Method and system for monitoring stress state of large recreation facility
By establishing twin models in large amusement facilities and using electromagnetic stress monitoring sensors, combined with machine learning models, the problem of low stress detection accuracy in the existing technology is solved, more accurate stress monitoring is achieved, and the safe operation of the facilities is ensured.
Patent Information
- Application Number
- CN202410962461.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
AI Technical Summary
In the existing stress monitoring technology of large-scale amusement facilities, the detection accuracy of strain gauge and fiber grating sensors is greatly affected by the bonding process and ambient temperature, resulting in inaccurate detection results.
By establishing a twin model of the amusement facility, conduct simulation experiments, determine stress monitoring points, and set up electromagnetic stress monitoring sensors at these points, use excitation coils and induction coils to detect stress, combine machine learning models to identify stress values, avoid the influence of bonding processes and temperatures .
Improves the accuracy of stress detection and ensures the safe operation of large-scale ride facilities, suitable for various sizes and types of ride facilities components.
Smart Images

Figure CN118913488A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of amusement facility monitoring, and more particularly to a method and system for monitoring the stress state of a large-scale amusement facility. Background Art
[0002] In order to attract tourists, amusement parks and theme parks usually have various large-scale amusement facilities, such as roller coasters, Ferris wheels, pirate ships, etc. The main structure of large-scale amusement facilities is usually made of metal materials such as steel, with complex structures, various movement forms and stress-strain conditions. Therefore, monitoring the stress conditions of the metal structures of large-scale amusement facilities is of great significance to the safe operation of large-scale amusement facilities and avoiding accidents.
[0003] Existing large-scale amusement facilities usually use strain gauges, fiber Bragg grating sensors and other stress strain sensors to monitor the stress of the main structure. However, in the actual application of strain gauges, the strain gauges need to be tightly bonded to the structure to be measured, which is greatly affected by the adhesive and bonding process, resulting in poor detection accuracy; fiber Bragg grating sensors detect the strain of the sensor by monitoring the displacement of the detection light wavelength in the optical fiber, and its detection accuracy is also affected by the bonding process between the fiber Bragg grating and the structure to be detected, and the detection light wavelength of the fiber Bragg grating sensor is also affected by the ambient temperature, resulting in inaccurate detection results.
[0004] Therefore, how to detect the stress of the structure of large-scale amusement facilities and improve the detection accuracy so as to ensure the safe operation of large-scale amusement facilities is a problem that technical personnel in this field urgently need to solve. Summary of the invention
[0005] In view of this, the present invention provides a stress state monitoring method and system for large-scale amusement facilities. By establishing a twin model of the large-scale amusement facilities and conducting simulation experiments, the stress monitoring points of the amusement facility structure are determined, and two stress detection coils are set at each monitoring point to monitor the stress at the monitoring point, thereby avoiding the problem of inaccurate detection accuracy of strain gauges and fiber grating sensors being affected by bonding process and temperature.
[0006] In order to achieve the above object, the present invention provides the following technical solutions:
[0007] The present invention discloses a stress state monitoring method for a large-scale amusement facility, and the specific steps are as follows:
[0008] Constructing an amusement facility model according to geometric parameters, material parameters and connection relationships of components of the amusement facility; conducting simulation experiments using the amusement facility model to determine stress monitoring points of the amusement facility;
[0009] Arrange electromagnetic stress monitoring sensors at each of the stress monitoring points, and input the static parameters of each sensor into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor, under the action of the excitation magnetic field generated by the excitation coil and the components of the stress monitoring point, generates an induced voltage and an induced current and sends them to the monitoring host as dynamic monitoring data;
[0010] The monitoring host identifies the stress of each stress monitoring point using a machine learning model based on the static parameters and the dynamic monitoring data of each stress monitoring point.
