Real-time fault prediction method and system for central air-conditioning system based on digital twinning

Abstract

The invention provides a central air-conditioning system real-time fault prediction method and system based on digital twinning, and the method comprises the steps: deploying various types of sensors for multi-dimensional detection, and collecting the data of the sensors; edge computing nodes are deployed around the central air conditioning unit; a digital twinborn model based on the central air conditioning unit is constructed according to the preprocessed data, a dynamic view of key performance indexes is provided, and the state of the central air conditioning unit is mapped and simulated in real time; designing a fault prediction model, and predicting and identifying faults; deploying the fault prediction model at an edge computing node and operating the fault prediction model, and performing real-time fault monitoring and diagnosis; when a fault is found, the AI-drive maintains decision support, and automatically adjusts a fault prediction model parameter or arranges a maintenance strategy; and constructing a learning loop based on performance feedback, and optimizing a fault prediction model and a maintenance strategy. Quantum calculation, the digital twinning technology and edge calculation are fused, and the operation efficiency and reliability of the central air conditioning system are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of air conditioning fault monitoring, and in particular to a real-time fault prediction method and system for a central air conditioning system based on digital twins. Background Art

[0002] The efficient and reliable operation of central air-conditioning systems is essential to modern building environments. With the advancement of technology, traditional monitoring and maintenance methods have gradually failed to meet the high standards of energy efficiency and system reliability required by modern buildings. Existing air-conditioning system fault monitoring methods are mainly based on preset rules and simple fault diagnosis algorithms, such as temperature and pressure anomaly detection. These methods often rely on static data thresholds, which are not only limited in effectiveness in complex fault situations, but also difficult to adapt to situations with large data volumes and fast environmental changes.

[0003] In addition, the data processing of existing technologies mostly relies on cloud computing platforms, and data needs to be uploaded to the cloud for analysis and processing, which not only increases latency but also introduces risks to data security and privacy. These technical limitations lead to untimely response to air conditioning system fault handling, high maintenance costs, and inability to effectively predict and prevent potential faults.

[0004] In this context, in order to meet the needs of high-efficiency operations and the exploration of advanced technologies, there is an urgent need to develop a new type of fault prediction system that can achieve efficient, accurate and real-time fault monitoring and prediction while ensuring the security and timeliness of data processing. Summary of the invention

[0005] The purpose of the present invention is to design a real-time fault prediction method and system for a central air-conditioning system based on digital twins, which not only significantly improves the operating efficiency and reliability of the central air-conditioning system, but also ensures the efficiency and security of data processing. The development and application of this invention will promote the intelligent and digital transformation of the HVAC industry and provide strong technical support for future building environment management.

[0006] In order to achieve the above object, in a first aspect of the present invention, a real-time fault prediction method for a central air-conditioning system based on digital twin is provided, comprising the following steps:

[0007] Deploy various types of sensors around the central air conditioning units to conduct multi-dimensional detection, collect sensor data, and perform preliminary processing on the data;

[0008] Deploy edge computing nodes around central air conditioning units to pre-process the data after preliminary processing;

[0009] Build a digital twin model based on the central air-conditioning unit based on the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time;

[0010] Design fault prediction models based on quantum computing and machine learning algorithms to predict and identify faults;

[0011] Deploy the fault prediction model on the edge computing node and run it for real-time fault monitoring and diagnosis. When a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, automatically adjusting fault prediction model parameters or arranging maintenance strategies.

[0012] Build a learning loop based on performance feedback to optimize failure prediction models and maintenance strategies;

[0013] The design is based on a fault prediction model of quantum computing and machine learning algorithms to predict and identify faults, including:

[0014] The quantum support vector machine is selected to build the fault prediction model, which is expressed as:

[0015]

[0016] Among them, α represents the Lagrange multiplier vector, Q represents the matrix calculated by the training data and the quantum kernel function, and e represents the all-1 vector;

[0017] The preprocessed data is encoded onto the qubit using amplitude coding, expressed as:

[0018]

[0019] Among them, x i represents the components of the normalized data vector, |i> represents the basis vector of the quantum state;

[0020] Construct a quantum kernel function to calculate the similarity between data points on a quantum device, expressed as:

[0021] K(x i ,x j )=|<ψ(x i )∣ψ(x j )>| 2

[0022] Among them, |ψ(x i )> and |ψ(x j )> respectively represent the quantum states after encoding the two data points, K(x i ,x j ) represents the value of quantum kernel function;

[0023] A quantum optimization algorithm is run to train the fault prediction model and find the best hyperplane to distinguish between normal operating states and potential fault states.

[0024] As an improvement of the above solution, the quantum optimization algorithm includes:

[0025] Initialize quantum states and quantum circuits;

[0026] Apply quantum gates to simulate the training process of fault prediction models;

[0027] Perform quantum measurements to obtain model parameters.

