Method for building a digital twin system for elevator fault diagnosis

Abstract

A method for building a digital twin system for elevator fault diagnosis belongs to the field of digital twin and elevator fault diagnosis. The method comprises the following steps: the first step is to build a four-dimensional model framework for elevator digital twin fault diagnosis; the second step is to collect elevator data; the third step is to build a virtual elevator model; the fourth step is to build a numerical model of the vertical direction of the elevator; the fifth step is to improve the PINNs model, introduce parameters as physical information regulators in the loss function, and introduce a reactivation algorithm to optimize the loss function; the sixth step is to obtain simulated vibration data, input a small amount of collected elevator fault vibration signals into the PINNs model with motion equation constraints, and obtain simulated elevator vibration data; the seventh step is to improve the GCN model for elevator fault diagnosis; the eighth step is to establish an elevator fault diagnosis digital twin platform. The present invention integrates elevator fault diagnosis and digital twin technology to effectively improve the accuracy of fault diagnosis.

Description

Technical Field

[0001] The present invention belongs to the field of digital twin and elevator fault diagnosis, and specifically relates to a method for building a digital twin system for elevator fault diagnosis. Background Art

[0002] Elevators are a convenient and indispensable means of transportation in high-rise buildings. Studies have shown that 54% of the world's population currently lives in urban areas. With the rapid development of urbanization, the number of elevators is also growing. Faced with the severe situation of elevator safety management, various places have built comprehensive elevator emergency response platforms, deployed telephone hotlines or call systems, and configured professional response teams, providing important support for the rapid response and proper handling of elevator safety accidents. However, with the further increase in the number of elevators and the arrival of aging equipment, the phenomenon of abnormal vibration of elevators has become more and more common. The contradiction between the number of responders and the number of elevators, and the inability of accumulated professional experience to meet the needs of the surge in elevator abnormalities have become prominent. Therefore, it is urgent to carry out research and application of new intelligent processing technologies for elevators. However, most of the current technical work has certain obstacles, mainly because real elevators are rarely connected to virtual models, which leads to the inability to effectively utilize the large amount of operating data generated by real elevators. At present, with the development of cyber-physical systems (CPS), it is crucial to pay attention to virtual space and realize the seamless integration of physical space and virtual space to improve elevator fault diagnosis. In this context, it is important to introduce digital models of devices and their data to accurately depict the behavior of real entities in virtual space.

[0003] With the rapid development of technologies such as the Internet, artificial intelligence, and digital twins, these technologies have gradually become valuable opportunities in elevator fault diagnosis. Digital twin technology can realize the digital expression of real scenes, and through the perception analysis and model construction of scene elements, it can realize the two-way real-time mapping between physical entities and virtual models. With the support of simulation verification methods, it helps to deduce the trend of real scene changes, and also shows great application advantages in remote monitoring, state monitoring and control optimization of special equipment. In order to solve the problems of inaccurate characteristics and slow convergence speed of traditional elevator braking performance prediction models, some people have proposed elevator braking performance prediction and evaluation based on digital twins. By deploying sensors to monitor the multi-dimensional data of the elevator itself, and based on the intelligent gateway to collect real-time dynamic data of the elevator, a high-fidelity model of the elevator is built to realize the two-way mapping and dynamic interaction between the elevator physical entity and the twin. Some people have proposed a digital twin reference model for rotating machinery fault diagnosis, and discussed the construction requirements of the digital twin model, and proposed a model correction scheme based on parameter sensitivity analysis to enhance the adaptability of the digital twin model.

[0004] Deep learning is a novel intelligent method that can autonomously extract effective features from raw signals without relying on manual feature extraction. At present, deep learning technology has been widely used in speech classification, language translation, image classification and other fields, and has achieved great success. Deep learning has also made great progress in the field of elevator fault diagnosis. Such technology can diagnose the operating health of mechanical equipment based on raw monitoring data. In view of the sample imbalance and hyperparameter tuning problems of the XGBoost model, a method for elevator fault diagnosis based on recursive feature elimination (RFE)-SMOTE-Tomek and IAO-XGBoost was proposed, which can better predict potential faults and take preventive maintenance measures, thereby extending the service life of the elevator and improving its reliability; ZHAO M et al. proposed and developed a deep residual contraction network model that borrowed the design idea of ​​the residual block in the deep learning model Res Net to improve the learning ability of the high noise vibration signal characteristics of the elevator, and used the data for fault diagnosis; Mishra et al. applied deep learning to the prediction of elevator faults based on the time series data of elevator operation, and used neural networks to extract features from the time series vibration data of the elevator for corresponding research.

[0005] However, the elevator diagnosis model requires a large amount of labeled fault data that contains all possible fault conditions. In actual operation, the collection and recording of elevator fault data is often limited by time, cost and feasibility. Therefore, the acquisition of fault data is limited by many factors, making it impossible for many diagnostic methods to be fully trained in the absence of data, resulting in performance degradation. Summary of the invention

[0006] In order to overcome the shortcomings of the existing technology, the present invention provides a method for building a digital twin system for elevator fault diagnosis that improves the accuracy of elevator fault diagnosis. By establishing an elevator digital twin model that is highly consistent with the physical entity and combining it with a deep learning algorithm, elevator fault diagnosis is performed. Ultimately, an effective method for solving the problem of elevator fault diagnosis is proposed, which has good practical value.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] A method for building a digital twin system for elevator fault diagnosis includes the following steps:

[0009] The first step is to build a four-dimensional model framework for elevator digital twin fault diagnosis. The four-dimensional model of the elevator digital twin fault diagnosis system consists of physical space, virtual space, interactive space and service space. The model framework construction formula is as follows:

[0010] ;

[0011] Among them, PS represents physical space, VS represents virtual space, IS represents interactive space, and SS represents service space;

[0012] The second step is to collect elevator data, including the vibration signal of the elevator, the temperature inside the host, the elevator operation status, the air pressure inside the elevator, whether the elevator is level, the door opening and closing status of the elevator, as well as real-time monitoring of the elevator, remote intercom and video recording functions, and detection of whether there are people, and remote data acquisition of various elevator data;

[0013] The third step is to build a virtual elevator model. By measuring the size data of the elevator, use 3Dmax to build a virtual elevator model that is similar to the physical elevator.

[0014] Step 4: Construct a numerical model of the elevator in the vertical direction;

[0015] Step 5: Improve the Physics-Informed Neural Networks (PINNs) model and introduce it into the loss function As a physical information regulator, the reactivation algorithm is introduced to optimize the loss function. The process is as follows:

[0016] Step (5.1) designs the loss function of the PINNs model into two parts, one of which is the loss of the neural network. , which is used to represent the error between the predicted value and the true value based on the network, and the other part is the loss of physical information , the loss of physical information includes the initial value loss , Boundary loss and the residual of the ordinary differential equation ;

[0017] Step (5.2) Design a reactivation algorithm to obtain the optimal solution of the model;

[0018] Step 6: Acquire simulated vibration data. Input a small amount of elevator fault vibration signals collected in the first step into the PINNs model with motion equation constraints to obtain simulated elevator vibration data.

