Rock mass risk assessment model training method, electronic device and readable storage medium

Abstract

Embodiments of the present application disclose a rock mass risk assessment model training method, an electronic device and a readable storage medium. The rock mass risk assessment model is trained by the rock mass risk assessment model training method. In subsequent input of to-be-evaluated data through an input layer of the rock mass risk assessment model, after processing through a hidden layer and an output layer, a risk assessment result is output. The rock mass engineering system can be accurately perceived in a comprehensive manner, so that relevant personnel can learn the risk of the rock mass engineering, and problems such as lack of risk perception and difficulty in timely and accurate early warning of the scene in the related art are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a rock mass risk assessment model training method, an electronic device and a readable storage medium. BACKGROUND

[0002] With the development of economy, the construction quantity of deep tunnels, chambers and other projects has increased rapidly. Under the conditions of "three high and one disturbance", the projects often face engineering disasters such as tunnel collapse, rock burst, large deformation of soft rock, sudden mud gushing and water gushing, which brings great challenges to engineering construction. Therefore, it is particularly important to form a scientific and reasonable monitoring scheme under different geological conditions, to master the dynamic information of the surrounding rock supporting structure within the volume range, and to build a disaster warning system for geotechnical engineering. However, there are problems such as low sensor adaptability, mutual independence between devices, insufficient depth of monitoring data mining, lack of risk perception and the like in the existing monitoring technology, which makes it difficult to timely and accurately warn the site. Therefore, how to assess the risk of the project has become a problem to be solved. SUMMARY

[0003] Embodiments of the present application provide a rock mass risk assessment model training method, an electronic device and a readable storage medium, wherein the rock mass risk assessment model trained by the rock mass risk assessment model training method can assess the risk of the rock mass of the project according to the rock mass data, realize accurate perception of the rock mass of the project in a systematic and all-round way, and enable relevant personnel to learn about the risk of the rock mass of the project, thereby avoiding the problem in the related art that the risk of the rock mass cannot be assessed, resulting in failure to warn the site.

[0004] Other characteristics and advantages of the present application will become apparent from the following detailed description, or will be learned by practice of the present application.

[0005] According to an aspect of an embodiment of the present application, a rock mass risk assessment model training method is provided, which comprises: initializing an initial rock mass risk assessment model according to input rock mass data, the initial rock mass risk assessment model comprising an input layer, a hidden layer and an output layer; determining a hidden layer output corresponding to the hidden layer according to the rock mass data, and determining an output layer output corresponding to the output layer according to the hidden layer output; performing prediction network error calculation according to the output layer output and the rock mass data; updating the initial rock mass risk assessment model according to the prediction network error to obtain a rock mass risk assessment model, which is used for risk assessment of the rock mass.

[0006] In some examples, the rock mass data comprises: excavated region data; initializing an initial rock mass risk assessment model according to input rock mass data comprises: dimensioning a spatial vector corresponding to the excavated region data; initializing the number of nodes of the input layer, the number of nodes of the output layer, and the number of nodes of the initial hidden layer based on the spatial vector.

[0007] In some examples, after initializing the initial rock mass risk assessment model according to the input rock mass data, the method further comprises: initializing first connection weights between the input layer and the hidden layer, and initializing second connection weights between the hidden layer and the output layer; initializing thresholds of the hidden layer and the output layer, and giving a neuron excitation function corresponding to the hidden layer.

[0008] In some examples, determining a hidden layer output corresponding to the hidden layer according to the rock mass data comprises: inputting the excavated region data, the first connection weights, and the thresholds of the hidden layer into a first preset formula to determine the hidden layer output; the first preset formula is:

[0009] (j = 1, 2,..., l),

[0010] wherein, l is the number of nodes of the hidden layer, g is an excitation function corresponding to the hidden layer, a i is a threshold of the i-th node of the hidden layer, ω ij is the first connection weight, (k-date i ) is a data matrix at the i-th monitoring point in the excavated region data, H j is the output of the j-th node of the hidden layer.

[0011] In some examples, determining an output layer output corresponding to the output layer according to the hidden layer output comprises: inputting the hidden layer output, the second connection weights, and the thresholds of the output layer into a second preset formula to determine the output layer output, the second preset formula is:

[0012] (k = 1, 2,..., m),

[0013] wherein, (uk-date) is the output layer output, H j is the output of the j-th node of the hidden layer, k is the number of hidden nodes, ω jk is the second connection weight, b k is a threshold of the k-th node of the output layer.

[0014] In some examples, the prediction network error calculation is performed according to the output layer output and the rock mass data, including: inputting the output layer output and the rock mass data into a third preset formula to obtain the prediction network error, the third preset formula being as follows:

[0015] e k = (UK-date k ) - (uk-date k ), (k = 1, 2,..., m)

[0016] wherein e k is the prediction network error at the kth node of the output layer, (UK-date k ) is the excavated area data, and (uk-date k ) is the predicted output value of the kth node of the output layer in the excavated area.

[0017] In some examples, the initial rock mass risk assessment model is updated according to the prediction network error, including: updating the first connection weight and the second connection weight according to the prediction network error; and updating the threshold values of the hidden layer and the output layer according to the prediction network error.

[0018] In some examples, after the initial rock mass risk assessment model is updated according to the prediction network error to obtain a rock mass risk assessment model, the method further includes: inputting to-be-evaluated data into the rock mass risk assessment model to obtain predicted data corresponding to the to-be-evaluated data; and generating a stress nephogram in front of a working face based on the predicted data, the stress nephogram being used to judge the risk of the to-be-evaluated data.

[0019] According to an aspect of an embodiment of the present application, an electronic device is provided, including one or more processors; a storage device configured to store one or more computer programs, when the one or more computer programs are executed by the one or more processors, the electronic device is caused to implement the method as described above.