[0011] Furthermore, the arrangement of electromagnetic stress monitoring sensors at each of the stress monitoring points includes:
[0012] The excitation coil and the induction coil are wound on the component at the stress monitoring point with a fixed number of turns, and the distance between the excitation coil and the induction coil is measured; the excitation coil and the induction coil are waterproofed and fixed by an adhesive, and an electromagnetic shielding layer is attached; the excitation coil is connected to an excitation generating unit, and the induction coil is connected to a voltage and current detection unit.
[0013] Furthermore, the static parameters include: the diameter of the excitation coil, the diameter of the induction coil, the distance between the excitation coil and the induction coil, and the cross-sectional area, magnetic permeability, and elastic modulus of the stress monitoring point component.
[0014] Furthermore, the excitation magnetic field is generated by passing an excitation current into the excitation coil; the excitation current is composed of a fixed number of multiple pulse currents in sequence, and the current magnitude of each pulse current is different; the time length and frequency of the pulse current are fixed values, and the excitation current is composed of the pulse currents arranged in order from small to large or from large to small.
[0015] Furthermore, the machine learning model includes a convolutional neural network and a fully connected neural network;
[0016] The method of using a machine learning model to identify the stress of each stress monitoring point includes:
[0017] Preprocessing the dynamic parameters to obtain an induced voltage matrix and an induced current matrix;
[0018] Inputting the induced voltage matrix and the induced current matrix into the convolutional neural network to obtain a voltage feature matrix and a current feature matrix;
[0019] Expand the current characteristic matrix and the voltage characteristic matrix in a fixed direction and order to obtain a first vector and a second vector respectively; splice the first vector and the second vector row by row to obtain a third vector; and splice the values of the static parameters at the end of the third vector in order to obtain a fourth vector;
[0020] The fourth vector is input into the fully connected neural network, and the stress value of the stress monitoring point is output.
[0021] Furthermore, the preprocessing includes: arranging the voltage sequence and the current sequence based on the values of the induced voltage and the induced current and the corresponding time; dividing the voltage sequence and the current sequence into a plurality of segments according to the time length of the pulse current, so that each segment corresponds to the magnitude of the pulse current; arranging the segmented voltage sequence and the current sequence according to the arrangement order of the pulse current to obtain the induced voltage matrix and the induced current matrix. The rows and columns of the voltage matrix correspond to the magnitude and time of the pulse current, respectively.
[0022] Furthermore, the convolutional neural network includes a voltage convolution branch and a current convolution branch, and the voltage convolution branch and the current convolution branch are formed by sequentially connecting a set number of convolution modules; the induced voltage matrix is convolved by the voltage convolution branch, and the voltage feature matrix is obtained after extracting features; the induced current matrix is convolved by the current convolution branch, and the current feature matrix is obtained after extracting features;
[0023] The convolution module includes a convolution layer, an activation function, and a pooling layer connected in sequence;
[0024] The convolution layer performs convolution calculation on the input matrix data by using a convolution operator with a set size and moving step, performs weighted summation on the values within the convolution operator window range, and outputs the weighted summation result to the corresponding position of the convolution layer output matrix;
[0025] The activation function is a ReLU activation function, which adds nonlinearity to the data of the output matrix of the convolution layer and then inputs it into the pooling layer;
[0026] The pooling layer performs a maximum pooling operation on the matrix data processed by the activation function according to a set step size through a pooling operator of a set size to obtain the output matrix of the convolution module.
[0027] Furthermore, the fully connected neural network is composed of a set number of fully connected layers, and each of the fully connected layers contains a set number of neurons;
[0028] The neuron multiplies each element in the input vector by the corresponding weight and adds the sum, and then adds the constant term bias and calculates the output value of the current neuron through the activation function;
[0029] The outputs of all the neurons in the same fully connected layer are sequentially formed into a vector, which is used as the input vector of each neuron in the next fully connected layer;
[0030] The last fully connected layer contains only one neuron, and the output value of the neuron is the stress value of the stress monitoring point.