[0019] Step 7: Improve the Graph Convolutional Networks (GCN) model for elevator fault diagnosis. The process is as follows: Combine the Recursive Neural Network (RNN) and the Graph Convolutional Network (GCN) to propose the e-RGCN (Epsilon Recurrent Network-Graph Convolutional Network) fault diagnosis algorithm, set a recursive radius, and construct a correlation graph through the recursive radius. Then, GCN can effectively identify the elevator fault status through the constructed correlation graph.

[0020] Step 8. Establish a digital twin platform for elevator fault diagnosis.

[0021] Preferably, the process of the eighth step is: the collected operating data of the real-world elevator is sent to the database through the network, Unity3D then reads the operating data from the database and displays it on the visualization interface, the status monitoring part is connected to the mobile device to realize real-time monitoring of the status of the real elevator; the user can observe the operating status of the real elevator while sitting in front of the designed system; the equipment fault warning interface will display the elevator operating status in real time, and will alarm when an abnormality is detected in the elevator; the elevator fault is diagnosed through the collected real-time signal data, and the equipment fault analysis interface will analyze the data of the abnormal time period and diagnose the specific fault type.

[0022] Further, the process of the first step is as follows:

[0023] Step (1.1) PS is the basis of elevator EDTM. It is a collection of physical entities involved in the operation of the elevator digital twin fault diagnosis system. It is responsible for collecting measurement activities to provide the physical size data and physical operation data of the elevator. The formula is as follows:

[0024] ;

[0025] in, Represents physical entities in the physical space, including various parts of the elevator and various sensors.

[0026] Step (1.2) VS is a digital depiction and description of PS, which aims to be highly similar to PS and provide visualization and interactive functions for the service. The formula is as follows:

[0027] ;

[0028] in, Represents virtual entities in virtual space, and One-to-one correspondence, in EDTM’s VS, VE is a dynamically running geometric model, which is continuously updated as real-time twin data is generated, thereby achieving synchronization with the physical elevator in PS;

[0029] Step (1.3) IS is a bridge between PS, VS and SS, used to realize data exchange among the three. The formula is as follows:

[0030] ;

[0031] Among them, TD stands for twin data and PL stands for platform. TD is the core driver for the operation of the elevator digital twin geometric model;

[0032] Step (1.4) SS provides elevator fault diagnosis technology to establish a digital twin geometric model for fault diagnosis during elevator operation. The formula is as follows:

[0033] ;

[0034] in, Indicates elevator operation data monitoring. Indicates elevator fault warning report. Indicates elevator fault diagnosis. Indicates the generation of elevator simulation fault data.

[0035] Preferably, the process of step (1.3) is:

[0036] Step (1.3.1) TD is the core driver of the elevator digital twin geometric model operation. The formula is as follows:

[0037] ;

[0038] in, Represents virtual entity data, which is the geometric data related to the digital model, including the geometric dimension data involved in three-dimensional modeling and the vibration data obtained by measurement; Represents physical entity data, including rated data and geometric information data of each elevator component; Relevant data representing the service space, including computer-aided design (CAD) model data, SolidWorks model data, elevator vibration simulation data obtained by algorithms, and elevator fault diagnosis result data obtained by neural networks;

[0039] Steps (1.3.2) It provides tools and third-party libraries for supporting services, which are used for modeling, visualization, interaction, measurement and measurement data processing. The PS modeling is realized by using SolidWorks software. The geometric modeling of the virtual entity of the elevator is realized by 3DMax. The visualization and interaction of the three-dimensional data and models are realized by combining Python with Unity3D. The application visualization, namely the Graphical User Interface (GUI) design is realized by Unity3Dl. The accelerometer for vibration measurement is connected to the development software to perform the measurement and control functions. The measurement data processing is realized based on the algorithm provided by Python.

[0040] In the second step, the vibration signal of the elevator is collected at a set frequency by the acceleration sensor installed in the car top box on the top of the elevator car; the temperature inside the host is collected by the temperature sensor installed in the host on the top of the machine; the elevator stop, up and down operation and other operating states are collected by the leveling / base sensor installed on the top of the car; the air pressure in the elevator is collected by the atmospheric pressure sensor installed in the host on the top of the car; whether the elevator is level is collected by the photoelectric sensor installed on the top of the car; the door opening and closing state of the elevator is collected by the door magnetic sensor installed on the top of the car; the real-time monitoring, remote intercom and video recording functions of the elevator are collected by the monitoring camera installed inside the car; the infrared sensor installed on the elevator door is used to detect whether someone passes through the elevator; various data are remotely acquired through the TCP / IP communication protocol, the sensor, elevator and gateway are connected by serial port connection, and Modbus and RS485 protocols are used for data communication.

[0041] The process of the fourth step is as follows:

[0042] Step (4.1) establishes the dynamic equation of the system, the formula is as follows,

[0043] ;

[0044] Where T is the kinetic energy of the system; U is the potential energy of the system; D is the dissipation coefficient of the system; x is the generalized displacement (or angular displacement); is the generalized velocity (or angular velocity); Q is the generalized force (exciting force or exciting torque);

[0045] Step (4.2) calculates the system's kinetic energy, the system's potential energy and the system's dissipation coefficient,

[0046] ;

[0047] ;

[0048] ;

[0049] in, is the equivalent mass of the car and the loaded car; is the counterweight mass; , are the equivalent stiffness and damping of the wire rope between the car and the traction sheave respectively; , are the equivalent stiffness and damping of the wire rope between the counterweight and the traction sheave respectively; , and They are the car vibration linear displacement, counterweight vibration linear displacement, and traction sheave vibration angular displacement; , are the moment of inertia of the traction sheave and the radius of the rope groove of the traction sheave respectively;

[0050] Step (4.3) Arrange the formulas in step (4.2) and step (4.1) as follows:

[0051] Step (4.3.1) for generalized coordinates ,have,

[0052] ;

[0053] Step (4.3.2) Since the elevator car is only affected by the vertical wire rope tension and gravity, substitute the result of step (4.3.1) into the Lagrange equation of step (4.1) to obtain,

[0054] ;

[0055] Step (4.3.3) for generalized coordinates ,have,

[0056] ;

[0057] Step (4.3.4) The elevator counterweight is only affected by the vertical wire rope tension and gravity. Substituting the result of step (4.3.3) into the Lagrange equation of step (4.1), we get:

[0058] ;

[0059] Step (4.3.5) for generalized coordinates ,have,

[0060] ;

[0061] Step (4.3.6) for generalized coordinates , the system inertia force in the elevator start-brake stage is used as the excitation source, and the exciting force is expressed as,

[0062] ;

[0063] Step (4.4) Among them, step (4.3.2), step (4.3.2) and step (4.3.2) constitute the differential equation of motion, which can be expressed in matrix form as follows,

[0064] ;

[0065] in,

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] Among them, M is the system mass matrix, C is the system damping matrix, K is the system stiffness matrix, and F is the exciting force matrix; , , They represent the acceleration, velocity and displacement of the system respectively.

