5G-based intelligent communication method, computer device, and storage medium for cableway safety.

Abstract

This invention discloses a 5G-based intelligent communication method, computer device, and storage medium for cableway safety. The 5G-based intelligent communication method for cableway safety includes two types of terminals categorized by data type: terminals with built-in 5G modules and terminals attached to a 5G industrial gateway. Sensors for cable grips, bearings, and wire rope vibration belong to terminals with built-in 5G modules, while cameras, temperature, humidity, and logs belong to terminals attached to the 5G industrial gateway. Terminals with built-in 5G modules complete the registration process, while attached terminals do not participate in the registration signaling process; the registration process is completed by the 5G industrial gateway. The terminals upload data, which reaches the 5G base station, where the 5G base station calculates a risk score based on an anomaly scoring algorithm. The 5G base station outputs scheduling actions based on a deep reinforcement learning algorithm, adjusting slice resources, and the terminals transmit data according to these slice resources. This 5G-based intelligent communication method for cableway safety, by acquiring device data and calculating the device's risk level, adjusts slice resources based on the risk level, ensuring the stability of the communication link. During data transmission, it obtains the best slice resources, significantly reducing latency and increasing communication quality.

Description

Technical Field

[0001] This invention relates to the field of cableway safety, and in particular to a 5G-based intelligent communication method, computer device, and storage medium for cableway safety. Background Technology

[0002] With the development of cableway technology and the maturity of information networks, the focus has shifted from simply ensuring normal operation to prioritizing cableway safety. Real-time, visualized, and predictable monitoring across six key dimensions—equipment status, structural health, operational control, personnel and environment, emergency command, and network and data security—is crucial to completely resolve issues such as numerous blind spots, slow response times, and reliance on manual intervention in cableway monitoring. Narrowband IoT can only collect data at the second level, completely missing millisecond-level impact and vibration waveforms, thus failing to provide early warnings of fatigue, wear, and jamming. Furthermore, bandwidth and latency limitations prevent the support of high-frequency concurrent monitoring points across the entire line, resulting in sparse deployment and numerous blind spots around cable grips, wheel sets, and wire ropes. A 5G-based intelligent communication system for cableway safety can solve these problems; currently, there is no 5G-based intelligent communication system for cableway safety. Summary of the Invention

[0003] In view of at least one of the above-mentioned technical problems, the purpose of this invention is to provide a 5G-based intelligent communication method, computer device, and storage medium for cableway safety.

[0004] On one hand, embodiments of the present invention include a 5G-based intelligent communication method for cableway safety. During the same time period, cableway sensors, video acquisition data, and log data are connected to a 5G network. The 5G network serves as a service network, allowing real-time transmission of sensor data, video data, and log data. The 5G-based intelligent communication method for cableway safety comprises:

[0005] Based on data type, terminals are divided into two types: terminals with built-in 5G modules and terminals attached to 5G industrial gateways. Sensors for vibration of cable clamps, bearings, and wire ropes belong to terminals with built-in 5G modules, while cameras, temperature, humidity, and logs belong to terminals attached to 5G industrial gateways.

[0006] Terminals with built-in 5G modules complete the registration process, while connected terminals do not participate in the registration signaling process; the registration process is completed by the 5G industrial gateway.

[0007] The terminal uploads data, which arrives at the 5G base station, where the 5G base station calculates a risk score based on an anomaly scoring algorithm.

[0008] 5G base stations output scheduling actions based on deep reinforcement learning algorithms to adjust slice resources, and terminals transmit data according to slice resources.

[0009] Furthermore, the aforementioned 5G-based intelligent communication method for cableway safety is characterized in that the risk score calculated by the 5G base station according to the anomaly scoring algorithm includes:

[0010] Calculate time-domain and frequency-domain characteristics based on the data;

[0011] The dispersion is calculated based on the time-domain and frequency-domain characteristics;

[0012] The anomaly score of the grouped multi-measurement point is calculated based on the dispersion, thereby obtaining the global risk score of the entire domain;

[0013] Furthermore, the aforementioned 5G-based intelligent communication method for cableway safety is characterized in that the calculation of time-domain and frequency-domain features based on data includes:

[0014] The following was obtained by discrete sampling based on the original data:

[0015]

[0016] The following time-domain features were obtained based on the sampled data:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024] The above This is the average value. RMS root mean square Peak value, Peak-to-peak value For ravine, For waveform factor, The pulse factor;

[0025] Performing a Fourier transform on the sampled data yields:

[0026]

[0027] Based on the Fourier transform data, the frequency domain characteristics are obtained:

[0028]

[0029]

[0030] The above The mean of the spectrum, This refers to frequency band energy.

