A method and system for detecting derailment of an underground rail car

Abstract

The present application relates to the technical field of derailment detection, and discloses a kind of underground rail car derailment detection method and system, the method comprises: acquisition underground node to be monitored position information constructs underground position network, according to the distance of different node to be monitored and generates redundant node connected graph;Energy-saving robust routing strategy is generated, and the optimal transmission path of monitoring data of node to be monitored is determined using energy-saving robust routing strategy, and underground monitoring data transmission is carried out;According to the model parameter obtained by optimization solution, the optimal underground rail car derailment detection model is formed, and underground rail car derailment detection is carried out.The present application selects the next hop transmission path node in combination with the sensing radius of sensor and mutual distance, obtains the optimal transmission path of monitoring data with the fastest transmission speed and lower interference degree, extracts effective feature information from historical operation data to carry out fusion operation on current operation data, obtains the probability value of underground rail car derailment at current time, and realizes rail car derailment detection.

Description

Technical Field

[0001] This invention relates to the field of derailment detection, and more particularly to a method and system for detecting derailment of underground railcars. Background Technology

[0002] Mining car derailment refers to the phenomenon of a mining car deviating from its track during underground mining operations such as coal and metal mines due to track failure or other reasons. Such accidents not only severely impact production but can also lead to casualties and property damage. Therefore, derailment detection of underground tracked mining cars is of great practical significance. Traditional underground mining car derailment monitoring mainly relies on personnel inspection, visual observation, and experience-based judgment. However, these methods suffer from personnel risks and reduced efficiency. Manual inspection may not be able to cover the entire mining area in a timely manner, and misjudgments are more likely to occur in dimly lit or high-risk areas. Furthermore, the accuracy of personnel-based derailment warnings is limited. Real-world environments place high demands on the automation of underground tracked mining car derailment detection. However, the mining environment in real-world operations is typically complex and harsh, with high levels of dust, high humidity, and abnormal gas concentrations. These factors significantly affect the normal operation of sensors, making them prone to failure and causing derailment detection errors. Therefore, there is an urgent need to develop environmentally adaptive underground environment derailment detection methods to improve the accuracy and efficiency of derailment monitoring. Based on the above background, this invention proposes a method and system for detecting derailment of underground railcars. By introducing sensor technology and environmental reasoning perception technology, it realizes real-time monitoring and anomaly detection of the status of underground minecars, which helps to provide early warning of derailment risks, reduce the probability of accidents, ensure the safety of mine workers, and improve production efficiency and economic benefits. Summary of the Invention

[0003] In view of this, the present invention provides a method for detecting derailment of underground railcars, the purpose of which is to: 1) calculate the sensing overlap ratio of different monitoring nodes based on the distance between the monitoring nodes underground, mark two nodes with a high sensing overlap ratio as mutually redundant nodes, realize the redundancy marking in the sensor network, form a redundant node connectivity graph, and calculate the neighboring node density based on the redundancy relationship of the nodes, and calculate the importance of the neighboring nodes based on the neighboring node density and the distance between the neighboring nodes and the monitoring data receiving center, and select the neighboring node with higher importance as the next hop node in the optimal transmission path of the monitoring data, realizing the combination of sensor sensing radius and mutual... 1) Optimize the sensor data transmission path for distance measurement; 2) Collect the running data sequence of underground railcars and split the data sequence to obtain the historical running data sequence and the current running data of underground railcars. Extract effective feature information from the historical running data and perform fusion calculation on the current running data to obtain the probability value of the underground railcar derailment at the current moment. During the model training process, update the learning rate according to the change value of the training objective function each time the parameters are iterated to realize the dynamic change of the learning rate. As the number of parameter iterations increases, the learning rate becomes smaller and smaller to avoid skipping the optimal parameters due to an excessively large learning rate or poor training efficiency due to an excessively small learning rate, thus ensuring the training efficiency of the model.

[0004] To achieve the above objectives, the present invention provides a method for detecting derailment of underground railcars, comprising the following steps:

[0005] S1: Collect the location information of the nodes to be monitored in the well to construct the well location network, search and calculate the distance between the nodes to be monitored and other nodes in the well location network one by one, and generate a redundant node connectivity graph based on the distance of different nodes to be monitored.

[0006] S2: Generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, determine the optimal transmission path for monitoring data of the nodes to be monitored using the energy-saving and robust routing strategy, and transmit downhole monitoring data based on the optimal transmission path.

[0007] Continuous collection of underground monitoring data forms a sequence of underground railcar operation data;

[0008] S3: Construct an underground railcar derailment detection model, wherein the underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output;

[0009] S4: Optimize and solve the parameters of the underground railcar derailment detection model, construct the optimal underground railcar derailment detection model based on the optimized model parameters, and perform underground railcar derailment detection. The optimization algorithm based on dynamic learning rate is the main implementation method for optimizing and solving the model parameters.

[0010] As a further improvement of the present invention:

[0011] Optionally, in step S1, the collection of location information of the nodes to be monitored downhole to construct an downhole location network, and the calculation of the distance between the node to be monitored and other nodes in the downhole location network, include:

[0012] The location information of the nodes to be monitored downhole is collected to construct a downhole location network, which is as follows:

[0013] R = {R n =(x n y n , z n |n∈[1,N]}

[0014] in:

[0015] R represents the downhole location network, which contains the location information of N nodes to be monitored.

[0016] R n This represents the location information of the monitored node numbered n in the downhole location network, (x n y n , z n (x) represents the location information of the node n to be monitored. n y n () represents the coordinates of the node to be monitored, n, in a ground coordinate system centered at the wellhead, z n This represents the vertical distance between the node n to be monitored and the plane where the wellhead is located;

[0017] The distances between the node to be monitored and other nodes in the downhole location network are calculated one by one, where the distance between node n and node i to be monitored is:

[0018]

[0019] in:

[0020] i∈[1,N],(x i y i , z i ) represents the location information of the node i to be monitored;

[0021] d ni This represents the distance between node n and node i to be monitored;

[0022] Sensors are deployed at the locations of the nodes to be monitored to collect underground monitoring data from the rail-guided mining trucks.