[0020] According to an aspect of an embodiment of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program, when the computer program is executed by a processor of an electronic device, the electronic device is caused to implement the method as described above.

[0021] The rock mass risk assessment model training method provided in the example comprises: initializing an initial rock mass risk assessment model according to input rock mass data, wherein the initial rock mass risk assessment model comprises an input layer, a hidden layer and an output layer; determining a hidden layer output corresponding to the hidden layer according to the rock mass data, and determining an output layer output corresponding to the output layer according to the hidden layer output; performing prediction network error calculation according to the output layer output and the rock mass data; updating the initial rock mass risk assessment model according to the prediction network error to obtain a rock mass risk assessment model, wherein the rock mass risk assessment model is used for risk assessment of a rock mass. The rock mass risk assessment model trained by the above method can accurately perceive the rock mass engineering systemically and comprehensively, so that relevant personnel can learn the risk of the rock mass engineering, and the problem that the rock mass cannot be assessed for risk in the related art, resulting in the problem that the site cannot be warned, is avoided.

[0022] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0023] The drawings incorporated into the specification and forming a part of the specification, show embodiments consistent with the present application, and together with the specification, serve to explain the principles of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without creative labor.

[0024] Figure 1 The basic flowchart of the rock mass risk assessment model training method provided for the first embodiment;

[0025] Figure 2 The basic structure schematic diagram of the vibrating string type wireless sensor provided for the second embodiment;

[0026] Figure 3 The structure explosion schematic diagram of the vibrating string type wireless sensor provided for the second embodiment;

[0027] Figure 4 The basic schematic diagram of the vibrating string type wireless sensor communication mode provided for the second embodiment;

[0028] Figure 5 The basic schematic diagram of the monitoring and arrangement scheme determination flow provided for the second embodiment;

[0029] Figure 6 The three-dimensional data generation principle schematic diagram provided for the second embodiment;

[0030] Figure 7 A support structure deformation early warning schematic diagram provided for the second embodiment;

[0031] Figure 8 A digital twin result schematic diagram provided for the second embodiment;

[0032] Figure 9 A neural network structure perception prediction model basic schematic diagram provided for the second embodiment;

[0033] Figure 10 A stress cloud diagram of the front of the working face generated based on the above prediction data provided for the second embodiment;

[0034] Figure 11 A structural schematic diagram of a computer system of an electronic device provided for the embodiment. DETAILED DESCRIPTION

[0035] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, the same numbers refer to the same or similar elements throughout the drawings. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they only represent examples of apparatuses and methods consistent with some aspects of the present application, as detailed in the appended claims.

[0036] The block diagrams shown in the drawings are merely functional entities, and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0037] The flowcharts shown in the drawings are merely exemplary descriptions, and do not necessarily include all contents and operations, nor do they necessarily be executed in the described order. For example, some operations can be further divided, and some operations can be combined or partially combined, so the actual execution order can be changed according to the actual situation.

[0038] It should also be noted that "multiple" as mentioned in the present application refers to two or more. The association relationship of "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the three cases of A alone, A and B together, and B alone. The character " / " generally represents an "or" relationship between the associated objects before and after it.

[0039] Embodiment I

[0040] To solve the above technical problems, the present application provides a rock mass risk assessment model training method, as shown in Figure 1As shown, the method comprises:

[0041] S101, initializing an initial rock mass risk assessment model according to input rock mass data, the initial rock mass risk assessment model comprising: an input layer, a hidden layer and an output layer;

[0042] S102, determining a hidden layer output corresponding to the hidden layer according to the rock mass data, and determining an output layer output corresponding to the output layer according to the hidden layer output;

[0043] S103, performing prediction network error calculation according to the output layer output and the rock mass data;

[0044] S104, updating the initial rock mass risk assessment model according to the prediction network error to obtain a rock mass risk assessment model, the rock mass risk assessment model being used for risk assessment of a rock mass.

[0045] It can be understood that the rock mass risk assessment model training method provided in the embodiment is applied to a terminal, and the terminal can be implemented in various forms. For example, the terminal described in the present application can include mobile terminals such as mobile phones, tablet computers, notebook computers, palmtop computers, personal digital assistants (Personal Digital Assistant, PDA), portable media players (Portable Media Player, PMP), navigation devices, wearable devices, smart bracelets, pedometers, etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0046] Among them, the above-mentioned rock mass data is the supporting structure data of the rock mass obtained by sensors and the like, and the rock mass includes but is not limited to tunnel rock mass; specifically, the sensor is arranged in the tunnel to obtain the supporting structure data of the tunnel rock mass.

[0047] Among them, the sensor is a vibrating wire wireless sensor, which includes but is not limited to: a bolt, a cover plate, a PCB board, a power supply, an information receiving device, an automatic acquisition module, a module box, and a wiring cap.

[0048] The relationship between each part: the PCB board, the power supply, the information receiving device, and the automatic acquisition module are all fixed and packaged in the module box; the module box and the cover plate are connected by the bolt into a closed whole; the wiring cap is on one side of the module box, which is a reserved hole position.

[0049] The role of each part: the PCB board is responsible for providing various electronic components such as integrated circuits, realizing the wiring and electrical connection between various electronic components such as integrated circuits, and enabling the normal operation of each working element; the information receiving device is responsible for converting the received vibrating string sensor fluctuation signal into an electrical signal that the system can recognize; the automatic acquisition module acquires and processes the vibrating frequency change of the vibrating string sensor; the wiring cap is a reserved hole for external vibrating string sensors.

