Predictive analysis method and system for laboratory hvac system based on digital twinning

Abstract

The application relates to the technical field of clean laboratory environments, and discloses a predictive analysis method and system for a laboratory heating and ventilation system based on digital twinning. First, a plurality of first target data and a first reference value are acquired in a first time period, and first trend data and first frequency spectrum data are calculated through a first trend algorithm and a first conversion algorithm respectively. Then, a plurality of second target data and a second reference value are acquired in a second time period, and second trend data and second frequency spectrum data are calculated. Next, the difference between the first and second trend data is calculated as first state information based on a digital twinning platform, and the difference frequency band between the first and second frequency spectrum data is calculated as second state information. Then, corresponding equipment state data is matched according to the state information, and the similarity value of the two is calculated. If the similarity value is less than a preset equipment reference similarity value, an equipment early warning prompt is sent. The application can effectively promote hardware linkage, realize accurate prediction of equipment state, and reduce equipment failure risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of clean laboratory environment, and in particular to a predictive analysis method and system for a laboratory heating and ventilation system based on digital twinning. BACKGROUND

[0002] A clean laboratory refers to a laboratory that controls dust particles, microorganisms and other pollutants in indoor air to a certain level or below through specific design, construction and equipment configuration to meet the environmental cleanliness requirements of specific experiments or production processes. The laboratory needs various work equipment to maintain a clean working environment, so the work equipment in the laboratory needs to be operated uninterruptedly for 24 hours, and intelligent hardware is needed to monitor the work equipment.

[0003] Intelligent hardware in the laboratory refers to hardware devices and systems that use sensors, Internet of Things and other technologies to intelligently monitor, control and manage various equipment, instruments, environments and other equipment in the laboratory. At present, sensor technology is relatively mature and can accurately measure various physical quantities, chemical quantities and biological quantities in the laboratory, such as temperature, humidity, air pressure, gas concentration, light intensity, pH value, etc. At the same time, the accuracy, stability and reliability of sensors are continuously improved, and the volume is getting smaller and smaller, and the power consumption is getting lower and lower, which can meet the needs of different laboratory environments and application scenarios. These intelligent hardware can realize automatic data collection, real-time state monitoring and other functions, aiming to improve the operation efficiency, safety and reliability of the laboratory, reduce labor costs and experimental errors, and provide a more convenient and efficient experimental environment for researchers.

[0004] However, intelligent hardware in the laboratory involves multiple hardware and subsystems of different brands, such as sensor systems, communication systems, control systems, etc., and the system complexity is high. In the integration process, more data of each subsystem is collected and summarized in the background, and the data of each subsystem is relatively independent, and the linkage between hardware cannot be realized, and the device state prediction between the hardware of the subsystem cannot be realized. SUMMARY

[0005] In order to realize the linkage between the hardware and the device state prediction between the hardware of the subsystem, the present application provides a predictive analysis method and system for a laboratory heating and ventilation system based on digital twinning.

[0006] In a first aspect, the present application provides a predictive analysis method for a laboratory heating and ventilation system based on digital twinning, which adopts the following technical solution:

[0007] A predictive analysis method for a laboratory heating and ventilation system based on digital twinning, comprising the following steps:

[0008] acquire a plurality of first target data and a plurality of first reference values in a first time period;

[0009] calculate first trend data according to a plurality of the first target data by a first trend algorithm, and calculate first frequency spectrum data according to a plurality of the first reference values by a first conversion algorithm;

[0010] acquire a plurality of second target data and a plurality of second reference values in a second time period;

[0011] calculate second trend data according to a plurality of the second target data by a first trend algorithm, and calculate second frequency spectrum data according to a plurality of the second reference values by a first conversion algorithm;

[0012] calculate a difference between the first trend data and the second trend data as first state information based on a digital twin platform, and calculate a difference frequency band between the first frequency spectrum data and the second frequency spectrum data as second state information;

[0013] match first equipment state data according to the first state information, match second equipment state data according to the second state information, calculate a similarity value between the first equipment state data and the second equipment state data, and if the similarity value is less than a preset equipment reference similarity value, issue an equipment early warning prompt.

[0014] By adopting the above technical solution, the digital twin intelligent management and control platform can realize real-time monitoring of the health state of the equipment by collecting key operating parameters of the equipment such as temperature, vibration, noise, etc. in real time. Real-time remote monitoring of the clean room is realized, and environmental abnormalities and equipment failures are discovered and warned in a timely manner, so as to protect the living environment of experimental animals and improve the accuracy and reliability of experimental data.

[0015] Optionally, based on a temperature sensor and a vibration sensor installed on the working equipment, the temperature sensor is electrically connected with a first edge computing unit, and the vibration sensor is electrically connected with a second edge computing unit, and the first edge computing unit and the second edge computing unit are both remotely connected with the digital twin platform.

[0016] The method further comprises the following steps:

[0017] The first edge computing unit acquires a plurality of the first target data and a plurality of the second target data from the temperature sensor, calculates the first trend data and the second trend data by a first trend algorithm, and the first trend algorithm is used to calculate the average fluctuation value of a plurality of values.

[0018] The second edge computing unit obtains a plurality of first reference values and a plurality of second reference values from the vibration sensor, and calculates the first frequency spectrum data and the second frequency spectrum data by using a first conversion algorithm, wherein the first conversion algorithm is used to convert a plurality of values in a time domain into frequency spectrum data in a frequency domain;

[0019] The digital twin platform obtains the first trend data, the second trend data, the first frequency spectrum data and the second frequency spectrum data, calculates a difference value or an average difference value between the first trend data and the second trend data as first state information, and calculates an energy difference value of a corresponding frequency band in the first frequency spectrum data and the second frequency spectrum data, and calculates a continuous frequency band with the energy difference value greater than a preset energy reference value as second state information.

[0020] By using the above technical solution, the edge computing unit is an AI edge server, the AI edge server is used to analyze data, abnormal conditions are found in time, early warning information is sent to an operator of a management and control platform in time, personnel and spare parts are prepared when actual faults of equipment have not occurred, the service life of the equipment is ensured to achieve greater utilization, and the inventory preparation of personnel and spare parts is better controlled, so that an optimal balance between practicality and economy is achieved.

[0021] Optionally, in the method for calculating the similarity value between the first device state data and the second device state data, the following sub-steps are further included:

[0022] A first dynamic gain coefficient is introduced for the first device state data, the first dynamic gain coefficient is dynamically adjusted according to a non-linear decay function K1(t)=e^(-αt) according to a device uncalibrated duration t, wherein α is a preset decay coefficient;

[0023] A second dynamic gain coefficient is introduced for the second device state data, the second dynamic gain coefficient is dynamically adjusted according to an inverse proportional function K2(P)=β / (P+γ) according to a real-time monitored working device power P, wherein β and γ are power compensation parameters;

[0024] Based on the first dynamic gain coefficient and the second dynamic gain coefficient, a matching degree index of a fluctuation amplitude of weighted device state data is calculated:

[0025] Match=Σ[K1(t)×ΔS1(i)-K2(P)×ΔS2(i)]^2;

[0026] Wherein ΔS1(i) and ΔS2(i) are respectively an i-th fluctuation component of the first device state data and the second device state data;

[0027] Similarity value = 1 / (1+Match).

[0028] By adopting the above technical solution, the larger the value of the similarity value, the higher the degree of similarity. Accurately match the device status data: By matching the fluctuation of the first device status data with the change amplitude of the second device data, it is possible to gain a more accurate insight into the relationship between the status of the devices. This feature helps to quickly and accurately judge the synergy of the operating status of different devices in complex equipment systems, thereby providing a solid data foundation for the overall optimized operation of the equipment. Gain value that adapts to dynamic changes: The setting of the first gain value becoming smaller as the uncalibrated time increases, and the second gain value decreasing as the power of the working equipment increases, enables this patented technology to dynamically adapt to different operating conditions of the equipment. When the equipment is operated for a long time without calibration, the change in the first gain value can effectively reduce the impact of errors accumulated over time on the accuracy of the data; and the second gain value is adjusted according to the equipment power, which can always maintain accurate measurement of the equipment status data when the equipment power fluctuates, greatly improving the reliability of equipment status monitoring under different working conditions.

[0029] Optionally, based on an image sensor and a vibration sensor installed on the working equipment, the image sensor is electrically connected to a first edge computing unit, the vibration sensor is electrically connected to a second edge computing unit, and both the first edge computing unit and the second edge computing unit are remotely connected to the digital twin platform;

[0030] The method further comprises the steps of:

[0031] The first edge computing unit acquires a plurality of first target data and a plurality of second target data from the image sensor, and calculates the first trend data and the second trend data by using a first trend algorithm, wherein the first trend algorithm is used to calculate a target state value corresponding to a preset template in the plurality of image data;

[0032] The second edge computing unit obtains a plurality of first reference values ​​and a plurality of second reference values ​​from the vibration sensor, and calculates the first spectrum data and the second spectrum data by using a first conversion algorithm, wherein the first conversion algorithm is used to convert a plurality of values ​​in the time domain into spectrum data in the frequency domain;

[0033] The digital twin platform obtains the first trend data, the second trend data, the first frequency spectrum data and the second frequency spectrum data, calculates a difference or an average difference between the first trend data and the second trend data as first state information, and calculates an energy difference of a corresponding frequency band in the first frequency spectrum data and the second frequency spectrum data, and calculates a continuous frequency band with the energy difference greater than a preset energy reference value as second state information.