[0028] As an improvement of the above solution, the sensors include a temperature sensor, a pressure sensor, a humidity sensor, a meteor sensor and a vibration sensor.

[0029] As an improvement of the above solution, the preliminary processing includes performing low-pass filtering on the data collected by the sensor to eliminate high-frequency noise, including:

[0030] Let x(t) be the original signal, and the filtered signal y(t) be expressed as follows:

[0031] y(t)=θ·x(t)+(1-α)·y(t-1)

[0032] Wherein, θ is the filter coefficient, and its value range is 0<α<1; t represents time t.

[0033] As an improvement to the above solution, edge computing nodes are deployed around the central air conditioning unit to pre-process the data after preliminary processing, including:

[0034] Dynamically adjust edge computing deployment based on the actual operating load and data generation of the central air-conditioning unit system;

[0035] Deploy network connections with failover and multi-path transmission capabilities to ensure automatic switching to a backup connection when one connection fails; where the network connections are deployed using multi-protocol label switching or software-defined networking;

[0036] Implement data preprocessing functions at each edge node and deploy security measures;

[0037] The data preprocessing function implemented at each edge node includes:

[0038] Convert the data collected by different sensors into a unified standard unit, then perform time series analysis on the data, identify periodic changes and trends in the data, and use the autoregressive moving average model for modeling;

[0039] Extract key features from the data stream of each sensor and combine relevant data from different sensors to form comprehensive features.

[0040] As an improvement of the above scheme, the digital twin model based on the central air-conditioning unit is constructed according to the preprocessed data, including:

[0041] Define the main components of the central air conditioning unit, set the physical parameters for each component, establish the state equation for each component, describe its dynamic behavior and use transfer functions to simulate the interaction between components;

[0042] The preprocessed sensor data is integrated into the digital twin model, and the status of the digital twin model is updated in real time, as shown below:

[0043] θ new =θ old +ε(z i -θ old )

[0044] Among them, θ new represents the new model parameters, ε is the learning rate, θ old represents the old model parameters, z i represents the preprocessed data points;

[0045] Run historical data and real-time data on the digital twin model to perform simulations, calculate the difference between the model's predictions and the actual system operation, and dynamically adjust the parameters of the digital twin model based on the differences.

[0046] As an improvement of the above solution, the fault prediction model is deployed on the edge computing node and runs, including:

[0047] Use the message queue management system to manage the data received by edge computing nodes in real time;

[0048] The fault detection threshold is dynamically adjusted according to the operating status and environmental conditions of the central air-conditioning unit, as shown below:

[0049] θ t+1 =θ t +β(x t -θ t )

[0050] Among them, θ t+1 represents the threshold at time t+1, θ t represents the threshold at time t, x t represents the latest observation data, and β represents the adjustment coefficient, which is used to smoothly update the threshold;

[0051] Based on the real-time updated fault prediction model and dynamic thresholds, the system status is continuously monitored to detect anomalies; where detection is represented as follows:

[0052]

[0053] Among them, f(x) represents the prediction function, x represents the current input data, α i ,y i ,x i and b represent the Lagrange multiplier, the label of the training sample and the bias term;

[0054] When a potential fault is detected, diagnostic information is generated and sent to the maintenance team.

[0055] As an improvement of the above solution, the AI-driven maintenance decision support and fault response simultaneously include:

[0056] Design a rules engine that contains various failure scenarios and corresponding action guidelines;

[0057] Build a fault database and corresponding strategies, and automatically trigger corresponding maintenance measures when a fault is detected;

[0058] The results of each maintenance are recorded as a maintenance log, and machine learning technology is used to analyze the maintenance log for the improvement of future strategies.

[0059] As an improvement of the above solution, the construction of a learning loop based on performance feedback to optimize the fault prediction model and maintenance strategy includes:

[0060] Continuously collect the operating data of central air-conditioning units, conduct statistical analysis on the operating data, and use statistical control charts to track changes in operating data;

[0061] Update the parameters of the digital twin model and fault prediction model based on the results of statistical analysis, and dynamically adjust the fault response strategy and maintenance plan based on minimizing the delay and cost of maintenance response;

[0062] Integrate newly acquired knowledge and experience into the fault database and rule engine, and update the fault database and maintenance strategy;

[0063] Apply machine learning algorithms to periodically retrain the fault prediction model.