[0050] The data acquisition and wireless communication mode with the vibrating string wireless sensing module as the core has a flowchart as Figure 4 The vibrating string sensor, which is mature in technology and low in cost, can monitor anchor shaft force, steel stress, and concrete strain gauge. The sensor is connected to the wireless sensing module through a signal line. After the installation and debugging of the sensing device are completed, it is buried in the primary lining structure to prevent disturbance caused by factors such as blasting and construction on site. The collected data is transmitted to the DTU through the Lora wireless communication mode. The DTU transmits the data to the cloud server through the GPRS network signal. Finally, the data from the cloud server is displayed on the web-based monitoring platform and the mobile app.

[0051] To better obtain geotechnical data, the method provided in the embodiment further includes determining a geotechnical data monitoring scheme. Specifically, the monitoring scheme is determined by AHP multi-level analysis. The hierarchical structure model mainly includes the target layer, the element layer, the standard layer, and the scheme layer. The target layer is first refined to the standard layer according to the element layer, and the optimal scheme in the scheme layer is derived according to the standard layer, thereby determining the most reasonable arrangement scheme. The target layer of the patent is the monitoring arrangement scheme; the element layer is the surrounding rock grade u1, the tunnel burial depth u2, the hydrological condition u3, the adverse geology u4, the support system u5, and the excavation step distance u6 elements; taking the acquisition of the geotechnical data of the tunnel as an example, the scheme layer is the arrangement scheme, mainly including the arrangement position (① three positions of the vault and the left and right sidewalls, ② five positions of the vault, the left and right arch feet, and the left and right sidewalls), the cross-section distance (15 m, 25 m), and the monitoring frequency (3 h / time, 12 h / time), totaling eight arrangement and monitoring schemes. The method provides a reasonable, effective, scientific, and feasible monitoring scheme for the design of roadways and tunnels under different working conditions. The arrangement position, frequency, and step distance of the sensor are determined according to the actual situation of each factor, Q = (x, f, d...), where Q represents the scheme, x represents the arrangement position, f represents the monitoring frequency, and d represents the step distance. The selection and calculation process of the scheme by the AHP is as follows:

[0052] (1) Establish a hierarchical model;

[0053] (2) Construct an element importance comparison matrix;

[0054] (3) Consistency check;

[0055] (4) Overall ranking and scheme selection. Since the analytic hierarchy process is highly subjective when constructing the element importance comparison matrix, 1 to 3 schemes can be obtained by comprehensively considering the opinions of relevant parties when making scheme arrangements. In order to determine the rationality of the scheme, the Topsis comprehensive evaluation method is used to make a decision on the scheme.

[0056] The evaluation indicators for the monitoring scheme are: sensor cost, amount of data collected, and impact on construction. The process is as follows: (1) Establish the scheme decision evaluation matrix r i×3 (2) After normalizing the original data matrix, perform standardization b. i×3 (3) Considering the comprehensive weighting of subjective and objective factors, the loss evaluation matrix C = wb is obtained. i3 (4) Close reading calculation, where the positive ideal solution is Negative ideal solution is Proximity Calculate Q for different schemes respectively i The corresponding V i + The optimal solution is Q = max(V) i + ).

[0057] The weights w of the indicators in the calculation process are mainly determined by the entropy weight method, based on the standardized evaluation matrix b. i×3 The entropy value of the evaluation index is calculated. in but

[0058] After determining the geotechnical data monitoring scheme, the method further includes: installing sensors and setting up a wireless sensor module data transmission system according to the sensor arrangement method in the geotechnical data monitoring scheme.

[0059] The mechanical information of the support structure, such as the axial force of anchor bolts and the stress of reinforcing bars, is obtained from monitoring points. Based on the inverse distance weighted (IDW) difference method, the information of unknown points is inferred from the information of known points, and a three-dimensional mechanical model of the support structure is constructed. The information of the unknown measuring point z0 is comprehensively determined based on the five closest known measuring points z1 to z5. in The final three-dimensional surface region data matrix of the support structure is formed:

[0060]

[0061] Where a is the a-th plane of the tunnel in the x-direction in three-dimensional coordinates;

[0062] i represents the i-th plane in the y-direction in three-dimensional coordinates;

[0063] j is the jth plane in the z direction under three-dimensional coordinates;

[0064] is the stress value of the support structure at coordinates i, z, a under three-dimensional coordinates. In addition, to prevent false alarms caused by individual sensor failure and poor measurement accuracy during monitoring, based on the three-dimensional data matrix S, regional evaluation rules are established to ensure the accuracy of the warning from a three-dimensional space perspective. The rules are as follows:

[0065] (1) Establish a warning principle based on hierarchical double control, which is divided into three levels of yellow warning, orange warning, and red warning. Double control indicators refer to the change amount (reinforcement yield strength, anchor yield strength) and the change rate (1 kN / h). When both double control indicators exceed the monitoring measurement control value, or the measured change rate increases sharply, it is a red warning. When both double control indicators exceed 85% of the monitoring measurement control value, or one of the double control indicators exceeds the monitoring measurement control value, it is an orange warning. When both double control indicators exceed 70% of the monitoring measurement control value, or one of the double control indicators exceeds 85% of the monitoring measurement control value, it is a yellow warning (specific numerical values).

[0066] (2) After the warning occurs, to ensure the accuracy of the warning, regional evaluation criteria are established, and the judgment rule is as follows: taking the warning point as the center, search the space range of 5x5x5, when the number of warning points g(x) in the space range meets then this point is identified as a warning point, otherwise it is treated as a false alarm point.

[0067] (3) Transmit the field data to the cloud platform, generate a three-dimensional model of the support structure based on digital twin technology, and view it on the monitoring and warning platform. Through VR virtual technology, the virtual information of the support structure stress is superimposed with the surrounding rock mass, realizing the interaction between virtual and reality. Real-time visualization model can be viewed through VR glasses, which can more intuitively and visually judge the stress change of the support structure, and truly realize the artificial intelligence of monitoring and warning.

