A limited space monitoring robot monitoring system based on 5G communication

Abstract

The application discloses a limited space monitoring robot monitoring system based on 5G communication, and the system comprises a data acquisition module, a risk assessment module, a man-machine interaction module and a 5G base station transmission module, wherein the data acquisition module is used for collecting environment characteristic data and harmful gas concentration data of the limited space, the environment characteristic data comprises temperature, air flow, biomass and space volume; the risk assessment module is used for analyzing the data collected by the data acquisition module, and assessing the risk level and occurrence probability of the limited space; the man-machine interaction module is used for displaying the risk level and occurrence probability monitoring result of the limited space; and the 5G base station transmission module is used for transmitting the risk level and occurrence probability monitoring result of the limited space to a working end outside the limited space; the problems that it is difficult to judge the probability of harmful gas hazards and the monitoring result lacks credibility in the prior art are solved, and the interpretation capability of the monitoring result of the limited space robot is improved.

Description

Technical Field

[0001] This invention belongs to the field of robotics technology and relates to a monitoring system for a confined space monitoring robot based on 5G communication. Background Technology

[0002] Confined spaces have always been a threat to human safety, especially in summer when high temperatures and humidity create conditions where toxic and harmful gases can easily evaporate and accumulate. If people enter a poorly ventilated confined space, they are highly susceptible to poisoning and asphyxiation. Against the backdrop of rapid advancements in robotics and big data technologies, developing a confined space monitoring robot is of significant practical value. A confined space monitoring robot system based on 5G communication can not only effectively reduce monitoring costs but also mitigate the risks faced by humans entering hazardous environments to a certain extent.

[0003] Currently, the monitoring of hazardous gases in confined spaces primarily involves setting fixed concentration thresholds, triggering an alarm when the concentration exceeds these thresholds. However, due to differences in the internal characteristics of various confined spaces (such as volume, temperature, airflow, and biomass), the degree of impact from hazardous gases is difficult to determine. Furthermore, relying solely on whether the concentration exceeds a threshold to trigger an alarm fails to assess the probability of hazard, resulting in unreliable monitoring results. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a monitoring robot system for confined space based on 5G communication, so as to solve the problem that it is difficult to determine the probability of harmful gases causing harm and the monitoring results lack credibility in the prior art.

[0005] Technical solution of the present invention:

[0006] A confined space monitoring robot monitoring system based on 5G communication includes a data acquisition module, a risk assessment module, a human-computer interaction module, and a 5G base station transmission module, wherein:

[0007] The data acquisition module is used to collect environmental characteristic data and harmful gas concentration data of a confined space. The environmental characteristic data includes temperature, air flow rate, biomass, and space volume.

[0008] The risk assessment module is used to analyze the data collected by the data acquisition module to assess the risk level and probability of occurrence of the confined space.

[0009] The human-computer interaction module is used to display the monitoring results of the risk level and probability of occurrence in a confined space;

[0010] The 5G base station transmission module is used to transmit the monitoring results of the risk level and probability of occurrence of a confined space to the working terminal outside the confined space.

[0011] In the risk assessment module, the data collected by the analysis data acquisition module is used to assess the risk level and probability of a confined space, including the following steps:

[0012] S1. Detect the presence of harmful gases in a confined space;

[0013] S2. When there are no harmful gases in the confined space, it is determined that there is no risk in the confined space;

[0014] S3. When harmful gases are present in a limited space, input the environmental characteristic data into the trained Bayesian network model, and output the risk level of the harmful gas concentration and its probability of occurrence.

[0015] In step S3, the trained Bayesian network model is trained based on monitoring data from a limited space over a historical period. The training steps are as follows:

[0016] S31. Model Structure Determination: Determine the variables for building the model, as well as the initial Bayesian network model structure;

[0017] S32. Model parameter estimation: The K-means method is used to cluster environmental feature data and harmful gas concentrations. Based on the clustering results, the parameters of the probability distribution of the Bayesian network model are estimated, and a conditional probability table is constructed based on the parameter estimation results.

[0018] S33. Model Structure Learning: Using structure learning algorithms, the network structure is further adjusted and optimized to obtain a more stable network structure.

[0019] S34. Model Validation: Validate the accuracy of the model using a known prediction dataset.

[0020] In step S31, the model structure is determined. The initial Bayesian network model structure uses environmental feature data as the parent node and harmful gas concentration as the child node, and connects the parent node and the child node.

[0021] In this embodiment, the parent node is equivalent to the independent variable, and the child node is equivalent to the dependent variable; that is, the parent node is the cause of changes in the child node. A Bayesian network is a causal probabilistic model that can not only quantify the causal relationship between parent and child nodes but also analyze the probability of various parameter states within the child node.