[0064] In a second aspect of the present invention, a real-time fault prediction system for a central air-conditioning system based on digital twin is provided, the system comprising:

[0065] Data acquisition device, used to deploy various types of sensors around the central air conditioning unit to perform multi-dimensional detection, collect sensor data, and perform preliminary processing on the data;

[0066] Edge computing devices are used to deploy edge computing nodes around central air conditioning units to pre-process the data after preliminary processing;

[0067] Fault prediction device, used to build a digital twin model based on the central air-conditioning unit according to the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time; design a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults; deploy and run the fault prediction model on the edge computing node to perform real-time fault monitoring and diagnosis; when a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, automatically adjusting fault prediction model parameters or arranging maintenance strategies;

[0068] System update device for building a learning loop based on performance feedback to optimize fault prediction models and maintenance strategies;

[0069] The design is based on a fault prediction model of quantum computing and machine learning algorithms to predict and identify faults, including:

[0070] The quantum support vector machine is selected to build the fault prediction model, which is expressed as:

[0071]

[0072] Among them, α represents the Lagrange multiplier vector, Q represents the matrix calculated by the training data and the quantum kernel function, and e represents the all-1 vector;

[0073] The preprocessed data is encoded onto the qubit using amplitude coding, expressed as:

[0074]

[0075] Among them, x i represents the components of the normalized data vector, |i> represents the basis vector of the quantum state;

[0076] Construct a quantum kernel function to calculate the similarity between data points on a quantum device, expressed as:

[0077] K(x i ,x j )=|<ψ(x o )∣ψ(x j )>| 2

[0078] Among them, |ψ(x i )> and |ψ(x j )> respectively represent the quantum states after encoding the two data points, K(x i ,x j ) represents the value of quantum kernel function;

[0079] A quantum optimization algorithm is run to train the fault prediction model and find the best hyperplane to distinguish between normal operating states and potential fault states.

[0080] The beneficial technical effects of the present invention are at least as follows:

[0081] (1) The present invention uses quantum computing to optimize the calculation process of the fault prediction model. In particular, quantum computing can significantly improve the calculation speed and accuracy when processing large-scale data sets and complex algorithms. Traditional algorithms are slow to operate in a big data environment and have difficulty responding in real time. The addition of quantum computing enables the system to process complex data in real time, improving the timeliness and accuracy of fault detection.

[0082] (2) The present invention constructs a digital twin model of the central air conditioning system, which reflects the physical state of the system in real time. Through continuous simulation and analysis, it can predict and identify potential fault points. Traditional monitoring technology cannot accurately simulate the overall picture and dynamic changes of the system. The use of digital twin technology provides a comprehensive dynamic monitoring platform for the system, enhancing the comprehensiveness and accuracy of fault prediction.

[0083] (3) The present invention integrates edge computing nodes at key parts of the system to achieve local rapid data processing, reduce dependence on central servers, and improve data processing speed and security. It reduces the delay and security risks caused by data transmission, making fault detection and response more timely and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0085] Figure 1 This is a flow chart of the real-time fault prediction method for a central air-conditioning system based on digital twins of the present invention.

[0086] Figure 2 This is a framework diagram of the real-time fault prediction system of the central air-conditioning system based on digital twins of the present invention. DETAILED DESCRIPTION

[0087] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0088] AI-driven refers to the artificial intelligence driven module.

[0089] In one or more embodiments, Figure 1As shown, a real-time fault prediction method for a central air-conditioning system based on digital twin is disclosed, and the method includes the following steps S1-S6:

[0090] S1. Deploy various types of sensors around the central air-conditioning unit to perform multi-dimensional detection, collect sensor data, and perform preliminary processing on the data;

[0091] S2, deploy edge computing nodes around the central air conditioning units to pre-process the data after preliminary processing;

[0092] S3. Build a digital twin model based on the central air-conditioning unit based on the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time;

[0093] S4. Design a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults;

[0094] S5. Deploy the fault prediction model on the edge computing node and run it to perform real-time fault monitoring and diagnosis. When a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, automatically adjusting fault prediction model parameters or arranging maintenance strategies.

[0095] S6. Build a learning loop based on performance feedback to optimize fault prediction models and maintenance strategies.

[0096] Exemplarily, the specific working process of the method for monitoring the operation failure of the assembled central air-conditioning unit is as follows:

[0097] (1) Providing raw input data. By installing a variety of sensors at key locations, this step ensures that comprehensive and detailed operating data is collected from every important component of the system. This is a basic step because without accurate and comprehensive data, subsequent analysis and prediction are impossible.

[0098] (2) Perform preliminary processing based on the data collected in (1). The application of edge computing reduces the dependence on central servers, reduces the delay and risk of data transmission, and ensures the timeliness and security of data processing. This step provides high-quality input data for the digital twin model and subsequent data analysis.

[0099] (3) Using the data processed in (2), a virtual copy of the air conditioning system is established and updated in real time. This model can accurately reflect the real-time status of the physical system and provide a dynamic analysis basis for fault prediction. The application of digital twin technology has significantly improved the accuracy and foresight of fault diagnosis.

[0100] (4) Utilize the powerful computing power of quantum computing to optimize and accelerate the training process of machine learning models, so that the models can efficiently process large-scale complex data. This step directly improves the calculation speed and accuracy of the fault prediction model and provides technical support for real-time fault monitoring.