[0034] By adopting the technical solution, based on the image sensor and the vibration sensor installed on the working equipment, the first and second edge computing units electrically connected with the image sensor and the vibration sensor process the acquired data, that is, the first edge computing unit calculates the target state value of the image data to obtain trend data through a specific trend algorithm, and the second edge computing unit converts the vibration sensor data from time domain to frequency domain to obtain frequency spectrum data. The differences between different parameters on the working equipment are compared through the detection of the equipment state in the image. The digital twin platform calculates the state information based on the data, can effectively analyze the multi-dimensional data of the working equipment, timely find the abnormal trend and potential fault in the equipment operation process, realize the remote real-time monitoring and early warning of the equipment, help to ensure the stable operation of the equipment, improve the working efficiency and reliability, and at the same time, fully utilize the edge computing to reduce the data transmission pressure and improve the overall performance of the system.

[0035] Optionally, the method further comprises the following steps:

[0036] Real-time acquisition of the three-phase current, voltage and power factor of the working equipment, and extraction of the time domain feature vector of the power parameter based on a sliding time window;

[0037] Adopting a dynamic threshold algorithm to detect the time domain feature vector, and the dynamic threshold algorithm calculates an abnormal index according to the following formula:

[0038] ;

[0039] Wherein, X i (t) is the i-th power parameter at time t, μ i (t), σ i (t) are the mean and standard deviation of the previous N-minute sliding window, and ω i is a preset weight coefficient;

[0040] When the abnormal index exceeds a preset adaptive threshold, the frequency domain harmonic component of the corresponding power parameter is extracted, and is matched with the fault mode feature corresponding to the frequency domain harmonic component in the historical energy consumption database to generate an abnormal fluctuation spectrum diagram;

[0041] Input the abnormal fluctuation spectrum diagram into a pre-trained deep residual network model, output a prediction result, the prediction result includes a device health degree prediction value and a remaining life prediction interval, and store the prediction result into a blockchain storage database after associating the prediction result with a timestamp of the prediction result.

[0042] By adopting the technical scheme, the energy monitoring system accurately monitors the energy use of each room and device through monitoring equipment, statistically analyzes the energy consumption abnormality, and finds unreasonable parts in the energy distribution scheme in the system. Through the learning analysis of artificial intelligence, the comprehensive performance energy monitoring of all systems can not only generate an optimized energy utilization scheme, but also take emergency plans in advance according to historical energy consumption data and energy consumption peaks in time and season.

[0043] Optionally, the method further comprises the following steps:

[0044] Synchronously collecting heterogeneous data of the physical entity by the multi-source sensor, the heterogeneous data including stress-strain distribution data of a mechanical structure, three-dimensional geometric point cloud data, and turbulent intensity spectrum data and multiphase flow pressure field data of a fluid system;

[0045] Constructing a finite element model generator based on a graph neural network, inputting the stress-strain distribution data and the geometric point cloud data into the finite element model generator, and iteratively optimizing grid division:

[0046] ;

[0047] Wherein, K i is an element stiffness matrix, u i is a node displacement vector, f i is an external force vector acting on the node, TV(u) is a total variation regularization term, the above parameters are calculated through the stress-strain distribution data and the geometric point cloud data, and λ is a balance coefficient;

[0048] A fluid dynamics model generator is constructed by adopting a space-time convolution network, the turbulent intensity spectrum data and the multiphase flow pressure field data are mapped to a fluid dynamics grid, and a Navier-Stokes equation residual is solved by a differentiable solver:

[0049] ;

[0050] Wherein, v is a fluid velocity vector, t is time, is the kinematic viscosity, p is the pressure; (v·∇)v is the convection term; is the viscous diffusion term; ∇p is the pressure gradient;

[0051] The finite element model provides output gridded data, which has a matching data interface for coupling with the fluid dynamics grid, for updating the dynamic grid boundary conditions in the fluid dynamics grid for dynamic correction of the R NS ;

[0052] An online model update engine is deployed to receive a data stream of a multi-source sensor of a physical entity update in real time, and dynamically adjust the weight parameters of the finite element model generator and the fluid dynamics model generator through an adaptive gradient descent algorithm:

[0053] ;

[0054] wherein, is the measured data at time t; M t is the model prediction value; JS is the Jensen-Shannon divergence, and α and β are multi-objective optimization weight coefficients; η is the learning rate; θ t+1 is the new weight parameter of the finite element model and the fluid dynamics model after the tth iteration update; θ t is the weight parameter at the current time t; ∇ θ is the gradient operator with respect to the weight parameter θ.

[0055] By adopting the above technical solutions, the digital twin model construction method for different complex systems is researched, such as automatically generating accurate models based on machine learning and deep learning algorithms, including finite element models of mechanical structures and computational fluid dynamics models of fluid systems.

[0056] Optionally, the method further comprises the following steps:

[0057] Marking data containing sensitive attributes in the digital twin system;

[0058] Real-time encryption of the marked sensitive data using a preset asymmetric encryption algorithm;

[0059] Identity verification of the access request through a biometric recognition-based verification algorithm;

[0060] According to the identity verification result, performing periodic incremental backup of the encrypted sensitive data that passes the verification.

[0061] By adopting the above technical solutions, the data security and privacy protection technology in the digital twin system is researched to prevent data leakage, tampering, and malicious attacks, such as using encryption algorithms and access control technology to protect data security. A security protection system for constructing a digital twin model is built to ensure the integrity and reliability of the model and prevent the model from being illegally used or damaged.

[0062] Optionally, the method further comprises the following steps:

[0063] Decrypting and checking the target sensitive data by using a decryption algorithm corresponding to the asymmetric encryption algorithm to generate a data integrity check value;

[0064] Inputting the data integrity check value and the feature vector of the trend data into a preset neural network check model to verify the logical consistency of the digital twin model;

[0065] When the logical consistency verification fails, a multi-level early warning mechanism is triggered: first, a device-level alarm is sent to the local terminal, and if there is no response within a preset time, a system-level alarm is sent to the central platform.

[0066] By using the above technical solutions, the operation behavior of the user in the digital twin system and various events generated by the system are audited for security, and relevant operation and event information is recorded to form a log. Subsequently, through analysis of the audit log, abnormal behavior and potential security vulnerabilities can be detected in a timely manner, and targeted measures can be taken for repair and prevention; for the digital twin model, techniques such as digital signature and hash check are used to regularly check the integrity of the model, and once the model is found to be illegally modified or damaged, an alarm can be sent in a timely manner and recovery measures can be taken to ensure the integrity and reliability of the model and prevent the model from being illegally used.

[0067] In a second aspect, the present application provides a predictive analysis system for a laboratory HVAC system based on digital twinning, which adopts the following technical solutions:

[0068] A predictive analysis system for a laboratory HVAC system based on digital twinning, comprising a processor, wherein the processor executes the steps of the predictive analysis method for a laboratory HVAC system based on digital twinning as described in any one of the above.

[0069] In a third aspect, the present application provides a storage medium, which adopts the following technical solutions:

[0070] A storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the steps of the predictive analysis method for a laboratory HVAC system based on digital twinning as described in any one of the above.

[0071] In summary, the present application includes at least one of the following beneficial technical effects: BRIEF DESCRIPTION OF DRAWINGS

[0072] Figure 1 is a step diagram of the predictive analysis method for a laboratory HVAC system based on digital twinning.

[0073] Figure 2 is a step diagram of the calculation of the first state information and the second state information based on the temperature sensor and the vibration sensor installed on the working device. DETAILED DESCRIPTION

[0074] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings.

[0075] In the description of the present specification, the description of the terms "certain embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in connection with the described embodiments or examples are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Also, the described specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0076] The embodiment of the present application discloses a predictive analysis method for a laboratory heating and ventilation system based on digital twinning, referring to Figure 1 , comprising the following steps:

[0077] In the first time period, such as from 9 am to 10 am for one hour, a plurality of first target data and a plurality of first reference values are obtained. The first target data is derived from the values of temperature sensors uniformly distributed at key positions of the heating and ventilation system, and these temperature sensors record data every 1 minute, so that 60 first target data are obtained in this hour. The first reference value is from the values of vibration sensors or sound sensors, which are used to measure sound data. For example, a vibration sensor is installed near the fan of the heating and ventilation system, which collects vibration data at a frequency of 100 times per second, and a large amount of vibration data is obtained as the first reference value in this hour; a sound sensor is arranged near the air duct to collect sound data generated by the operation of the fan and the airflow passing through the air duct.