[0022] In step S32, the K-means method is used to cluster environmental feature data and harmful gas concentrations, and the parameters of the probability distribution of the Bayesian network model are estimated based on the clustering results. This includes the following steps:

[0023] S321. Data normalization processing: Normalize the environmental characteristic data and the hazardous gas concentration data together;

[0024] S322. Initialize cluster centers: Randomly select several data objects in the data space as initial cluster centers;

[0025] S323. Initialize data object clusters: Calculate the Euclidean distance between all data objects and the initial cluster center, and divide each data object into the category with the smallest Euclidean distance from the initial cluster center to form the initial cluster.

[0026] S324. Update cluster centers: Calculate the average value of the data objects in each initial cluster, use this average value as the new cluster center, and recalculate the Euclidean distance between all data objects and the new cluster center.

[0027] S325. Determine the final cluster: Repeat steps S321-S324 until the cluster centers no longer change, and the iteration is complete;

[0028] S326. Estimate the parameters of the probability distribution of the Bayesian network model based on the range of environmental characteristic data and harmful gas concentration data in the cluster.

[0029] In step S321, the normalization process is expressed by the following formula:

[0030]

[0031] In the formula, g(X) is the normalization function; X i This represents environmental characteristic data, where 1 ≤ i ≤ n, and n is the number of samples in the data. X min X represents the minimum value in the dataset. max This represents the maximum value in the dataset.

[0032] In step S323, the Euclidean distance is calculated using the following formula:

[0033]

[0034] In the formula, d(x,C) i ) represents the Euclidean distance function; x is the normalized data object; C i Let x represent the i-th cluster center, 1 ≤ i ≤ k, where k is the number of cluster centers; n is the number of samples in the dataset; j Represents the data object of the j-th variable in the dataset; C ij This represents the cluster center of the j-th variable.

[0035] In step S32, the construction of the conditional probability table involves: constructing a conditional probability table for each variable based on the parameter values ​​obtained from parameter estimation; for each variable, given its parent node, calculating its joint probability across multiple parent nodes, using the following formula:

[0036]

[0037]

[0038] In the formula, P(U) represents a set of nodes U = B1, B2, ..., B n The joint probability distribution of Pa(B) i ) represents node B in a Bayesian network. i The set of parent nodes; P(B) i ) represents the target node B i The prior probability; n is the number of nodes.

[0039] In step S33, the model structure learning method is the Bayesian Information Criterion (BIC), which enumerates all network structures and selects the network structure with the smallest BIC value as the optimal model structure.

[0040] The beneficial effects of this invention are:

[0041] This invention constructs a Bayesian network model based on historical limited space monitoring data, quantifies the relationship between environmental characteristics and harmful gas concentrations within a limited space, solves the problem that existing technologies cannot determine the degree of influence of harmful gases based on the environmental characteristics of different limited spaces, and improves the monitoring accuracy of a limited space monitoring robot system based on 5G communication.

[0042] This invention uses the K-means method to cluster environmental features and harmful gas concentrations to estimate the probability distribution parameters of a Bayesian network model. It also uses a conditional probability table to quantify the probability of child node parameters occurring under the conditions of parent node parameters. This solves the problem in existing technologies that it is difficult to determine the probability of harmful gases causing damage and that the monitoring results lack credibility, thus improving the interpretability of monitoring results for robots in confined spaces. Attached Figure Description

[0043] Figure 1 This is a structural diagram of a limited space monitoring robot monitoring system based on 5G communication according to the present invention. Detailed Implementation

[0044] Please see Figure 1 This invention provides a monitoring system for a confined space monitoring robot based on 5G communication, including a data acquisition module, a risk assessment module, a human-computer interaction module, and a 5G base station transmission module, wherein:

[0045] The data acquisition module is used to collect environmental characteristic data and harmful gas concentration data of a confined space. The environmental characteristic data includes temperature, air flow rate, biomass, and space volume.

[0046] The risk assessment module is used to analyze the data collected by the data acquisition module and assess the risk level and probability of occurrence of the confined space.

[0047] The human-computer interaction module is used to display the risk level and probability of occurrence monitoring results of a confined space;

[0048] The 5G base station transmission module is used to transmit the monitoring results of the risk level and probability of occurrence of the confined space to the working terminal outside the confined space.

[0049] Furthermore, in the risk assessment module, the data collected by the analysis data acquisition module is used to assess the risk level and probability of a confined space, including the following steps:

[0050] S1. Detect the presence of harmful gases in a confined space;

[0051] S2. When there are no harmful gases in the confined space, it is determined that there is no risk in the confined space;

[0052] S3. When harmful gases are present in a limited space, input the environmental characteristic data into the trained Bayesian network model, and output the risk level of the harmful gas concentration and its probability of occurrence.

