical cable splice closure in the coastal area of Sanya, Hainan as the application scenario, and elaborates the specific configuration of the predictive testing system based on digital twin and environmental Monte Carlo simulation, the implementation steps of the testing method, and the meaning of the core algorithm parameters. Through the climate Monte Carlo engine, dynamic evolution of multi-dimensional environmental parameters is generated, combined with real-time monitoring of the internal state of the device by the online perception network, and then the health state evaluation and feedback control module realizes closed-loop feedback regulation, effectively solving the technical defects of traditional testing methods that cannot simulate the nonlinear coupling effect of multiple environmental factors and the open-loop of the testing process. The implementation results show that the system and method can reproduce the aging process of the device in the actual environment with high fidelity, accurately predict the remaining useful life, and provide a reliable technical means for environmental adaptability evaluation of critical infrastructure equipment.

[0142] Embodiment Two

[0143] Based on the system configuration described in Embodiment 1, the present embodiment elaborates the workflow of a predictive testing system and method based on digital twin and environmental Monte Carlo simulation. The specific steps are as follows:

[0144] 1. System initialization and target environment model loading:

[0145] Start the climate Monte Carlo engine and load the preset target environment model, including meteorological statistical data of the target geographic location, historical extreme event data, and accelerated aging model.

[0146] Set the test parameters, such as the total test duration, termination conditions (such as device failure or reaching the preset time), and initialize the actuators and sensors.

[0147] 2. Generate initial environmental parameter sequence:

[0148] The climate Monte Carlo engine generates a set of initial multi-dimensional environmental control parameter sequences based on the target environment model, including time evolution data of temperature, humidity, vibration, salt spray concentration, electromagnetic field strength, etc.

[0149] 3. Environmental stress application:

[0150] The health state evaluation and feedback control module converts the generated environmental parameter sequence into control instructions and sends them to the environmental parameter actuators.

[0151] The environmental parameter actuators (including temperature controllers, humidity controllers, mechanical stress loading devices, electromagnetic interference generators, and biochemical micro-injection systems) synchronously apply corresponding environmental stresses in the test cabin.

[0152] 4. Real-time state monitoring:

[0153] The online sensing network collects the internal state parameters of the device under test in real time at a high sampling rate, including strain, temperature, humidity difference, acoustic emission signals, etc.

[0154] The collected data is transmitted to the health state assessment and feedback control module through the data acquisition card.

[0155] 5. Health state assessment and threshold judgment:

[0156] The health state assessment and feedback control module calculates the device health index based on the pre-set degradation model, and judges whether the internal state parameters have reached the cascade failure threshold (such as strain exceeding limit, sealing performance degradation, etc.).

[0157] At the same time, check whether the test termination condition is met (such as reaching the pre-set time length or the device health index being lower than the critical value).

[0158] 6. Feedback control and parameter adjustment:

[0159] If the termination condition is not met and the cascade failure threshold is not reached, the system continues to perform the test according to the current environmental parameter sequence, and returns to step 3.

[0160] If the cascade failure threshold is reached, the health state assessment and feedback control module dynamically adjusts the evolution algorithm parameters of the climate Monte Carlo engine (such as state feedback adjustment coefficient), and corrects the environmental parameter sequence to simulate the multi-factor coupling acceleration effect. After adjustment, return to step 3 to continue testing.

[0161] 7. Test termination and data integration:

[0162] When the termination condition is reached, stop applying environmental stress, and the health state assessment and feedback control module integrates all collected internal state parameter time series data.

[0163] 8. Construction of aging digital twin model and life prediction:

[0164] Use the integrated data to train the aging digital twin model, which uses a neural network structure with environmental parameter sequence as input and device state parameter as output.

[0165] Based on the trained model, predict the remaining useful life of the device under test in the real environment.

[0166] The whole process forms a closed-loop feedback control, realizes adaptive environmental testing based on real-time device state, and effectively reproduces the device aging process under the coupling effect of multiple environmental factors.

[0167] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed with the preferred embodiments as above, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, as long as the changes or modifications do not deviate from the technical solution of the present application. Any modification, change, equivalent change and modification of the above embodiments made according to the technical essence of the present application still belong to the scope of the technical solution of the present application.

Claims

1. A predictive testing system based on digital twin and environmental Monte Carlo simulation, comprising a test chamber (200) for housing a device under test (300), and at least one environmental parameter actuator (210) connected to said test chamber (200), characterized in that, The system also includes a climate Monte Carlo engine (100), an online sensing network (310) deployed inside the device under test (300), and a health status assessment and feedback control module (400). The climate Monte Carlo engine (100) is configured to generate a set of multidimensional environmental control parameters that evolve dynamically over time and are interconnected, based on a preset target environment model. The online sensing network (310) is used to monitor the internal status parameters of the device under test (300) in real time; One end of the health status assessment and feedback control module (400) is connected to the online sensing network (310), and the other end is connected to the climate Monte Carlo engine (100) and the environmental parameter actuator (210); The health status assessment and feedback control module (400) is configured to receive the internal status parameters and assess the health status of the device under test (300) based on a preset degradation model, and at the same time dynamically adjust the evolution algorithm of the climate Monte Carlo engine (100) according to the health status to correct the multidimensional environmental control parameter sequence to form a closed-loop feedback control.

2. The predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 1, characterized in that, The target environment model includes meteorological statistics of the target geographical location, historical extreme event data, and a preset accelerated aging model; The meteorological statistics data shall include at least the probability distribution and correlation matrix of temperature, humidity, solar radiation intensity, rainfall, and typhoon events; The historical extreme event data refers to the record of extreme weather conditions that occurred in the target area within a preset historical period. The accelerated aging model is a model framework that compresses actual environmental stress into laboratory accelerated test stress through mathematical transformation, and it is constructed based on the principle of material degradation kinetics.

