Digital Twin-Based Fault Prediction and Health Management System for Quick-Freezing Equipment

Abstract

This invention discloses a fault prediction and health management system for quick-freezing equipment based on digital twins, specifically relating to the field of quick-freezing equipment operation and maintenance. It includes a data acquisition module to obtain a full lifecycle operational dataset; a theoretical fault index generation module to generate a multi-dimensional theoretical fault index matrix; a real-time fault index calculation module to generate a real-time fault index; a deviation threshold library establishment unit to form a dynamically updated deviation threshold library; a remaining lifespan prediction module to calculate the remaining operating time of the equipment from its current state to the occurrence of a fault; a health management module to calculate the equipment health score and classify the equipment health status into four levels; and a visualization module to provide users with a visual interface. Compared to traditional operation and maintenance methods that cannot predict remaining lifespan, this invention allows maintenance personnel to plan repairs and component replacements in advance, rationally arrange production plans, reduce the risk of sudden failures, and transform equipment operation and maintenance from a passive response to proactive prevention.

Description

Technical Field

[0001] This invention relates to the field of quick-freezing equipment operation and maintenance, and more specifically, to a quick-freezing equipment fault prediction and health management system based on digital twins. Background Technology

[0002] Against the backdrop of intelligent upgrading in the food processing industry and increasingly stringent food safety regulations, quick-freezing equipment serves as a core guarantee for food preservation and production continuity. In the field of quick-freezing equipment operation and maintenance, the main existing technologies employed include manual inspection and maintenance, traditional PLC monitoring, and simple fault alarm systems. Manual inspection and maintenance relies on the experience of technicians, who assess the equipment's status by periodically checking instrument panel parameters, listening to compressor operation sounds, and touching the equipment surface temperature. Traditional PLC monitoring only collects basic operating data such as compressor current and suction pressure, storing it in the local control system, and can only provide passive alarms when the rated threshold is exceeded. While some companies have introduced simple fault alarm systems that can trigger alarms for single parameters, these systems rely solely on fixed thresholds and lack analysis of parameter correlations and fault evolution trends.

[0003] However, in actual use, it still has some shortcomings. First, manual inspection and maintenance is limited by subjective experience and inspection frequency, making it impossible to capture instantaneous anomalies in equipment operating parameters and difficult to detect hidden faults such as micro-leakage and early bearing wear. Second, traditional PLC monitoring data is stored in isolation and cannot integrate basic equipment information, historical fault records and maintenance data, resulting in the inability to correlate historical operating patterns when tracing faults and low troubleshooting efficiency. Third, simple fault alarm systems lack multi-dimensional parameter collaborative analysis and are prone to false alarms due to fluctuations in a single parameter or missed alarms due to cross-influence of parameters, making it difficult to meet the high-precision and low-false-alarm requirements of the food processing industry for equipment maintenance.

[0004] Secondly, in practical applications, the factors influencing the failure of quick-freezing equipment are complex, involving mechanical wear, thermodynamic anomalies, and electrical system fluctuations. Existing technologies cannot achieve a closed-loop process from real-time monitoring of fault diagnosis to health assessment to lifespan prediction. For example, existing technologies can detect abnormal compressor current, but cannot combine vibration and energy efficiency parameters to determine whether the abnormality is caused by bearing wear or refrigerant leakage, nor can they predict the remaining safe operating time. This leads to passive operation and maintenance decisions, either resulting in over-maintenance that increases costs or delayed maintenance that causes serious failures. At the same time, existing technologies lack the integration and sharing of data for equipment of the same model. Historical failure cases, operation and maintenance plans, and spectral characteristic data are scattered across different enterprises or departments, failing to provide data support for equipment fault diagnosis and process optimization. This leads to the recurrence of similar failures, shortens the overall service life of equipment, and restricts the improvement of intelligent operation and maintenance levels in the food processing industry. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a fault prediction and health management system for quick-freezing equipment based on digital twins. The present invention provides the following technical solution:

[0006] The data acquisition module is used to acquire the full lifecycle operation dataset of the target quick-freezing equipment and its counterparts.