[0078] According to the 60 first target data, first trend data is calculated by a first trend algorithm, which can be a moving average method, etc. The trend of the current temperature change is obtained by processing these data, for example, showing a trend of temperature rising 0.2℃ every 10 minutes. For a plurality of first reference values, first frequency spectrum data is calculated by a first conversion algorithm, which uses FFT (Fast Fourier Transform) here. Taking vibration data as an example, FFT analysis is performed on the collected vibration signal to convert time domain signal to frequency domain signal, so that the first frequency spectrum data of vibration energy distribution at different frequencies is obtained, and it may be found that there is a significant vibration energy peak at 500 Hz frequency.

[0079] At the second time period, assume it is the same time of the next day, from 9:00 am to 10:00 am, again acquire a plurality of second target data and a plurality of second reference values. Similarly, the temperature sensor and the vibration, sound sensor collect data according to the previous acquisition frequency and mode.

[0080] According to the plurality of second target data, a second trend data is calculated by the first trend algorithm, for example, the temperature shows a trend of rising 0.3℃ every 10 minutes. According to the plurality of second reference values, a second frequency spectrum data is calculated by the first conversion algorithm, if the vibration data is analyzed by FFT again, it may be found that the vibration energy peak value at 500Hz frequency has increased, and a new energy peak value at 1000Hz frequency appears.

[0081] Based on the digital twin platform, the difference between the first trend data and the second trend data is calculated as the first state information, which can clearly represent the difference in temperature trend between the two time periods, such as the temperature rising rate being accelerated. At the same time, the difference frequency band between the first frequency spectrum data and the second frequency spectrum data is calculated as the second state information, which reflects the difference in vibration frequency change between the two time periods, such as the newly added 1000Hz frequency energy peak value.

[0082] According to the first state information, the first device state data is matched in the digital twin platform, for example, in the digital twin model, when the temperature rising rate exceeds a certain threshold, the corresponding device may have a heat dissipation problem, and the data representing the degree of abnormality of such device is matched. According to the second state information, the second device state data is matched, for example, in the vibration frequency spectrum, some frequency energy abnormally increases, which may indicate that the corresponding device has a problem such as loose parts, and the data representing the degree of abnormality of the vibration device is matched in the digital twin platform. Finally, the similarity value between the first device state data and the second device state data is calculated, if the similarity value is greater than the preset device reference similarity value, such as the preset similarity value is 0.7, when the calculated similarity value is 0.8, a device warning prompt is sent out, reminding the staff that the heating system device may have hidden faults.

[0083] The digital twin intelligent management and control platform has powerful functions, it can collect key operating parameters of the equipment in real time, such as temperature, vibration, noise, etc., and monitor the health status of the equipment in real time. Taking a clean room for experimental animals as an example, the platform monitors the temperature of the heating system equipment in real time, and once the temperature exceeds the range suitable for the survival of experimental animals, it judges whether the equipment is running normally in combination with vibration and noise data. If the equipment is found to be abnormal, a warning can be sent out in time to avoid affecting the survival environment of experimental animals due to environmental temperature or equipment failure, so as to ensure the accuracy and reliability of experimental data and prevent experimental results from being biased due to environmental factors.

[0084] Refer toFigure 2 In the construction of the laboratory's working equipment monitoring system, an efficient and intelligent data collection and analysis architecture has been formed based on temperature sensors and vibration sensors installed in key parts of the working equipment. The temperature sensor is electrically connected to the first edge computing unit through a dedicated circuit, and the vibration sensor is electrically connected to the second edge computing unit in the same way. Both the first edge computing unit and the second edge computing unit establish a remote connection with the remote digital twin platform via a high-speed, stable wireless network. For example, on the air-conditioning unit equipment in a large clean laboratory, temperature sensors are installed in parts that are prone to heat, such as the compressor housing and motor windings, to accurately measure the real-time temperature of key parts of the equipment; vibration sensors are installed near parts that are prone to vibration, such as fan bearings and compressor pistons, to capture vibration conditions during equipment operation.

[0085] The method further includes the following detailed steps:

[0086] The first edge computing unit acquires multiple first target data and multiple second target data from the temperature sensor in real time. Assuming that the temperature sensor collects data once per minute during the first time period from 9:00 AM to 10:00 AM, the first edge computing unit will acquire 60 first target data points. During the second time period from 9:00 AM to 10:00 AM the next day, the first edge computing unit will also acquire 60 second target data points. The first edge computing unit uses the first trend algorithm to conduct an in-depth analysis of this data to calculate the first trend data and the second trend data. The first trend algorithm here is a tool specifically used to calculate the average fluctuation value of multiple values. The average fluctuation value can be calculated in two ways. One method is to calculate the average absolute value of the increase per unit time. For example, if the temperature rises from an initial 25°C to 28°C within the first hour, then the average absolute value of the increase per unit time (1 hour) is (28-25) ÷ 1 = 3°C / hour. Another method is to calculate the absolute value of the increase over N consecutive data points. Assuming N is 10 consecutive data points, with the temperature increasing from 25.1°C in the first data point to 25.6°C in the tenth data point, then the absolute value of the increase over these 10 data points is 25.6-25.1 = 0.5°C. Both methods can reflect the growth trend.

[0087] The second edge computing unit obtains a plurality of first reference values and a plurality of second reference values from the vibration sensor. In the first hour, the vibration sensor collects vibration data at a frequency of 100 times per second, so the number of first reference values is as many as 360,000; the number of second reference values in the same time period on the second day is also the same. The second edge computing unit processes these data through the first conversion algorithm to calculate the first frequency spectrum data and the second frequency spectrum data. The core function of the first conversion algorithm is to convert a plurality of values in the time domain into frequency spectrum data in the frequency domain. Taking vibration data as an example, through the first conversion algorithm (such as Fast Fourier Transform, FFT), the vibration amplitude change data originally presented on the time axis can be converted into frequency spectrum data of vibration energy distribution at different frequencies, so that it can be clearly seen that the device has greater vibration energy at which frequency band, for example, it is found that there is a significant vibration energy peak at 500Hz frequency.

[0088] After the digital twin platform obtains the first trend data, the second trend data, the first frequency spectrum data and the second frequency spectrum data, it carries out in-depth analysis. It calculates the difference or average difference between the first trend data and the second trend data as the first state information. For example, if the first trend data shows that the temperature rises by an average of 2℃ per hour, and the second trend data shows that the temperature rises by an average of 3℃ per hour, then the difference of 1℃ or the average difference ((3-2) ÷ 2 = 0.5℃) between them becomes the first state information, which directly reflects the difference in temperature change trend between the two time periods. The digital twin platform also calculates the energy difference of the corresponding frequency band in the first frequency spectrum data and the second frequency spectrum data, and takes the continuous frequency band with an energy difference greater than a preset energy reference value as the second state information. Assuming that the preset energy reference value is 100, the energy value of the 500Hz frequency band in the first frequency spectrum data is 150, and the energy value of the frequency band in the second frequency spectrum data is 200, the energy difference is 50, if similar energy difference changes greater than the preset value appear in a plurality of continuous frequency bands, these continuous frequency bands constitute the second state information, which reflects the change of the vibration frequency characteristics of the device in the two time periods.

[0089] The edge computing unit here adopts an advanced AI edge server, which has powerful artificial intelligence algorithms and data analysis models built-in. The AI edge server can perform fast and efficient analysis of data locally, without the need to transmit all data to the cloud, greatly reducing data transmission delays. For example, when the AI edge server analyzes temperature data and vibration data, it can detect abnormal conditions in time once it finds that the temperature rise rate is abnormally fast or a new high-energy abnormal frequency band appears in the vibration spectrum. Moreover, it will quickly send warning information to the platform operator through a reliable communication link. Taking a chip manufacturing clean room as an example, when the AI edge server detects that the temperature of the lithography equipment is abnormally high and the vibration spectrum has an abnormal frequency band, it sends warning information to the operator before the equipment actually fails, such as the lithography precision has not been significantly affected. In this way, relevant personnel can immediately prepare for personnel and spare parts, arrange maintenance personnel on standby, and prepare spare parts that may need to be replaced. In this way, not only does it ensure that the service life of the equipment is maximized, but it also better controls the inventory of personnel and spare parts, preventing excessive storage from wasting resources, achieving an optimal balance between practicality and economy. At the same time, for the experimental clean room, the digital twin platform will integrate environmental parameters (such as temperature, humidity, air pressure, etc.) and equipment operating conditions (such as temperature, vibration, speed, etc.) to establish a highly realistic digital twin model. This model can reflect the real state of the experimental clean room in real time, providing strong support for the efficient operation and precise management of the laboratory.