[0031] The present invention also discloses a stress state monitoring system for large-scale amusement facilities, comprising:
[0032] Model building experiment module: construct an amusement facility model according to the geometric parameters, material parameters and connection relationship of the components of the amusement facility; use the amusement facility model to conduct simulation experiments to determine the stress monitoring points of the amusement facility;
[0033] Sensor module: an electromagnetic stress monitoring sensor is arranged at each of the stress monitoring points, and the static parameters of each sensor are input into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor generates an induced voltage and an induced current under the action of the excitation magnetic field generated by the excitation coil and the components of the stress monitoring point, and sends the induced voltage and induced current as dynamic monitoring data to the monitoring host;
[0034] Stress monitoring module: The monitoring host uses a machine learning model to identify the stress of each stress monitoring point based on the static parameters and dynamic monitoring data of each stress monitoring point.
[0035] Further, the electromagnetic stress monitoring sensor includes: an excitation generating unit, an excitation coil, an induction coil, a voltage and current detection unit, a communication unit, and a power supply unit;
[0036] The excitation generating unit is connected to the communication unit by signal and electrically connected to the excitation coil; the excitation current is generated by receiving the control command of the monitoring host through the communication unit and inputted into the excitation coil; the excitation coil generates an excitation magnetic field by using the excitation current;
[0037] The voltage and current detection unit is signal-connected to the communication unit and electrically connected to the induction coil; the induction coil generates an induced voltage and an induced current under the action of the excitation magnetic field, and the voltage and current detection unit detects and sends the detected induced voltage and induced current values to the monitoring host through the communication unit;
[0038] The power supply unit is electrically connected to the excitation generating unit, the voltage and current detecting unit, and the communication unit respectively, and supplies power to the excitation generating unit, the voltage and current detecting unit, and the communication unit.
[0039] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a stress state monitoring method and system for large-scale amusement facilities, which determines stress monitoring points by simulating large-scale amusement facilities; based on the principle that the magnetic properties of ferromagnetic materials will change under stress, the correlation between magnetic properties and stress is utilized to set excitation coils and induction coils as monitoring sensors at stress monitoring points, and the detection accuracy is insensitive to the fixing method of the coils and is not affected by the bonding force between the coils and the components to be detected, thereby avoiding the problem of poor stress monitoring accuracy caused by the fixing process and long-term use of strain gauges and fiber grating sensors; through trained machine learning, based on the static parameters and dynamic parameters of the electromagnetic monitoring sensor, the stress of the stress monitoring point is monitored, which can adapt to various sizes and types of amusement facility components, and accurately monitor the stress value of each monitoring point, thereby ensuring the safe operation of large-scale amusement facilities. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0041] Figure 1 It is a schematic diagram of the overall process of an embodiment of the present invention. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] The embodiment of the present invention discloses a stress state monitoring method for a large-scale amusement facility, such as Figure 1 As shown, the specific steps are as follows:
[0044] Construct an amusement facility model based on the geometric parameters, material parameters and connection relationship of the components of the amusement facility; use the amusement facility model to conduct simulation experiments and determine the stress monitoring points of the amusement facility;
[0045] Electromagnetic stress monitoring sensors are arranged at each stress monitoring point, and the static parameters of each sensor are input into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor generates an induced voltage and an induced current under the action of the excitation magnetic field generated by the excitation coil and the components at the stress monitoring point, and the induced voltage and induced current are sent to the monitoring host as dynamic monitoring data;
[0046] The monitoring host uses a machine learning model to identify the stress of each stress monitoring point based on the static parameters and dynamic monitoring data of each stress monitoring point.
[0047] In a specific embodiment, an electromagnetic stress monitoring sensor is arranged at each stress monitoring point, including:
[0048] The excitation coil and the induction coil are wound on the component at the stress monitoring point with a fixed number of turns, and the distance between the excitation coil and the induction coil is measured; the number of turns of the excitation coil and the induction coil, as well as the distance between the two, may not be the same, but in actual operation, the number of turns and the distance are usually set to fixed values, thereby simplifying the parameters involved in the training of the machine learning model and reducing the amount of calculation. The excitation coil and the induction coil are waterproofed and fixed with an adhesive, and an electromagnetic shielding layer is attached to shield the interference of the external electromagnetic environment. The adhesive fixes the coils mainly to ensure that the distance between the two coils remains unchanged. Since the stress of the component is detected by the coil, it is based on the correlation between the change of magnetic properties and the magnitude of stress. The detection current is mainly affected by the excitation magnetic field and the magnetic permeability of the component under stress. The interaction force between the coil and the component can be ignored for the detection, thereby avoiding the problem of inaccurate detection results caused by the use of strain gauges and fiber grating sensors. The excitation coil is connected to the excitation generating unit, and the induction coil is connected to the voltage and current detection unit.