[0053] In this embodiment, the degree of influence of harmful gases is determined by monitoring the environmental characteristics of a confined space. These environmental characteristics include temperature, airflow, biomass, and space volume, all of which are related to the concentration of harmful gases. For example, temperature affects the movement of harmful gas molecules; airflow determines the ability of a confined space to dilute harmful gases; biomass is used to characterize the degree of harm caused by harmful gases and is a common means of biological monitoring; and space volume is used to further determine the total amount of harmful gases.

[0054] Furthermore, in step S3, the trained Bayesian network model is trained based on monitoring data from a limited space over a historical period. The training steps are as follows:

[0055] S31. Model Structure Determination: Determine the variables for building the model, as well as the initial Bayesian network model structure;

[0056] S32. Model parameter estimation: The K-means method is used to cluster environmental feature data and harmful gas concentrations. Based on the clustering results, the parameters of the probability distribution of the Bayesian network model are estimated, and a conditional probability table is constructed based on the parameter estimation results.

[0057] S33. Model Structure Learning: Using structure learning algorithms, the network structure is further adjusted and optimized to obtain a more stable network structure.

[0058] S34. Model Validation: Validate the accuracy of the model using a known prediction dataset.

[0059] Further, in step S31, the model structure is determined, wherein the initial Bayesian network model structure uses environmental feature data as the parent node and harmful gas concentration as the child node, and connects the parent node and the child node.

[0060] Further, in step S32, the step of using the K-means method to cluster environmental feature data and harmful gas concentrations, and estimating the parameters of the probability distribution of the Bayesian network model based on the clustering results, includes the following steps:

[0061] S321. Data normalization processing: Normalize the environmental characteristic data and the hazardous gas concentration data together;

[0062] S322. Initialize cluster centers: Randomly select several data objects in the data space as initial cluster centers;

[0063] S323. Initialize data object clusters: Calculate the Euclidean distance between all data objects and the initial cluster center, and divide each data object into the category with the smallest Euclidean distance from the initial cluster center to form the initial cluster.

[0064] S324. Update cluster centers: Calculate the average value of the data objects in each initial cluster, use this average value as the new cluster center, and recalculate the Euclidean distance between all data objects and the new cluster center.

[0065] S325. Determine the final cluster: Repeat steps S321-S324 until the cluster centers no longer change, and the iteration is complete;

[0066] S326. Estimate the parameters of the probability distribution of the Bayesian network model based on the range of environmental characteristic data and harmful gas concentration data in the cluster.

[0067] In this embodiment, the K-means method is used to cluster environmental feature data and hazardous gas concentration data into different clusters. The environmental feature data and hazardous gas concentration data in the same cluster have similar data characteristics. Therefore, each cluster is defined as a risk level. This process is also the parameter estimation process of the Bayesian network.

[0068] Further, in step S321, the normalization process is expressed by the following formula:

[0069]

[0070] In the formula, g(X) is the normalization function; X i This represents environmental characteristic data, where 1 ≤ i ≤ n, and n is the number of samples in the dataset. X min X represents the minimum value in the dataset. max This represents the maximum value in the dataset.

[0071] Further, in step S323, the Euclidean distance is calculated using the following formula:

[0072]

[0073] In the formula, d(x,C) i ) represents the Euclidean distance function; x is the normalized data object; C i Let x represent the i-th cluster center, 1 ≤ i ≤ k, where k is the number of cluster centers; n is the number of samples in the dataset; j Represents the data object of the j-th variable in the dataset; C ij This represents the cluster center of the j-th variable.

[0074] Further, in step S32, the construction of the conditional probability table involves: constructing a conditional probability table for each variable based on the parameter values ​​obtained from parameter estimation; for each variable, given its parent node, calculating its joint probability across multiple parent nodes, using the following formula:

[0075]

[0076]

[0077] In the formula, P(U) represents a set of nodes U = B1, B2, ..., B n The joint probability distribution of Pa(B) i ) represents node B in a Bayesian network. i The set of parent nodes; P(B) i ) represents the target node B i The prior probability; n is the number of nodes.

[0078] In this embodiment, constructing a conditional probability table can quantify the influence of parent nodes on child nodes. Combined with the K-means method for classifying risk levels, the Bayesian network model can predict the risk level of harmful gas concentration and its probability of occurrence through environmental feature data, thereby improving the interpretability of monitoring.