[0090] The method of calculating the similarity value between the first device state data and the second device state data further includes the following sub-steps:

[0091] A first dynamic gain coefficient is introduced for the first device state data, which is dynamically adjusted according to a non-linear decay function K1(t) = e^(-αt) based on the uncalibrated duration t, where α is a preset decay coefficient. For example, in a certain laboratory, after a long time of running a set of key reaction equipment, the measurement accuracy of its temperature sensor, pressure sensor and other equipment may gradually decrease due to various factors, and this decrease will continue to exist until recalibration. Assuming that α is set to 0.01, when the uncalibrated duration t reaches 100 hours, the first dynamic gain coefficient K1(100) = e^(-0.01×100) = e^(-1) ≈ 0.368. This means that as the uncalibrated time continues to grow, the first dynamic gain coefficient will decrease exponentially, effectively reducing the impact of data errors caused by long-term uncalibration of the equipment on the overall equipment state judgment, making subsequent analysis of the first device state data more accurate.

[0092] The second dynamic gain coefficient is introduced for the second equipment state data, and the second dynamic gain coefficient is dynamically adjusted according to the real-time monitored working equipment power P according to an inverse proportional function K2(P)=β / (P+γ), wherein β and γ are power compensation parameters. Taking a large high-precision processing equipment in a laboratory as an example, the power of the equipment will change greatly in different working modes. When the equipment is fine processing, the power P may be low, assuming that β is 100 and γ is 50, if the power P is 100 W at this time, then the second dynamic gain coefficient K2(100)=100 / (100+50)=2 / 3; and when the equipment is rough processing, the power is greatly increased to 300 W, at this time the second dynamic gain coefficient K2(300)=100 / (300+50)=2 / 7. The second dynamic gain coefficient which is adjusted in real time according to the equipment power can accurately measure the equipment state data when the equipment power fluctuates, avoid the data measurement deviation caused by the power change, and provide guarantee for accurate analysis of the equipment state.

[0093] Based on the first dynamic gain coefficient and the second dynamic gain coefficient, the matching degree index of the weighted equipment state data fluctuation amplitude is calculated:

[0094] Match=Σ[K1(t)×ΔS1(i)-K2(P)×ΔS2(i)]^2.

[0095] Wherein ΔS1(i) and ΔS2(i) are the i-th fluctuation component of the first equipment state data and the second equipment state data respectively. Assuming that at a certain moment, the fluctuation component of the first equipment state data ΔS1(1)=0.5, ΔS1(2)=0.3, the fluctuation component of the second equipment state data ΔS2(1)=0.4, ΔS2(2)=0.2, the first dynamic gain coefficient K1(t)=0.5, and the second dynamic gain coefficient K2(P)=0.6.

[0096] Then Match=[0.5×0.5-0.6×0.4]^2+[0.5×0.3-0.6×0.2]^2=(0.25-0.24)^2+(0.15-0.12)^2=0.001. Through this formula, the two dynamic gain coefficients considering the equipment uncalibrated time and real-time power are integrated into the calculation of the equipment state data fluctuation amplitude, so that the matching degree index more comprehensively and accurately reflects the relationship between the equipment state data.

[0097] Finally, the similarity value = 1 / (1+Match). According to the Match value calculated above, at this time the similarity value = 1 / (1+0.001) = 0.999. The greater the value of the similarity value, the higher the degree of similarity. This way of accurately matching equipment state data, by matching the fluctuation of the first equipment state data with the change range between the second equipment data, can more accurately understand the correlation between the states of the equipment. For example, in an automated production experiment system containing multiple devices, this method can quickly and accurately judge the coordination of the running states of different devices. When the temperature change of one device and the vibration change of another device have a high similarity value, it may mean that there is some potential relationship between them, such as sharing the same cooling system or mutual influence on the mechanical structure, etc. This feature provides a solid data foundation for the overall optimization of the operation of the equipment, and engineers can determine whether the equipment is running normally and whether the operating parameters of the equipment need to be adjusted, etc. according to these similarity values.

[0098] The setting that the first gain value decreases with the increase of uncalibrated time and the second gain value decreases with the increase of the power of the working equipment makes this patent technology dynamically adapt to different running conditions of the equipment. When the equipment is running for a long time without calibration, the change of the first gain value can effectively reduce the influence of the error caused by time accumulation on the accuracy of the data. For example, for the chemical production laboratory equipment mentioned earlier, as the uncalibrated time increases, the first dynamic gain coefficient decreases, avoiding the interference of inaccurate measurement data on the judgment of the equipment state; and the second gain value is adjusted according to the power of the equipment, which can always maintain accurate measurement of the equipment state data when the power of the equipment fluctuates. For example, for the processing equipment in an electronic manufacturing laboratory, no matter how the power changes, the second dynamic gain coefficient can accurately reflect the equipment state, greatly improving the reliability of equipment state monitoring under different working conditions.

[0099] In other embodiments, in a heating and ventilation laboratory, a data acquisition and analysis system based on image sensors and vibration sensors is used to achieve efficient monitoring and accurate management of the working equipment. These sensors are installed on key working equipment such as air conditioner compressors, fans, etc., and can capture real-time running state information of the equipment.

[0100] The image sensors and vibration sensors are electrically connected to the first edge computing unit and the second edge computing unit respectively. This electrical connection ensures that the data collected by the sensors can be quickly and stably transmitted to the edge computing unit for preliminary processing. The first edge computing unit and the second edge computing unit are remotely connected to the digital twin platform through a high-speed network, and the digital twin platform performs in-depth analysis and processing on the data from the edge computing unit.

[0101] The first edge computing unit is responsible for processing image sensor data. In a heating and ventilation laboratory, image sensors can be installed on the exterior of air conditioning units, inside air ducts, on heat exchanger surfaces, and other locations to capture images of the device's appearance, the operating status of components, and the flow of air currents.

[0102] The first edge computing unit obtains multiple first target data and multiple second target data from the image sensors. For example, within a specific time period, the image sensors take pictures of the device at certain time intervals. The first target data can be a series of images taken between 9 am and 10 am, while the second target data can be images taken between 2 pm and 3 pm.

[0103] The first edge computing unit processes these image data using a first trend algorithm. The core of this algorithm is to calculate the target state value corresponding to the preset template in multiple image data. The preset template is constructed based on the image features of the device in a normal operating state. For example, for an air conditioning unit heat exchanger, the preset template may include features such as the neat arrangement of heat exchanger fins and the absence of dirt accumulation on the surface. The first trend algorithm analyzes each image, extracts features related to the preset template, such as the inclination angle of the fins and the coverage area of the dirt, and converts these features into target state values.

[0104] By processing the first target data and the second target data separately, the first edge computing unit obtains first trend data and second trend data. For example, the first trend data shows that the average inclination angle of the heat exchanger fins is 5 degrees during the morning period, while the second trend data shows that the average inclination angle changes to 8 degrees during the afternoon period. This indicates that the state of the heat exchanger fins may have changed, possibly indicating loosening or deformation.

[0105] The second edge computing unit focuses on processing data from vibration sensors. Vibration sensors are usually installed in critical parts of the device, such as the compressor shell, fan bearings, and other locations, to monitor the vibration of the device during operation.

[0106] The second edge computing unit obtains multiple first reference values and multiple second reference values from the vibration sensors. These reference values are the amplitude values of the vibration signals collected by the vibration sensors in different time periods. For example, the first reference value is the vibration amplitude collected within the first 30 minutes after the device starts, while the second reference value is the vibration amplitude collected within the 30 minutes after the device has been running continuously for 2 hours.

[0107] The second edge computing unit converts multiple values in the time domain into frequency spectrum data in the frequency domain using a first conversion algorithm. Time domain data reflects the changes of vibration signals over time, while frequency domain data shows the distribution of vibration signals at different frequencies. A commonly used conversion algorithm is the Fast Fourier Transform (FFT). Through the FFT algorithm, the second edge computing unit converts the time domain vibration signals collected by the vibration sensor into frequency spectrum data in the frequency domain, obtaining first frequency spectrum data and second frequency spectrum data.

[0108] For example, the first frequency spectrum data shows that during the device startup phase, the vibration signal is mainly concentrated at two frequencies of 50Hz and 100Hz, while the second frequency spectrum data shows that after the device has been running continuously for 2 hours, there is a clear vibration peak at the frequency of 200Hz. This may mean that a component of the device has failed during operation, causing a change in vibration frequency.

[0109] The digital twin platform receives the first trend data and the second trend data from the first edge computing unit, and the first frequency spectrum data and the second frequency spectrum data from the second edge computing unit. Through comprehensive analysis of these data, the digital twin platform can gain a deep understanding of the running state of the device.

[0110] The digital twin platform first calculates the difference or average difference between the first trend data and the second trend data as the first state information. Taking the trend data of the inclination angle of the heat exchanger fins as an example, the calculated difference is 3 degrees, which becomes the first state information. This information can intuitively reflect the state changes of the device at different time periods, helping the maintenance personnel to judge whether there are potential problems in the device.

[0111] The digital twin platform also calculates the energy difference of the corresponding frequency bands in the first frequency spectrum data and the second frequency spectrum data. For example, for the frequency band of 50Hz, the energy value in the first frequency spectrum data is 100 Joules, and the energy value in the second frequency spectrum data is 150 Joules, with an energy difference of 50 Joules. When the energy difference is greater than a preset energy reference value, the digital twin platform will further calculate the continuous frequency bands with energy differences greater than the reference value as the second state information. Assuming that the preset energy reference value is 30 Joules, and the energy differences in the frequency band of 50Hz-100Hz are all greater than 30 Joules, then the continuous frequency band of 50Hz-100Hz constitutes the second state information.