[0023] Optionally, the generation of a redundant node connectivity graph based on the distance between different nodes to be monitored in step S1 includes:

[0024] A redundant node connectivity graph is generated based on the distances between different nodes to be monitored. The generation process of the redundant node connectivity graph is as follows:

[0025] S11: Based on the sensing radius r of the sensors deployed at the location of the node to be monitored and the distance between different monitoring nodes, the sensing overlap ratio between different monitoring nodes is calculated, where the sensing overlap ratio between monitoring node n and monitoring node i is:

[0026]

[0027] in:

[0028] p ni This represents the sensing overlap ratio between node n and node i to be monitored;

[0029] S12: Mark two nodes to be monitored with a sensing overlap ratio higher than a preset ratio threshold as mutually redundant nodes;

[0030] S13: Construct a redundant node connectivity graph based on the labeling results of mutually redundant nodes. The nodes in the redundant node connectivity graph are nodes to be monitored, and the edges are the redundancy relationships between different nodes to be monitored. The degree of the node to be monitored in the redundant node connectivity graph is the number of nodes that are mutually redundant with the node to be monitored. Redundancy relationships include mutually redundant nodes and mutually non-redundant nodes.

[0031] S14: Select the sensor at the node to be monitored in the redundant node connected graph where the node degree is greater than the preset degree threshold and put it into sleep mode.

[0032] Optionally, step S2 involves generating an energy-efficient and robust routing strategy based on the redundant node connectivity graph, and using this strategy to determine the optimal transmission path for the monitoring data of the node to be monitored, including:

[0033] An energy-saving and robust routing strategy is generated based on the redundant node connectivity graph. This strategy is then used to determine the optimal transmission path for the monitoring data of the node to be monitored. The process for generating the optimal transmission path for the monitoring data of node n, combined with the energy-saving and robust routing strategy, is as follows:

[0034] S21: Initialize the set of nodes for the optimal transmission path of monitoring data for node n to be monitored, where the set of nodes for the optimal transmission path of monitoring data is empty;

[0035] S22: Add the node n to be monitored to the set of nodes along the optimal transmission path of the monitoring data, as L n(1) represents the first transmission node in the optimal transmission path of the monitoring data;

[0036] S23: Initialize k=1, calculate L in the optimal transmission path node set for monitoring data. n (k) The neighboring nodes to be monitored of the node to be monitored. * Neighborhood density:

[0037]

[0038] in:

[0039] L represents n (k) The neighboring nodes to be monitored of the node to be monitored. * The neighborhood density;

[0040] Num(k * ) represents the neighboring node k to be monitored. * The degree of a node in a redundant node connectivity graph;

[0041] Indicates the nearest monitored node k * The set of neighboring nodes; in this embodiment of the invention, if the distance between two nodes to be monitored is less than a preset distance threshold, then the two nodes to be monitored are neighboring nodes.

[0042] Indicates the nearest monitored node k * set of neighboring nodes The distance to node c to be monitored;

[0043] Indicates the nearest monitored node k * set of neighboring nodes The sensing overlap ratio of the node c to be monitored;

[0044] Sum(k * ) represents the set of neighboring nodes The number of nodes in;

[0045] S24: Calculate the nearest monitored node k * Importance:

[0046]

[0047] in:

[0048] ε = 0.0001;

[0049] α(k * ) represents the neighboring node k to be monitored. *The state of the sensor at the location, α(k) * ) = 1 indicates that the sensor is in operation, α(k) * ) = 0 indicates that the sensor is in sleep mode;

[0050] Indicates the nearest monitored node k * Distance to the monitoring data receiving center;

[0051] Indicates the nearest monitored node k * The importance of;

[0052] S25: Select the nearest monitored node with the highest importance and add it to the set of nodes along the optimal transmission path for monitoring data, as L. n (k+1), until the selected neighboring node to be monitored becomes the data receiving center, forming the optimal data transmission path L for node n to be monitored. n =(L n (1), L n (2), ..., L n (k), ..., end), where end represents the monitoring data receiving center. In this embodiment of the invention, the monitoring data receiving center is a node to be monitored that does not require the deployment of sensors.

[0053] Optionally, in step S2, underground monitoring data is transmitted based on the optimal transmission path to obtain an underground railcar operation data sequence, including:

[0054] An energy-efficient and robust routing strategy is used to determine the optimal transmission path for monitoring data of the node to be monitored, where the optimal transmission path for monitoring data of node n is L. n Based on the optimal transmission path for monitoring data, underground monitoring data transmission is performed to obtain the underground railcar operation data sequence, which is as follows:

[0055] H(t)=(h1(t),h2(t),...,h n (t), ..., h N (t)), t∈[t0, t end ]

[0056] in:

[0057] H(t) represents the underground railcar operation data sequence, t represents the time series information, t0 represents the initial time of underground railcar operation data acquisition, and t end This indicates the current moment of data collection for the underground railcar operation;

[0058] h n(t) represents the monitoring data collected by the sensor at the node n to be monitored, wherein the monitoring data includes the vibration amplitude and frequency of the ground where the sensor is located, as well as the ambient audio signal. In this embodiment of the invention, the time interval between adjacent data collection times is Δt.

[0059] Optionally, the construction of the underground railcar derailment detection model in step S3 includes:

[0060] A derailment detection model for underground railcars is constructed. The underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output. The underground railcar derailment detection model includes an input layer, a feature information calculation layer, a feature fusion layer combining feature information, and an output layer.

[0061] The input layer is used to receive the underground railcar operation data sequence and split the data sequence to obtain the historical operation data sequence of the underground railcar and the current operation data of the underground railcar;

[0062] The feature information calculation layer is used to calculate feature information from the historical operation data sequence of underground railcars;

[0063] The feature fusion layer, which combines feature information, is used to fuse and calculate the feature information and the current operating data of the underground railcar to generate the probability of the underground railcar derailing at the current moment.

[0064] The output layer is used to output the probability of the underground mine car derailing at the current moment.