[0101] (5) The model developed in (4) is deployed on the edge computing node to achieve real-time monitoring of system status and instant fault diagnosis. This step ensures that system faults can be detected and handled in the shortest possible time, greatly reducing system downtime and maintenance costs. Maintenance strategies are automatically formulated and executed based on the diagnosis results. This automated maintenance response mechanism reduces manual intervention and improves maintenance efficiency and accuracy.

[0102] (6) Establish a continuous learning mechanism based on performance feedback to continuously optimize fault prediction models and maintenance strategies. Through dynamic learning and adjustment, the system can adapt to the impact of environmental changes and equipment aging and maintain long-term high performance.

[0103] The following is a detailed description of each step:

[0104] S1. Deploy various types of sensors around the central air-conditioning units to conduct multi-dimensional detection, collect sensor data, and perform preliminary processing on the data.

[0105] These include the following sensors:

[0106] Temperature sensors: located near the air outlet, return air outlet and compressor of the air conditioner. The temperature data at these locations is crucial for monitoring the thermal efficiency of the system and potential overheating problems.

[0107] Pressure sensor: Installed at the inlet and outlet of the refrigerant pipeline. Pressure changes can indicate whether there is a leak or other flow problem in the refrigeration system.

[0108] Humidity sensor: It is installed at the air inlet and outlet of the air conditioner. The monitoring of humidity level helps to evaluate the dehumidification performance of the air conditioner.

[0109] Meteor Sensor: Placed in cooling water system and refrigerant circulation system, flow data can reflect the cooling efficiency and potential blockage of the system.

[0110] Vibration sensor: Installed on the bearings of compressors and fans, abnormal increase in vibration is usually a precursor to mechanical failure.

[0111] Furthermore, the data is collected and preliminarily processed, including:

[0112] All sensors are set to collect data once a second to capture instant state changes in system operation;

[0113] The time stamp method is used to ensure the temporal consistency of data collected from different sensors, which is convenient for subsequent analysis;

[0114] The sensor data is low-pass filtered to remove high-frequency noise. Let x(t) be the original signal, and the filtered signal y(t) can be calculated by the following formula:

[0115] y(t)=θ·x(t)+(1-α)·y(t-1) (1)

[0116] Among them, θ is the filter coefficient, and its value range is 0<α<1;

[0117] Perform anomaly detection on each sensor reading to eliminate abnormal data caused by equipment failure or external interference. Set upper and lower thresholds and set a reasonable threshold range based on historical data.

[0118] S2. Deploy edge computing nodes around the central air-conditioning units to pre-process the data after preliminary processing.

[0119] Specifically, the deployment of edge computing resources is dynamically adjusted according to the actual operating load and data generation of the air-conditioning system. By analyzing data traffic and processing requirements in real time, the system automatically decides whether to add more edge nodes or adjust the computing resource allocation of existing nodes, using a resource management algorithm based on a predictive model, which uses historical data analysis and real-time performance indicators to predict future data processing needs and adjust resources accordingly. t represents the computational load at time t, C t For the currently available computing resources, the adjustment strategy is defined by the following relationship:

[0120] C t+1 =f(L t ,C t ) (2)

[0121] Where f is the resource adjustment function, which optimizes resource allocation according to predetermined performance criteria and cost-benefit analysis.

[0122] Among them, network connections with failover and multi-path transmission capabilities are deployed to ensure automatic switching to backup connections when one connection fails, thereby maintaining the continuity of data transmission and high availability of the system.

[0123] It should be noted that the implementation of Multi-Protocol Label Switching (MPLS) or Software Defined Networking (SDN) allows for intelligent routing and management of network traffic, improving the stability and efficiency of data transmission.

[0124] Furthermore, data preprocessing is implemented at each edge node so that data can be processed at the place closest to the source, reducing transmission delays and improving response speed. Distributed processing modules are developed so that each node not only collects data but also performs preliminary cleaning, standardization, and feature extraction. For example, if x i is the original data point, and the preprocessed data z i This can be obtained through a standardized process that is performed locally:

[0125]

[0126] where μ local and σ local It is the local mean and standard deviation calculated based on the data collected by the current node.

[0127] As a preference, advanced security measures are deployed at edge computing nodes, including data encryption, access control, and intrusion detection systems to ensure data security during collection, processing, and transmission. TLS / SSL encryption is used for data transmission, role-based access control (RBAC) policies are implemented to manage data access rights, and AI-based intrusion detection systems are deployed to automatically identify and respond to potential security threats.

[0128] Among them, the preprocessing includes the following steps:

[0129] Convert data collected by different sensors into a unified standard unit, such as converting temperature from Fahrenheit to Celsius and pressure from psi to Pascal. i The calculation formula is:

[0130]

[0131] where x i is the original data, μ i and σ i are the mean and standard deviation of the data respectively;

[0132] Perform time series analysis on the data to identify periodic changes and trends in the data and use the autoregressive moving average model (ARMA) for modeling:

[0133]

[0134] Where X t is the data point at time t, φ and θ are model parameters, ∈ t It's white noise.