[0112] Through the first state information and the second state information, the digital twin platform can comprehensively and accurately evaluate the running state of the equipment. If the first state information shows that a certain parameter of the equipment has changed greatly, and the second state information shows that an abnormal frequency band appears in the vibration spectrum, it can be judged that the equipment may have a fault, and needs to be repaired and maintained in time. In this way, the data acquisition and analysis system based on image sensors and vibration sensors, combined with the intelligent processing capability of the digital twin platform, provides strong support for the equipment management of the heating and ventilation laboratory, ensuring the stable operation and efficient work of the equipment.

[0113] The method further comprises the following steps:

[0114] During the operation of the equipment, the energy monitoring system will collect real-time power parameters such as three-phase current, voltage, and power factor of the working equipment. These parameters reflect the power consumption of the equipment. For example, in a large heating and ventilation laboratory, a continuously running motor has three-phase currents I1, I2, and I3, a voltage U, and a power factor cosφ. The system processes these data based on a sliding time window, which is like a moving observation frame that continuously captures data within a certain time range. In this way, the time domain feature vector corresponding to the power parameters is extracted. Assuming that the sliding time window is set to 10 minutes, every 1 minute, the window slides forward by 1 minute, continuously collecting power parameter data within the 10 minutes, and extracting representative time domain feature vectors from them, which contain the change information of the power parameters in the time dimension.

[0115] A dynamic threshold algorithm is used to detect anomalies in the extracted time domain feature vectors. The dynamic threshold algorithm calculates an anomaly index according to the formula Here, taking multiple devices in a large heating and ventilation laboratory as an example, it can be the current parameter of the i-th device at time t (it can also be the voltage, power factor, or other power parameters).

[0116] X i (t) is the i-th power parameter at time t, μ i (t), σ i (t) are the mean and standard deviation of the sliding window of the previous N minutes (assuming N = 15 minutes), and ω i is a preset weight coefficient. For example, for the current parameter of a key device, a higher weight coefficient is assigned to highlight its importance in anomaly detection. When the calculated anomaly index A(t) exceeds the preset adaptive threshold, it indicates that the power parameter of the equipment has an abnormal fluctuation.

[0117] When the abnormal index exceeds the preset adaptive threshold, the system further extracts the frequency domain harmonic components of the corresponding power parameters. Taking a variable frequency device in a large HVAC laboratory as an example, its frequency domain harmonic components are within a certain range during normal operation. Once the device fails, such as the aging of internal components of the frequency converter, the frequency domain harmonic components will change. The system matches the extracted frequency domain harmonic components with the fault mode characteristics corresponding to the frequency domain harmonic components in the historical energy consumption database. The historical energy consumption database stores the power parameter characteristics of various devices under different fault conditions. Through matching, an abnormal fluctuation spectrum graph is generated. For example, if the harmonic components of the device match the fault mode characteristics of the motor winding short circuit in the database, the corresponding abnormal fluctuation spectrum graph is generated, which intuitively shows the specific situation of the device anomaly.

[0118] The generated abnormal fluctuation spectrum graph is input into a pre-trained deep residual network model. The deep residual network model is an intelligent model trained on a large amount of data, which can analyze the abnormal fluctuation spectrum graph. The model outputs a prediction result, which includes a device health degree prediction value and a remaining life prediction interval. For example, after model analysis, the health degree prediction value of a certain device is 70%, indicating that the device is currently in a relatively healthy but potentially hazardous state, and the remaining life prediction interval is given as 3-6 months, prompting the operation and maintenance personnel to pay close attention to the device or arrange maintenance within this time range. The prediction result is associated with a timestamp of the prediction result and stored in a blockchain evidence database. The blockchain evidence database has the characteristics of non-tamperability, ensuring the authenticity and reliability of the data, facilitating subsequent queries and traceability.

[0119] The energy monitoring system accurately monitors the energy use of each region and device by monitoring the equipment. For example, in a HVAC laboratory, the system can monitor the power consumption of air conditioners, lighting, and other devices in each HVAC region. Statistical analysis of energy consumption anomalies can identify unreasonable parts of the energy distribution plan in the system. Suppose it is found that the air conditioning energy consumption in a certain HVAC region is abnormally high during off-peak hours. Through analysis, it may be found that the air conditioning temperature control system is not set reasonably. Comprehensive performance monitoring of all systems through artificial intelligence learning and analysis can not only generate an optimized energy utilization plan, but also, based on historical energy consumption data, take emergency plans in advance to address time-based and seasonal energy consumption peaks. For example, based on energy consumption data from previous summers, it is predicted that there will be power consumption peaks on certain days this summer, and energy distribution strategies are adjusted in advance, such as reasonable dispatching of standby power sources, to ensure rational use of energy and stable operation of equipment.

[0120] The method further includes the following steps:

[0121] In the HVAC laboratory, multiple sensors work together to collect heterogeneous data of physical entities. Heterogeneous data includes stress-strain distribution data of mechanical structures, three-dimensional geometric point cloud data, and turbulent intensity spectrum data and multiphase flow pressure field data of fluid systems. For example, for the laboratory air conditioning unit, in terms of mechanical structure, stress-strain sensors are installed on the key components of the compressor to synchronously collect stress-strain distribution data. When the compressor is running at high speed, these sensors can accurately capture the stress changes that the components are subjected to, helping engineers understand whether the components have fatigue damage due to long-term high-load operation. Three-dimensional laser scanning equipment is used to obtain three-dimensional geometric point cloud data of the air conditioning unit, accurately depicting the outline, pipe layout, and internal space structure of the unit, providing accurate geometric information basis for subsequent airflow simulation and structure optimization.

[0122] In terms of fluid systems, multiple turbulent intensity sensors and spectrum analysis equipment are arranged in the air duct to collect turbulent intensity spectrum data. These data can reflect the turbulent characteristics of airflow in the air duct in detail, such as the degree of turbulence, the generation and development of vortexes, etc., which is crucial for optimizing air duct design, reducing airflow noise and energy loss. Pressure sensors are installed near components such as heat exchangers and humidifiers to collect multiphase flow pressure field data, monitor the pressure changes of air and water and other multiphase fluids during heat exchange and humidification, and ensure that the system operates stably under appropriate pressure conditions, ensuring the heat and humidity processing effect of the HVAC system.

[0123] Taking the frame structure of the large combined air conditioning unit in the HVAC laboratory as an example, a finite element model generator based on graph neural networks is constructed. Stress-strain distribution data and geometric point cloud data are input into the finite element model generator. Stress-strain distribution data records the stress-strain conditions of the frame due to factors such as vibration and pressure during unit operation. Geometric point cloud data accurately presents the shape, size, and connection relationship of each component of the frame.

[0124] Through iterative optimization of grid division, a finite element model is obtained, which is calculated according to the following formula:

[0125] .

[0126] Where K i is the element stiffness matrix, u i is the node displacement vector, TV(u) is the total variation regularization term, and λ is the balance coefficient. Among them, the element stiffness matrix K i determines the ability of each element in the frame structure to resist deformation, just like the strength of each beam and column in a building. The node displacement vector u iThe displacement of the frame node under force is described, for example, the swing amplitude of the node under the vibration excitation of the fan. The external force vector fi is the external force acting on the node, such as the vibration force generated by the operation of the fan. The total variation regularization term TV(u) guarantees the stability and smoothness of the model, preventing unreasonable mutations of the model. The balance coefficient λ adjusts the weight of each term in the optimization process, so that the model can more accurately simulate the mechanical behavior of the frame in actual operation, predict the structural stability of the frame, avoid deformation, fracture and other problems caused by long-term stress, and ensure the safe and reliable operation of the air conditioning unit.

[0127] In the finite element model generator based on the graph neural network, the corresponding relationship of each parameter and the stress-strain distribution data and the geometric point cloud data is as follows:

[0128] The unit stiffness matrix K i : The geometric point cloud data provides the geometric shape and spatial distribution of the structure, which is used to determine the geometric parameters of the unit (such as node coordinates, unit size), which directly affects the construction of K i (for example, calculate the stiffness matrix through material properties and geometric shape). The stress-strain distribution data reflect the mechanical properties of the material (such as elastic modulus, Poisson's ratio), which together with the geometric data determine the value of K i .

[0129] The node displacement vector u i : The geometric point cloud data generates node positions through meshing, and u i is the displacement calculation result of these nodes. The stress-strain distribution data indirectly affect u i , because the material mechanical properties (reflected by the stress-strain data) determine the stiffness matrix K i , which in turn solves the displacement through the equation K i u i =f i .

[0130] The external force vector f i : The stress-strain distribution data directly provides the external force information acting on the node (such as mechanical load, thermal load), which is used to define the value of f i .