[0065] Optionally, step S4 involves optimizing the parameters of the underground railcar derailment detection model, including:

[0066] The training dataset, consisting of the sequence of underground railcar operation data for Group U and the corresponding derailment status of the underground railcars, is as follows:

[0067] data={(H u ,β u |u∈[1,U]}

[0068] in:

[0069] (H u ,β u H represents the u-th training data set in the training dataset data. u This represents the sequence of underground railcar operation data for the u-th group, β u This represents the data sequence H of underground railcar operation. u The corresponding derailment situation of underground railcars, β u ={0,1},β u =0 indicates that the underground mine car did not derail, βu =1 indicates that the underground railcar has derailed;

[0070] The objective function for training the parameters of the underground mine car derailment detection model is as follows:

[0071]

[0072] in:

[0073] F(θ) represents the training objective function of the underground railcar derailment detection model parameters, and θ represents the underground railcar derailment detection model parameters to be trained. In this embodiment of the invention, the underground railcar derailment detection model parameters include the bias parameters of the feature information calculation layer, the convolution matrix parameters, and the convolution matrix parameters of the feature fusion layer.

[0074] This indicates the sequence of underground railcar operation data H. u As a model for detecting derailment of underground railcars based on model parameter θ, the model outputs the probability of derailment of underground railcars.

[0075] Generate initial parameters θ0 for the underground mine car derailment detection model. Set the current iteration number of the model parameters to z, with an initial value of 0 and a maximum value of Max. Then, the parameters of the underground mine car derailment detection model obtained in the z-th iteration are θ0. z ;

[0076] The parameters of the underground railcar derailment detection model are iterated, and the iterative formula is as follows:

[0077]

[0078]

[0079] in:

[0080] η z Let η0 represent the learning rate for the z-th iteration, where η0 = 0.6; in this embodiment of the invention, η0... -1 =0;

[0081] This represents the gradient of the training objective function; in this embodiment of the invention, This means taking the partial derivative of F(θ) and subtracting it into θ. z ;

[0082] Let z = z + 1, and iterate the parameters according to the iterative formula until the maximum number of iterations is reached, so as to obtain the optimized solution of the underground railcar derailment detection model parameters.

[0083] Optionally, in step S4, the optimal underground railcar derailment detection model is constructed based on the model parameters obtained from the optimization solution to perform underground railcar derailment detection, including:

[0084] The optimal underground railcar derailment detection model is constructed based on the model parameters obtained from the optimization solution. This model is then used to detect underground railcar derailment. The underground railcar derailment detection process for the underground railcar running data sequence H(t) is as follows:

[0085] S41: The input layer receives the underground railcar operation data sequence H(t) and splits the data sequence to obtain the underground railcar historical operation data sequence H(t′) and the underground railcar current operation data H(t′). end ):

[0086] H(t′)=(h1(t′),h2(t′),...,h n (t′), ..., h N (t′)), t′∈[t0, t end-1 ]

[0087] H(t end )=(h1(t end h2(t) end ), ..., h n (t end ), ..., h N (t end ))

[0088] in:

[0089] h n (t end ) indicates that the sensor at node n to be monitored is at time t end Monitoring data collected in real time;

[0090] S42: The feature information calculation layer calculates feature information for the historical operation data sequence H(t′) of the underground railcar.

[0091]

[0092] in:

[0093] b represents the bias parameter of the feature information calculation layer, and W represents the convolution matrix parameter in the feature information calculation layer.

[0094] f(t′) represents the feature information of the historical operation data sequence H(t′) of the underground railcar; in this embodiment of the invention, WH(t) e-1 ) = 1;

[0095] S43: The feature fusion layer combines the feature information f(t′) and the current operating data H(t′) of the underground railcar. end Perform fusion calculations to generate the current time t. end The probability of a mine car derailing in an underground mine is calculated using the following formula:

[0096] P end =Softmax(sigmoid(W * H(t end ))tanh(f(t′)))

[0097] in:

[0098] P end Represents the current time t end The probability of underground railcars derailing;

[0099] W * This represents the parameters of the convolution matrix in the feature fusion layer;

[0100] S44: Output layer outputs the current time t end The probability P of underground railcar derailment end If the probability P of a mine car derailing underground is... end If the probability exceeds the preset probability threshold, it indicates that the underground railcar is derailed.

[0101] To address the above problems, the present invention provides a derailment detection system for underground railcars, characterized in that the system comprises:

[0102] The downhole location network construction module is used to collect the location information of the nodes to be monitored in the downhole to construct the downhole location network, search and calculate the distance between the nodes to be monitored and other nodes in the downhole location network one by one, and generate a redundant node connectivity graph based on the distances of different nodes to be monitored.

[0103] The underground data monitoring module is used to generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, use the energy-saving and robust routing strategy to determine the optimal transmission path of the monitoring data of the node to be monitored, transmit the underground monitoring data based on the optimal transmission path of the monitoring data, and continuously collect underground monitoring data to form an underground railcar operation data sequence.

[0104] The mine car derailment detection device is used to construct an optimal underground mine car derailment detection model based on the model parameters obtained from the optimization solution, and to detect underground mine car derailment.

[0105] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0106] Memory, storing at least one instruction;

[0107] Communication interfaces enable communication between electronic devices; and

[0108] The processor executes the instructions stored in the memory to implement the above-described method for detecting derailment of underground railcars.

[0109] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the aforementioned method for detecting derailment of underground railcars.

[0110] Compared with existing technologies, this invention proposes a method for detecting derailment of underground railcars, which has the following advantages:

[0111] First, this scheme proposes a downhole data monitoring method. Based on the sensing radius r of the sensors deployed at the location of the monitored node and the distance between different monitoring nodes, the sensing overlap ratio between different monitored nodes is calculated, where the sensing overlap ratio between monitored node n and monitored node i is:

[0112]

[0113] Where: p ni The sensing overlap ratio between monitored node n and monitored node i is represented. Two monitored nodes with a sensing overlap ratio higher than a preset threshold are marked as mutually redundant nodes. A redundant node connectivity graph is constructed based on the marking results of mutually redundant nodes. In the redundant node connectivity graph, the nodes are monitored nodes, the edges are the redundancy relationships between different monitored nodes, and the degree of a monitored node in the redundant node connectivity graph is the number of nodes that are mutually redundant with it. Redundancy relationships include mutually redundant nodes and mutually non-redundant nodes. Sensors at monitored nodes with a degree greater than a preset degree threshold in the redundant node connectivity graph are selected for sleep processing. An energy-saving robust routing strategy is generated based on the redundant node connectivity graph. The optimal transmission path of the monitored data of the monitored node is determined using the energy-saving robust routing strategy. The process of generating the optimal transmission path of the monitored data of monitored node n in combination with the energy-saving robust routing strategy is as follows: Initialize the node set of the optimal transmission path of the monitored data of monitored node n, where the node set of the optimal transmission path of the monitored data is empty; add the monitored node n to the node set of the optimal transmission path of the monitored data as L. n (1) represents the first transmission node in the optimal transmission path of monitoring data; initialize k=1, and calculate the L in the node set of the optimal transmission path of monitoring data. n (k) The neighboring nodes to be monitored of the node to be monitored. * Neighborhood density:

[0114]

[0115] in: L represents n (k) The neighboring nodes to be monitored of the node to be monitored. * The neighborhood density; Num(k) * ) represents the neighboring node k to be monitored. * The degree of a node in a redundant node connected graph; Indicates the nearest monitored node k * The set of neighboring nodes; Indicates the neighboring node k to be monitored * set of neighboring nodes The distance to node c to be monitored; Indicates the neighboring node k to be monitored * set of neighboring nodes The sensing overlap ratio of the node to be monitored, c; Sum(k) * ) represents the set of neighboring nodes The number of nodes in the data; calculate the k of the nearest monitored nodes. * Importance:

[0116]

[0117] Where: ε=0.0001; α(k * ) represents the neighboring node k to be monitored. * The state of the sensor at the location, α(k) * ) = 1 indicates that the sensor is in operation, α(k) * ) = 0 indicates that the sensor is in sleep mode; Indicates the nearest monitored node k * Distance to the monitoring data receiving center; Indicates the nearest monitored node k * Importance; select the nearest monitored node with the highest importance and add it to the set of nodes along the optimal transmission path of the monitoring data, as L. n (k+1), until the selected neighboring node to be monitored becomes the data receiving center, forming the optimal data transmission path L for node n to be monitored. n =(L n (1), L n (2), ..., L n(k), ..., end), where end represents the monitoring data receiving center. This scheme calculates the sensing overlap ratio of different monitoring nodes based on the distance between the nodes to be monitored underground. Two nodes with a high sensing overlap ratio are marked as mutually redundant nodes, thus achieving redundancy marking in the sensor network and forming a redundant node connectivity graph. The neighborhood density of neighboring nodes is calculated based on the redundancy relationship of the nodes. The importance of neighboring nodes is calculated by combining the neighborhood density of neighboring nodes and the distance between neighboring nodes and the monitoring data receiving center. The neighboring nodes with higher importance are selected as the next-hop nodes in the optimal transmission path of the monitoring data, thus achieving the optimization solution for sensor data transmission paths that combine sensor sensing radii and mutual distances.

[0118] Meanwhile, this solution proposes an iterative method for model parameters and a method for detecting derailment of underground railcars by combining historical data. The method iterates the parameters of the underground railcar derailment detection model, with the iterative formula being:

[0119]

[0120]

[0121] Where: η z Let η0 represent the learning rate for the z-th iteration, where η0 = 0.6; The gradient of the training objective function is represented. Based on the model parameters obtained from the optimization solution, an optimal underground mine car derailment detection model is constructed. This optimal model is then used to detect underground mine car derailments. The derailment detection process for the underground mine car running data sequence H(t) is as follows: The input layer receives the underground mine car running data sequence H(t) and splits the data sequence to obtain the historical running data sequence H(t′) and the current running data H(t′). end ):

[0122] H(t′)=(h1(t′),h2(t′),...,h n (t′), ..., h N (t′)), t′∈[t0, t end-1 ]

[0123] H(t end )=(h1(t end h2(t) end ), ..., h n (t end ), ..., h N (t end ))

[0124] Where: h n (tend ) indicates that the sensor at node n to be monitored is at time t end The monitoring data is collected in real time; the feature information calculation layer calculates the feature information of the historical operation data sequence H(t′) of the underground railcar:

[0125]

[0126] Where: b represents the bias parameter of the feature information calculation layer, W represents the convolution matrix parameter in the feature information calculation layer; f(t′) represents the feature information of the historical operation data sequence H(t′) of the underground railcar; the feature fusion layer combines the feature information f(t′) and the current operation data H(t′) of the underground railcar. end Perform fusion calculations to generate the current time t. end The probability of a mine car derailing in an underground mine is calculated using the following formula:

[0127] P end =Softmax(sigmoid(W * H(t end ))tanh(f(t′)))

[0128] Where: P end Represents the current time t end The probability of underground railcars derailing; W * The convolution matrix parameters of the feature fusion layer are represented; the output layer outputs the current time t. end The probability P of underground railcar derailment end If the probability P of a mine car derailing underground is... end If the probability exceeds a preset probability threshold, it indicates that the underground mine car is derailed. This scheme collects and splits the underground mine car operation data sequence to obtain historical operation data sequences and current operation data. Effective feature information is extracted from the historical operation data and fused with the current operation data to obtain the probability value of derailment at the current moment. During model training, the learning rate is updated based on the change in the training objective function during each parameter iteration, achieving dynamic change of the learning rate. As the number of parameter iterations increases, the learning rate decreases, avoiding either excessively large learning rates that skip optimal parameters or excessively small learning rates that result in poor training efficiency, thus ensuring the model's training efficiency. Attached Figure Description

[0129] Figure 1 This is a flowchart illustrating a method for detecting derailment of underground railcars according to an embodiment of the present invention.

[0130] Figure 2 This is a functional block diagram of an underground railcar derailment detection system provided in an embodiment of the present invention;

[0131] Figure 2 In the middle: 100 Underground rail car derailment detection system, 101 Underground location network construction module, 102 Underground data monitoring module, 103 Mine car derailment detection device;

[0132] Figure 3 This is a schematic diagram of the structure of an electronic device for implementing a method for detecting derailment of underground railcars, according to an embodiment of the present invention.

[0133] Figure 3 In Chinese: 1. Electronic device; 10. Processor; 11. Memory; 12. Program; 13. Communication interface;

[0134] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0135] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0136] This application provides a method for detecting derailment of underground railcars. The execution entity of the method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0137] Example 1:

[0138] S1: Collect the location information of the nodes to be monitored in the well to construct the well location network, search and calculate the distance between the nodes to be monitored and other nodes in the well location network one by one, and generate a redundant node connectivity graph based on the distance of different nodes to be monitored.