[0072] Furthermore, the process of step (5.1) is as follows:

[0073] Step (5.1.1) The loss of the neural network is as follows,

[0074] ;

[0075] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the predicted value of the neural network, Indicates the actual measured value;

[0076] Step (5.1.2) A core idea of ​​PINNs is to make the neural network satisfy the given physical laws. The physical laws are expressed in the form of ODE. The ODE residual loss is used to measure whether the solution of the network satisfies these equations. The residual of the ordinary differential equation is as follows,

[0077] ;

[0078] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the actual measured value, for any time;

[0079] Step (5.1.3) The boundary conditions define the behavior of the solution at the boundary of the problem region, and the initial conditions define the state of the system at time t = 0. The boundary conditions and initial conditions of PINNs are obtained through actual measurement and estimation as follows:

[0080] ;

[0081] Step (5.1.4) The boundary loss function compares the boundary conditions predicted by the model with the actual physical boundary conditions, so that the solution of the model in the boundary area conforms to the physical laws. The boundary loss is as follows:

[0082] ;

[0083] in, is the number of sampled data, The operator representing the mean square error MSE, represents the predicted value of the neural network, Indicates the actual measured value;

[0084] Step (5.1.5) The initial value loss function measures the difference between the model prediction value and the actual initial conditions to ensure that the network output meets the physical constraints at the initial moment. The initial value loss is as follows:

[0085] ;

[0086] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the predicted value of the neural network, Indicates the actual measured value;

[0087] Step (5.1.6) adds the losses mentioned in steps (5.1.1) to (5.1.6) to get the final loss function of the PINNs model. The formula is as follows:

[0088] ;

[0089] in, It is a physical information regulator; by introducing λ, PINNs can accurately estimate the output of the equation. By introducing physical information into the loss function, λ plays an important role in determining the contribution of physics to the final loss function. As λ increases, the output of PINNs will gradually rely on the constraints of the physical equation, while reducing λ will make the actual measured data more influential on the output; if the value of λ is set to 0, the entire loss function does not contain the physical information term and becomes a traditional ANN, then the network only relies on the data during training.

[0090] In the step (5.2), using PINNs to solve the differential equation is an optimization problem of minimizing the loss function. For the vibration dynamics equation of the elevator, when using an optimization algorithm with line search and convergence criteria, a local optimal solution can be predicted. The local optimal solution can be regarded as the exact solution obtained. However, sometimes the local optimal solution obtained may have a large deviation. In order to avoid the situation where the solution has a large deviation, a reactivation technology is proposed to reactivate the optimization algorithm so that it can get the optimal solution again; the process is as follows:

[0091] Step (5.2.1) PINNs uses the Adma optimizer with strong Wolfe. In the kth iteration, a step size αk is found that satisfies the following conditions:

[0092] Function value decreasing condition,

[0093] ;

[0094] Strong curvature condition,

[0095] ;

[0096] in, is the search direction, and is a constant of the strong Wolfe condition, 0< < <1, is the current point, is the gradient of the current point. If the optimization falls into a local optimal solution at the beginning of the kth iteration, there is no αk that satisfies the equation. If no αk is found to satisfy the equation, the optimization is stopped.

[0097] Step (5.2.2) In order to restart the optimization process, introduce The constant makes the amplification factor slightly greater than zero, giving the following formula,

[0098] ;

[0099] The above formula is used to replace the function reduction condition. The strong curvature condition is only used in the first iteration of reactivation. The introduction of the above formula allows Greater than , making it easier to find αk that satisfies the equation; once found, the reactivation optimization algorithm will continue to use the standard strong Wolfe condition, that is, step (5.2.1).

[0100] The process of the seventh step is as follows:

[0101] Step (7.1) constructs the association graph. e-RNN is one of the methods for complex network analysis of time series. Recursive network is an intuitive method that uses the recursiveness of mechanical systems to convert time series into graphs. Unlike traditional time series analysis that focuses more on the sequence of samples, recursive network analysis focuses more on the spatial topological structure of samples and the relationship between samples.

[0102] Step (7.2) Construction of the e-RGCN fault diagnosis network. The proposed e-RGCN classification network consists of two GCN layers, one fully connected layer and an activation function. The GCN layer extracts graph features from the association graph, and the fully connected layer is used as a classifier for elevator fault diagnosis.

[0103] Step (7.3) inputs the graph into the network, extracts features through two GCN layers, and obtains deep features; then classifies the deep features through a softmax classification layer, and outputs the fault type classification result;

[0104] Among them, the Softmax function is a commonly used activation function, which is often used in multi-category classification problems. The Softmax function can convert a real number vector into a vector representing a probability distribution, so that each element in the vector is between 0 and 1, and the sum of all elements is 1.

[0105] The processing process of step (7.1) is as follows:

[0106] Step (7.1.1) constructs a graph that requires a relationship between the central vertex and other vectors. Two vertices can only be There is an edge between two vertices only when is the Euclidean norm between two vectors, is the recursive radius, which is the global threshold of RNN and can adjust the criteria of recursive pairs; two vertices are recursive pairs only when the conditions are met; in addition, the topological structure of all samples is obtained by performing pairwise operations on all vectors, and the structure of the network is encoded into a recursive matrix expressed as follows,

[0107] ;

[0108] in, is the Heaviside function, expressed as follows,

[0109] ;

[0110] in, is the Euclidean norm of a vector, if and There is an edge between The value of is 1, otherwise The value of is 0;

[0111] Step (7.1.2) The adjacency matrix of the graph can be constructed from the recursive matrix as follows,

[0112] ;

[0113] in is the Kronecker function, expressed as,

[0114] ;

[0115] The graph of the vibration signal is obtained through the obtained adjacency matrix.

[0116] The processing process of step (7.2) is as follows:

[0117] Step (7.2.1) uses the cross-entropy loss function as the loss function of the network output. The cross-entropy loss function is a loss function in machine learning and deep learning. Especially in classification problems, it is used to measure the difference between the model's predicted value and the true label. It is particularly suitable for classification tasks (such as binary classification and multi-classification tasks). The loss function of the e-RGCN model is Indicates the degree of match between the predicted label and the actual label, as follows,

[0118] ;

[0119] Where N is the number of samples, C is the number of categories, is the probability of the jth category in the true label of the i-th sample, is the jth category in the model's predicted probability value for the i-th sample;

[0120] Step (7.2.2) During the e-RGCN training process, multi-layer GCN mixes the features of the node itself and its neighbors. Even if an edge is deleted, information about the node can still be obtained from its neighbors. Drop Edge randomly deletes some edges in the original graph, which increases the generalization of the network.

[0121] ;

[0122] Among them, A is the graph structure input of GCN, This is the graph structure after randomly selecting some edges.