[0007] The virtual modeling module is used to construct a multi-physical domain coupled model of quick-freezing equipment;

[0008] The theoretical fault index generation module is used to simulate operating scenarios with different fault types and different fault degrees by injecting fault parameters into the digital twin model, and generate a multi-dimensional theoretical fault index matrix.

[0009] The real-time fault index calculation module is used to generate a real-time fault index from the real-time operating data acquired by the data acquisition module.

[0010] The index comparison and analysis module includes: a deviation threshold library establishment unit and a real-time status determination unit;

[0011] The deviation threshold database establishment unit is used to analyze the deviation patterns before the occurrence of faults in historical fault data and form a dynamically updated deviation threshold database.

[0012] The real-time status determination unit is used to calculate the deviation between the real-time fault index and the theoretical fault index in real time, and output the status of the device in combination with the deviation threshold library.

[0013] The remaining life prediction module is used to calculate the remaining operating time of the equipment from its current state to the occurrence of a fault, based on the fault type identification results output by the index comparison analysis module and the current operating status of the equipment.

[0014] The health management module is used to build a health evaluation index system for quick-freezing equipment, calculate the equipment health score, and divide the equipment health status into four levels.

[0015] The prediction and optimization module dynamically updates the index weights of each dimension in the index comparison and analysis module based on historical early warning results and actual fault occurrences.

[0016] The visualization module is used to provide users with a visual interface.

[0017] The technical effects and advantages of this invention are as follows:

[0018] 1. This invention aggregates data from the entire lifecycle of quick-freezing equipment through a data acquisition module, and constructs a multi-physical domain coupled model using virtual modeling to simulate equipment operating status. A theoretical fault index generation module injects fault parameters, simulates different fault scenarios, and generates a multi-dimensional theoretical fault index matrix. A real-time fault index calculation module generates a real-time fault index based on real-time operating data. The deviation threshold library establishment unit of the index comparison and analysis module establishes a dynamic deviation threshold library based on historical fault data. The real-time status determination unit compares the real-time and theoretical fault index deviations with the thresholds to accurately determine the equipment status. This effectively improves fault diagnosis accuracy, provides early warning of potential faults, effectively reduces the annual average equipment failure rate, and minimizes production interruption losses caused by faults.

[0019] 2. This invention constructs a health evaluation index system based on a health management module, calculates equipment health scores, and divides them into four levels. It progresses from calculating health scores based on a single indicator to weighted fusion of comprehensive health scores, and then introduces remaining lifespan health scores to finally obtain an overall health score. This intuitively reflects the instantaneous operating status and long-term lifespan loss of the equipment. Operators can clearly understand the real-time health status of the equipment through the visualization module, providing a clear basis for daily operation and maintenance and regular inspections, changing the previous vague state of judging equipment health based on experience.

[0020] 3. This invention uses the remaining life prediction module based on the output of the index comparison analysis module, combined with the equipment's full life cycle data and fatigue damage theory, to accurately calculate the current cumulative damage and remaining operating time. Compared with traditional operation and maintenance, which cannot predict the remaining life, this invention enables operation and maintenance personnel to plan maintenance and replacement of parts in advance, rationally arrange production plans, reduce the risk of sudden failures, and transform equipment operation and maintenance from passive response to proactive prevention. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0022] Figure 2 This is a flowchart of the health calculation process of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] As attached Figure 1 The digital twin-based fault prediction and health management system for quick-freezing equipment, as shown, includes the following steps in its specific implementation:

[0025] The data acquisition module is used to acquire the full lifecycle operation dataset of the target quick-freezing equipment and its counterparts.

[0026] As a preferred and feasible embodiment, it should be explained that the full lifecycle dataset includes basic information data of the target quick-freezing equipment and equipment of the same model, real-time operation data, historical fault data, and operation and maintenance record data.

[0027] Basic information data includes equipment model, compressor / evaporator core parameters, refrigerant type, and rated operating threshold; real-time operating data includes compressor stator current, casing vibration acceleration, suction pressure, evaporator surface temperature, and compressor outlet air velocity; historical fault data includes fault type and occurrence time, key parameter change curves before and after the fault, fault repair plan and effect; operation and maintenance record data includes cumulative running time, component replacement time and model, regular maintenance content and frequency, and cleaning and defrosting records.