[0139] In step S1, the location information of the nodes to be monitored downhole is collected to construct an downhole location network, and the distance between the node to be monitored and other nodes in the downhole location network is calculated one by one, including:

[0140] The location information of the nodes to be monitored downhole is collected to construct a downhole location network, which is as follows:

[0141] R = {R n =(x n y n , z n |n∈[1,N]}

[0142] in:

[0143] R represents the downhole location network, which contains the location information of N nodes to be monitored.

[0144] R n This represents the location information of the monitored node numbered n in the downhole location network, (x n y n , z n (x) represents the location information of the node n to be monitored. n y n () represents the coordinates of the node to be monitored, n, in a ground coordinate system centered at the wellhead, z n This represents the vertical distance between the node n to be monitored and the plane where the wellhead is located;

[0145] The distances between the monitored node and other nodes in the downhole location network are calculated one by one, where the distance between monitored node n and monitored node i is:

[0146]

[0147] in:

[0148] i∈[1,N],(x i y i , z i ) represents the location information of the node i to be monitored;

[0149] d ni This represents the distance between node n and node i to be monitored;

[0150] Sensors are deployed at the locations of the nodes to be monitored to collect underground monitoring data from the rail-guided mining trucks.

[0151] Step S1, which generates a redundant node connectivity graph based on the distances between different nodes to be monitored, includes:

[0152] A redundant node connectivity graph is generated based on the distances between different nodes to be monitored. The generation process of the redundant node connectivity graph is as follows:

[0153] S11: Based on the sensing radius r of the sensors deployed at the location of the node to be monitored and the distance between different monitoring nodes, the sensing overlap ratio between different monitoring nodes is calculated, where the sensing overlap ratio between monitoring node n and monitoring node i is:

[0154]

[0155] in:

[0156] p ni This represents the sensing overlap ratio between node n and node i to be monitored;

[0157] S12: Mark two nodes to be monitored with a sensing overlap ratio higher than a preset ratio threshold as mutually redundant nodes;

[0158] S13: Construct a redundant node connectivity graph based on the labeling results of mutually redundant nodes. The nodes in the redundant node connectivity graph are nodes to be monitored, and the edges are the redundancy relationships between different nodes to be monitored. The degree of the node to be monitored in the redundant node connectivity graph is the number of nodes that are mutually redundant with the node to be monitored. Redundancy relationships include mutually redundant nodes and mutually non-redundant nodes.

[0159] S14: Select the sensor at the node to be monitored in the redundant node connected graph where the node degree is greater than the preset degree threshold and put it into sleep mode.

[0160] S2: Generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, use the energy-saving and robust routing strategy to determine the optimal transmission path of monitoring data of the node to be monitored, transmit the monitoring data underground based on the optimal transmission path of the monitoring data, and continuously collect the underground monitoring data to form the underground railcar operation data sequence.

[0161] In step S2, an energy-saving and robust routing strategy is generated based on the redundant node connectivity graph. This strategy is then used to determine the optimal transmission path for the monitoring data of the node to be monitored, including:

[0162] An energy-saving and robust routing strategy is generated based on the redundant node connectivity graph. This strategy is then used to determine the optimal transmission path for the monitoring data of the node to be monitored. The process for generating the optimal transmission path for the monitoring data of node n, combined with the energy-saving and robust routing strategy, is as follows:

[0163] S21: Initialize the set of nodes for the optimal transmission path of monitoring data for node n to be monitored, where the set of nodes for the optimal transmission path of monitoring data is empty;

[0164] S22: Add the node n to be monitored to the set of nodes along the optimal transmission path of the monitoring data, as L n (1) represents the first transmission node in the optimal transmission path of the monitoring data;

[0165] S23: Initialize k=1, calculate L in the optimal transmission path node set for monitoring data. n (k) The neighboring nodes to be monitored of the node to be monitored. * Neighborhood density:

[0166]

[0167] in:

[0168] L representsn (k) The neighboring nodes to be monitored of the node to be monitored. * The neighborhood density;

[0169] Num(k * ) represents the neighboring node k to be monitored. * The degree of a node in a redundant node connected graph;

[0170] Indicates the nearest monitored node k * The set of neighboring nodes; in this embodiment of the invention, if the distance between two nodes to be monitored is less than a preset distance threshold, then the two nodes to be monitored are neighboring nodes.

[0171] Indicates the nearest monitored node k * set of neighboring nodes The distance to node c to be monitored;

[0172] Indicates the nearest monitored node k * set of neighboring nodes The sensing overlap ratio of the node c to be monitored;

[0173] Sum(k * ) represents the set of neighboring nodes The number of nodes in;

[0174] S24: Calculate the nearest monitored node k * Importance:

[0175]

[0176] in:

[0177] ε = 0.0001;

[0178] α(k * ) represents the neighboring node k to be monitored. * The state of the sensor at the location, α(k) * ) = 1 indicates that the sensor is in operation, α(k) * ) = 0 indicates that the sensor is in sleep mode;

[0179] Indicates the nearest monitored node k * Distance to the monitoring data receiving center;

[0180] Indicates the nearest monitored node k * The importance of;

[0181] S25: Select the nearest monitored node with the highest importance and add it to the set of nodes along the optimal transmission path for monitoring data, as L. n (k+1), until the selected neighboring node to be monitored becomes the data receiving center, forming the optimal data transmission path L for node n to be monitored. n =(L n (1), L n (2), ..., L n (k), ..., end), where end represents the monitoring data receiving center.

[0182] In step S2, underground monitoring data is transmitted based on the optimal transmission path to obtain the underground railcar operation data sequence, including:

[0183] An energy-efficient and robust routing strategy is used to determine the optimal transmission path for monitoring data of the node to be monitored, where the optimal transmission path for monitoring data of node n is L. n Based on the optimal transmission path for monitoring data, underground monitoring data transmission is performed to obtain the underground railcar operation data sequence, which is as follows:

[0184] H(t)=(h1(t),h2(t),...,h n (t), ..., h N (t)), t∈[t0, t end ]

[0185] in:

[0186] H(t) represents the underground railcar operation data sequence, t represents the time series information, t0 represents the initial time of underground railcar operation data acquisition, and t end This indicates the current moment of data collection for the underground railcar operation;

[0187] h n (t) represents the monitoring data collected by the sensor at node n to be monitored, including the vibration amplitude and frequency of the ground where the sensor is located, as well as the ambient audio signal.