[0131] The total variation regularization term TV(u): The geometric point cloud data constructs the regularization term through the node adjacency relationship (defined by the geometric structure), which constrains the smoothness of the displacement field and avoids mesh distortion. The stress-strain distribution data implies physical constraints (such as material continuity), and the regularization term optimizes the displacement field by balancing the residual and physical rationality.

[0132] The balance coefficient λ is a hyperparameter determined through experiments or optimization algorithms, which is used to adjust the residual term ||K i u i -f i||2 and the weight of the regularization term TV(u) ensure that the model balances between geometric accuracy (dependent on point cloud data) and physical plausibility (dependent on stress-strain data).

[0133] The geometric point cloud data dominates the K i geometric part and the spatial constraint of TV(u) support the solution of nodal displacement u i . The stress-strain distribution data provide material properties and external force information, which directly affect the material part of K i and f i , and indirectly constrain u i through the mechanical equilibrium equation. The parameter λ acts as a global coordination factor to ensure that the physical and geometric characteristics of the two types of data work together during the optimization process.

[0134] For the air handling system of a HVAC laboratory, a spatio-temporal convolutional network is used to construct a fluid dynamics model generator. Taking the simulation of airflow flow in the duct as an example, the turbulence intensity spectrum data and multiphase flow pressure field data are mapped to CFD, which is a fluid dynamics grid. The turbulence intensity spectrum data reflects the turbulence characteristics of the airflow in the duct, such as the vortex frequency and intensity distribution of the airflow, which will affect the delivery efficiency and mixing effect of the air. The pressure field data shows the pressure at different positions in the duct, which plays a key role in the direction and speed of the airflow.

[0135] The Navier-Stokes equation residual is solved by a differentiable solver using the formula:

[0136] .

[0137] Where the fluid velocity vector v describes the speed and direction of the airflow; time t is used to analyze the dynamic changes of the airflow over time; kinematic viscosity reflects the viscosity characteristics of the air; pressure p determines the pressure driving condition of the airflow. (v·∇) v is the convection term; is the viscous diffusion term; ∇p is the pressure gradient. By solving the equation residual, the flow, heat transfer and mass transfer processes of the air in the duct can be more accurately simulated, helping engineers optimize the shape, size and layout of the duct, reduce airflow resistance, improve the energy efficiency of the air handling system, and ensure that the temperature, humidity and air quality in the laboratory meet the standard.

[0138] The cross-scale coupling of finite element models and CFD is based on structure-fluid interaction:

[0139] The finite element model calculates the deformation or vibration of the mechanical structure, generating a displacement field u i ;

[0140] The displacement field is input as a dynamic boundary condition into CFD, updating the flow field grid and affecting the velocity v and pressure p of the fluid.

[0141] Data transfer interface: use interpolation algorithms such as radial basis functions to map displacement data from finite element model nodes to corresponding locations on the CFD mesh; achieve bidirectional real-time data exchange between FEM and CFD through coupling solvers such as Co-Simulation.

[0142] Real-time data assimilation: deploy an online update engine to receive multi-source sensor data (such as real-time pressure, flow rate), adjust CFD model parameters (such as , boundary conditions) through adaptive gradient descent algorithm; combined with FEM structural state feedback (such as material aging), dynamically correct fluid-structure coupling boundary conditions.

[0143] Multi-objective optimization: the objective function simultaneously considers the fluid residual ||RNs|| and the structural residual ||K i u i -f i ||, balancing the convergence of both through weight coefficients α, β.

[0144] The combination of finite element model and CFD is essentially a multi-physics coupling and data-driven collaboration. The finite element model provides structural deformation data as dynamic boundary conditions for CFD, and the turbulence intensity spectrum and multiphase flow pressure field data are embedded in the CFD solving process through feature extraction and machine learning. Online updating mechanism ensures real-time consistency between the model and the physical entity. Realize the full-dimensional digital twin of the heating system from structural health to fluid performance, support fault warning and energy efficiency optimization.

[0145] In more detail, the meshing data provided by the finite element model and the coupling of CFD provide a consistent data interface, so the finite element model can update the CFD dynamic mesh boundary conditions, dynamically correct Rns, and realize fluid-structure coupling simulation.

[0146] In fluid-structure coupling simulation, combining finite element model to update CFD dynamic mesh boundary conditions and dynamically correct Navier-Stokes equation residual (RNS) can be achieved through the following steps:

[0147] Adopt explicit coupling: solve finite element model and CFD independently in steps, and realize interaction through alternating data transmission.

[0148] Displacement interpolation: map the node displacement u i calculated by the finite element model to the CFD boundary mesh nodes through radial basis function (RBF) or nearest neighbor interpolation. Ensure the continuity of displacement at the structure-fluid interface, avoid geometric mismatch.

[0149] Dynamic mesh: treat the fluid domain as an elastic medium and adjust the internal mesh according to the boundary displacement.

[0150] Adaptive time step: dynamically adjust the time step according to the residual rate of change, such as reducing the step size when the residual increases.

[0151] Relaxation iteration: introduce a sub-relaxation factor Control the update speed of the variable, improve the convergence stability, the formula is: ;

[0152] Data-driven correction: use the measured flow field data from sensors such as pressure and velocity as constraints to correct the residual term through adjoint optimization, making the simulation results approach the real data.

[0153] Coupled solution: based on the established finite element model and CFD, define the interface such as the pipe wall. The finite element model solves the structure displacement u i , output the interface displacement. Update the CFD dynamic mesh boundary conditions, transfer the displacement to the fluid domain. CFD solves the updated flow field, which includes velocity v and pressure p, and calculates the fluid load ff. Apply the fluid load ff as an external force to the finite element model, and enter the next time step.

[0154] Convergence judgment: monitor the structure residual and fluid residual at the same time, until both are below the set threshold.

[0155] Take the vibration analysis of HVAC system fan blades as an example: the finite element model calculates the vibration displacement of the blades under the action of airflow, and updates the flow field shape around the blades in CFD through dynamic mesh. CFD simulates the updated airflow pressure distribution, which is fed back to the finite element model as dynamic load, and iterates until the vibration and flow field reach a steady state, predicting the fatigue life of the blade. By combining finite element model and CFD, and dynamically updating the mesh boundary conditions and correcting the RNS residual, the fluid-structure coupling simulation can accurately simulate the interaction between structure deformation and fluid response, and improve the dynamic adaptability of the model through data-driven and machine learning, providing a reliable tool for fault warning and design optimization in HVAC systems and other fields.

[0156] In the HVAC laboratory, deploy an online model update engine to monitor the running state of air conditioning units and other equipment in real time. Take the compressor of an air conditioning unit as an example, real-time receive update data stream from various sensors on the compressor, including temperature sensor measured compressor shell temperature, vibration sensor detected vibration amplitude and frequency, current sensor feedback working current and other data.

[0157] Adjust the weight parameters of the finite element model and fluid dynamics model dynamically through the adaptive gradient descent algorithm, according to the formula:

[0158] ;

[0159] Where, is the measured data at time t, such as the actual temperature value; Mt is the model prediction value; Jensen-Shannon divergence JS measures the difference between the distribution of the measured data and the model prediction data, to judge the accuracy of the model prediction. The multi-objective optimization weight coefficients a, b adjust the optimization direction of the model under different objectives, such as assigning different weights when focusing on temperature changes and vibration abnormalities. η is the learning rate, which controls the step size of model parameter update, ensuring that the model can quickly and accurately adapt to changes in the compressor operating state. t+1 is the new weight parameter of the finite element model and the fluid dynamics model after the t-th iteration update; θ t is the weight parameter at the current time t; ∇ θ is the gradient operator with respect to the weight parameter θ. By continuously adjusting the model weight parameter, the model can timely reflect the actual operation of the compressor, and potential faults such as component wear and poor lubrication can be discovered in advance, providing a scientific basis for equipment maintenance and management, and ensuring the stable operation of the heating and ventilation laboratory equipment.

[0160] The method further comprises the following steps:

[0161] Data containing sensitive attributes in the digital twin system is marked. In the digital twin system of the heating and ventilation laboratory, there are a large amount of data of different types. Data containing sensitive attributes need special attention. For example, data collected by high-precision temperature and humidity sensors in the laboratory, which reflect the precise temperature and humidity changes of the experimental environment, are crucial for some experiments that are extremely sensitive to environmental conditions (such as biological sample culture and high-precision chemical reaction experiments), and belong to the category of sensitive data. For another example, operating parameters of heating and ventilation equipment during some special experimental processes, such as start-stop time and frequency adjustment range of the compressor, which may affect the normal progress of the experiment or even cause the experiment to fail if leaked, also have sensitive attributes.

[0162] Through a specially designed marking algorithm, data in the system is screened by category. When the algorithm detects that a certain category of data matches the preset sensitive attribute characteristics, it automatically marks it. For example, when a group of data about real-time monitoring of temperature and humidity in a specific experimental area is identified, a "sensitive-temperature and humidity experimental data" mark is immediately added to the metadata of the data, facilitating subsequent targeted processing.