[0068] In some examples of the embodiment, the rock mass data includes: excavated area data; initializing an initial rock mass risk assessment model according to input rock mass data, including: through the spatial vector dimension corresponding to the excavated area data; initializing the number of nodes of the input layer, the number of nodes of the output layer, and the number of nodes of the initial hidden layer based on the spatial vector dimension. Taking tunnel rock mass data as an example, the excavated area data is the data transmitted by the sensors in the tunnel area that has started construction, which includes but is not limited to the data sequence of the surrounding rock grade u1, the tunnel burial depth u2, the hydrological condition u3, the adverse geology u4, the support system u5, the excavation step distance u6, etc. in the data matrix S. i

[0069] ​The number of nodes of the input layer is initialized according to the spatial vector dimension corresponding to the excavated region data, specifically, the number of nodes of the input layer is initialized to the spatial vector dimension corresponding to the excavated region data, the number of nodes m of the output layer (the type number of the support structure stress data to be output) i ), and the number of nodes of the hidden layer is initialized to l, where l is set by relevant personnel according to actual needs;

[0070] In some examples of the embodiment, initializing the initial rock mass risk assessment model according to the input rock mass data further includes: initializing first connection weights between the input layer and the hidden layer, and initializing second connection weights between the hidden layer and the output layer; initializing thresholds of the hidden layer and the output layer, and giving a neuron excitation function corresponding to the hidden layer.

[0071] where the first connection weights between the input layer and the hidden layer are initialized and denoted as ω ij , the second connection weights between the hidden layer and the output layer are initialized and denoted as ω jk , and finally the thresholds of the hidden layer a and the output layer b are initialized, and a neuron excitation function corresponding to the hidden layer is given, for example, the neuron excitation function is given as a Sigmoid function.

[0072] In some examples of the embodiment, determining the hidden layer output corresponding to the hidden layer according to the rock mass data includes: inputting the excavated region data, the first connection weights, and the thresholds of the hidden layer into a first preset formula to determine the hidden layer output; the first preset formula is:

[0073] (j = 1, 2,..., l),

[0074] where l is the number of nodes of the hidden layer, g is an excitation function corresponding to the hidden layer, a i is the threshold of the i th node of the hidden layer, ω ij is the first connection weight, (k-date i ) is a data matrix at the i th monitoring point in the excavated region data, H j is the output of the j th node of the hidden layer, and specifically, g is a Sigmoid function The excavated region data, the first connection weights, and the thresholds of the hidden layer are input into the first preset formula to obtain the output of the hidden layer.

[0075] In some examples of the embodiment, determining the output layer output corresponding to the output layer according to the hidden layer output comprises: inputting the hidden layer output, the second connection weight and a threshold value of the output layer into a second preset formula to determine the output layer output, the second preset formula being:

[0076] (k = 1, 2,..., m),

[0077] wherein, uk-date is the output layer output, H j is an output of the jth node of the hidden layer, k is a number of hidden nodes, ω jk is the second connection weight, and b k is a threshold value of the kth node of the output layer ω jk .

[0078] In some examples of the embodiment, the prediction network error is calculated according to the output layer output and the rock mass data, comprising: inputting the output layer output and the excavated area data contained in the rock mass data into a third preset formula to obtain the prediction network error, the third preset formula being as follows:

[0079] e k = (UK-date k ) - (uk-date k ), (k = 1, 2,..., m)

[0080] wherein, e k is the prediction network error at the kth node of the output layer, (UK-date k ) is data of the excavated area, (uk-date k ) is a predicted output value of the kth node of the output layer in the excavated area e k = (UK-date k ) - (uk-date e )(k = 1, 2,..., m). Wherein, UK-date k is a stress value of the actual supporting structure of the excavated area, which can be a rod axial force, a steel stress, a concrete strain, etc.

[0081] In some examples of the embodiment, the initial rock mass risk assessment model is updated according to the prediction network error, comprising: updating the first connection weight and the second connection weight according to the prediction network error; and updating the threshold values of the hidden layer and the output layer according to the prediction network error.

[0082] Specifically, the first connection weight ω ijand the second connection weight ω jk , comprising: performing weighted calculation on the first connection weight ω ij according to the prediction network error e to obtain an updated first connection weight, the weighted calculation being as follows:

[0083]

[0084] wherein H j is an output of the jth node of the hidden layer, and e k is the prediction network error at the kth node of the output layer;

[0085] updating the first connection weight ω ij and the second connection weight ω jk according to the prediction network error e, comprising: performing weighted calculation on the second connection weight ω jk according to the prediction network error e to obtain an updated second connection weight, the weighted calculation being as follows:

[0086]

[0087] wherein H j is an output of the jth node of the hidden layer, and e k is the prediction network error at the kth node of the output layer.

[0088] wherein the first connection weight and the second connection weight are updated according to the prediction network error; after the threshold values of the hidden layer and the output layer are updated according to the prediction network error, the method further comprises:

[0089] determining whether the initial rock mass risk assessment model is iterated to completion, if not, returning to step S102 and re-executing step S102 according to the updated data, that is, executing the determination of the hidden layer output corresponding to the hidden layer according to the rock mass data, and the determination of the output layer output corresponding to the output layer according to the hidden layer output according to the updated data.

[0090] when the initial rock mass risk assessment model is iterated to completion, the initial rock mass risk assessment model iterated to completion is taken as the rock mass risk assessment model.

[0091] in some examples of the embodiment, after the initial rock mass risk assessment model is updated according to the prediction network error to obtain a rock mass risk assessment model, the method further comprises: inputting the to-be-evaluated data into the rock mass risk assessment model to obtain prediction data corresponding to the to-be-evaluated data; generating a stress nephogram in front of the working face based on the prediction data, the stress nephogram being used to determine the risk of the to-be-evaluated data.