[0049] In a specific embodiment, the static parameters include: the diameter of the excitation coil, the diameter of the induction coil, the distance between the excitation coil and the induction coil, the cross-sectional area of the stress monitoring point component, the magnetic permeability, and the elastic modulus. The coil diameter is calculated based on the model geometry parameters and the diameter of the winding. The component cross-sectional area can also be obtained based on the model geometry parameters. The magnetic permeability and elastic modulus are obtained by querying or testing the model of the component material.
[0050] In a specific embodiment, an excitation magnetic field is generated by passing an excitation current into an excitation coil; the excitation current is composed of a fixed number of multiple pulse currents in sequence, and the current magnitude of each pulse current is different; the time length and frequency of the pulse current are fixed values, and the excitation current is composed of the pulse currents arranged in order from small to large or from large to small.
[0051] Specifically, the time length of the pulse current is t, the number is n, and the frequency is mHz, then the time length of the excitation current is T=t×n; the pulse current can be a square wave, a sine wave or a sawtooth wave, and the current value is the equivalent current size of the pulse current.
[0052] In a specific embodiment, the machine learning model includes a convolutional neural network and a fully connected neural network;
[0053] Use machine learning models to identify stress at each stress monitoring point, including:
[0054] Preprocess the dynamic parameters to obtain the induced voltage matrix and the induced current matrix;
[0055] Input the induced voltage matrix and the induced current matrix into the convolutional neural network to obtain the voltage feature matrix and the current feature matrix;
[0056] Expand the current characteristic matrix and the voltage characteristic matrix in a fixed direction and order to obtain a first vector and a second vector respectively; splice the first vector and the second vector row by row to obtain a third vector; splice the values of the static parameters at the end of the third vector in order to obtain a fourth vector;
[0057] The fourth vector is input into the fully connected neural network, and the stress value of the stress monitoring point is output.
[0058] In a specific embodiment, the preprocessing includes: based on the values of the induced voltage and the induced current and the corresponding time, the voltage sequence and the current sequence are sorted; according to the time length of the pulse current, the voltage sequence and the current sequence are divided into multiple segments, so that each segment corresponds to the size of the pulse current; according to the arrangement order of the pulse current, the segmented voltage sequence and current sequence are sorted to obtain the induced voltage matrix and the induced current matrix. The rows and columns of the induced voltage matrix correspond to the time and size of the pulse current respectively; the rows and columns of the induced current matrix correspond to the time and size of the pulse current respectively. Since the rows of the matrix correspond to the time of the pulse current, the elements in the first vector and the second vector are spliced row by row, so that the vector can better express the time characteristics of the feature.
[0059] In a specific embodiment, the convolutional neural network includes a voltage convolution branch and a current convolution branch, and the voltage convolution branch and the current convolution branch are formed by sequentially connecting a set number of convolution modules; the induced voltage matrix is convolved by the voltage convolution branch, and the voltage feature matrix is obtained after the feature is extracted; the induced current matrix is convolved by the current convolution branch, and the current feature matrix is obtained after the feature is extracted;
[0060] The convolution module includes a convolution layer, an activation function, and a pooling layer connected in sequence;
[0061] The convolution layer performs convolution calculation on the input matrix data by setting the convolution operator of the size and moving step, performs weighted summation on the values within the convolution operator window range, and outputs the weighted summation result to the corresponding position of the convolution layer output matrix;
[0062] The activation function is the ReLU activation function, which adds nonlinearity to the data of the convolutional layer output matrix and then inputs it to the pooling layer;
[0063] The pooling layer uses a pooling operator of a set size to perform a maximum pooling operation on the matrix data processed by the activation function according to the set step size to obtain the output matrix of the convolution module.