[0028] The acquisition methods are as follows: Historical fault records and basic information data are acquired in batches by connecting the equipment control system and the maintenance management platform of the same equipment manufacturer via API interface; compressor stator current and suction pressure are directly read from the PLC's IO module through an industrial gateway using the standard monitoring points built into the equipment PLC; vibration acceleration is collected by installing an ICP-type accelerometer, evaporator surface temperature is collected by installing a PT100 platinum resistance thermometer, and compressor outlet air velocity is collected by installing a thermal anemometer outputting a Modbus digital signal; structured maintenance data is formed by recording equipment maintenance, component replacement, and other operation records by maintenance personnel; finally, all data with high-precision timestamps are aligned by timestamp and integrated into a full lifecycle dataset, supporting rapid access to subsequent modeling and analysis modules.

[0029] The virtual modeling module is used to construct a multi-physical domain coupled model of quick-freezing equipment;

[0030] As a preferred and feasible embodiment, it should be explained that the multi-physics domain coupling model includes:

[0031] Mechanical domain sub-model: encompasses compressor rotor dynamics model and evaporator structural mechanics model; the former is built based on rotor material characteristics, bearing friction coefficient and historical vibration data, and can simulate torque and vibration response under normal rotor operation, jamming, imbalance and other states; the latter combines evaporator fin size and material strength parameters to simulate the influence of frost thickness change on fin deformation and air duct resistance, and realize the dynamic correlation between mechanical vibration and structural deformation.

[0032] Thermodynamic domain sub-model: includes a refrigeration cycle model and an evaporator heat exchange model; the refrigeration cycle model, based on the thermodynamic properties of the refrigerant type, couples compressor suction / discharge pressure and refrigerant flow parameters, and dynamically modifies key parameters such as refrigerant mass flow rate and compressor isentropic efficiency in the model through a software interface, simulating the system pressure and temperature response under various operating conditions such as normal operation, refrigerant leakage, and compressor wear; the heat exchange model, combined with evaporator surface temperature and air velocity data, simulates the dynamic process of evaporator frosting and scaling leading to a decrease in heat exchange efficiency by adjusting the heat transfer coefficient and duct resistance coefficient, reflecting its quantitative impact on the cooling effect. The thermodynamic domain sub-model is also linked with the mechanical domain model to achieve precise coupling of mechanical and thermodynamic processes;

[0033] Control domain sub-model: Based on the original equipment manufacturer's control logic, it integrates control rules such as grid voltage regulation and load adaptation. It can simulate the coordinated response of each physical domain sub-model under different operating commands, realize the full-domain coupling of mechanical, thermodynamic and control logic, and ensure that the virtual model is completely consistent with the operating logic of the physical equipment.

[0034] The theoretical fault index generation module is used to simulate operating scenarios with different fault types and different fault degrees by injecting fault parameters into the digital twin model, and generate a multi-dimensional theoretical fault index matrix.

[0035] It should be explained that the steps for generating the multi-dimensional theoretical fault index matrix are as follows:

[0036] A1. Fault Injection and Simulation Execution;

[0037] It should be noted that the type of fault must be identified before performing this step. and degree of failure Then, the fault state is simulated in the digital twin model: This defines typical faults in quick-freezing equipment, such as... =Refrigerant leak, =Compressor wear, etc.; degree of failure Different severity levels are set for each type of fault, in order to Taking refrigerant leakage as an example, we can see that =Slight leakage, =Moderate leakage, etc.

[0038] It needs to be further explained that when performing this step, the corresponding physical parameters in the digital twin model are modified to the fault state values, the simulation is run, and the output data is recorded.

[0039] A2. Multi-dimensional feature extraction to generate core indices;

[0040] It should be explained that the core indices include the vibration index, thermodynamic performance index, and electrical index, and the specific formulas are as follows:

[0041] The vibration index K is calculated according to the statistical definition of kurtosis and is a classic indicator for extracting impact-type fault characteristics in mechanical fault diagnosis. The formula is: , This refers to the instantaneous value of the compressor's vibration acceleration at the nth vibration sampling point. Let σ be the mean of the vibration data, σ be the standard deviation of the vibration data, and N be the total amount of vibration data collected. The numerator of the formula is... It is the fourth central moment, and the denominator is... It is the fourth power of the standard deviation, with the numerator and denominator having matching dimensions to ensure that K is a dimensionless index that can be compared across equipment and operating conditions.