[0123] The beneficial effects of the present invention are mainly manifested in:

[0124] 1. Integrate elevator fault diagnosis and digital twin technology, build an elevator digital twin system by building an elevator numerical model combined with a deep learning algorithm. This method can effectively obtain data that is difficult to obtain in reality, facilitate various subsequent studies, and provide effective materials for similar studies;

[0125] 2. The traditional PINNs model is improved. The present invention introduces λ as a physical information regulator in the entire loss function to adjust the proportion of physical information in the entire loss function to obtain more accurate prediction results;

[0126] 3. The e-RGCN fault diagnosis algorithm is proposed, and the ε constraint is introduced to enhance the robustness and fault tolerance of the algorithm. The recursive network is organically combined with the graph convolutional network, and the vibration data is processed graphically through the ε constraint. It can effectively improve the accuracy of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0127] Figure 1 It is the framework diagram of the four-dimensional digital twin model for elevator fault diagnosis.

[0128] Figure 2 This is a graph of fault diagnosis accuracy with and without the participation of digital twins.

[0129] Figure 3 This is a comparison chart of the accuracy of seven models for seven experiments on four labels. DETAILED DESCRIPTION

[0130] The present invention will be further described below in conjunction with the accompanying drawings.

[0131] Reference Figure 1~Figure 3 , a method for building a digital twin system for elevator fault diagnosis, comprising the following steps:

[0132] The first step is to build a four-dimensional model framework for elevator digital twin fault diagnosis. The elevator digital twin fault diagnosis system is essentially an integrated system that combines physical models, virtual models, interactive data, diagnostic algorithms, software platforms and other components. On this basis, a four-dimensional model of the elevator digital twin fault diagnosis system is proposed. Figure 1 As shown in the figure, the four-dimensional model of the elevator digital twin fault diagnosis system consists of physical space, virtual space, interactive space and service space. The model framework construction formula is as follows:

[0133] ;

[0134] Among them, PS represents physical space, VS represents virtual space, IS represents interactive space, and SS represents service space;

[0135] The process of the first step is as follows:

[0136] Step (1.1) PS is the basis of elevator EDTM, such as Figure 1 As shown in Figure 1, it is a collection of physical entities involved in the operation of the elevator digital twin fault diagnosis system, which is responsible for collecting measurement activities to provide the physical size data and physical operation data of the elevator. The formula is as follows:

[0137] ;

[0138] in, Represents physical entities in the physical space, including various parts of the elevator and various sensors;

[0139] Step (1.2) is as follows Figure 1 As shown, VS is a digital depiction and description of PS, which aims to be highly similar to PS and provide visualization and interactive functions for the service. The formula is as follows:

[0140] ;

[0141] in, Represents virtual entities in virtual space, and One-to-one correspondence, in EDTM’s VS, VE is a dynamically running geometric model, which is continuously updated as real-time twin data is generated, thereby achieving synchronization with the physical elevator in PS;

[0142] Step (1.3) is as follows Figure 1 As shown, IS is a bridge between PS, VS and SS, used to realize data exchange among the three. The formula is as follows:

[0143] ;

[0144] Among them, TD stands for twin data and PL stands for platform. TD is the core driver for the operation of the elevator digital twin geometric model;

[0145] The process of step (1.3) is:

[0146] Step (1.3.1) TD is the core driver of the elevator digital twin geometric model operation. The formula is as follows:

[0147] ;

[0148] in, Represents virtual entity data, which is the geometric data related to the digital model, including the geometric dimension data involved in three-dimensional modeling and the vibration data obtained by measurement; Represents physical entity data, including rated data and geometric information data of each elevator component; Relevant data representing the service space, including computer-aided design (CAD) model data, SolidWorks model data, elevator vibration simulation data obtained by algorithms, and elevator fault diagnosis result data obtained by neural networks;

[0149] Steps (1.3.2) It provides tools and third-party libraries for supporting services, which are used for modeling, visualization, interaction, measurement and measurement data processing. PS modeling is implemented using SolidWorks software. The geometric modeling of virtual elevator entities is implemented through 3DMax. The visualization and interaction of 3D data and models are implemented through Python combined with Unity3D. Application visualization, i.e. Graphical User Interface (GUI) design is implemented through Unity3Dl. The accelerometer for vibration measurement is connected to the development software to perform measurement and control functions. The measurement data processing is implemented based on the algorithm provided by Python.

[0150] Step (1.4) is as follows Figure 1 As shown, SS provides elevator fault diagnosis technology to establish a digital twin geometric model for fault diagnosis during elevator operation. The formula is as follows:

[0151] ;

[0152] in, Indicates elevator operation data monitoring. Indicates elevator fault warning report. Indicates elevator fault diagnosis. Indicates the generation of elevator simulation fault data.

[0153] The second step is to collect elevator data, including the vibration signal of the elevator, the temperature inside the host, the elevator operation status, the air pressure inside the elevator, whether the elevator is level, the door opening and closing status of the elevator, as well as real-time monitoring of the elevator, remote intercom and video recording functions, and detection of whether there are people, and remote data acquisition of various elevator data;

[0154] In the second step, if Figure 1As shown, the vibration signal of the elevator is collected by the acceleration sensor installed in the car top box on the top of the elevator car at a set frequency; the temperature inside the host is collected by the temperature sensor installed in the host on the top of the machine; the elevator stop, up and down operation and other operating states are collected by the leveling / base sensor installed on the top of the car; the air pressure in the elevator is collected by the atmospheric pressure sensor installed in the host on the top of the car; whether the elevator is level is collected by the photoelectric sensor installed on the top of the car; the door opening and closing state of the elevator is collected by the door magnetic sensor installed on the top of the car; the real-time monitoring, remote intercom and video recording functions of the elevator are collected by the monitoring camera installed inside the car; the infrared sensor installed on the elevator door is used to detect whether someone passes through the elevator; various data are remotely acquired through the TCP / IP communication protocol, the sensor, elevator and gateway are connected by serial port connection, and Modbus and RS485 protocols are used for data communication.

[0155] The third step is to build a virtual elevator model. By measuring the size data of the elevator, use 3Dmax to build a virtual elevator model that is similar to the physical elevator.