[0031] Furthermore, the aforementioned 5G-based intelligent communication method for cableway safety is characterized in that the calculation of discreteness based on time-domain and frequency-domain characteristics includes:

[0032]

[0033] The maximum deviation above is ,when =0, characteristics are completely normal, when = If the characteristics deviate significantly from the baseline, it is considered high risk.

[0034] Furthermore, the aforementioned 5G-based intelligent communication method for cableway safety is characterized in that the step of calculating the anomaly score of multiple measurement points based on dispersion to obtain a global risk score includes:

[0035] We weight the deviations of all features for each group to obtain:

[0036]

[0037] The above For a certain set of characteristic numbers, For a certain group, in ,when When the equipment is healthy,

[0038] At that time, the equipment malfunctioned slightly. At that time, the equipment issued a moderate alarm. At that time, the equipment experienced a high-risk malfunction.

[0039] The maximum risk for each group is:

[0040]

[0041] above The number of measurement points in each group is given, and the maximum risk is the maximum value of the risks of all measurement points.

[0042] The overall risk score is:

[0043]

[0044] The above This represents the number of all types of vibrating equipment.

[0045] Furthermore, the aforementioned 5G-based intelligent communication method for cableway safety is characterized in that the 5G base station outputs scheduling actions according to a deep reinforcement learning algorithm to adjust slice resources, and the terminal transmits data according to the slice resources, including:

[0046] According to the resource adjustment method Identify the following four behaviors:

[0047] Action={

[0048] 0: "Maintain current resource allocation"

[0049] 1: "Slightly increase URLLC priority and bandwidth".

[0050] 2: "Significantly expand URLLC, limit eMMB bandwidth".

[0051] 3: "URLLC dedicated resources, eMBB / eMTC rate limiting and hibernation (emergency)"

[0052] }

[0053] reward function The rules to be followed are:

[0054] If the device is functioning correctly and action=0, then Positive values ​​are given for all others, while negative values ​​are given for the others.

[0055] If the equipment has a minor malfunction and action=1, then... Positive values ​​are given for all others, while negative values ​​are given for the others.

[0056] If the equipment has a moderate fault and action=2, then Positive values ​​are given for all others, while negative values ​​are given for the others.

[0057] If the equipment has a high-risk malfunction and action=3, then... One is a positive value, and the others are negative values.

[0058] Fit the Q-value function using a neural network; Q-value function In the state Next, execute the action. The training objective here is to obtain long-term cumulative rewards:

[0059]

[0060] above This is a discount factor used to weigh immediate rewards against future rewards. Assuming the state for the next time step, the network continuously approaches the Q-value, eventually converging to obtain the optimal scheduling strategy. Then, based on the input state... Output the Q-values ​​of all actions, and select the action with the highest Q-value to execute:

[0061]

[0062] Adjust the slice resources based on this action.

[0063] On the other hand, embodiments of the present invention also include a 5G network, characterized in that the 5G network is used to execute a cableway safety intelligent communication method.

[0064] On the other hand, embodiments of the present invention also include a computer device, characterized in that it includes a memory and a processor, the memory being used to store at least one program, and the processor being used to load the low-Earth orbit satellite switching control method in the embodiments.

[0065] On the other hand, embodiments of the present invention also include a storage medium storing a processor-executable program, characterized in that the processor-executable program, when executed by a processor, is used for the low-Earth orbit satellite handover control method in the embodiments.

[0066] The beneficial effects of the present invention are as follows: The cableway safety intelligent communication method in the embodiment obtains equipment data, calculates the risk level of the equipment, and adjusts the slice resources according to the risk level of the equipment, which ensures the stability of the communication link. During data transmission, the best slice resources can be obtained for data transmission, which greatly reduces latency and increases communication quality. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of a 5G-based cableway communication network.

[0068] Figure 2 The timing diagram of the registration signaling for the 5G module terminal of the 5G-based cableway safety intelligent communication method;

[0069] Figure 3 This is a timing diagram of the registration signaling for a gateway terminal in a 5G-based intelligent communication method for cableway safety.