[0092] Wherein, input the known data sequence (u1…u6) of the unexcavated rock mass, automatically generate the stress state prediction data z of the supporting structure i Finally, the stress nephogram in front of the working face is generated, and the stress characteristics of the nephogram are compared with the stress characteristics of the supported section to determine the risk of the unexcavated rock mass. If there is a stress concentration feature at the left arch foot of the working face, supporting reinforcement measures can be taken in advance.

[0093] The rock mass risk assessment model training method provided in the example includes: initializing an initial rock mass risk assessment model according to input rock mass data, the initial rock mass risk assessment model including: an input layer, a hidden layer and an output layer; determining a hidden layer output corresponding to the hidden layer according to the rock mass data, and determining an output layer output corresponding to the output layer according to the hidden layer output; performing prediction network error calculation according to the output layer output and the rock mass data; updating the initial rock mass risk assessment model according to the prediction network error to obtain a rock mass risk assessment model, which is used for risk assessment of a rock mass. The rock mass risk assessment model trained by the above method can accurately perceive the rock mass engineering systemically and comprehensively, so that relevant personnel can learn the risk of the rock mass engineering, and the problems of lack of risk perception and difficulty in timely and accurate early warning of the site in the related art are avoided.

[0094] Embodiment Two

[0095] In order to better understand the present application, a more specific example is provided in the present embodiment:

[0096] As shown in Figure 2 , as Figure 2 shown, a basic structure schematic diagram of a vibrating string type wireless sensor is provided in the example, as Figure 3 shown, Figure 3 an exploded schematic diagram of the vibrating string type wireless sensor is provided in the example, the vibrating string type wireless sensor provided in the example mainly includes: a bolt 1, a cover plate 2, a PCB board 3, a power supply 4, an information receiving device 5, an automatic acquisition module 6, a module box 7, and a wiring cap 8.

[0097] Relationships between the parts: the PCB board 3, the power supply 4, the information receiving device 5, and the automatic acquisition module 6 are fixed and packaged in the module box 7; the module box 7 and the cover plate 2 are connected into a closed whole by the bolt 1; and the wiring cap 8 is on one side of the module box 7, which is a reserved hole position.

[0098] The role of each part: the PCB board 3 is responsible for providing various electronic components such as integrated circuits, realizing the wiring and electrical connection between various electronic components such as integrated circuits, and enabling the normal operation of each working element; the information receiving device 5 is responsible for converting the received vibrating string sensor fluctuation signal into an electrical signal that the system can recognize; the automatic acquisition module 6 acquires and processes the vibrating frequency change of the vibrating string sensor; the wiring cap 8 is a reserved hole for external vibrating string sensors.

[0099] The data acquisition and wireless communication mode with the vibrating string wireless sensing module as the core has a flowchart as Figure 4 The vibrating string sensor, which is mature in technology and low in cost, can monitor the anchor rod axial force, steel stress, and concrete strain meter. The sensor is connected to the wireless sensing module through a signal line. After the installation and debugging of the sensing device are completed, it is buried in the primary lining structure to prevent disturbance from factors such as blasting and construction on site. The collected data is transmitted to the DTU through the Lora wireless communication mode. The DTU transmits the data to the cloud server through the GPRS network signal. Finally, the data from the cloud server is displayed on the web-based monitoring platform and the mobile app, as shown in Figure 4 , Figure 4 which is the basic schematic diagram of the vibrating string wireless sensing communication mode provided in this example.

[0100] Workflow:

[0101] 1. Determine the monitoring scheme by AHP multi-level analysis method. The hierarchical structure model mainly includes target layer, element layer, standard layer, and scheme layer. The target layer is first refined to the standard layer according to the element layer, and the optimal scheme in the scheme layer is obtained according to the standard layer, thereby determining the most reasonable arrangement scheme. The target layer of this patent is the monitoring arrangement scheme; the element layer is the surrounding rock grade u1, the tunnel burial depth u2, the hydrological condition u3, the adverse geology u4, the support system u5, and the excavation step distance u6; the scheme layer is the arrangement scheme, mainly including the layout position (① three positions of the vault and the left and right side walls, ② five positions of the vault, the left and right arch feet, and the left and right side walls), the section distance (15m, 25m), and the monitoring frequency (3h / time, 12h / time), totaling 8 arrangement and monitoring schemes. This provides a reasonable, effective, scientific, and feasible monitoring scheme for the design of roadways and tunnels under different conditions. The core design scheme of the sensor arrangement position, frequency, and step distance is determined according to the actual situation of each factor, Q = (x, f, d...), where Q represents the scheme, x represents the arrangement position, f represents the monitoring frequency, and d represents the step distance. The selection and calculation process of the scheme by the AHP is as follows:

[0102] (1) Establish a hierarchical model;

[0103] (2) Construct an element importance comparison matrix;

[0104] (3) Consistency test;

[0105] (4) Total ranking and scheme selection. Since the importance comparison matrix of elements in the analytic hierarchy process is subjective, the opinions of three experts are considered to obtain 1-3 schemes when the scheme is laid out. To determine the rationality of the scheme, the Topsis comprehensive evaluation method is used to make decisions on multiple schemes, as shown in Figure 5 Figure 5 The basic schematic diagram of the monitoring and arrangement scheme determination process provided in this example is shown in

[0106] The evaluation index of the monitoring scheme is three types of indexes: sensor cost, data acquisition amount, and influence on construction. The process is as follows: (1) establish a scheme decision evaluation matrix r i×3 ; (2) normalize the original data matrix b i×3 after positive direction; (3) consider the subjective and objective comprehensive weighting to obtain the loss evaluation matrix C=wb i3 ; (4) calculate the closeness degree, wherein the positive ideal solution is and the negative ideal solution is The closeness degree is calculated for different schemes Q i corresponding to V i + , and the optimal scheme is Q=max(V i + ).