[0042] Thermodynamic performance index: , The cooling capacity of the digital twin model is calculated through numerical simulation based on thermodynamic principles and equipment physical parameters, and expressed using formulas. Calculations show that Let A be the heat transfer coefficient, A be the heat transfer area, and ΔT be the temperature difference. The compressor power consumption is calculated using the power balance equation based on the compressor characteristic curve and simulation parameters in the digital twin model. The formula is as follows: The calculations show that U is the voltage and I is the current. The power factor (COP) directly reflects the efficiency of the refrigeration system. Faults such as refrigerant leakage, evaporator frosting, and compressor wear can all lead to a decrease in COP.

[0043] Electrical Index: , The amplitude of the fundamental current wave. Let H be the current amplitude of the h-th harmonic, where H is the upper limit of the harmonic order (usually taken as 30), and h is the harmonic order variable. It is the total harmonic distortion rate of the current.

[0044] A3. Construct a multi-dimensional theoretical fault index matrix.

[0045] It should be explained that a set of feature indices [K, COP, THD] can be extracted for each type of fault scenario, and the index results of all fault scenarios can be integrated to form a multi-dimensional theoretical fault index matrix.

[0046] The real-time fault index calculation module is used to generate a real-time fault index from the real-time operating data acquired by the data acquisition module.

[0047] It should be explained that the real-time fault index is a dynamic characteristic index generated based on the timestamped real-time operating data obtained by the data acquisition module, according to the calculation logic that is completely consistent with the theoretical fault index. The specific generation process is as follows: First, the real-time operating data is preprocessed: the original compressor vibration acceleration signal is subjected to a 50Hz low-pass filter to remove high-frequency interference from the power grid; the stator current signal is subjected to a Fourier transform to decompose the fundamental wave and each harmonic component; parameters such as suction pressure and evaporator temperature are processed by moving average to eliminate instantaneous fluctuations.

[0048] Subsequently, the calculations were performed synchronously according to the formulas for the three core indices: the vibration index K was calculated using a real-time collected vibration acceleration sequence, with 50 sampling points per second, to determine the ratio of the fourth-order central moment to the fourth power of the standard deviation at the current moment; the thermodynamic performance index COP was calculated by back-calculating the cooling capacity using real-time suction and discharge pressures. By combining the compressor's real-time power data, the voltage and current read from the PLC are calculated to obtain... Then calculate the ratio of the two; electrical index The distortion rate is calculated based on the real-time amplitudes of the fundamental and harmonic components of the current, using the distortion rate formula.

[0049] The final generated real-time fault index is automatically associated with the theoretical fault index of the corresponding timestamp, forming a real-time index-theoretical index pair. This ensures that the two are completely matched in terms of calculation dimension, scale, and time scale, providing dynamic feature data that can be directly compared for the subsequent index comparison and analysis module.

[0050] The index comparison and analysis module includes: a deviation threshold library establishment unit and a real-time status determination unit;

[0051] The deviation threshold database establishment unit is used to analyze the deviation patterns before the occurrence of faults in historical fault data and form a dynamically updated deviation threshold database.

[0052] As a preferred feasible embodiment, it should be explained that the deviation pattern is based on the full lifecycle historical data accumulated by the data acquisition module, extracting the real-time fault index sequence and theoretical fault index sequence corresponding to different severity levels of the fault 1 to 7 days before the fault occurs; for each type of index, the real-time theoretical index deviation in the fault precursor stage is calculated: The normal distribution of the deviation was fitted using statistical process control methods, and two types of thresholds were calculated: the normal deviation threshold and the normal deviation threshold. The upper limit of the deviation band during normal equipment operation is calculated by adding three times the standard deviation to the mean deviation during the fault-free steady-state operation phase; the warning deviation threshold is also defined. Take the 25th percentile of the deviation data during the early stage of the fault as the critical deviation value at which the fault is about to occur.

[0053] It should be explained that the threshold library is not static, but continuously optimized as the equipment operates. When the index comparison and analysis module determines that the equipment has entered the warning state, if the physical equipment does not fail within the next 24 hours, the warning threshold for the corresponding fault type will be automatically adjusted; if the equipment actually fails, the current threshold will be strengthened.