[0163] The marked sensitive data is encrypted in real time using a preset asymmetric encryption algorithm. The marked sensitive data is encrypted in real time using a preset asymmetric encryption algorithm, such as the RSA algorithm. In the digital twin system of the HVAC laboratory, data generation and transmission are continuous. When the temperature and humidity data with sensitive labels are transmitted from the sensor to the system server, the system will automatically call the encryption module. Using the asymmetric encryption algorithm, a corresponding public key and private key pair is generated for each set of sensitive data. The public key is used to encrypt the data, and the encrypted ciphertext is transmitted within the system internal network. For example, a set of sensitive data representing the current accurate temperature value in the laboratory is encrypted using the public key before transmission, becoming a string of ciphertext in the form of random code. Only authorized devices or users with the corresponding private key can decrypt the ciphertext to restore the real temperature value, effectively preventing data from being stolen and illegally viewed during transmission and storage.

[0164] The access request is identity verified through a biometric recognition-based verification algorithm. The access request is identity verified through a biometric recognition-based verification algorithm. In the HVAC laboratory scenario, personnel entering the laboratory to operate the digital twin system may include researchers, maintenance engineers, etc. Taking fingerprint recognition as an example, when each authorized person enters the laboratory for the first time and registers the permission, their fingerprint information is recorded into the system. When the researcher wants to access the sensitive experimental environment data stored in the digital twin system, fingerprint collection is required on a specific identity verification device (such as a fingerprint-recognized access control or computer login device). The system compares the collected fingerprint information with the pre-stored authorized fingerprint information and calculates the similarity through a complex biometric recognition algorithm. Only when the similarity reaches a preset high threshold (such as more than 95%) is the identity verification considered to be passed, allowing access to related sensitive data. In addition, facial recognition, iris recognition, and other biometric recognition technologies can also be applied to this identity verification process, further enhancing the accuracy and security of identity verification.

[0165] According to the identity verification result, the encrypted sensitive data that passes the verification is periodically backed up. According to the identity verification result, the encrypted sensitive data that passes the verification is periodically backed up. In the HVAC laboratory, the amount of encrypted sensitive data can be very large, such as long-term accumulation of device operating parameters, experimental environment monitoring data, etc. In order to ensure the security and recoverability of the data, the system will perform incremental backup according to the preset time period (such as 2 o'clock in the morning every day). Incremental backup only backs up the data that has changed since the last backup, rather than repeatedly backing up all data. For example, within a day, only part of the HVAC device operating parameters have changed, and the system intelligently identifies these changed data and backs them up with the relevant encryption key to a dedicated data storage device (such as a secure server in a remote location). This not only saves storage resources, but also quickly and accurately recovers to the most recent correct state in the event of data loss or damage, ensuring the normal operation of the digital twin system and the integrity of the data.

[0166] Data security and privacy protection are crucial research topics in the digital twin system of the HVAC laboratory. In addition to the encryption algorithm mentioned above, access control technology is also an important means of protecting data security. For example, different user roles are set in the system, such as super administrator, ordinary researcher, and operation and maintenance personnel, and different permissions are assigned to each role. The super administrator has the highest permission and can operate all data and functions of the system; the ordinary researcher can only access the temperature and humidity, device operating status, and other data related to his own experiment; the operation and maintenance personnel are mainly responsible for monitoring and adjusting part of the device operating parameters and can only access the data related to device operation and maintenance. Through this strict access control strategy, the access range of data by different personnel is limited, preventing unauthorized access and tampering of data.

[0167] At the same time, a security protection system for building digital twin models is constructed to ensure the integrity and reliability of the models from multiple aspects. At the network level, firewalls, intrusion detection systems, and other devices are used to monitor and filter network traffic entering the digital twin system in real time, preventing external malicious network attacks such as denial of service attacks (DoS) and port scanning. At the system level, the software of the digital twin system is periodically scanned for vulnerabilities and repaired, and security patches are updated to prevent hackers from exploiting software vulnerabilities to illegally use or damage the model. At the physical level, the server and other hardware devices are placed in a secure computer room environment, and access control and monitoring measures are taken to prevent physical devices from being illegally accessed and damaged. Through the comprehensive use of these technologies and measures, a comprehensive security protection barrier for the HVAC laboratory digital twin system is constructed, providing a solid data security guarantee for the laboratory's scientific research work and stable operation of the equipment.

[0168] The method further comprises the following steps:

[0169] In the digital twin system of the HVAC laboratory, it is necessary to ensure the integrity of the data and the logical consistency of the model. When sensitive data needs to be processed or analyzed, the decryption algorithm corresponding to the asymmetric encryption algorithm is used to decrypt and verify the data first.

[0170] Taking the temperature and humidity data of a key experimental area in the laboratory as an example, these data are encrypted and protected by using an asymmetric encryption algorithm during transmission and storage. When researchers need to view real-time temperature and humidity data in this area, the system will call the corresponding decryption algorithm to decrypt the encrypted temperature and humidity data. During decryption, the system will generate a data integrity check value at the same time. For example, a hash algorithm such as SHA-256 is used to calculate the decrypted data to obtain a unique hash value as the data integrity check value. This hash value can be used to verify whether the data has been tampered with during transmission and storage.

[0171] Next, the data integrity check value and the feature vector of the trend data are input into the pre-set neural network verification model to verify the logical consistency of the digital twin model. The feature vector of the trend data is extracted from historical data, reflecting the trend and pattern of data changes over time. For example, in the HVAC laboratory, temperature and humidity data usually have certain periodic and seasonal variation rules. By analyzing historical temperature and humidity data, the feature vector of these variation rules can be extracted. The neural network verification model learns the relationship between these feature vectors and data integrity check values to determine whether the digital twin model can accurately reflect the actual physical system. If the temperature and humidity trend in the digital twin model is significantly different from the actual collected data, or the data integrity check value is inconsistent with the expected result, it means that the logical consistency of the digital twin model may be a problem.

[0172] When the logical consistency verification fails, in order to discover and solve the problem in time, the system will trigger a multi-level warning mechanism. Taking the air conditioning unit in the laboratory as an example, if the energy consumption of the air conditioning unit predicted by the digital twin model is significantly different from the actual collected energy consumption data, and the logical consistency is found to be inconsistent after verification by the neural network verification model, the system will take immediate action.

[0173] The system sends a device-level alert to the local terminal. The local terminal can be a monitoring screen on the laboratory site, a handheld device of the operation and maintenance personnel, etc. The alert information will specify the device that has a problem (such as a specific air conditioning unit), the type of problem (such as abnormal energy consumption), and the possible impact. After receiving the device-level alert, the operation and maintenance personnel need to respond within a preset time (such as 15 minutes). If the operation and maintenance personnel do not respond within the preset time, the system considers that the problem is serious and requires higher-level intervention, at which time the system sends a system-level alert to the central platform. The central platform is usually monitored by the laboratory managers or technical experts, who can organize professional personnel to conduct in-depth investigation and treatment according to the system-level alert information, to ensure that the problem is solved in a timely manner and to avoid greater impact on the normal operation of the laboratory.

[0174] Security auditing of the operation behavior of users within the digital twin system and various events generated by the system is an important measure to ensure system security. In the HVAC laboratory, the operation behavior of users includes control of devices, query and modification of data, etc., and system events include start, stop, fault alarm of devices, etc. The system records all these operation and event information to form detailed audit logs.

[0175] For example, when a researcher modifies the temperature and humidity set value of an experimental area, the system records the time of the operation, the identity of the operator, the set value before and after the modification, etc. Similarly, when a device has a fault alarm, the system records the time of the fault occurrence, the fault type, the related parameters of the device, etc. Through analysis of these audit logs, abnormal behavior and potential security vulnerabilities can be detected in a timely manner. For example, if it is found that a user frequently operates a device outside working hours, or a device frequently has a fault alarm within a short period of time, it may mean that there is an abnormal situation. In response to these abnormal situations, the system administrator can further investigate the cause and take targeted measures to repair and prevent, such as strengthening user permission management, maintaining and repairing the device, etc.

[0176] For the digital twin model, technologies such as digital signature and hash check are used to periodically check the integrity of the model. Digital signature can ensure the source and integrity of the model. For example, when creating a digital twin model, a digital signature is generated using an encryption algorithm and stored together with the model. When checking the integrity of the model, the system verifies the validity of the digital signature. If the digital signature is invalid, it means that the model may have been illegally modified.

[0177] Hash verification verifies the integrity of the model by calculating its hash value. For example, the code and data of the digital twin model are hashed regularly, and the calculated hash value is compared with the previously saved hash value. If the two hash values ​​are inconsistent, it means that the model may have been modified. Once it is discovered that the model has been illegally modified or damaged, the system will promptly issue an alarm and take recovery measures. For example, the original version of the model can be restored from a backup, or the model can be retrained and optimized to ensure the integrity and reliability of the model, prevent the model from being illegally used, and ensure the normal operation of the HVAC laboratory digital twin system and the accuracy of the experimental data.