[0107] In the calculation process, the determination of the index weight w is mainly determined by the entropy weight method. According to the standardized evaluation matrix b i×3 , the entropy value of the evaluation index is calculated as wherein

[0108] 2. According to the arrangement scheme of the sensor, the sensor is installed at the specified measuring point, and the wireless sensor module data transmission system is arranged.

[0109] 3. Obtain the support structure mechanical information of the anchor rod axial force, steel stress, and other monitoring points, and based on the inverse distance weighted (IDW) difference method, the information of unknown points is inferred from the information of known points, and a three-dimensional mechanical model of the support structure is constructed, as shown in Figure 6 Figure 6 is a schematic diagram of three-dimensional data generation principle. The information of the unknown measuring point z0 is determined based on the five nearest known measuring points z1-z5, i.e. wherein Finally, the three-dimensional surface domain data matrix of the support structure is formed:

[0110] ​​​

[0111] Where a is the a-th plane of the tunnel in the x-direction in three-dimensional coordinates;

[0112] i represents the i-th plane in the y-direction in three-dimensional coordinates;

[0113] j represents the j-th plane in the z-direction in three-dimensional coordinates;

[0114] Let i be the force value of the support structure at coordinates i, z, a in three-dimensional coordinates;

[0115] 4. To prevent false alarms caused by individual sensor failures or poor measurement accuracy during monitoring, regional evaluation rules are established based on the three-dimensional data matrix S to ensure the accuracy of early warnings from a three-dimensional spatial perspective. The rules are as follows:

[0116] (1) Based on the principle of graded dual control, an early warning system is established, divided into three levels: yellow, orange, and red. The dual control indicators refer to the change quantity (yield strength of reinforcing steel bars and yield strength of anchor bolts) and the change rate (1 kN / h). A red warning is issued when both dual control indicators exceed the monitored control values, or when the measured change rate shows a sharp increase. An orange warning is issued when both dual control indicators exceed 85% of the monitored control values, or when one of the dual control indicators exceeds the monitored control value. A yellow warning is issued when both dual control indicators exceed 70% of the monitored control values, or when one of the dual control indicators exceeds 85% of the monitored control value. For example, Figure 7 As shown, Figure 7 This is a schematic diagram for early warning of deformation of the support structure.

[0117] (2) After an early warning occurs, to ensure its accuracy, a regional evaluation standard is established. The judgment rule is as follows: with the early warning point as the center, search a 5×5×5 spatial range. When the number of early warning points g(x) within the spatial range satisfies If the point is not identified as a warning point, it will be treated as a false alarm point.

[0118] (3) Transmit on-site data to the cloud platform and generate a three-dimensional model of the support structure based on digital twin technology, such as... Figure 8 As shown, Figure 8 This is a schematic diagram of the digital twin results; it can be viewed on the monitoring and early warning platform. Through VR virtual technology, virtual information about the stress on the support structure is superimposed on the surrounding rock mass, achieving interaction between the virtual and real worlds. The visualized model can be viewed in real time using VR glasses, allowing for a more intuitive and vivid assessment of stress changes in the support structure, truly realizing the three-dimensional artificial intelligence of monitoring and early warning.

[0119] 5. Conduct risk assessment and prediction of the unexcavated rock mass in front of the tunnel face. For example... Figure 9 As shown,Figure 9 This is a basic schematic diagram of a neural network structure perception prediction model. It is trained using sample data k-date from the support structure data matrix, where k-date contains z from the data matrix S. i The data sequence, including the surrounding rock grade u1, tunnel depth u2, hydrological conditions u3, adverse geological conditions u4, support system u5, and excavation step distance u6, is processed through an input layer, a hidden layer, and finally an output layer. The network weights ω are continuously adjusted through training with sample data. ij ω jk The threshold (initializing the hidden layer threshold a, output layer threshold b) causes the error function to decrease along the negative gradient direction, gradually approaching the expectation, forming a system with predictive function. By inputting information about the unexcavated rock mass, it can predict its stress characteristics and dangerous areas.

[0120] The process is as follows:

[0121] (1) Perform network initialization. Based on the system input data matrix and the known information sequence of each point (z... i Determine the number of network input layer nodes n (the spatial vector dimension corresponding to the excavated sensor data k-date, which is 6 in this system), the number of hidden layer nodes l, and the number of output layer nodes m (the force data z of the support structure to be output). i (Number of types), initialize the connection values ​​ω between input layer-hidden layer and hidden layer-output layer neurons. ij ω jk Initialize the hidden layer threshold 'a' and the output layer threshold 'b', and select the Sigmoid function for the given neuron activation function.

[0122] (2) Hidden layer output calculation. Based on the input sensor data k-date of the excavated portion, the connection weight ω between the input layer and the hidden layer is calculated. ij And the hidden layer threshold 'a', outputting the hidden layer output H:

[0123] (j = 1, 2, ..., l), where: l is the number of nodes in the hidden layer, g is the activation function corresponding to the hidden layer, and a i ω is the threshold value of the i-th node in the hidden layer. ij For the first connection weight, (k-date) i H is the data matrix at the i-th monitoring point within the excavated area. j Let l be the output of the j-th node of the hidden layer; g is the activation function of the hidden layer, which is a sigmoid function.

[0124] (3) Output layer output calculation. Based on the hidden layer output H and connection weights ω... jkand threshold value b, the neural network prediction of unexcavated region data uk-date is calculated.