[0188] S3: Construct an underground railcar derailment detection model. The underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output.

[0189] The S3 step involves constructing an underground railcar derailment detection model, including:

[0190] A derailment detection model for underground railcars is constructed. The underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output. The underground railcar derailment detection model includes an input layer, a feature information calculation layer, a feature fusion layer combining feature information, and an output layer.

[0191] The input layer is used to receive the underground railcar operation data sequence and split the data sequence to obtain the historical operation data sequence of the underground railcar and the current operation data of the underground railcar;

[0192] The feature information calculation layer is used to calculate feature information from the historical operation data sequence of underground railcars;

[0193] The feature fusion layer, which combines feature information, is used to fuse and calculate the feature information and the current operating data of the underground railcar to generate the probability of the underground railcar derailing at the current moment.

[0194] The output layer is used to output the probability of the underground mine car derailing at the current moment.

[0195] S4: Optimize and solve the parameters of the underground railcar derailment detection model, construct the optimal underground railcar derailment detection model based on the optimized model parameters, and perform underground railcar derailment detection.

[0196] Step S4 involves optimizing the parameters of the underground railcar derailment detection model, including:

[0197] The training dataset, consisting of the sequence of underground railcar operation data for Group U and the corresponding derailment status of the underground railcars, is as follows:

[0198] data={(H u ,β u |u∈[1,U]}

[0199] in:

[0200] (H u ,β u H represents the u-th training data set in the training dataset data. u This represents the sequence of underground railcar operation data for the u-th group, β u This represents the data sequence H of underground railcar operation. u The corresponding derailment situation of underground railcars, β u ={0,1},β u =0 indicates that the underground mine car did not derail, β u =1 indicates that the underground railcar has derailed;

[0201] The objective function for training the parameters of the underground mine car derailment detection model is as follows:

[0202]

[0203] in:

[0204] F(θ) represents the training objective function of the underground railcar derailment detection model parameters, and θ represents the underground railcar derailment detection model parameters to be trained. In this embodiment of the invention, the underground railcar derailment detection model parameters include the bias parameters of the feature information calculation layer, the convolution matrix parameters, and the convolution matrix parameters of the feature fusion layer.

[0205] This indicates the sequence of underground railcar operation data H. u As a model for detecting derailment of underground railcars based on model parameter θ, the model outputs the probability of derailment of underground railcars.

[0206] Generate initial parameters θ0 for the underground mine car derailment detection model. Set the current iteration number of the model parameters to z, with an initial value of 0 and a maximum value of Max. Then, the parameters of the underground mine car derailment detection model obtained in the z-th iteration are θ0. z ;

[0207] The parameters of the underground railcar derailment detection model are iterated, and the iterative formula is as follows:

[0208]

[0209]

[0210] in:

[0211] η z Let η0 represent the learning rate for the z-th iteration, where η0 = 0.6; in this embodiment of the invention, η0... -1 =0;

[0212] This represents the gradient of the training objective function; in this embodiment of the invention, This means taking the partial derivative of F(θ) and subtracting it into θ. z ;

[0213] Let z = z + 1, and iterate the parameters according to the iterative formula until the maximum number of iterations is reached, so as to obtain the optimized solution of the underground railcar derailment detection model parameters.

[0214] In step S4, the optimal underground mine car derailment detection model is constructed based on the model parameters obtained from the optimization solution to perform underground mine car derailment detection, including:

[0215] The optimal underground railcar derailment detection model is constructed based on the model parameters obtained from the optimization solution. This model is then used to detect underground railcar derailment. The underground railcar derailment detection process for the underground railcar running data sequence H(t) is as follows:

[0216] S41: The input layer receives the underground railcar operation data sequence H(t) and splits the data sequence to obtain the underground railcar historical operation data sequence H(t′) and the underground railcar current operation data H(te′). nd ):

[0217] H(t′)=(h1(t′),h2(t′),...,h n (t′), ..., h N (t′)), t′∈[t0, t end-1 ]

[0218] H(t end )=(h1(t end h2(t) end ), ..., h n (t end ), ..., h N (t end ))

[0219] in:

[0220] h n (t end ) indicates that the sensor at node n to be monitored is at time t end Monitoring data collected in real time;

[0221] S42: The feature information calculation layer calculates feature information for the historical operation data sequence H(t′) of the underground railcar.

[0222]

[0223] in:

[0224] b represents the bias parameter of the feature information calculation layer, and W represents the convolution matrix parameter in the feature information calculation layer.

[0225] f(t′) represents the feature information of the historical operation data sequence H(t′) of the underground railcar; in this embodiment of the invention, WH(t) e-1 ) = 1;

[0226] S43: The feature fusion layer combines the feature information f(t′) and the current operating data H(t′) of the underground railcar. end Perform fusion calculations to generate the current time t. endThe probability of a mine car derailing in an underground mine is calculated using the following formula:

[0227] P end =Softmax(sigmoid(W * H(t end ))tanh(f(t)))

[0228] in:

[0229] P end Represents the current time t end The probability of underground railcars derailing;

[0230] W * This represents the parameters of the convolution matrix in the feature fusion layer;

[0231] S44: Output layer outputs the current time t end The probability P of underground railcar derailment end If the probability P of a mine car derailing underground is... end If the probability exceeds the preset probability threshold, it indicates that the underground railcar is derailed.

[0232] Example 2:

[0233] like Figure 2 The diagram shown is a functional block diagram of an underground railcar derailment detection system provided in an embodiment of the present invention, which can implement the underground railcar derailment detection method in Embodiment 1.

[0234] The underground mine car derailment detection system 100 of this invention can be installed in an electronic device. Depending on the functions implemented, the underground mine car derailment detection system may include an underground location network construction module 101, an underground data monitoring module 102, and a mine car derailment detection device 103. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and which are stored in the memory of the electronic device.

[0235] The downhole location network construction module 101 is used to collect the location information of the nodes to be monitored in the downhole to construct the downhole location network, search and calculate the distance between the nodes to be monitored and other nodes in the downhole location network one by one, and generate a redundant node connectivity graph based on the distance of different nodes to be monitored.