[0070] Figure 4 This is a timing diagram of data transmission for a 5G module terminal based on a 5G-based intelligent communication method for cableway safety.

[0071] Figure 5 The timing diagram for data transmission of the gateway terminal in a 5G-based intelligent communication method for cableway safety is shown.

[0072] Figure 6 Algorithm flowchart of a 5G-based intelligent communication method for cableway safety Detailed Implementation

[0073] In this embodiment, the 5G-based cableway safety intelligent communication method is applied to... Figure 1 The diagram shows a 5G-based cableway communication network. (See reference) Figure 1 The communication network system includes 5G modules, 5G industrial gateways, 5G base stations, and a core network. To maintain... Figure 1 For simplicity, only the individual base stations in the network are shown; other components such as servers and optical front-end networks in the mobile network are not shown. Similarly... Figure 1 The core network elements are also not shown. The 5G core network mainly includes the Access Management Element (AMF), Authentication Management Element (AUSF), Slicing Management Element (NSSF), Session Management Element (SMF), and Data Plane Management Element (UPF), etc. (See reference...) Figure 2 The registration process for 5G terminal modules is clearly displayed:

[0074] 1. Community Search: Target communities with dedicated 5G private networks for cable car routes;

[0075] 2. Random access: Establishing an underlying signaling channel;

[0076] 3. Registration request, including SUPI, NASSAI, and URLLC capabilities.

[0077] 4. Identity authentication: Private network access verification to prevent unauthorized terminals from accessing the network;

[0078] 5. Slice binding: Fixed allocation of URLLC slices;

[0079] 6. PDU Session Establishment Request

[0080] 7. SMF selects the local UPF and establishes a tunnel.

[0081] 8. Registration complete: Continuously upload vibration data, no sleep mode, short DRX.

[0082] Reference Figure 3 This demonstrates the registration process for terminals connected to a 5G industrial gateway:

[0083] 1. The downmount sensor only performs local short-range access and does not participate in 5G signaling;

[0084] 2. The gateway's built-in 5G module powers on, completes cell search, and randomly connects to the network;

[0085] 3. The gateway initiates 5G registration;

[0086] 4. The gateway initiates authentication and dual-slice allocation, including eMBB and mMTC;

[0087] 5. Establish two PDU sessions and complete gateway registration.

[0088] Reference Figure 4This demonstrates the data transmission process of the 5G module terminal:

[0089] 1. The 5G module terminal has completed registration and remains online.

[0090] 2. The raw vibration data is transmitted to the 5G base station gNB;

[0091] 3. RIC executes the anomaly scoring algorithm to calculate the risk score;

[0092] 4. RIC executes a deep reinforcement learning algorithm to dynamically instruct gNB to adjust slice resources;

[0093] 5. Data is transmitted to the core network and MEC.

[0094] Reference Figure 5 This demonstrates the data transmission process of the terminal connected to the 5G industrial gateway:

[0095] 1. Terminals connected to the gateway complete local access and registration;

[0096] 2. Terminals connected to the gateway transmit service data to the gateway;

[0097] 3. The gateway forwards the data to the 5G base station gNB;

[0098] 4. RIC executes anomaly scoring algorithm to calculate risk score;

[0099] 5. The RIC executes a deep reinforcement learning algorithm to dynamically instruct the gNB to adjust slice resources;

[0100] 6. Data is transmitted to the core network and MEC.

[0101] Reference Figure 6 The method for safe and intelligent communication of cableways includes the following steps:

[0102] S1. Acquire raw sensor data and perform sampling;

[0103] S2. Time-domain and frequency-domain characteristics are calculated from the sampled data;

[0104] S3. Calculate the anomaly risk score using time-domain and frequency-domain features;

[0105] S4. Use anomaly risk scores for deep reinforcement learning to adjust slice resources.

[0106] In step S1, the raw sensor data is acquired and sampled. The types of sensor data are consistent with the previous description. For simplicity, only vibration data is shown here. An example of the sampled triaxial vibration data is shown in Table 1:

[0107] Table 1

[0108]

[0109] In the above table, , , They are respectively , , Shaft vibration acceleration data.