[0064] In a specific embodiment, the fully connected neural network is composed of a set number of fully connected layers, each of which contains a set number of neurons;
[0065] The neuron multiplies each element in the input vector by the corresponding weight and adds the sum, then adds the constant term bias and calculates the output value of the current neuron through the activation function;
[0066] The outputs of all neurons in the same fully connected layer form a vector in sequence, which serves as the input vector of each neuron in the next fully connected layer;
[0067] The last fully connected layer contains only one neuron, and the output value of this neuron is the stress value of the stress monitoring point.
[0068] In a specific embodiment, multiple experimental samples with different cross-sectional areas, magnetic permeabilities, and elastic moduli are selected, and electromagnetic stress sensors are set on the experimental samples to record the distance between the two coils and the coil diameter; then, stresses of different magnitudes are applied to the experimental samples through a stress loading device, and an excitation magnetic field is generated by applying the same excitation current as in actual monitoring to obtain corresponding detection current and voltage data; multiple sets of cross-sectional area, magnetic permeability, elastic modulus, coil distance, coil diameter, detection current, detection voltage data, and corresponding stress magnitude are obtained through experiments, and then sorted (with static parameters and dynamic parameters of the experiment as sample data, and stress value magnitude as sample label) into multiple sample data and divided into training data sets and test data sets. Initialize the network parameters and weight values of the machine learning model, use the samples in the training data set to perform iterative training of back propagation, guide to reach the set number of iterations or the error loss meets the set threshold, and end the training; finally, test the test data set to determine whether the error of the machine learning model for stress recognition meets the requirements.
[0069] In a specific embodiment, a metal fatigue control group of facility experimental samples is also included, and experiments are conducted on experimental samples with metal fatigue, thereby obtaining a data set of samples with metal fatigue. A second machine learning model is established and initialized, and the metal fatigue sample data set is used as a training data set for training, so that the second machine learning model can identify whether a component has metal fatigue; wherein the second machine learning model has the same overall structure as the machine learning model trained with non-metal fatigue samples, and the last fully connected layer of its fully connected neural network contains only one neuron, and the output value of the neuron is whether metal fatigue exists. Specifically, metal fatigue usually produces tiny cracks inside the material, and the cracks usually start at some microscopic defects or grain boundaries in the material, and gradually expand and deepen after repeated loading or stress, causing metal fatigue in the material, and eventually leading to the fracture and failure of the material. The tiny cracks generated inside the material will cause a significant change in the magnetic properties of the material, so that the detection current and detection voltage of the induction coil obviously show current and voltage characteristics that are different from normal materials. Through the second machine model to learn the current and voltage characteristics of the metal fatigue component, it is possible to further identify whether there is metal fatigue at the stress monitoring point of the large amusement facility. Therefore, when the monitoring host of the monitoring system detects that the stress at the stress monitoring point is greater than the preset threshold, it will issue a first alarm message to indicate stress abnormality; when the monitoring host detects that metal fatigue exists at the stress monitoring point, it will issue a first alarm message to indicate the existence of metal fatigue and immediately stop the operation of the amusement facility.
[0070] The embodiment of the present invention also discloses a stress state monitoring system for a large-scale amusement facility, comprising:
[0071] Model building experiment module: construct the model of the amusement facility according to the geometric parameters, material parameters and connection relationship of the components of the amusement facility; use the amusement facility model to conduct simulation experiments and determine the stress monitoring points of the amusement facility;
[0072] Sensor module: electromagnetic stress monitoring sensors are arranged at each stress monitoring point, and the static parameters of each sensor are input into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor generates an induced voltage and an induced current under the action of the excitation magnetic field generated by the excitation coil and the components at the stress monitoring point, and the induced voltage and induced current are sent to the monitoring host as dynamic monitoring data;
[0073] Stress monitoring module: The monitoring host uses a machine learning model to identify the stress of each stress monitoring point based on the static parameters and dynamic monitoring data of each stress monitoring point.