[0054] The real-time status determination unit is used to calculate the deviation between the real-time fault index and the theoretical fault index in real time, and output the status of the device in combination with the deviation threshold library.

[0055] It should be explained that the deviation value is calculated based on the dynamic index output by the real-time fault index calculation module. , , The theoretical fault index generation module stores the same operating conditions and simultaneous theoretical indices. , , Calculate the deviation dimension by dimension: , , After determining the degree to which each index deviates from the theoretical value, the dynamic threshold library is called to compare the deviation value with the dynamic threshold. If... This proves that the single-index state determination is normal and the equipment is operating stably without fault characteristics; if If the single-index state is determined as a warning, the equipment is a precursor to failure, and a certain physical domain characteristic begins to malfunction; if If a single index indicates a fault, a device malfunction is imminent.

[0056] After determining the initial state corresponding to each index, a sensitivity weight for each index to faults is introduced. Through a weighted fusion algorithm, the state of a single index is integrated into the global state of the equipment, and three types of results are finally output: if all deviations are within the normal threshold or the comprehensive score is low, it is judged as normal; if a single index triggers an early warning or the comprehensive score reaches the mild range, it is judged as a mild early warning and a single domain anomaly is indicated; if an index triggers a fault or the comprehensive score reaches the severe range, it is judged as a severe early warning and a multi-domain fault requiring emergency intervention is indicated. At the same time, the output results are associated with the source of abnormal indices. For example, an abnormal vibration index points to a mechanical fault in the compressor and maintenance suggestions, realizing a complete closed loop from data comparison to fault diagnosis and maintenance guidance. This makes the equipment state determination accurate and traceable, providing direct support for maintenance decisions.

[0057] The remaining life prediction module is used to calculate the remaining operating time of the equipment from its current state to the occurrence of a fault, based on the fault type identification results output by the index comparison analysis module and the current operating status of the equipment.

[0058] As a preferred and feasible embodiment, it should be explained that the calculation of remaining runtime requires the fault type and current state output by the exponential comparison analysis module as the core input, combined with the equipment's full life cycle data and fatigue damage theory, and is achieved through quantitative mapping, cumulative calculation, and lifespan back-calculation, as detailed below:

[0059] First, based on historical fault data of the same model of equipment, a mapping relationship of "fault type - status - damage rate" is constructed. For example, in compressor wear faults, a minor warning corresponds to a bearing damage rate of 0.008 per hour, and a severe warning corresponds to 0.025 per hour; in refrigerant leakage faults, a minor warning corresponds to a seal damage rate of 0.005 per hour, and a severe warning corresponds to 0.018 per hour, ensuring that the damage rate closely matches the actual wear pattern. Next, the current cumulative damage is calculated using the following formula: , It is the runtime of the device in the i-th state. This refers to the damage rate corresponding to the current state. For example, if the bearing has been running for 1200 hours since the last replacement, including 300 hours in the mild warning stage, then... .

[0060] The damage growth pattern is determined by using a digital twin model, and then based on the component's rated life parameters, the following formula is applied: Calculate the remaining running time. The rated damage threshold for the component. The damage rate is the current state. and These are the environmental and maintenance correction factors, respectively. If the current level is a mild warning, the remaining tolerable damage is 10 - 2.4 = 7.6. Hours, final output results and maintenance suggestions.

[0061] The health management module is used to build a health evaluation index system for quick-freezing equipment, calculate the equipment health score, and divide the equipment health status into four levels.

[0062] It needs to be explained that, by appendix Figure 2 As shown, the steps for calculating the equipment health score are as follows:

[0063] B1. Calculate the health score of a single indicator;

[0064] The deviation value calculated in real time based on the index comparison analysis module Deviation thresholds provided by the deviation threshold library establishment unit The instantaneous health status of each core failure index is calculated using the following formula: ,when When it is 0, =100 points, when the deviation reaches the warning threshold hour, point.

[0065] B2. Calculate the preliminary comprehensive health score;

[0066] The preliminary comprehensive health score is obtained by weighting and integrating the health scores of individual indicators. , The weights of the vibration index K, The weight of the thermodynamic performance index COP. The weight of the electrical index THD is dynamically adjusted by the prediction optimization module based on the historical accuracy of early warnings.