[0135] Extract key features such as maximum, minimum, mean, and standard deviation from each sensor’s data stream. These features provide a concise summary of the data for each time frame and facilitate subsequent fault analysis:

[0136]

[0137] Combine relevant data from different sensors (such as temperature and humidity) to form comprehensive features to provide a more complete representation of the system status. The calculation of the comprehensive feature F can be expressed as a weighted sum:

[0138]

[0139] Among them, z k is the standardized single sensor data, w k is the corresponding weight, reflecting the importance of the sensor data in the comprehensive features.

[0140] S3. Build a digital twin model based on the central air-conditioning unit based on the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time.

[0141] Specifically, we first define the main components of the central air conditioning system, such as compressors, cooling towers, fans, etc., and represent these components in the form of nodes and edges to build the topology of the system. Set physical parameters for each component, such as mass, volume, heat capacity, and resistance. These parameters will be used in subsequent dynamic simulations.

[0142] Among them, a state equation is established for each component in the system to describe its dynamic behavior. For example, for a compressor, the following energy conservation equation can be used to represent its thermodynamic process:

[0143]

[0144] Among them, U represents internal energy, Q represents heat exchange, and W represents power output.

[0145] Use transfer functions to model interactions between components, for example:

[0146]

[0147] Where s is the Laplace transform variable, m, b, k represent the mass, damping and stiffness coefficients respectively, and Y(s) and U(s) are the Laplace changes of the output and input respectively.

[0148] Furthermore, the data from each sensor (preprocessed in S2) is integrated into the model to update the state of the model in real time. i , Update the parameters of the corresponding model components:

[0149] θ new =θ old +ε(z i -θ old ) (10)

[0150] Among them, θ represents the model parameters and ε is the learning rate. Ensure that all data is synchronized in time and use system timestamps to ensure data consistency.

[0151] Preferably, historical data and real-time data are run on the model to perform simulations and observe the consistency between the model's predictions and the actual system operation. Model parameters are adjusted based on simulation results to optimize the model's accuracy and response speed.

[0152] S4. Design a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults, including:

[0153] (1) Select a quantum algorithm suitable for fault prediction. Here, the quantum support vector machine (QSVM) is used to build a fault prediction model because it can effectively handle large-scale feature spaces and classification problems. Formula:

[0154]

[0155] Where α is the Lagrange multiplier vector, Q is the matrix calculated by the training data and the quantum kernel function, and e is the all-1 vector.

[0156] (2) Encode the preprocessed data onto the quantum bit. Amplitude encoding is used because it can efficiently map classical data onto the amplitude of the quantum state. The calculation is as follows:

[0157]

[0158] where x i is the component of the normalized data vector, and |i> is the basis vector of the quantum state.

[0159] (3) Run the quantum optimization algorithm to train the fault prediction model and find the best hyperplane to distinguish between normal operation status and potential fault status, including:

[0160] A. Initialize quantum states and quantum circuits;

[0161] B. Apply quantum gates to simulate the training process of QSVM;

[0162] C. Perform quantum measurements to obtain model parameters.

[0163] S5. Deploy the fault prediction model on the edge computing node and run it for real-time fault monitoring and diagnosis. When a fault is found, AI-driven maintenance decision support is provided while responding to the fault, automatically adjusting the fault prediction model parameters or arranging maintenance strategies.

[0164] In a preferred embodiment, real-time fault monitoring and diagnosis are as follows:

[0165] (1) Use a message queue management system (such as Kafka or RabbitMQ) to process real-time data streams. Data is uploaded at a fixed frequency and distributed to processing modules through the message system.

[0166] (2) The fault detection threshold is dynamically adjusted according to the system's operating status and environmental conditions, and is calculated as follows:

[0167] θ t+1 =θ t +β(x t -θ t ) (14)

[0168] Among them, θ t is the threshold at time t, x t is the latest observation data, and β is the adjustment coefficient used to smoothly update the threshold.

[0169] (3) Based on the real-time updated model and dynamic thresholds, the system status is continuously monitored to detect anomalies. The quantum computing optimized fault prediction model developed will evaluate the deviation of the current data from the normal operating mode, and the calculation is as follows:

[0170]

[0171] Where f(x) is the prediction function, x is the current input data, and α i ,y i ,x i and b are parameters obtained from the QSVM model.

[0172] (4) Define the mapping relationship between faults and their possible causes. When the system detects an anomaly, it automatically generates a detailed fault report, including possible causes and recommended response measures. Use a secure API or messaging system to send the fault report to relevant personnel or systems in real time.

[0173] In a preferred embodiment, AI-driven maintenance decision support, while performing fault response, automatically adjusts fault prediction model parameters or arranges maintenance strategies, including:

[0174] (1) Develop a rule engine that contains various failure scenarios and corresponding action guidelines. Each rule is in the form of if (condition) then (action), where condition describes a specific failure symptom and action specifies the corresponding maintenance action.