[0178] An embodiment of the present application also discloses a predictive analysis system for a laboratory HVAC system based on a digital twin, including a processor that executes the steps of the predictive analysis method for a laboratory HVAC system based on a digital twin as described in any one of the above.

[0179] An embodiment of the present application also discloses a storage medium, in which a program is stored. When the program is executed by a processor, the steps of the predictive analysis method of the laboratory HVAC system based on digital twin are implemented.

[0180] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A predictive analysis method for laboratory HVAC systems based on digital twins, characterized in that: The steps include: Acquiring a plurality of first target data and a plurality of first reference values ​​within a first time period; Calculating first trend data using a first trend algorithm based on the plurality of first target data, and calculating first spectrum data using a first conversion algorithm based on the plurality of first reference values; Acquiring a plurality of second target data and a plurality of second reference values ​​within a second time period; Calculating second trend data using a first trend algorithm based on the plurality of second target data; Calculating second spectrum data using a first conversion algorithm according to the plurality of second reference values; Calculate the difference between the first trend data and the second trend data based on the digital twin platform as first state information, and calculate the difference frequency band between the first spectrum data and the second spectrum data as second state information; Match first device status data based on the first status information, match second device status data based on the second status information, calculate a similarity value between the first device status data and the second device status data, and issue a device warning prompt if the similarity value is less than a preset device reference similarity value.

2. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1 is characterized in that: Based on a temperature sensor and a vibration sensor installed on the working equipment, the temperature sensor is electrically connected to a first edge computing unit, the vibration sensor is electrically connected to a second edge computing unit, and the first edge computing unit and the second edge computing unit are both remotely connected to the digital twin platform; The method further comprises the steps of: The first edge computing unit acquires a plurality of first target data and a plurality of second target data from the temperature sensor, and calculates the first trend data and the second trend data by using a first trend algorithm, wherein the first trend algorithm is used to calculate an average fluctuation value of the plurality of values; The second edge computing unit obtains a plurality of first reference values ​​and a plurality of second reference values ​​from the vibration sensor, and calculates the first spectrum data and the second spectrum data by using a first conversion algorithm, wherein the first conversion algorithm is used to convert a plurality of values ​​in the time domain into spectrum data in the frequency domain; The digital twin platform obtains the first trend data, the second trend data, the first spectrum data, and the second spectrum data, and calculates a difference or an average difference between the first trend data and the second trend data as first state information; The energy difference between the first spectrum data and the energy difference between the corresponding frequency bands in the second spectrum data is calculated, and the continuous frequency bands in which the energy difference is greater than a preset energy reference value are calculated as the second state information.

3. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1 is characterized in that: The method for calculating the similarity value between the first device status data and the second device status data further includes the following sub-steps: Introducing a first dynamic gain coefficient for the first device state data, the first dynamic gain coefficient being dynamically adjusted according to a nonlinear attenuation function K1(t)=e^(-αt) based on the duration t of the device being uncalibrated, where α is a preset attenuation coefficient; A second dynamic gain coefficient is introduced for the second device status data. The second dynamic gain coefficient is dynamically adjusted according to the inverse proportional function K2(P)=β / (P+γ) based on the real-time monitored working device power P, where β and γ are power compensation parameters; Based on the first dynamic gain coefficient and the second dynamic gain coefficient, a weighted matching index of the fluctuation amplitude of the device status data is calculated: Match=Σ[K1(t)×ΔS1(i)-K2(P)×ΔS2(i)]^2; Wherein ΔS1(i) and ΔS2(i) are the i-th fluctuation components of the first device status data and the second device status data respectively; Similarity value = 1 / (1+Match).

4. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1, characterized in that: Based on an image sensor and a vibration sensor installed on the working equipment, the image sensor is electrically connected to a first edge computing unit, the vibration sensor is electrically connected to a second edge computing unit, and the first edge computing unit and the second edge computing unit are both remotely connected to the digital twin platform; The method further comprises the steps of: The first edge computing unit acquires a plurality of first target data and a plurality of second target data from the image sensor, and calculates the first trend data and the second trend data by using a first trend algorithm, wherein the first trend algorithm is used to calculate a target state value corresponding to a preset template in the plurality of image data; The second edge computing unit obtains a plurality of first reference values ​​and a plurality of second reference values ​​from the vibration sensor, and calculates the first spectrum data and the second spectrum data by using a first conversion algorithm, wherein the first conversion algorithm is used to convert a plurality of values ​​in the time domain into spectrum data in the frequency domain; The digital twin platform obtains the first trend data, the second trend data, the first spectrum data, and the second spectrum data, and calculates a difference or an average difference between the first trend data and the second trend data as first state information; The energy difference between the first spectrum data and the energy difference between the corresponding frequency bands in the second spectrum data is calculated, and the continuous frequency bands in which the energy difference is greater than a preset energy reference value are calculated as the second state information.

5. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1, characterized in that: The method further comprises the steps of: Collect the power parameters of the three-phase current, voltage and power factor of the working equipment in real time, and extract the time domain feature vector of the power parameters based on the sliding time window; A dynamic threshold algorithm is used to perform anomaly detection on the time domain feature vector. The dynamic threshold algorithm calculates the anomaly index according to the following formula: ； Among them, X i (t) is the i-th power parameter at time t, μ i (t), σ i (t) are the mean and standard deviation of the sliding window in the first N minutes, ω i is the preset weight coefficient; When the abnormal index exceeds a preset adaptive threshold, the frequency domain harmonic component of the corresponding power parameter is extracted and matched with the fault mode feature corresponding to the frequency domain harmonic component in the historical energy consumption database to generate an abnormal fluctuation spectrum; The abnormal fluctuation spectrum is input into a pre-trained deep residual network model to output a prediction result, which includes a predicted value of equipment health and a predicted interval of remaining life. The prediction result is associated with a timestamp generated for the prediction result and then stored in a blockchain evidence database.

6. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1, characterized in that: The method further comprises the steps of: Synchronously collect heterogeneous data of physical entities through multi-source sensors, including stress and strain distribution data of mechanical structures, three-dimensional geometric point cloud data, and turbulence intensity spectrum data and multiphase flow pressure field data of fluid systems; Construct a finite element model generator based on graph neural network, input the stress and strain distribution data and the geometric point cloud data into the finite element model generator, and obtain the finite element model by iteratively optimizing the mesh division: ； Among them, K i is the element stiffness matrix, u i is the node displacement vector, f i is the external force vector acting on the node, TV(u) is the total variation regularization term, the above parameters are calculated through the stress-strain distribution data and the geometric point cloud data, and λ is the equilibrium coefficient; A fluid dynamics model generator is constructed using a spatiotemporal convolutional network. The turbulence intensity spectrum data and the multiphase flow pressure field data are mapped to a fluid dynamics grid, and the residual of the Navier-Stokes equation is solved by a differentiable solver: ； Where v is the fluid velocity vector, t is the time, is the kinematic viscosity, p is the pressure; (v·∇)v is the convection term; is the viscous diffusion term; ∇p is the pressure gradient; The finite element model provides output grid data, which has a matching data interface for coupling with the fluid dynamics grid. The finite element model is used to update the dynamic grid boundary conditions in the fluid dynamics grid for dynamic correction of R NS ; Deploy an online model update engine to receive data streams from multiple source sensors for physical entity updates in real time, and dynamically adjust the weight parameters of the finite element model generator and the fluid dynamics model generator through an adaptive gradient descent algorithm: ； in, is the measured data at time t; M t is the model prediction value; JS is the Jensen-Shannon divergence, α and β are the multi-objective optimization weight coefficients; η is the learning rate; θ t+1 is the new weight parameter in the finite element model and fluid dynamics model after the tth iteration update; θ t is the weight parameter at the current time t; ∇ θ is the gradient operator with respect to the weight parameter θ.

7. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 1, characterized in that: The method further comprises the steps of: Labeling data containing sensitive attributes in digital twin systems; Use a preset asymmetric encryption algorithm to encrypt the marked sensitive data in real time; Verify the identity of the access request through a biometric-based verification algorithm; Based on the identity verification results, periodic incremental backups are performed on verified encrypted sensitive data.

8. The predictive analysis method for laboratory HVAC systems based on digital twins according to claim 7, characterized in that: The method further comprises the steps of: Decrypt the target sensitive data using a decryption algorithm corresponding to the asymmetric encryption algorithm to generate a data integrity check value; Inputting the data integrity check value and the characteristic vector of the trend data into a preset neural network verification model to verify the logical consistency of the digital twin model; When the logical consistency verification fails, a multi-level warning mechanism is triggered: first, a device-level alarm is sent to the local terminal, and if there is no response within the preset time, a system-level alarm is sent to the central platform.

9. A predictive analysis system for laboratory HVAC systems based on digital twins, characterized in that: It includes a processor that executes the steps of the predictive analysis method for a laboratory HVAC system based on a digital twin as described in any one of claims 1 to 8.

10. A storage medium, characterized in that: The medium stores a program, which, when executed by a processor, implements the steps of the predictive analysis method for a laboratory HVAC system based on a digital twin as described in any one of claims 1 to 8.

Images