[0156] Step 4: Construct a numerical model of the elevator in the vertical direction;

[0157] The process of the fourth step is as follows:

[0158] Step (4.1) establishes the dynamic equation of the system, the formula is as follows,

[0159] ;

[0160] Where T is the kinetic energy of the system; U is the potential energy of the system; D is the dissipation coefficient of the system; x is the generalized displacement (or angular displacement); is the generalized velocity (or angular velocity); Q is the generalized force (exciting force or exciting torque);

[0161] Step (4.2) calculates the system's kinetic energy, the system's potential energy and the system's dissipation coefficient,

[0162] ;

[0163] ;

[0164] ;

[0165] in, is the equivalent mass of the car and the loaded car; is the counterweight mass; , are the equivalent stiffness and damping of the wire rope between the car and the traction sheave respectively; , are the equivalent stiffness and damping of the wire rope between the counterweight and the traction sheave respectively; , and They are the car vibration linear displacement, counterweight vibration linear displacement, and traction sheave vibration angular displacement; , are the moment of inertia of the traction sheave and the radius of the rope groove of the traction sheave respectively;

[0166] Step (4.3) Arrange the formulas in step (4.2) and step (4.1) as follows:

[0167] Step (4.3.1) for generalized coordinates ,have,

[0168] ;

[0169] Step (4.3.2) Since the elevator car is only affected by the vertical wire rope tension and gravity, substitute the result of step (4.3.1) into the Lagrange equation of step (4.1) to obtain,

[0170] ;

[0171] Step (4.3.3) for generalized coordinates ,have,

[0172] ;

[0173] Step (4.3.4) The elevator counterweight is only affected by the vertical wire rope tension and gravity. Substituting the result of step (4.3.3) into the Lagrange equation of step (4.1), we get:

[0174] ;

[0175] Step (4.3.5) for generalized coordinates ,have,

[0176] ;

[0177] Step (4.3.6) for generalized coordinates , the system inertia force in the elevator start-brake stage is used as the excitation source, and the exciting force is expressed as,

[0178] ;

[0179] Step (4.4) Among them, step (4.3.2), step (4.3.2) and step (4.3.2) constitute the differential equation of motion, which can be expressed in matrix form as follows,

[0180] ;

[0181] in,

[0182] ;

[0183] ;

[0184] ;

[0185] ;

[0186] ;

[0187] Among them, M is the system mass matrix, C is the system damping matrix, K is the system stiffness matrix, and F is the exciting force matrix; , , They represent the acceleration, velocity and displacement of the system respectively.

[0188] Step 5: Improve the PINNs model and introduce it into the loss function As a physical information regulator, the reactivation algorithm is introduced to optimize the loss function. The process is as follows:

[0189] Step (5.1) designs the loss function of the PINNs model into two parts, one of which is the loss of the neural network. , which is used to represent the error between the predicted value and the true value based on the network, and the other part is the loss of physical information , the loss of physical information includes the initial value loss , Boundary loss and the residual of the ordinary differential equation ;

[0190] The process of step (5.1) is as follows:

[0191] Step (5.1.1) The loss of the neural network is as follows,

[0192] ;

[0193] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the predicted value of the neural network, Indicates the actual measured value;

[0194] Step (5.1.2) A core idea of ​​PINNs is to make the neural network satisfy the given physical laws. The physical laws are expressed in the form of ODE. The ODE residual loss is used to measure whether the solution of the network satisfies these equations. The residual of the ordinary differential equation is as follows,

[0195] ;

[0196] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the actual measured value, for any time;

[0197] Step (5.1.3) The boundary conditions define the behavior of the solution at the boundary of the problem region, and the initial conditions define the state of the system at time t = 0. The boundary conditions and initial conditions of PINNs are obtained through actual measurement and estimation as follows:

[0198] ;

[0199] Step (5.1.4) The boundary loss function compares the boundary conditions predicted by the model with the actual physical boundary conditions, so that the solution of the model in the boundary area conforms to the physical laws. The boundary loss is as follows:

[0200] ;

[0201] in, is the number of sampled data, The operator representing the mean square error MSE, represents the predicted value of the neural network, Indicates the actual measured value;

[0202] Step (5.1.5) The initial value loss function measures the difference between the model prediction value and the actual initial conditions to ensure that the network output meets the physical constraints at the initial moment. The initial value loss is as follows:

[0203] ;

[0204] in, is the number of sampled data, The operator representing the mean square error (MSE), represents the predicted value of the neural network, Indicates the actual measured value;

[0205] Step (5.1.6) adds the losses mentioned in steps (5.1.1) to (5.1.6) to get the final loss function of the PINNs model. The formula is as follows:

[0206] ;

[0207] in, It is a physical information regulator; by introducing λ, PINNs can accurately estimate the output of the equation; by introducing physical information into the loss function, λ plays an important role in determining the contribution of physics to the final loss function. As λ increases, the output of PINNs will gradually rely on the constraints of the physical equations, while reducing λ will make the actual measured data more influential on the output; if the value of λ is set to 0, the entire loss function does not contain the physical information term and becomes a traditional ANN, then the network only relies on the data during training.

[0208] Step (5.2) Design a reactivation algorithm to obtain the optimal solution of the model;

[0209] In the step (5.2), using PINNs to solve the differential equation is an optimization problem of minimizing the loss function. For the vibration dynamics equation of the elevator, when using an optimization algorithm with line search and convergence criteria, a local optimal solution can be predicted. The local optimal solution can be regarded as the exact solution obtained. However, sometimes the local optimal solution obtained may have a large deviation. In order to avoid the situation where the solution has a large deviation, a reactivation technology is proposed to reactivate the optimization algorithm so that it can get the optimal solution again; the process is as follows:

[0210] Step (5.2.1) PINNs uses the Adma optimizer with strong Wolfe. In the kth iteration, a step size αk is found that satisfies the following conditions:

[0211] Function value decreasing condition,

[0212] ;

[0213] Strong curvature condition,

[0214] ;

[0215] in, is the search direction, and is a constant of the strong Wolfe condition, 0< < <1, is the current point, is the gradient of the current point. If the optimization falls into a local optimal solution at the beginning of the kth iteration, there is no αk that satisfies the equation. If no αk is found to satisfy the equation, the optimization is stopped.

[0216] Step (5.2.2) In order to restart the optimization process, introduce The constant makes the amplification factor slightly greater than zero, giving the following formula,

[0217] ;

[0218] The above formula is used to replace the function reduction condition. The strong curvature condition is only used in the first iteration of reactivation. The introduction of the above formula allows Greater than , making it easier to find αk that satisfies the equation; once found, the reactivation optimization algorithm will continue to use the standard strong Wolfe condition, that is, step (5.2.1).

[0219] Step 6: Acquire simulated vibration data. Input a small amount of elevator fault vibration signals collected in the first step into the PINNs model with motion equation constraints to obtain simulated elevator vibration data.

[0220] Step 7: The improved graph convolutional network (GCN) model is used for elevator fault diagnosis. The process is as follows: the e-RGCN fault diagnosis algorithm is proposed by combining the recursive neural network (RNN) and the graph convolutional network GCN, a recursive radius is set, and a correlation graph is constructed through the recursive radius. Then, GCN can effectively identify the elevator fault state through the constructed correlation graph;

[0221] The process of the seventh step is as follows:

[0222] Step (7.1) constructs the association graph. e-RNN is one of the methods for complex network analysis of time series. Recursive network is an intuitive method that uses the recursiveness of mechanical systems to convert time series into graphs. Unlike traditional time series analysis that focuses more on the sequence of samples, recursive network analysis focuses more on the spatial topological structure of samples and the relationship between samples.