3. The predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 1, characterized in that, The climate Monte Carlo engine (100) uses an adaptive Monte Carlo evolution algorithm to generate the multidimensional environmental control parameter sequence. This algorithm realizes the dynamic evolution of environmental parameters through the following mathematical formula: in, Indicates the time step The multidimensional environmental control parameter vector is determined by the number of factors in the target environment model. Indicates the time step The updated multidimensional environmental control parameter vector; This is the environmental evolution rate coefficient, a positive real number used to control the overall rate of change of parameters; The environmental factor coupling matrix is ​​a symmetric square matrix, and the matrix elements represent the correlation strength between different environmental parameters. The external environment disturbance vector is randomly generated based on historical data in the target environment model. This is the status feedback adjustment coefficient, which is a real number and is dynamically adjusted by the output of the health status assessment and feedback control module (400); The device state response function takes as input the internal state parameter vector monitored by the online sensing network (310). The output is a standardized adjustment calculated based on the degradation model.

4. The predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 1, characterized in that, The environmental parameter actuator (210) includes at least two of the following: a temperature controller, a humidity controller, a mechanical stress loading device, an electromagnetic interference generator, and a biochemical micro-injection system; The temperature controller can realize cooling and heating functions and accurately adjust within a preset temperature range; The humidity controller maintains the relative humidity within the test chamber (200) within a set range through humidification and dehumidification modules; The mechanical stress loading device is a multi-degree-of-freedom vibration table that can simulate various mechanical environments. The electromagnetic interference generator produces a controllable electromagnetic field to simulate electromagnetic interference conditions. The biochemical micro-injection system consists of a storage tank, a precision metering pump, and a nozzle, and is used to inject corrosive media into the test chamber (200).

5. A predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 4, characterized in that, The biochemical micro-injection system is configured to inject acidic media, alkaline media, and salt spray media into the test chamber (200) according to the instructions of the health status assessment and feedback control module (400); The acidic medium is a solution with a pH value less than 7, the alkaline medium is a solution with a pH value greater than 7, and the salt spray medium is formed by atomizing a sodium chloride solution; The injection process is controlled by a precision metering pump with a flow rate of 0.1 ml per minute, and the injection timing and duration are dynamically triggered by the health status assessment and feedback control module (400) based on the cascade failure threshold.

6. The predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 1, characterized in that, The online sensing network (310) includes a fiber optic grating sensor array embedded in the material of the device under test (300), a piezoelectric sensor attached to the surface of the device under test (300), a chemical corrosion probe embedded in the device under test (300), and a miniature acoustic probe installed inside the device under test (300). The fiber grating sensor array is used to monitor strain and temperature changes; The piezoelectric sensor is used to detect vibration and stress signals; The chemical corrosion probe is used to measure local chemical environment parameters; The microacoustic probe is used to capture acoustic emission events generated by microcracks in the material.

7. A predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 1, characterized in that, The health status assessment and feedback control module (400) is also configured to: construct an aging digital twin model of the device under test (300) based on the time series data of the internal status parameters, and further output the predicted value of the remaining effective lifespan of the device under test; The aging digital twin model is a dynamic mathematical model whose input is a sequence of external environmental parameters and whose output is an estimated value of key internal state parameters of the equipment. The predicted remaining effective lifetime is calculated by integrating the cumulative damage amount of the degradation model.

8. A predictive testing method based on digital twin and environmental Monte Carlo simulation, applicable to the predictive testing system based on digital twin and environmental Monte Carlo simulation as described in any one of claims 1-7, for testing the device under test (300) in a test chamber (200) containing an environmental parameter actuator (210), characterized in that, The method includes the following steps: S100, a set of initial multidimensional environmental control parameter sequences are generated by a climate Monte Carlo engine (100) based on the target environment model; S200, the environmental parameter actuator (210) applies environmental stress in the test chamber (200) according to the parameter sequence; S300: The internal state parameters of the device under test (300) are collected in real time through an online sensing network (310) deployed inside the device under test (300); S400, A health status assessment and feedback control module (400) determines whether the internal status parameters have reached the preset cascading failure threshold; S500 If the cascading failure threshold is reached, the health status assessment and feedback control module (400) dynamically adjusts the evolution algorithm of the climate Monte Carlo engine (100) or corrects the parameter sequence to simulate the coupling acceleration effect of environmental factors. S600: If the threshold is not reached, continue to execute the test according to the original parameter sequence. If the test reaches the preset duration or triggers the termination condition, the test process ends and proceeds to S700.

9. A predictive testing system based on digital twins and Monte Carlo simulation of the environment according to claim 8, characterized in that, The cascading failure threshold in S400 is defined as the strain value when the first microcrack appears inside the material and the leakage rate when the sealing performance decreases by more than a predetermined percentage. The strain value is measured by a fiber optic grating sensor and the unit is microstrain. The leakage rate is detected by the pressure attenuation method.

10. A predictive testing system based on digital twins and environmental Monte Carlo simulation according to claim 8, characterized in that, The method also includes the step of integrating all collected internal status parameters to generate an aging digital twin model of the device under test (300) after the test process is completed; S700 The aging digital twin model is trained using a neural network structure. The input layer receives the sequence of environmental parameters, the hidden layer contains multiple fully connected layers, and the output layer predicts the device state parameters. The training process uses time series data and optimizes the weights through the backpropagation algorithm.

Images