[0175] (2) Create a detailed fault database for each possible fault in the system. The fault database contains the fault code, possible cause, impact, historical maintenance data, and repair steps. The data structure is as follows:

[0176] {Fault_Code:"F001",Symptoms:["hightemp","lowpressure"],Causes:["leak","blockage"],Actions:["inspecthoses","cleanfilter"]}

[0177] (3) Automatically trigger corresponding maintenance measures when specific faults are detected, including:

[0178] Assume R(t) represents the response action at time t, and S(t) represents the corresponding fault symptom. The selection of the response action is based on the matching calculation:

[0179] R(t)=argmax R Match(S(t),R) (16)

[0180] Among them, the Match function evaluates the matching degree between the fault symptoms and each response strategy.

[0181] (4) Record the results of each maintenance and use them to improve future strategies. After each maintenance activity, update the fault database and rule engine, including new fault data and maintenance effects. Use machine learning technology to analyze maintenance logs and continuously optimize response strategies.

[0182] S6. Build a learning loop based on performance feedback to optimize fault prediction models and maintenance strategies.

[0183] Specifically, the update steps are as follows:

[0184] (1) Continuously collect system operation data and the results of its response maintenance activities. Set the data collection focus, including fault detection data, response time, maintenance effect and user feedback. Perform statistical analysis on the collected data to identify patterns and trends.

[0185] Preferably, statistical control charts are used to track changes in key performance indicators (KPIs), using the formula:

[0186]

[0187] in, is the mean of the indicator, σ is the standard deviation, and z is the z-score corresponding to the desired confidence level.

[0188] (2) Adjust the fault detection and prediction model based on the analysis results. Update the model parameters based on the fault data, such as updating the α i and b. Adjust the fault response strategy and maintenance plan, and optimize the threshold θ based on the analysis results t And response strategy. Optimize the operation and maintenance strategy by minimizing the delay and cost of maintenance response, and apply the optimization equation:

[0189]

[0190] Among them, c t is the cost, d t is the delay and λ is the weight factor.

[0191] (3) Integrate newly acquired knowledge and experience into the fault database and rule engine. Update the fault database and maintenance strategy based on the latest maintenance experience and failure data. Update the knowledge base in real time to reflect new failure modes and maintenance strategies. Apply machine learning algorithms to regularly retrain the prediction model. Implement periodic system reviews and technical assessments to ensure that the system is consistent with the latest technical standards and operating practices.

[0192] In one or more embodiments, Figure 2 As shown, a real-time fault prediction system for a central air-conditioning system based on digital twin is disclosed, and the system includes:

[0193] The data acquisition device 101 is used to deploy various types of sensors around the central air conditioning unit to perform multi-dimensional detection, collect sensor data, and perform preliminary processing on the data;

[0194] The edge computing device 102 is used to deploy edge computing nodes around the central air conditioning unit to pre-process the data after preliminary processing;

[0195] The fault prediction device 103 is used to build a digital twin model based on the central air-conditioning unit according to the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time; design a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults; deploy the fault prediction model on the edge computing node and run it to perform real-time fault monitoring and diagnosis; when a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, and fault prediction model parameters are automatically adjusted or maintenance strategies are arranged;

[0196] System update device 104, used to build a learning loop based on performance feedback to optimize the fault prediction model and maintenance strategy;

[0197] The design is based on a fault prediction model of quantum computing and machine learning algorithms to predict and identify faults, including:

[0198] The quantum support vector machine is selected to build the fault prediction model, which is expressed as:

[0199]

[0200] Among them, α represents the Lagrange multiplier vector, Q represents the matrix calculated by the training data and the quantum kernel function, and e represents the all-1 vector;

[0201] The preprocessed data is encoded onto the qubit using amplitude coding, expressed as:

[0202]

[0203] Among them, x i represents the components of the normalized data vector, |i> represents the basis vector of the quantum state;

[0204] Construct a quantum kernel function to calculate the similarity between data points on a quantum device, expressed as:

[0205] K(x i ,x j )=|<ψ(x i )∣ψ(x j )>| 2

[0206] Among them, |ψ(x i )> and |ψ(x j )> respectively represent the quantum states after encoding the two data points, K(x i ,x j ) represents the value of quantum kernel function;

[0207] A quantum optimization algorithm is run to train the fault prediction model and find the best hyperplane to distinguish between normal operating states and potential fault states.

[0208] It is worth noting that the specific workflow of the real-time fault prediction system for central air-conditioning system based on digital twin provided in the embodiment of the present invention is the same as the workflow of the real-time fault prediction method for central air-conditioning system based on digital twin described in the above embodiment, and will not be repeated here.