[0074] Further, the electromagnetic stress monitoring sensor includes: an excitation generating unit, an excitation coil, an induction coil, a voltage and current detection unit, a communication unit, and a power supply unit;
[0075] The excitation generating unit is connected to the communication unit signal and electrically connected to the excitation coil; the communication unit receives the control command of the monitoring host to generate an excitation current and inputs it into the excitation coil; the excitation coil generates an excitation magnetic field using the excitation current;
[0076] The voltage and current detection unit is connected to the communication unit signal and electrically connected to the induction coil; the induction coil generates an induced voltage and an induced current under the action of the excitation magnetic field, and the voltage and current detection unit detects and sends the detected induced voltage and induced current values to the monitoring host through the communication unit;
[0077] The power supply unit is electrically connected to the excitation generating unit, the voltage and current detecting unit, and the communication unit respectively, and supplies power to the excitation generating unit, the voltage and current detecting unit, and the communication unit.
[0078] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0079] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A stress state monitoring method for a large-scale amusement facility, characterized in that: The specific steps are as follows: Constructing an amusement facility model according to geometric parameters, material parameters and connection relationships of components of the amusement facility; conducting simulation experiments using the amusement facility model to determine stress monitoring points of the amusement facility; Arrange electromagnetic stress monitoring sensors at each of the stress monitoring points, and input the static parameters of each sensor into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor, under the action of the excitation magnetic field generated by the excitation coil and the components of the stress monitoring point, generates an induced voltage and an induced current and sends them to the monitoring host as dynamic monitoring data; The monitoring host identifies the stress of each stress monitoring point using a machine learning model based on the static parameters and the dynamic monitoring data of each stress monitoring point.
2. A stress state monitoring method for a large-scale amusement facility according to claim 1, characterized in that: The step of arranging electromagnetic stress monitoring sensors at each of the stress monitoring points comprises: The excitation coil and the induction coil are wound on the component at the stress monitoring point with a fixed number of turns, and the distance between the excitation coil and the induction coil is measured; the excitation coil and the induction coil are waterproofed and fixed by an adhesive, and an electromagnetic shielding layer is attached; the excitation coil is connected to an excitation generating unit, and the induction coil is connected to a voltage and current detection unit.
3. The stress state monitoring method of a large-scale amusement facility according to claim 1, characterized in that: The static parameters include: the diameter of the excitation coil, the diameter of the induction coil, the distance between the excitation coil and the induction coil, and the cross-sectional area, magnetic permeability, and elastic modulus of the stress monitoring point component.
4. The stress state monitoring method of a large-scale amusement facility according to claim 1, characterized in that: The excitation magnetic field is generated by passing an excitation current into the excitation coil; the excitation current is composed of a fixed number of multiple pulse currents in sequence, and the current magnitude of each pulse current is different; the time length and frequency of the pulse current are fixed values, and the excitation current is composed of the pulse currents arranged in order from small to large or from large to small.
5. A stress state monitoring method for a large-scale amusement facility according to claim 4, characterized in that: The machine learning model includes a convolutional neural network and a fully connected neural network; The method of using a machine learning model to identify the stress of each stress monitoring point includes: Preprocessing the dynamic parameters to obtain an induced voltage matrix and an induced current matrix; Inputting the induced voltage matrix and the induced current matrix into the convolutional neural network to obtain a voltage feature matrix and a current feature matrix; Expand the current characteristic matrix and the voltage characteristic matrix in a fixed direction and order to obtain a first vector and a second vector respectively; splice the first vector and the second vector row by row to obtain a third vector; and splice the values of the static parameters at the end of the third vector in order to obtain a fourth vector; The fourth vector is input into the fully connected neural network, and the stress value of the stress monitoring point is output.