[0110] In step S2, the time-domain features and frequency-domain features are calculated using the sampled data. The calculation process is as follows:

[0111] Obtained by discrete sampling based on a certain dimension of the original data:

[0112]

[0113] The following time-domain features were obtained based on the sampled data:

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120]

[0121] The above This is the average value. RMS root mean square Peak value, Peak-to-peak value For ravine, For waveform factor, The pulse factor;

[0122] Performing a Fourier transform on the sampled data yields:

[0123]

[0124] Based on the Fourier transform data, the frequency domain characteristics are obtained:

[0125]

[0126]

[0127] The above The mean of the spectrum, This refers to frequency band energy.

[0128] In step S3, the anomaly risk score is calculated using time-domain and frequency-domain features. The calculation process is as follows:

[0129] Calculate the dispersion based on time-domain and frequency-domain characteristics:

[0130]

[0131] The maximum deviation above is ,when The characteristics are completely normal, when If the characteristics deviate significantly from the baseline, it is considered high risk.

[0132] Based on the dispersion, the anomaly scores of multiple measurement points in the group are calculated, thereby obtaining the global risk score for the entire domain:

[0133] We weight the deviations of all features for each group to obtain:

[0134]

[0135] The above For a certain set of characteristic numbers, For a certain group, in ,when When the equipment is healthy, At that time, the equipment malfunctioned slightly. At that time, the equipment issued a moderate alarm. At that time, the equipment experienced a high-risk malfunction.

[0136] The maximum risk for each group is:

[0137]

[0138] above The number of measurement points in each group is given, and the maximum risk is the maximum value of the risks of all measurement points.

[0139] This yields the global risk score:

[0140]

[0141] The above This represents the total number of types of vibrating equipment.

[0142] In step S4, deep reinforcement learning is used to adjust the slice resources using anomaly risk scores, including the following steps:

[0143] S401. Generate a state vector using the anomaly risk vector:

[0144]

[0145] The state vectors above correspond to the abnormal risk scores of the cable clamp group, wheel group, wire rope group, gear group, carriage group, and motor group, respectively.

[0146] S402. Calculate the reward function using outlier scores:

[0147] According to the resource adjustment method Identify the following four behaviors:

[0148] Action={

[0149] 0: "Maintain current resource allocation"

[0150] 1: "Slightly increase URLLC priority and bandwidth".

[0151] 2: "Significantly expand URLLC, limit eMMB bandwidth".

[0152] 3: "URLLC dedicated resources, eMBB / eMTC rate limiting and hibernation (emergency)"

[0153] }

[0154] reward function The rules to be followed are:

[0155] If the device is functioning correctly and action=0, then Positive values ​​are given for all others, while negative values ​​are given for the others.

[0156] If the equipment has a minor malfunction and action=1, then Positive values ​​are given for all others, while negative values ​​are given for the others.

[0157] If the equipment has a moderate fault and action=2, then Positive values ​​are given for all others, while negative values ​​are given for the others.

[0158] If the equipment has a high-risk malfunction and action=3, then... One is a positive value, and the others are negative values.

[0159] S403. Calculate the Q-function and perform deep reinforcement learning to adjust the slice resources:

[0160]

[0161]

[0162] in The value ranges from 1 to N, where N is the number of dimensional transformation layers. .in For the first Number of layer outputs It is a non-linear transformation function. The output dimension of the last layer is the same as the dimension of the Action. The output of the last layer is the Q value.

[0163] Next calculation :

[0164]

[0165] above This is a discount factor used to weigh immediate rewards against future rewards. Assuming the state at the next time step, the network continuously approaches the Q-value, eventually converging to obtain the optimal scheduling strategy.

[0166] Then based on the input status Output the Q-values ​​of all actions, and select the action with the highest Q-value to execute:

[0167]

[0168] Adjust the slice resources based on this action.

Claims

1. A 5G-based intelligent communication method for cableway safety, wherein cableway sensors, video acquisition data, and log data are connected to a 5G network simultaneously, the 5G network serving as a service network, and sensor data, video data, and log data can be transmitted in real time, characterized in that... The 5G-based intelligent communication method for cableway safety includes: Based on data type, terminals are divided into two types: terminals with built-in 5G modules and terminals attached to 5G industrial gateways. Sensors for vibration of cable clamps, bearings, and wire ropes belong to terminals with built-in 5G modules, while cameras, temperature, humidity, and logs belong to terminals attached to 5G industrial gateways. Terminals with built-in 5G modules complete the registration process, while connected terminals do not participate in the registration signaling process; the registration process is completed by the 5G industrial gateway. The terminal uploads data, which arrives at the 5G base station, where the 5G base station calculates a risk score based on an anomaly scoring algorithm. 5G base stations output scheduling actions based on deep reinforcement learning algorithms to adjust slice resources, and terminals transmit data according to slice resources.