[0223] The processing process of step (7.1) is as follows:

[0224] Step (7.1.1) constructs a graph that requires a relationship between the central vertex and other vectors. Two vertices can only be There is an edge between two vertices only when is the Euclidean norm between two vectors, is the recursive radius, which is the global threshold of RNN and can adjust the criteria of recursive pairs; two vertices are recursive pairs only when the conditions are met; in addition, the topological structure of all samples is obtained by performing pairwise operations on all vectors, and the structure of the network is encoded into a recursive matrix expressed as follows,

[0225] ;

[0226] in, is the Heaviside function, expressed as follows,

[0227] ;

[0228] in, is the Euclidean norm of a vector, if and There is an edge between The value of is 1, otherwise The value of is 0;

[0229] Step (7.1.2) The adjacency matrix of the graph can be constructed from the recursive matrix as follows,

[0230] ;

[0231] in is the Kronecker function, expressed as,

[0232] ;

[0233] The graph of the vibration signal is obtained through the obtained adjacency matrix.

[0234] Step (7.2) Construction of the e-RGCN fault diagnosis network. The proposed e-RGCN classification network consists of two GCN layers, one fully connected layer and an activation function. The GCN layer extracts graph features from the association graph, and the fully connected layer is used as a classifier for elevator fault diagnosis.

[0235] The processing process of step (7.2) is as follows:

[0236] Step (7.2.1) uses the cross-entropy loss function as the loss function of the network output. The cross-entropy loss function is a loss function in machine learning and deep learning. Especially in classification problems, it is used to measure the difference between the model's predicted value and the true label. It is particularly suitable for classification tasks (such as binary classification and multi-classification tasks). The loss function of the e-RGCN model is Indicates the degree of match between the predicted label and the actual label, as follows,

[0237] ;

[0238] Where N is the number of samples, C is the number of categories, is the probability of the jth category in the true label of the i-th sample, is the jth category in the model's predicted probability value for the i-th sample;

[0239] Step (7.2.2) During the e-RGCN training process, multi-layer GCN mixes the features of the node itself and its neighbors. Even if an edge is deleted, information about the node can still be obtained from its neighbors. Drop Edge randomly deletes some edges in the original graph, which increases the generalization of the network.

[0240] ;

[0241] Among them, A is the graph structure input of GCN, This is the graph structure after randomly selecting some edges.

[0242] Step (7.3) inputs the graph into the network, extracts features through two GCN layers, and obtains deep features; then classifies the deep features through a softmax classification layer, and outputs the fault type classification result;

[0243] Among them, the Softmax function is a commonly used activation function, which is often used in multi-category classification problems. The Softmax function can convert a real number vector into a vector representing a probability distribution, so that each element in the vector is between 0 and 1, and the sum of all elements is 1.

[0244] Step 8: Establish a digital twin platform for elevator fault diagnosis. The process is as follows:

[0245] The collected real-world elevator operation data is sent to the database through the network. Unity3D then reads the operation data from the database and displays it on the visualization interface. The status monitoring part is connected to the mobile device to realize real-time monitoring of the status of the real elevator. The user can observe the operation status of the real elevator while sitting in front of the designed system. The equipment fault warning interface will display the elevator operation status in real time, and will alarm when an abnormality is detected in the elevator. The elevator fault is diagnosed through the collected real-time signal data, and the equipment fault analysis interface will analyze the data of the abnormal time period and diagnose the specific fault type.

[0246] An implementation process of this embodiment is as follows:

[0247] Step 1: Select 4 types and define the elevator data into four types: normal vibration, bouncing vibration, start-stop vibration, and severe vibration;

[0248] Step 2: Data processing. In the data preprocessing stage, 40 sets of data are extracted from the four data sets for experiments. The 40 sets of data are divided into faulty parts and normal parts with a step length of 300, and the training set and the test set are divided in a ratio of 3:1.

[0249] Step 3: Result analysis, such as Figure 2 The figure shows the impact of digital twins on fault diagnosis results. When the digital twin model is used for fault diagnosis, the effect is improved. The diagnostic accuracy without the digital twin model is generally less than 90%. The accuracy of the digital twin model can reach more than 90%.

[0250] The experimental results show that the digital twin system improves the accuracy of fault diagnosis. Although fault diagnosis can achieve relatively good results without using simulated data, the present invention only selects three types of situations with more fault data. There is not enough data to support the research of more fault situations of elevators, which greatly increases the importance of digital twin models. The advantage of digital twin models is that they can combine the respective advantages of simulated data and real data. Simulated data can help the network learn a wider and more diverse range of fault modes and characteristics, thereby improving the generalization ability and robustness of the model. On the other hand, real data provides information that is closer to the actual scene, which helps the network better adapt to the real environment. Therefore, using the digital twin simulation model of the elevator for fault diagnosis is a potential and practical method, which can balance the relationship between data availability and training effect to a certain extent, and can solve the shortage of elevator fault data sets to a certain extent, thereby providing a more reliable and feasible solution for fault diagnosis tasks.

[0251] Another implementation process of this embodiment is:

[0252] Step 1: Define 6 fault diagnosis models as follows:

[0253] LeNet-5: Through convolution operations, local image features are extracted and the number of parameters of the fully connected layer is reduced. The parameters of the same convolution kernel are shared across the entire input image, which greatly reduces the model parameters. The pooling layer is used to reduce the size of the feature map, retain important features, and reduce computational complexity.

[0254] ResNet: By introducing residual connections, the gradient vanishing problem in deep network training is solved, making it possible to stably train extremely deep network structures and improve performance.

[0255] DGCN: Enhances the flexibility and expressiveness of graph convolutional networks by dynamically adjusting graph structures and node features, thereby improving the modeling effect of complex graph data.

[0256] GAT: Dynamically assigns weights to different neighbors in the graph through the self-attention mechanism, thereby effectively capturing important information in the graph structure and the complex relationships between nodes.

[0257] SGCN: Improves the computational efficiency and storage performance of graph convolutional networks on large-scale graph data by introducing a sparse graph convolution mechanism while maintaining high expressive power.

[0258] GraphSAGE: It performs efficient feature learning in graphs by sampling and aggregating node neighbors. It is suitable for processing large-scale graph data and dynamic graph structures.

[0259] Step 2: Analyze and compare the results, refer to Figure 3 It can be seen from the results that the two methods without graph structure: LeNet-5 and ResNet did not perform very well in the experiment. Because these two models do not take into account the structural characteristics of the graph, they rely entirely on data-driven methods for training. Due to the limitation of training data, the model cannot fully realize its potential and is prone to overfitting. This may cause the model to perform poorly in identifying fault information, thereby affecting its overall performance. The other methods and e-RGCN are all based on graph structure. GCN captures the dependencies between time series from the perspective of graph topology, which enables them to learn and extract feature information of nodes and capture important information of edges between nodes. Therefore, the latter model is more accurate than the above two models. The reason why the e-RGCN model of the present invention is superior to several other GCN models is that the construction method of graph data directly determines the quality of graph data, so it is crucial to ensure high-quality fault sample graphs. In the present invention, e-RGCN is used to construct high-quality fault sample graphs, so that structured feature blocks can better identify faults. It can be shown from the figure that the method proposed by the present invention has good stability.