[0236] The underground data monitoring module 102 is used to generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, use the energy-saving and robust routing strategy to determine the optimal transmission path of the monitoring data of the node to be monitored, transmit the underground monitoring data based on the optimal transmission path of the monitoring data, and continuously collect the underground monitoring data to form an underground railcar operation data sequence.

[0237] The mine car derailment detection device 103 is used to construct an optimal underground rail mine car derailment detection model based on the model parameters obtained from the optimization solution, and to detect underground rail mine car derailment.

[0238] In detail, the modules in the underground railcar derailment detection system 100 described in this embodiment of the invention employ the same methods as described above during use. Figure 1 The method used is the same as the one described in the article for detecting derailment of underground railcars, and it can produce the same technical effect, so it will not be repeated here.

[0239] Example 3:

[0240] like Figure 3 The diagram shown is a structural schematic of an electronic device for implementing a method for detecting derailment of underground railcars, according to an embodiment of the present invention.

[0241] The electronic device 1 may include a processor 10, a memory 11, a communication interface 13 and a bus, and may also include a computer program, such as program 12, stored in the memory 11 and executable on the processor 10.

[0242] The memory 11 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. Furthermore, the memory 11 can include both internal and external storage units of the electronic device 1. The memory 11 can be used not only to store application software and various types of data installed on the electronic device 1, such as the code of program 12, but also to temporarily store data that has been output or will be output.

[0243] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (such as program 12 for detecting derailment of underground railcars) and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0244] The communication interface 13 may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the electronic device 1 and other electronic devices, and to enable communication between internal components of the electronic device.

[0245] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0246] Figure 3 Only electronic devices with components are shown; those skilled in the art will understand that... Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0247] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0248] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual user interface.

[0249] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0250] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0251] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0252] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for detecting derailment of underground railcars, characterized in that, The method includes: S1: Collect the location information of the nodes to be monitored in the well to construct the well location network, search and calculate the distance between the nodes to be monitored and other nodes in the well location network one by one, and generate a redundant node connectivity graph based on the distance of different nodes to be monitored. S2: Generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, use the energy-saving and robust routing strategy to determine the optimal transmission path of monitoring data of the node to be monitored, transmit the monitoring data underground based on the optimal transmission path of the monitoring data, and continuously collect the underground monitoring data to form the underground railcar operation data sequence. The step of generating an energy-efficient and robust routing strategy based on the redundant node connectivity graph, and using the energy-efficient and robust routing strategy to determine the optimal transmission path for monitoring data of the node to be monitored, includes: An energy-saving and robust routing strategy is generated based on the redundant node connectivity graph. This strategy is then used to determine the optimal transmission path for the monitoring data of the node to be monitored. The process for generating the optimal transmission path for the monitoring data of node n, combined with the energy-saving and robust routing strategy, is as follows: S21: Initialize the set of nodes for the optimal transmission path of monitoring data for node n to be monitored, where the set of nodes for the optimal transmission path of monitoring data is empty; S22: Add the node n to be monitored to the set of nodes along the optimal transmission path of the monitoring data, as... This indicates the first transmission node in the optimal transmission path for monitoring data; S23: Initialize k=1, calculate the optimal transmission path node set for monitoring data. The neighboring nodes of the corresponding node to be monitored Neighborhood density: in: express The neighboring nodes of the corresponding node to be monitored The neighborhood density; Indicates the neighboring nodes to be monitored The degree of a node in a redundant node connectivity graph; Indicates the neighboring nodes to be monitored The set of neighboring nodes; Indicates the neighboring nodes to be monitored set of neighboring nodes The distance to node c to be monitored; Indicates the neighboring nodes to be monitored set of neighboring nodes The sensing overlap ratio of the node c to be monitored; Represents the set of neighboring nodes The number of nodes in; S24: Calculate the nearest monitored nodes Importance: in: ； Indicates the neighboring nodes to be monitored The status of the sensor at the location. This indicates that the sensor is in operation. 0 indicates that the sensor is in sleep mode; Indicates the neighboring nodes to be monitored Distance to the monitoring data receiving center; Indicates the neighboring nodes to be monitored The importance of; S25: Select the nearest monitoring node with the highest importance and add it to the set of nodes along the optimal transmission path for monitoring data, as... The optimal transmission path for monitoring data of node n is formed until the selected nearest neighboring node to be monitored becomes the data receiving center. ,in Indicates the monitoring data receiving center; S3: Construct an underground railcar derailment detection model. The underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output. S4: Optimize and solve the parameters of the underground railcar derailment detection model, construct the optimal underground railcar derailment detection model based on the optimized model parameters, and perform underground railcar derailment detection.

2. The method for detecting derailment of underground railcars as described in claim 1, characterized in that, In step S1, the location information of the nodes to be monitored downhole is collected to construct an downhole location network, and the distance between the node to be monitored and other nodes in the downhole location network is calculated one by one, including: The location information of the nodes to be monitored downhole is collected to construct a downhole location network, which is as follows: in: This represents the downhole location network, which contains the location information of N nodes to be monitored. This represents the location information of the node to be monitored, numbered n, in the downhole location network. This represents the location information of the node n to be monitored. This represents the coordinates of the node n to be monitored in a ground coordinate system centered at the wellhead. This represents the vertical distance between the node n to be monitored and the plane where the wellhead is located; The distances between the node to be monitored and other nodes in the downhole location network are calculated one by one, where the distance between node n and node i to be monitored is: in: , This represents the location information of the node i to be monitored; This represents the distance between node n and node i to be monitored; Sensors are deployed at the locations of the nodes to be monitored to collect underground monitoring data from the rail-guided mining trucks.