2. The 5G-based intelligent communication method for cableway safety according to claim 1, characterized in that, The risk score calculated by the 5G base station based on the anomaly scoring algorithm includes: Calculate time-domain and frequency-domain characteristics based on the data; The dispersion is calculated based on the time-domain and frequency-domain characteristics; The anomaly score of the grouped multi-measurement point is calculated based on the dispersion, thereby obtaining the global risk score.

3. The 5G-based intelligent communication method for cableway safety according to claim 2, characterized in that, The calculation of time-domain and frequency-domain features based on the data includes: The following was obtained by discrete sampling based on the original data: The following time-domain features were obtained based on the sampled data: The above This is the average value. The root mean square (RMS) Peak value, Peak-to-peak value For ravine, For waveform factor, The pulse factor; Performing a Fourier transform on the sampled data yields: Based on the Fourier transform data, the frequency domain characteristics are obtained: The above The mean of the spectrum, This refers to frequency band energy.

4. The 5G-based intelligent communication method for cableway safety according to claim 2, characterized in that, The calculation of the discreteness based on time-domain and frequency-domain features includes: The maximum deviation above is ,when =0, characteristics are completely normal, when = If the characteristics deviate significantly from the baseline, it is considered high risk.

5. A 5G-based intelligent communication method for cableway safety according to claim 2, characterized in that, The process of calculating the anomaly score of multiple measurement points based on the dispersion to obtain the global risk score includes: We weight the deviations of all features for each group to obtain: The above For a certain set of characteristic numbers, For a certain group, in ,when When the equipment is healthy, At that time, the equipment malfunctioned slightly. At that time, the equipment issued a moderate alarm. At that time, the equipment experienced a high-risk malfunction. The maximum risk for each group is: above The number of measurement points in each group is given, and the maximum risk is the maximum value of the risks of all measurement points. The overall risk score is: The above This represents the number of all types of vibrating equipment.

6. The 5G-based intelligent communication method for cableway safety according to claim 1, characterized in that, The 5G base station outputs scheduling actions based on a deep reinforcement learning algorithm to adjust slice resources, and the terminal transmits data according to the slice resources, including: According to the resource adjustment method Identify the following four behaviors: Action={ 0: "Maintain current resource allocation" 1: "Slightly increase URLLC priority and bandwidth". 2: "Significantly expand URLLC, limit eMMB bandwidth". 3: "URLLC dedicated resources, eMBB / eMTC rate limiting and hibernation (emergency)" } reward function The rules to be followed are: If the device is functioning correctly and action=0, then Positive values ​​are given for all others, while negative values ​​are given for the others. If the equipment has a minor malfunction and action=1, then... Positive values ​​are given for all others, while negative values ​​are given for the others. If the equipment has a moderate fault and action=2, then Positive values ​​are given for all others, while negative values ​​are given for the others. If the equipment has a high-risk malfunction and action=3, then... One is a positive value, and the others are negative values. Fit the Q-value function using a neural network; Q-value function In the state Next, execute the action. The training objective here is to obtain long-term cumulative rewards: above This is a discount factor used to weigh immediate rewards against future rewards. Assuming the state for the next time step, the network continuously approaches the Q-value, eventually converging to obtain the optimal scheduling strategy. Then, based on the input state... Output the Q-values ​​of all actions, and select the action with the highest Q-value to execute: Adjust the slice resources based on this action.

7. A 5G-based intelligent communication method for cableway safety according to any one of claims 1 to 6, characterized in that, The resources of the slice are intelligently adjusted by the 5G base station.

8. A 5G network, characterized in that, The 5G network is used to execute the cableway safety intelligent communication method as described in any one of claims 1-7.

9. A computer device, characterized in that, The method includes a memory and a processor, the memory being used to store at least one program, and the processor being used to load the at least one program to perform the method according to any one of claims 1-7.

10. A storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to perform the method as described in any one of claims 1-7.

Images