[0125] (k = 1, 2,..., m), wherein: (uk-date) is the output of the output layer, H j is the output of the jth node of the hidden layer, k is the number of hidden nodes, ω jk is the second connection weight, b k is the threshold value of the kth node of the output layer.

[0126] (4) Error calculation. According to the network prediction of unexcavated region data (uk-date), the prediction network error e is calculated:

[0127] e k = (UK-date k ) - (uk-date k ), (k = 1, 2,..., m); wherein, e k is the prediction network error at the kth node of the output layer, (UK-date k ) is the excavated region data, (uk-date k ) is the predicted output value of the kth node of the output layer in the excavated region.

[0128] (5) Weight update. The network connection weights ω ij , ω jk are updated according to the network prediction error e:

[0129]

[0130]

[0131] (6) Threshold value update. The network node threshold values a, b are updated according to the network prediction error e:

[0132]

[0133] (j = 1, 2,..., l) (j = 1, 2,..., l);

[0134] b k = b k + e k b k = b k + e k , (k = 1, 2,..., m) (k = 1, 2,..., m)

[0135] (7) Determine whether the algorithm iteration is ended, if not, return to step (2).

[0136] After training, the network weight and threshold are adjusted, the associative memory and prediction ability are obtained, and the BP neural network has the prediction ability of the structure stress of the unexcavated rock mass in front of the working face. i The network weight and threshold are continuously adjusted to make the error function e reach the minimum, and the support stress cloud diagram after excavation is formed to determine the dangerous source area.

[0137] (8) Input the known data sequence (u1…u6) of the unexcavated rock mass, and automatically generate the stress state prediction data z of the support structure i As shown in Figure 10 , the stress cloud diagram in front of the working face is generated based on the above prediction data, and the stress characteristics of the cloud diagram are compared with the stress characteristics of the supported section to determine the danger of the unexcavated rock mass. Figure 10 The support structure stress concentration feature exists in the left arch foot position of the working face, and the position can be reinforced in advance.

[0138] The application forms a key point monitoring scheme, and the developed wireless sensing module realizes automatic acquisition and wireless data transmission in series to monitor the stress characteristics of the support structure.

[0139] Invention point one: a key monitoring point arrangement scheme is formed, and the construction of the surface area data matrix of the surrounding rock support structure is realized.

[0140] According to different engineering scenes and geological conditions, a scientific monitoring sensor arrangement scheme is designed by comprehensively considering geological conditions, construction conditions, economic cost and other factors through the analytic hierarchy process and Topsis comprehensive evaluation method.

[0141] Invention point two: a vibrating string type wireless sensing module is developed.

[0142] The sensing module is small in size, solid in shell, waterproof and shockproof, and meets the complex working conditions such as blasting and dewatering during construction.

[0143] The sensing module is buried in the lining, and is designed with low maintenance and low power consumption.

[0144] The breakthrough of unknown monitoring point information is realized by inverse distance weighted (IDW) difference method, a surface area three-dimensional matrix representing the deformation characteristics of the rock mass supporting structure is formed, and a machine learning algorithm for rock mass risk perception and hazard source positioning is developed based on this. Through the construction of a neuron topological structure learning model, real-time monitoring and early warning of the supporting structure, hazard perception in front of the working face, and instability warning of dangerous rock mass are realized.

[0145] Invention point four: a three-dimensional simulation model is constructed to realize the interaction between the engineering entity and virtual data.

[0146] Combined with digital twin technology, the engineering entity model is integrated with real-time monitoring data to form a deformation cloud chart of the tunnel in the monitoring system, realizing systematic and comprehensive accurate perception of the rock mass engineering; the monitoring data image is superimposed on the tunnel entity model to form a real-time monitoring image of the rock mass stress in the engineering site, and based on the high-definition holographic image, the surrounding environment information is summarized to determine the tunnel hazard source.

[0147] The embodiment of the application also provides an electronic device, including one or more processors, and a storage device, wherein the storage device is used to store one or more computer programs, when the one or more computer programs are executed by the one or more processors, the electronic device realizes the surrounding rock displacement visualization model generation method as above.

[0148] Figure 11 The structure schematic diagram of the computer system of the electronic device suitable for realizing the embodiment of the application is shown.

[0149] It should be noted that, Figure 11 The computer system 1800 of the electronic device shown is only an example, and should not impose any limitation on the functions and use range of the embodiment of the application.

[0150] As Figure 11 shown, the computer system 1800 includes a processor (Central Processing Unit, CPU) 1801, which can perform various appropriate actions and processes according to programs stored in a read-only memory (Read-Only Memory, ROM) 1802 or programs loaded from a storage part 1808 to a random access memory (Random Access Memory, RAM) 1803, such as performing the method in the above embodiment. In the RAM 1803, various programs and data required for system operation are also stored. The CPU 1801, the ROM 1802, and the RAM 1803 are connected to each other through a bus 1804. An input / output (Input / Output, I / O) interface 1805 is also connected to the bus 1804.

[0151] In some embodiments, the following components are connected to the I / O interface 1805: an input part 1806 including a keyboard, a mouse, etc.; an output part 1807 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc., and a speaker, etc.; a storage part 1808 including a hard disk, etc.; and a communication part 1809 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication part 1809 performs communication processing via a network such as the Internet. A drive 1810 is also connected to the I / O interface 1805 as necessary. A removable media 1811 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 1810 as necessary, so that a computer program read therefrom is installed in the storage part 1808 as necessary.

[0152] In particular, the processes described above with reference to the flowcharts can be implemented as a computer program according to embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing a computer program for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication part 1809, and / or installed from the removable media 1811. When the computer program is executed by the processor (CPU) 1801, various functions defined in the system of the present application are executed.