[0067] B3. Introduce remaining lifespan health;

[0068] Based on the output of the remaining life prediction module, a life depreciation factor is calculated to reflect the natural aging and depreciation of the equipment due to long-term use. The calculation formula is as follows: , This is the useful life depreciation factor, with a value ranging from 0 to 1. It is the current remaining runtime calculated by the remaining lifetime prediction module. This is the rated design life of the device, derived from basic data.

[0069] B4. Calculate the overall health score of the equipment;

[0070] The final overall equipment health score is obtained by multiplying the preliminary comprehensive score by the life depreciation factor. The calculation formula is as follows: The score ranges from 0 to 100 points and is a comprehensive indicator that reflects both the instantaneous operating status of the equipment and its long-term lifespan degradation.

[0071] B5. Classify health status levels.

[0072] Based on the final score, the device health status is divided into four levels: 90≤ ≤100 indicates Level 1 health, meaning the equipment is in optimal operating condition and all indicators are normal; 70≤ <90 indicates Level Two, Sub-health, where the device shows slight performance degradation or a single indicator triggers a warning, but overall function is normal; 50≤ <70 indicates a Level 3 warning, signifying significant equipment malfunction, performance degradation, or multiple warning indicators triggering a maintenance plan; 0≤ <50 indicates a Level 4 fault, meaning the equipment is in a severely abnormal state, and existing indicators have triggered a fault alarm, requiring immediate intervention.

[0073] The prediction and optimization module dynamically updates the index weights of each dimension in the index comparison and analysis module based on historical early warning results and actual fault occurrences.

[0074] As a preferred and feasible embodiment, it should be explained that the prediction and optimization module first periodically extracts historical early warning records from the index comparison and analysis module, correlates them with the actual fault occurrence time of the remaining life prediction module and the health level changes of the health management module, constructs an early warning fault matching degree model, calculates the false alarm rate, false negative rate, and lead time deviation, and iteratively optimizes the index weights of the real-time status judgment unit based on the analysis results. If a false alarm occurs, the corresponding index weight is lowered; if a false negative occurs, it is raised; if the lead time deviation is large, the fault injection parameters of the digital twin model are corrected. The adjustment results are fed back to the real-time status judgment unit to update the index weights, the theoretical fault index generation module to correct the simulation parameters, and the health management module to optimize the life depreciation factor calculation logic. By default, full optimization is performed automatically every week, and immediate optimization is triggered by major faults. After optimization, historical fault data is injected into the digital twin model for verification, ensuring that the false alarm rate and false negative rate are reduced, allowing the system's fault diagnosis capability to continuously evolve, adapt to the equipment's operating characteristics, and achieve the goal of intelligent operation and maintenance.

[0075] The visualization module is used to provide users with a visual interface.

[0076] As a preferred and feasible embodiment, it should be explained that the visualization module provides users with a visual interface. It deeply integrates data and analysis results from modules such as data collection, health management, and remaining lifespan prediction to construct a multi-dimensional and interactive panoramic view of equipment health. Based on core information such as full lifecycle data, health level, and remaining operating time, a three-level visualization architecture is built. The status overview page uses a 3D digital twin model of the equipment as a carrier, uses color coding for health level, and synchronously displays key indicators such as health score and remaining operating time, allowing users to quickly grasp the overall status. Clicking on key components can enter a detailed drill-down page to expand the health details of the corresponding subsystem, covering parameter curves and deviation trends of mechanical, thermodynamic, and electrical subsystems, and connecting to data from various modules. The historical traceability page supports querying full lifecycle data by time axis, reproducing parameter changes before failure, and linking maintenance records to assist in analyzing the evolution pattern of failure.