[0260] The contents described in the embodiments of this specification are merely enumerations of implementation forms of the inventive concept and are for illustrative purposes only. The protection scope of the present invention should not be considered to be limited to the specific forms described in this embodiment, and the protection scope of the present invention also extends to equivalent technical means that can be thought of by ordinary technicians in this field based on the inventive concept.

Claims

1. A method for building a digital twin system for elevator fault diagnosis, characterized in that: The method comprises the following steps: The first step is to build a four-dimensional model framework for elevator digital twin fault diagnosis. The four-dimensional model of the elevator digital twin fault diagnosis system consists of physical space, virtual space, interactive space and service space. The model framework construction formula is as follows: ； Among them, PS represents physical space, VS represents virtual space, IS represents interactive space, and SS represents service space; The second step is to collect elevator data, including the vibration signal of the elevator, the temperature inside the host, the elevator operation status, the air pressure inside the elevator, whether the elevator is level, the door opening and closing status of the elevator, as well as real-time monitoring of the elevator, remote intercom and video recording functions, and detection of whether there are people, and remote data acquisition of various elevator data; The third step is to build a virtual elevator model. By measuring the size data of the elevator, a virtual elevator model similar to the physical elevator is built using 3Dmax. Step 4: Construct a numerical model of the elevator in the vertical direction; Step 5: Improve the PINNs model and introduce it into the loss function As a physical information regulator, the reactivation algorithm is introduced to optimize the loss function. The process is as follows: Step (5.1) designs the loss function of the PINNs model into two parts, one of which is the loss of the neural network. , which is used to represent the error between the predicted value and the true value based on the network, and the other part is the loss of physical information , the loss of physical information includes the initial value loss , Boundary loss and the residual of the ordinary differential equation ; Step (5.2) Design a reactivation algorithm to obtain the optimal solution of the model; Using PINNs to solve differential equations is an optimization problem of minimizing the loss function. For the vibration dynamics equations of the elevator, a reactivation technique is proposed to reactivate the optimization algorithm so that it can obtain the optimal solution again; the process is as follows: Step (5.2.1) PINNs uses the Adma optimizer with strong Wolfe. In the kth iteration, it will find a step size that satisfies the following conditions , Function value decreasing condition, ； Strong curvature condition, ， in, is the search direction, and is a constant of the strong Wolfe condition, 0< < <1, is the current point, is the gradient of the current point. If the optimization falls into a local optimal solution at the beginning of the kth iteration, there is no Satisfies the equation, if not found If the equation is satisfied, the optimization is stopped; Step (5.2.2) In order to restart the optimization process, introduce The constant makes the amplification factor slightly greater than zero, giving the following formula, ； The above formula is used to replace the function value reduction condition. The strong curvature condition is only used in the first iteration of reactivation. The introduction of the above formula allows Greater than , so that the equation is satisfied It is easier to find; once found, the reactivation optimization algorithm will continue to use the standard strong Wolfe condition, that is, step (5.2.1); Step 6: Acquire simulated vibration data. Input a small amount of elevator fault vibration signals collected in the first step into the PINNs model with motion equation constraints to obtain simulated elevator vibration data. Step 7: Improve the graph convolutional network (GCN) model for elevator fault diagnosis. The process is as follows: combine the recursive neural network (RNN) and the graph convolutional network (GCN) to propose the e-RGCN fault diagnosis algorithm, set a recursive radius, and construct a correlation graph through the recursive radius. Then, GCN can effectively identify the elevator fault state through the constructed correlation graph. The process of the seventh step is as follows: Step (7.1) constructs the association graph; Step (7.2) Construction of the e-RGCN fault diagnosis network. The proposed e-RGCN classification network consists of two GCN layers, one fully connected layer and an activation function. The GCN layer extracts graph features from the association graph, and the fully connected layer is used as a classifier for elevator fault diagnosis. Step (7.3) inputs the association graph into the network, extracts features through two GCN layers, and obtains deep features; then classifies the deep features through a softmax classification layer, and outputs the fault type classification result; The processing process of step (7.1) is as follows: Step (7.1.1) constructs a graph that requires a relationship between the central vertex and other vectors. Two vertices can only be There is an edge between two vertices only when is the Euclidean norm between two vectors, is the recursive radius, which is the global threshold of RNN and can adjust the criteria of recursive pairs; two vertices are recursive pairs only when the conditions are met; in addition, the topological structure of all samples is obtained by performing pairwise operations on all vectors, and the structure of the network is encoded into a recursive matrix expressed as follows, ； in, is the Heaviside function, expressed as follows, ； in, is the Euclidean norm of a vector, if and There is an edge between The value of is 1, otherwise The value of is 0; Step (7.1.2) The adjacency matrix of the graph can be constructed from the recursive matrix as follows, ； in is the Kronecker function, expressed as, ； The graph of the vibration signal is obtained through the obtained adjacency matrix; The processing process of step (7.2) is as follows: Step (7.2.1) uses the cross-entropy loss function as the loss function of the network output, and the loss function of the e-RGCN model is Indicates the degree of match between the predicted label and the actual label, as follows, ； Where N is the number of samples, C is the number of categories, It is The true label of the samples The probability of the categories, The model is The predicted probability value of samples categories; Step (7.2.2) During the e-RGCN training process, multi-layer GCN mixes the features of the node itself and its neighbors. Even if an edge is deleted, information about the node can still be obtained from its neighbors. Drop Edge randomly deletes some edges in the original graph, which increases the generalization of the network. ； Among them, A is the graph structure input of GCN, The graph structure after randomly selecting some edges; Step 8. Establish a digital twin platform for elevator fault diagnosis.

2. The method for building a digital twin system for elevator fault diagnosis according to claim 1, characterized in that: The process of the first step is as follows: Step (1.1) PS It's an elevator EDTM The basis of is the collection of physical entities involved in the operation of the elevator digital twin fault diagnosis system, which is responsible for collecting measurement activities to provide the physical size data and physical operation data of the elevator. The formula is as follows: ； in, Represents physical entities in physical space, including various parts of the elevator and various sensors. K Indicates the total number of components and sensors of various types; Step (1.2) VS yes PS Digital depiction and description, aiming to PS It is highly similar in shape and provides visualization and interaction functions for the service. The formula is as follows: ； in, Represents virtual entities in virtual space, and One-to-one correspondence, in EDTM of VS middle, VE It is a dynamically running geometric model that is continuously updated as real-time twin data is generated, thus achieving PS Synchronization of physical elevators; Step (1.3) IS yes PS , VS as well as SS The bridge between them is used to realize data exchange among the three. The formula is as follows: ； in, TD represents twin data, PL Indicates the platform, TD It is the core driver for the operation of the elevator digital twin geometric model; Step (1.4) SS Provide elevator fault diagnosis technology to establish a digital twin geometric model for fault diagnosis during elevator operation. The formula is as follows: ； in, Indicates elevator operation data monitoring. Indicates elevator fault warning report. Indicates elevator fault diagnosis. Indicates the generation of elevator simulation fault data.