[0153] It should be noted that the computer-readable medium in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination thereof. The computer-readable storage medium may, for example, be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination thereof. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, the computer-readable signal medium can include a data signal propagated in a baseband or as a carrier wave in a propagated data signal, in which the computer-readable computer program is carried. Such a propagated data signal can take on many forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate, or transmit the program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted in any suitable medium, including but not limited to wireless, wired, or the like, or any suitable combination thereof.

[0154] The flowcharts and block diagrams in the drawings illustrate the possible architectures, functions, and operations of apparatuses, methods, and computer program products according to various embodiments of the present application. Each block in the flowcharts or block diagrams can represent a module, a program segment, or a portion of code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in a different order than that shown in the figures. For example, two blocks noted in succession can actually be executed substantially concurrently, or they can be executed in reverse order, depending on the functionality involved. It should also be noted that each block in the flowcharts or block diagrams, and combinations of blocks in the flowcharts or block diagrams, can be implemented by a dedicated hardware-based system, or can be implemented by a combination of dedicated hardware and computer programs.

[0155] The units or modules involved in the embodiments of the present application can be implemented in software or hardware, and the described units or modules can also be arranged in a processor. In some cases, the names of the units or modules do not constitute a limitation on the units or modules themselves.

[0156] Another aspect of the present application also provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the surrounding rock displacement visualization model generation method as described above. The computer readable storage medium can be included in the electronic device described in the above embodiments, or can exist separately without being assembled into the electronic device.

[0157] Another aspect of the present application also provides a computer program product, which includes a computer program stored in a computer readable storage medium. The processor of the electronic device reads the computer program from the computer readable storage medium. The processor executes the computer program to make the electronic device execute the surrounding rock displacement visualization model generation method as described above.

[0158] It should be noted that although several modules or units of the device for action execution are mentioned in the above detailed description, such division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into several modules or units.

[0159] Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the embodiments disclosed herein. The present application is intended to cover any variations, uses, or adaptations of the application following the general principles thereof and including such departures from the present disclosure as come within known use or custom in the art.

[0160] The above is only the preferred exemplary embodiments of the present application, and is not intended to limit the embodiments of the present application. Those skilled in the art can easily make corresponding modifications or changes according to the main idea and spirit of the present application, and therefore the protection scope of the present application should be subject to the protection scope required by the claims.

Claims

1. A method for training a rock mass risk assessment model, characterized in that, The method comprises: initializing an initial rock mass risk assessment model according to input rock mass data, the initial rock mass risk assessment model comprising: an input layer, a hidden layer, and an output layer; determining a hidden layer output corresponding to the hidden layer according to the rock mass data, and determining an output layer output corresponding to the output layer according to the hidden layer output; performing prediction network error calculation according to the output layer output and the rock mass data; updating the initial rock mass risk assessment model according to the prediction network error to obtain a rock mass risk assessment model, the rock mass risk assessment model being used for risk assessment of a rock mass; the rock mass data comprising: excavated area data; initializing the initial rock mass risk assessment model according to input rock mass data comprises: determining a spatial vector dimension corresponding to the excavated area data; initializing a node number of the input layer, a node number of the output layer, and a node number of the initial hidden layer based on the spatial vector dimension; The method further comprises determining the hidden layer output corresponding to the hidden layer according to the rock mass data, comprising: inputting the excavated region data, the first connection weight and a threshold value of the hidden layer into a first preset formula to determine the hidden layer output; the first preset formula is: , , wherein, n is a node number of the hidden layer, is an excitation function corresponding to the hidden layer, is a threshold value of an i-th node of the hidden layer, is the first connection weight, is a data matrix at an i-th monitoring point in the excavated region data, is an output of a j-th node of the hidden layer. The determining of the output layer output corresponding to the output layer according to the hidden layer output comprises: inputting the hidden layer output, the second connection weight and a threshold value of the output layer into a second preset formula to determine the output layer output, the second preset formula being: , wherein (uk-date) is the output layer output, uk is an output of the jth node of the hidden layer, is the second connection weight, is a threshold value of the kth node of the output layer. The prediction network error calculation is performed according to the output layer output and the rock mass data, including: inputting the output layer output and the excavated area data contained in the rock mass data into a third preset formula to obtain the prediction network error, and the third preset formula is as follows: , , , , , wherein, e k is the prediction network error at the kth node of the output layer, (e k ) is the excavated area data, (e k ) is the predicted output value of the kth node of the output layer in the excavated area. after obtaining the rock mass risk assessment model, the method further comprises: inputting to-be-evaluated data into the rock mass risk assessment model to obtain prediction data corresponding to the to-be-evaluated data; and generating a stress nephogram in front of a working face based on the prediction data, the stress nephogram being used for judging a risk of the to-be-evaluated data.

2. The method of claim 1, wherein, after initializing the initial rock mass risk assessment model according to input rock mass data, the method further comprises: initializing first connection weights between the input layer and the hidden layer, and initializing second connection weights between the hidden layer and the output layer; initializing threshold values of the hidden layer and the output layer, and giving a neuron excitation function corresponding to the hidden layer.

3. The method of claim 1, wherein, updating the initial rock mass risk assessment model according to the prediction network error comprises: updating the first connection weights and the second connection weights according to the prediction network error; updating the threshold values of the hidden layer and the output layer according to the prediction network error.

4. An electronic device, comprising: comprise one or more processors; a storage device for storing one or more computer programs, when the one or more computer programs are executed by the one or more processors, the electronic device realizes the method as claimed in any one of claims 1-3.

5. A computer readable storage medium, characterized in that, a computer program is stored thereon, when the computer program is executed by a processor of an electronic device, the electronic device executes the method as claimed in any one of claims 1-3.

Images