[0077] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0078] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A digital-twin based quick-freezing equipment failure prediction and health management system, characterized in that, The application relates to a quick-freezing equipment health management system, which comprises the following parts: a data acquisition module for acquiring a full-life-cycle operation data set of a target quick-freezing equipment and a same-model quick-freezing equipment; a virtual modeling module for constructing a quick-freezing equipment multi-physical-domain coupling model; a theoretical fault index generation module for generating a multi-dimensional theoretical fault index matrix by injecting fault parameters in a digital twin model to simulate operation scenes of different fault types and different fault degrees; a real-time fault index calculation module for generating a real-time fault index based on real-time operation data acquired by the data acquisition module; an index comparison and analysis module, which comprises a deviation threshold value library establishing unit and a real-time state judging unit; the deviation threshold value library establishing unit is used for analyzing deviation rules before faults occur in historical fault data to form a dynamically updated deviation threshold value library; the real-time state judging unit is used for calculating a deviation value of the real-time fault index and the theoretical fault index in real time, combining the deviation threshold value library, and outputting a state of the equipment; a residual life prediction module for calculating a residual operation time of the equipment from a current state to a fault occurrence based on a fault type identification result output by the index comparison and analysis module and a current operation state of the equipment; a health management module for constructing a quick-freezing equipment health degree evaluation index system, calculating a health degree score of the equipment, and dividing the health state of the equipment into four grades; a prediction optimization module for dynamically updating weight of each dimension index in the index comparison and analysis module based on historical early warning results and actual fault occurrence conditions; a visual display module for providing a visual interface for a user.

2. The digital-twin-based quick-freezing equipment failure prediction and health management system according to claim 1, characterized in that: The full-life-cycle operation data set comprises basic information data, real-time operation data, historical fault data and operation and maintenance record data of the target quick-freezing equipment and the same-model equipment.

3. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 1, wherein: The multi-dimensional theoretical fault index matrix is generated in the following steps: A1, fault injection and simulation operation; A2, multi-dimensional feature extraction to generate core indexes; A3, construction of a multi-dimensional theoretical fault index matrix.

4. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 3, wherein: The core indexes comprise a vibration index K, a thermodynamic performance index and an electrical index.

5. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 4, wherein: The calculation of the vibration index K follows the definition of statistical kurtosis, which is a classical index for extracting impact fault features in mechanical fault diagnosis, and the formula is: , is the nth vibration sampling point data, that is, the instantaneous value of compressor vibration acceleration, is the mean value of vibration data, σ is the standard deviation of vibration data, N is the total amount of collected vibration data, the formula numerator is the fourth order central moment, and the denominator is the fourth power of the standard deviation, and the numerator and denominator have the same dimension, so that K is a dimensionless index, which can be compared across devices and across operating conditions.

6. The digital twin based quick freeze equipment failure prediction and health management system as claimed in claim 1, wherein: The health degree score of the equipment is calculated in the following steps: B1, calculation of health degrees of single indexes; B2, calculation of a preliminary comprehensive health score; B3, introduction of a residual life health degree; B4, calculation of an overall health degree score of the equipment; B5, division of health state grades.

7. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 6, wherein: The calculation of the health degree of the single index is based on the deviation value calculated by the index contrast analysis module in real time The deviation threshold provided by the deviation threshold library establishment unit For each core fault index, its real-time health degree is calculated, and the calculation formula is: When =0, =100 points, when the deviation reaches the early warning threshold , Points.

8. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 6, wherein: The preliminary comprehensive health score is obtained by weighted fusion of single-index health degrees: , H Ik is the health degree of the vibration index K, ICOP is the health degree of the thermodynamic performance index COP, ITHD is the health degree of the electrical index THD, is the weight of the vibration index K, is the weight of the thermodynamic performance index COP, is the weight of the electrical index THD, which is dynamically adjusted by the prediction optimization module according to historical early warning accuracy.

9. The digital-twin-based quick-freeze equipment failure prediction and health management system of claim 6, wherein: The device overall health score is the multiplication of the preliminary comprehensive score and the life depreciation factor, and the final device overall health score is calculated by the formula: , is the device overall health score, is the life depreciation factor, and the device overall health score ranges from 0 to 100, which is a comprehensive index reflecting both the instantaneous running state of the device and the long-term life loss.

10. The digital twin based quick freeze equipment failure prediction and health management system as claimed in claim 6, wherein: The health status rating is a division of the device health status into four levels according to the final score: 90 ≤ 100, as level one health. 70 < 90, grade two, sub-health; 50 < 70, grade three, early warning; 0 < 50, grade four, failure.

Images