6. A stress state monitoring method for a large-scale amusement facility according to claim 5, characterized in that: The preprocessing includes: arranging the voltage sequence and the current sequence based on the values of the induced voltage and the induced current and the corresponding time to obtain a voltage sequence and a current sequence; dividing the voltage sequence and the current sequence into a plurality of segments according to the time length of the pulse current, so that each segment corresponds to the size of the pulse current one by one; and arranging the segmented voltage sequence and the current sequence according to the arrangement order of the pulse current to obtain the induced voltage matrix and the induced current matrix.
7. A stress state monitoring method for a large-scale amusement facility according to claim 5, characterized in that: The convolutional neural network includes a voltage convolution branch and a current convolution branch, and the voltage convolution branch and the current convolution branch are formed by sequentially connecting a set number of convolution modules; the induced voltage matrix is convolved by the voltage convolution branch, and the voltage feature matrix is obtained after extracting features; the induced current matrix is convolved by the current convolution branch, and the current feature matrix is obtained after extracting features; The convolution module includes a convolution layer, an activation function, and a pooling layer connected in sequence; The convolution layer performs convolution calculation on the input matrix data by using a convolution operator with a set size and moving step, performs weighted summation on the values within the convolution operator window range, and outputs the weighted summation result to the corresponding position of the convolution layer output matrix; The activation function is a ReLU activation function, which adds nonlinearity to the data of the output matrix of the convolution layer and then inputs it into the pooling layer; The pooling layer performs a maximum pooling operation on the matrix data processed by the activation function according to a set step size through a pooling operator of a set size to obtain the output matrix of the convolution module.
8. The stress state monitoring method of a large-scale amusement facility according to claim 5, characterized in that: The fully connected neural network is composed of a set number of fully connected layers, and each of the fully connected layers contains a set number of neurons; The neuron multiplies each element in the input vector by the corresponding weight and adds the sum, and then adds the constant term bias and calculates the output value of the current neuron through the activation function; The outputs of all the neurons in the same fully connected layer are sequentially formed into a vector, which is used as the input vector of each neuron in the next fully connected layer; The last fully connected layer contains only one neuron, and the output value of the neuron is the stress value of the stress monitoring point.
9. A stress state monitoring system for large-scale amusement facilities, characterized in that: include: Model building experiment module: construct an amusement facility model according to the geometric parameters, material parameters and connection relationship of the components of the amusement facility; use the amusement facility model to conduct simulation experiments to determine the stress monitoring points of the amusement facility; Sensor module: an electromagnetic stress monitoring sensor is arranged at each of the stress monitoring points, and the static parameters of each sensor are input into the monitoring host; the induction coil of the electromagnetic stress monitoring sensor generates an induced voltage and an induced current under the action of the excitation magnetic field generated by the excitation coil and the components of the stress monitoring point, and sends the induced voltage and induced current as dynamic monitoring data to the monitoring host; Stress monitoring module: The monitoring host uses a machine learning model to identify the stress of each stress monitoring point based on the static parameters and dynamic monitoring data of each stress monitoring point.
10. A stress state monitoring system for a large-scale amusement facility according to claim 9, characterized in that: The electromagnetic stress monitoring sensor comprises: an excitation generating unit, an excitation coil, an induction coil, a voltage and current detecting unit, a communication unit, and a power supply unit; The excitation generating unit is connected to the communication unit by signal and electrically connected to the excitation coil; the excitation current is generated by receiving the control command of the monitoring host through the communication unit and inputted into the excitation coil; the excitation coil generates an excitation magnetic field by using the excitation current; The voltage and current detection unit is signal-connected to the communication unit and electrically connected to the induction coil; the induction coil generates an induced voltage and an induced current under the action of the excitation magnetic field, and the voltage and current detection unit detects and sends the detected induced voltage and induced current values to the monitoring host through the communication unit; The power supply unit is electrically connected to the excitation generating unit, the voltage and current detecting unit, and the communication unit respectively, and supplies power to the excitation generating unit, the voltage and current detecting unit, and the communication unit.
Citation Information
Cited By
Axial force information acquisition gateway
CN120008783A