[0209] Compared with the prior art, the real-time fault prediction system for central air-conditioning system based on digital twin provided by the embodiment of the present invention deploys various types of sensors around the central air-conditioning unit for multi-dimensional detection, collects sensor data, and performs preliminary processing on the data; deploys edge computing nodes around the central air-conditioning unit to pre-process the data after preliminary processing; builds a digital twin model based on the central air-conditioning unit according to the pre-processed data, provides a dynamic view of key performance indicators, and maps and simulates the status of the central air-conditioning unit in real time; designs a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults; deploys the fault prediction model on the edge computing node and runs it to perform real-time fault monitoring and diagnosis; when a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, and fault prediction model parameters are automatically adjusted or maintenance strategies are arranged; a learning loop based on performance feedback is constructed to optimize fault prediction models and maintenance strategies;

[0210] The embodiment of the present invention also provides a real-time fault prediction device for a central air-conditioning system based on digital twins, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the above-mentioned embodiment of the real-time fault prediction method for a central air-conditioning system based on digital twins are implemented, for example Figure 1 or, the processor implements the functions of each module in the above-mentioned system embodiments when executing the computer program.

[0211] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the real-time fault prediction device for the central air-conditioning system based on digital twins.

[0212] The central air-conditioning system real-time fault prediction device based on digital twins can be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The central air-conditioning system real-time fault prediction device based on digital twins can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the central air-conditioning system real-time fault prediction device based on digital twins can also include input and output devices, network access devices, buses, etc.

[0213] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASAC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the real-time fault prediction device of the central air-conditioning system based on digital twins, and uses various interfaces and lines to connect various parts of the operation fault monitoring device of the entire assembled central air-conditioning unit.

[0214] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the real-time fault prediction device of the central air-conditioning system based on digital twin by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc.; the data storage area can store data created according to the operation of the air-conditioning controller, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedaaCard, SMC), a secure digital (SecureDagatal, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0215] Wherein, if the module integrated by the real-time fault prediction device of the central air-conditioning system based on digital twin is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0216] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0217] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A real-time fault prediction method for central air-conditioning system based on digital twin, characterized in that: The following steps are involved: Deploy various types of sensors around the central air conditioning units to conduct multi-dimensional detection, collect sensor data, and perform preliminary processing on the data; Deploy edge computing nodes around central air conditioning units to pre-process the data after preliminary processing; Build a digital twin model based on the central air-conditioning unit based on the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time; Design fault prediction models based on quantum computing and machine learning algorithms to predict and identify faults; Deploy the fault prediction model on the edge computing node and run it to perform real-time fault monitoring and diagnosis. When a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, automatically adjusting fault prediction model parameters or arranging maintenance strategies. Build a learning loop based on performance feedback to optimize failure prediction models and maintenance strategies.

2. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The design is based on a fault prediction model of quantum computing and machine learning algorithms to predict and identify faults, including: The quantum support vector machine is selected to build the fault prediction model, which is expressed as: Among them, α represents the Lagrange multiplier vector, Q represents the matrix calculated by the training data and the quantum kernel function, e represents the all-1 vector, and T represents the transpose; The preprocessed data is encoded onto the qubit using amplitude coding, expressed as: Among them, x i represents the components of the normalized data vector, |i> represents the basis vector of the quantum state, and N represents the total number of basis states; Construct a quantum kernel function to calculate the similarity between data points on a quantum device, expressed as: K(x i ,x j )=|<ψ(x i )∣ψ(x j )>| 2 Among them, |ψ(x i )> and |ψ(x j )> respectively represent the quantum states after encoding the two data points, K(x i ,x j ) represents the value of the quantum kernel function, x i and x j Represent two data points respectively; Run a quantum optimization algorithm to train the fault prediction model and find the best hyperplane to distinguish between normal operating states and potential fault states; The quantum optimization algorithm includes: Initialize quantum states and quantum circuits; Apply quantum gates to simulate the training process of fault prediction models; Perform quantum measurements to obtain model parameters.

3. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The sensors include temperature sensors, pressure sensors, humidity sensors, meteor sensors and vibration sensors.

4. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The preliminary processing includes performing low-pass filtering on the data collected by the sensor to eliminate high-frequency noise, including: Let x(t) be the original signal, and the filtered signal y(t) be expressed as follows: y(t)=θx(t)+(1-α)·y(t-1) Wherein, θ is the filter coefficient, and its value range is 0<α<1; t represents time.

5. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The edge computing nodes are deployed around the central air conditioning units to pre-process the data after preliminary processing, including: Dynamically adjust edge computing deployment based on the actual operating load and data generation of the central air-conditioning unit system; Deploy network connections with failover and multi-path transmission capabilities to ensure automatic switching to a backup connection when one connection fails; where the network connection is deployed using multi-protocol label switching or software-defined networking; Implement data preprocessing functions at each edge node and deploy security measures; The data preprocessing function implemented at each edge node includes: Convert the data collected by different sensors into a unified standard unit, then perform time series analysis on the data, identify periodic changes and trends in the data, and use the autoregressive moving average model for modeling; Extract key features from the data stream of each sensor and combine relevant data from different sensors to form comprehensive features.

6. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 5, characterized in that: The digital twin model based on the central air-conditioning unit is constructed according to the preprocessed data, including: Define the components of the central air conditioning unit, set the physical parameters for each component, establish the state equation for each component, describe its dynamic behavior and use transfer functions to simulate the interaction between components; The preprocessed sensor data is integrated into the digital twin model, and the status of the digital twin model is updated in real time, as shown below: i new =θ old +ε(z i -θ old ) Among them, θ new represents the new model parameters, ε is the learning rate, θ old represents the old model parameters, z i represents the preprocessed data points; Run historical data and real-time data on the digital twin model to perform simulations, calculate the difference between the model's predictions and the actual system operation, and dynamically adjust the parameters of the digital twin model based on the differences.

7. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The fault prediction model is deployed on the edge computing node and runs, including: Use the message queue management system to manage the data received by edge computing nodes in real time; The fault detection threshold is dynamically adjusted according to the operating status and environmental conditions of the central air-conditioning unit, as shown below: i t+1 =θ t +β(x t -θ t ) Among them, θ tt1 represents the threshold at time t+1, θ t represents the threshold at time t, x t represents the latest observation data, and β represents the adjustment coefficient, which is used to smoothly update the threshold; Based on the real-time updated fault prediction model and dynamic thresholds, the system status is continuously monitored to detect anomalies; where detection is represented as follows: Among them, f(x) represents the prediction function, x represents the current input data, α i ,y i ,x i and b represent the Lagrange multiplier, the label of the training sample and the bias term; When a potential fault is detected, diagnostic information is generated and sent to the maintenance team.

8. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 7, characterized in that: The AI-driven maintenance decision support, while performing fault response, includes: Design a rules engine that contains various failure scenarios and corresponding action guidelines; Build a fault database and corresponding strategies, and automatically trigger corresponding maintenance measures when a fault is detected; The results of each maintenance are recorded as a maintenance log, and machine learning technology is used to analyze the maintenance log for the improvement of future strategies.

9. The real-time fault prediction method for a central air-conditioning system based on digital twins according to claim 1, characterized in that: The described construction of a learning loop based on performance feedback to optimize the fault prediction model and maintenance strategy includes: Continuously collect the operating data of central air-conditioning units, conduct statistical analysis on the operating data, and use statistical control charts to track changes in operating data; Update the parameters of the digital twin model and fault prediction model based on the results of statistical analysis, and dynamically adjust the fault response strategy and maintenance plan based on minimizing the delay and cost of maintenance response; Integrate newly acquired knowledge and experience into the fault database and rule engine, and update the fault database and maintenance strategy; Apply machine learning algorithms to periodically retrain the fault prediction model.

10. The real-time fault prediction system of central air-conditioning system based on digital twin is characterized by: The system comprises: Data acquisition device, used to deploy various types of sensors around the central air conditioning unit to perform multi-dimensional detection, collect sensor data, and perform preliminary processing on the data; Edge computing devices are used to deploy edge computing nodes around central air conditioning units to pre-process the data after preliminary processing; Fault prediction device, used to build a digital twin model based on the central air-conditioning unit according to the preprocessed data, provide a dynamic view of key performance indicators, and map and simulate the status of the central air-conditioning unit in real time; design a fault prediction model based on quantum computing and machine learning algorithms to predict and identify faults; deploy and run the fault prediction model on the edge computing node to perform real-time fault monitoring and diagnosis; when a fault is found, AI-driven maintenance decision support is provided, and fault response is performed at the same time, automatically adjusting fault prediction model parameters or arranging maintenance strategies; System update device for building a learning loop based on performance feedback to optimize fault prediction models and maintenance strategies; The design is based on a fault prediction model of quantum computing and machine learning algorithms to predict and identify faults, including: The quantum support vector machine is selected to build the fault prediction model, which is expressed as: Among them, α represents the Lagrange multiplier vector, Q represents the matrix calculated by the training data and the quantum kernel function, and e represents the all-1 vector; The preprocessed data is encoded onto the qubit using amplitude coding, expressed as: Among them, x i represents the components of the normalized data vector, |i> represents the basis vector of the quantum state; Construct a quantum kernel function to calculate the similarity between data points on a quantum device, expressed as: K(x i ,x j )=|<ψ(x i )∣ψ(x j )>| 2 Among them, |ψ(x i )> and |ψ(x j )> respectively represent the quantum states after encoding the two data points, K(x i ,x j ) represents the value of quantum kernel function; A quantum optimization algorithm is run to train the fault prediction model and find the best hyperplane to distinguish between normal operating states and potential fault states.

Images