3. The method for detecting derailment of underground railcars as described in claim 2, characterized in that, Step S1, which generates a redundant node connectivity graph based on the distances between different nodes to be monitored, includes: A redundant node connectivity graph is generated based on the distances between different nodes to be monitored. The generation process of the redundant node connectivity graph is as follows: S11: Based on the sensing radius r of the sensors deployed at the location of the node to be monitored and the distance between different monitoring nodes, the sensing overlap ratio between different monitoring nodes is calculated, where the sensing overlap ratio between monitoring node n and monitoring node i is: in: This represents the sensing overlap ratio between node n and node i to be monitored; S12: Mark two nodes to be monitored with a sensing overlap ratio higher than a preset ratio threshold as mutually redundant nodes; S13: Construct a redundant node connectivity graph based on the labeling results of mutually redundant nodes. The nodes in the redundant node connectivity graph are nodes to be monitored, and the edges are the redundancy relationships between different nodes to be monitored. The degree of the node to be monitored in the redundant node connectivity graph is the number of nodes that are mutually redundant with the node to be monitored. Redundancy relationships include mutually redundant nodes and mutually non-redundant nodes. S14: Select the sensor at the node to be monitored in the redundant node connected graph where the node degree is greater than the preset degree threshold and put it into sleep mode.

4. The method for detecting derailment of underground railcars as described in claim 1, characterized in that, In step S2, underground monitoring data is transmitted based on the optimal transmission path to obtain the underground railcar operation data sequence, including: An energy-efficient and robust routing strategy is used to determine the optimal transmission path for monitoring data of the node to be monitored, where the optimal transmission path for monitoring data of node n is... Based on the optimal transmission path for monitoring data, underground monitoring data transmission is performed to obtain the underground railcar operation data sequence, which is as follows: in: This represents the data sequence of underground railcar operation, where t represents time-series information. This indicates the initial time of data collection for the underground railcar operation. This indicates the current moment of data collection for the underground railcar operation; This represents the monitoring data collected by the sensor at node n to be monitored. The monitoring data includes the vibration amplitude and frequency of the ground where the sensor is located, as well as the ambient audio signal.

5. The method for detecting derailment of underground railcars as described in claim 1, characterized in that, The S3 step involves constructing an underground railcar derailment detection model, including: A derailment detection model for underground railcars is constructed. The underground railcar derailment detection model takes the underground railcar operation data sequence as input and the underground railcar derailment probability as output. The underground railcar derailment detection model includes an input layer, a feature information calculation layer, a feature fusion layer combining feature information, and an output layer. The input layer is used to receive the underground railcar operation data sequence and split the data sequence to obtain the historical operation data sequence of the underground railcar and the current operation data of the underground railcar; The feature information calculation layer is used to calculate feature information from the historical operation data sequence of underground railcars; The feature fusion layer, which combines feature information, is used to fuse and calculate the feature information and the current operating data of the underground railcar to generate the probability of the underground railcar derailing at the current moment. The output layer is used to output the probability of the underground mine car derailing at the current moment.

6. The method for detecting derailment of underground railcars as described in claim 5, characterized in that, Step S4 involves optimizing the parameters of the underground railcar derailment detection model, including: The training dataset, consisting of the sequence of underground railcar operation data for Group U and the corresponding derailment status of the underground railcars, is as follows: in: This represents the u-th training data set in the training dataset data. This represents the sequence of underground railcar operation data for the u-th group. This represents the sequence of underground railcar operation data. The corresponding derailment situation of underground railcars. , This indicates that the underground mine car did not derail. This indicates that the underground mine car has derailed; The objective function for training the parameters of the underground mine car derailment detection model is as follows: in: The training objective function represents the parameters of the underground mine car derailment detection model. Indicates the parameters of the underground railcar derailment detection model to be trained; This indicates the sequence of underground railcar operation data. As based on model parameters In the underground railcar derailment detection model, the model outputs the probability of the underground railcar derailment. Generate initial parameters for the underground mine car derailment detection model. Let z be the current iteration number of the model parameters, with an initial value of 0 and a maximum value of Max. Then, the parameters of the underground railcar derailment detection model obtained in the z-th iteration are: ; The parameters of the underground railcar derailment detection model are iterated, and the iterative formula is as follows: in: This represents the learning rate in the z-th iteration. ; This represents the gradient of the training objective function; Let z = z + 1, and iterate the parameters according to the iterative formula until the maximum number of iterations is reached, so as to obtain the optimized solution of the underground railcar derailment detection model parameters.

7. The method for detecting derailment of underground railcars as described in claim 6, characterized in that, In step S4, the optimal underground mine car derailment detection model is constructed based on the model parameters obtained from the optimization solution to perform underground mine car derailment detection, including: The optimal underground mine car derailment detection model is constructed based on the model parameters obtained from the optimization solution. This model is then used to detect underground mine car derailments. The data sequence of the underground mine car's operation is also included. The derailment detection process for underground mine cars is as follows: S41: Input layer receives the underground railcar operation data sequence. The data sequence was then split to obtain the historical operation data sequence of the underground railcar. and current operating data of underground railcars : in: This indicates that the sensor at node n to be monitored is in Monitoring data collected in real time; S42: Feature information calculation layer for historical operation data sequence of underground railcars Perform feature information calculation: in: b represents the bias parameter of the feature information calculation layer, and W represents the convolution matrix parameter in the feature information calculation layer. This represents the historical operation data sequence of underground railcars. Feature information; S43: Feature fusion layer combining feature information and current operating data of underground railcars Perform fusion calculations to generate the current time step. The probability of a mine car derailing in an underground mine is calculated using the following formula: in: Indicates the current time The probability of underground railcars derailing; This represents the parameters of the convolution matrix in the feature fusion layer; S44: Output layer outputs the current time. The probability of underground mine cars derailing If the probability of a mine car derailing underground is... If the probability exceeds the preset probability threshold, it indicates that the underground railcar is derailed.

8. A derailment detection system for underground mine cars, characterized in that, The system includes: The downhole location network construction module is used to collect the location information of the nodes to be monitored in the downhole to construct the downhole location network, search and calculate the distance between the nodes to be monitored and other nodes in the downhole location network one by one, and generate a redundant node connectivity graph based on the distances of different nodes to be monitored. The underground data monitoring module is used to generate an energy-saving and robust routing strategy based on the redundant node connectivity graph, use the energy-saving and robust routing strategy to determine the optimal transmission path of the monitoring data of the node to be monitored, transmit the underground monitoring data based on the optimal transmission path of the monitoring data, and continuously collect underground monitoring data to form an underground railcar operation data sequence. A mine car derailment detection device is used to construct an optimal underground rail mine car derailment detection model based on the model parameters obtained by optimization solution, so as to realize an underground rail mine car derailment detection method as described in any one of claims 1-7.

Images