3. The method for building a digital twin system for elevator fault diagnosis according to claim 2, characterized in that: The process of step (1.3) is: Steps (1.3.1) TD It is the core driver of the elevator digital twin geometric model operation. The formula is as follows: ； in, Represents virtual entity data, which is the geometric data related to the digital model, including the geometric dimension data involved in three-dimensional modeling and the vibration data obtained by measurement; Represents physical entity data, including rated data and geometric information data of each elevator component; Relevant data representing the service space, including computer-aided design (CAD) model data, SolidWorks model data, elevator vibration simulation data obtained by algorithms, and elevator fault diagnosis result data obtained by neural networks; Steps (1.3.2) Provides tools and third-party libraries to support services for modeling, visualization, interaction, measurement, and measurement data processing; PS The modeling is implemented using SolidWorks software; the geometric modeling of the elevator virtual entity is realized through 3DMax; the visualization and interaction of 3D data and models are realized through Python combined with Unity3D; application visualization, that is, graphical user interface (GUI) design is realized through Unity3Dl; the accelerometer that performs vibration measurement is connected to the development software to perform measurement and control functions; the measurement data processing is implemented based on the algorithm provided by Python.

4. The method for building a digital twin system for elevator fault diagnosis according to any one of claims 1 to 3, characterized in that: In the second step, the vibration signal of the elevator is collected at a set frequency by an acceleration sensor installed in the car top box on the top of the elevator car; the temperature inside the host is collected by a temperature sensor installed in the host on the top of the machine; the operation status of the elevator such as stopping, going up and down is collected by a leveling / base sensor installed on the top of the car; The air pressure in the elevator is collected by the atmospheric pressure sensor installed in the main unit on the top of the car; Whether the elevator is level is detected by the photoelectric sensor installed on the top of the car; the door opening and closing status of the elevator is detected by the door magnetic sensor installed on the top step of the car; the real-time monitoring, remote intercom and video recording functions of the elevator are detected by the monitoring camera installed inside the car; the infrared sensor installed on the elevator door detects whether someone passes through the elevator; various data are remotely acquired through the TCP / IP communication protocol, the sensor, elevator and gateway are connected by serial port connection, and Modbus and RS485 protocols are used for data communication.

5. The method for building a digital twin system for elevator fault diagnosis according to any one of claims 1 to 3, characterized in that: The process of the fourth step is as follows: Step (4.1) establishes the dynamic equation of the system, the formula is as follows, ； Where T is the kinetic energy of the system; U is the potential energy of the system; D is the dissipation coefficient of the system; x is the generalized displacement; is the generalized velocity; Q is the generalized force; Step (4.2) calculates the system's kinetic energy, the system's potential energy and the system's dissipation coefficient, ； ； ； in, is the equivalent mass of the car and the loaded car; is the counterweight mass; , are the equivalent stiffness and damping of the wire rope between the car and the traction sheave respectively; , are the equivalent stiffness and damping of the wire rope between the counterweight and the traction sheave respectively; , and They are the car vibration linear displacement, counterweight vibration linear displacement, and traction sheave vibration angular displacement; , are the moment of inertia of the traction sheave and the radius of the rope groove of the traction sheave respectively; Step (4.3) Arrange the formulas in step (4.2) and step (4.1) as follows: Step (4.3.1) for generalized coordinates ,have, ； Step (4.3.2) Since the elevator car is only affected by the vertical wire rope tension and gravity, substitute the result of step (4.3.1) into the Lagrange equation of step (4.1) to obtain, ； Step (4.3.3) for generalized coordinates ,have, ； Step (4.3.4) The elevator counterweight is only affected by the vertical wire rope tension and gravity. Substituting the result of step (4.3.3) into the Lagrange equation of step (4.1), we get: ； Step (4.3.5) for generalized coordinates ,have, ； Step (4.3.6) for generalized coordinates , the system inertia force in the elevator start-brake stage is used as the excitation source, and the exciting force is expressed as, ； Step (4.4) Among them, step (4.3.2), step (4.3.2) and step (4.3.2) constitute the differential equation of motion, which can be expressed in matrix form as follows, ； in, ； ； ； ； ； Among them, M is the system mass matrix, C is the system damping matrix, K is the system stiffness matrix, and F is the exciting force matrix; They represent the acceleration, velocity and displacement of the system respectively.

6. The method for building a digital twin system for elevator fault diagnosis according to any one of claims 1 to 3, characterized in that: The process of step (5.1) is as follows: Step (5.1.1) The loss of the neural network is as follows, ； in, is the number of sampled data, The operator representing the mean square error MSE, represents the predicted value of the neural network, Indicates the actual measured value; Step (5.1.2) A core idea of ​​PINNs is to make the neural network satisfy the given physical laws. The physical laws are expressed in the form of ODE. The ODE residual loss is used to measure whether the solution of the network satisfies these equations. The residual of the ordinary differential equation is as follows, ； in, is the number of sampled data, The operator representing the mean square error MSE, represents the actual measured value, for any time; Step (5.1.3) The boundary conditions define the behavior of the solution at the boundary of the problem region, and the initial conditions define the state of the system at time t = 0. The boundary conditions and initial conditions of PINNs are obtained through actual measurement and estimation as follows: ； Step (5.1.4) The boundary loss function compares the boundary conditions predicted by the model with the actual physical boundary conditions, so that the solution of the model in the boundary area conforms to the physical laws. The boundary loss is as follows: ； in, is the number of sampled data, The operator representing the mean square error MSE, represents the predicted value of the neural network, Indicates the actual measured value; Step (5.1.5) The initial value loss function measures the difference between the model prediction value and the actual initial conditions to ensure that the network output meets the physical constraints at the initial moment. The initial value loss is as follows: ， in, is the number of sampled data, The operator representing the mean square error (MSE), represents the predicted value of the neural network, Indicates the actual measured value; Step (5.1.6) adds the losses mentioned in steps (5.1.1) to (5.1.6) to get the final loss function of the PINNs model. The formula is as follows: ； in, , It is a physical information regulator.

7. The method for building a digital twin system for elevator fault diagnosis according to any one of claims 1 to 3, characterized in that: The process of the eighth step is: the collected real-world elevator operation data is sent to the database through the network, Unity3D then reads the operation data from the database and displays it on the visualization interface, the status monitoring part is connected to the mobile device to realize real-time monitoring of the status of the real elevator; the user can observe the operation status of the real elevator while sitting in front of the designed system; the equipment fault warning interface will display the elevator operation status in real time, and will alarm when an abnormality is detected in the elevator; the elevator fault is diagnosed through the collected real-time signal data, and the equipment fault analysis interface will analyze the data of the abnormal time period and diagnose the specific fault type.

Images