[0079] Furthermore, in step S33, the model structure learning method is the Bayesian Information Criterion (BIC), which enumerates all network structures and selects the network structure with the smallest BIC value as the optimal model structure.

[0080] The beneficial effects of this invention are:

[0081] (1) A Bayesian network model was constructed based on historical limited space monitoring data, which quantified the relationship between environmental characteristics and harmful gas concentration in the limited space. This solved the problem that existing technologies could not determine the degree of influence of harmful gases based on the environmental characteristics of different limited spaces, and improved the monitoring accuracy of a limited space monitoring robot monitoring system based on 5G communication.

[0082] (2) The K-means method is used to cluster environmental features and harmful gas concentrations to estimate the probability distribution parameters of the Bayesian network model. The probability of child node parameters occurring under the condition of parent node parameters is quantified by the conditional probability table. This solves the problem that it is difficult to judge the probability of harmful gas causing harm in the existing technology and the monitoring results lack credibility. It also improves the ability to interpret the monitoring results of robots in confined spaces.

Claims

1. A monitoring system for a confined space monitoring robot based on 5G communication, characterized in that: The system includes a data acquisition module, a risk assessment module, a human-computer interaction module, and a 5G base station transmission module, wherein: The data acquisition module is used to collect environmental characteristic data and harmful gas concentration data of a confined space. The environmental characteristic data includes temperature, air flow rate, biomass, and space volume. The risk assessment module is used to analyze the data collected by the data acquisition module to assess the risk level and probability of occurrence of the confined space. The human-computer interaction module is used to display the monitoring results of the risk level and probability of occurrence in a confined space; The 5G base station transmission module is used to transmit the monitoring results of the risk level and probability of occurrence of a confined space to the working end outside the confined space; The risk assessment module, specifically the data collected by the data analysis and acquisition module, assesses the risk level and probability of a confined space, including the following steps: S1. Detect the presence of harmful gases in a confined space; S2. When there are no harmful gases in the confined space, it is determined that there is no risk in the confined space; S3. When harmful gases are present in a limited space, input the environmental characteristic data into the trained Bayesian network model and output the risk level of the harmful gas concentration and its probability of occurrence. In step S3, the trained Bayesian network model is trained based on monitoring data from a limited space over a historical period. The training steps are as follows: S31. Model Structure Determination: Determine the variables for building the model, as well as the initial Bayesian network model structure; S32. Model parameter estimation: The K-means method is used to cluster environmental feature data and harmful gas concentrations. Based on the clustering results, the parameters of the probability distribution of the Bayesian network model are estimated, and a conditional probability table is constructed based on the parameter estimation results. S33. Model Structure Learning: Using structure learning algorithms, the network structure is adjusted and optimized to obtain a more stable network structure. S34. Model Validation: Validate the accuracy of the model using a known prediction dataset.

2. The monitoring system for a confined space monitoring robot based on 5G communication according to claim 1, characterized in that: In step S31, the model structure is determined. The initial Bayesian network model structure uses environmental feature data as the parent node and harmful gas concentration as the child node, and connects the parent node and the child node.

3. The monitoring system for a confined space monitoring robot based on 5G communication according to claim 1, characterized in that: In step S32, the K-means method is used to cluster environmental feature data and harmful gas concentrations, and the parameters of the probability distribution of the Bayesian network model are estimated based on the clustering results. This includes the following steps: S321. Data normalization processing: Normalize the environmental characteristic data and the hazardous gas concentration data together; S322. Initialize cluster centers: Randomly select several data objects in the data space as initial cluster centers; S323. Initialize data object clusters: Calculate the Euclidean distance between all data objects and the initial cluster center, and divide each data object into the category with the smallest Euclidean distance from the initial cluster center to form the initial cluster. S324. Update cluster centers: Calculate the average value of the data objects in each initial cluster, use this average value as the new cluster center, and recalculate the Euclidean distance between all data objects and the new cluster center. S325. Determine the final cluster: Repeat steps S321-S324 until the cluster centers no longer change, and the iteration is complete; S326. Estimate the parameters of the probability distribution of the Bayesian network model based on the range of environmental characteristic data and harmful gas concentration data in the cluster.

4. A monitoring system for a confined space monitoring robot based on 5G communication according to claim 3, characterized in that: In step S321, the normalization process is expressed by the following formula: ， In the formula, g(X) is the normalization function; X i This represents environmental characteristic data, where 1 ≤ i ≤ n, and n is the number of samples in the data. X min X represents the minimum value in the dataset. max This represents the maximum value in the dataset.

5. A monitoring system for a confined space monitoring robot based on 5G communication according to claim 3, characterized in that: In step S323, the Euclidean distance is calculated using the following formula: ， In the formula, d(x, C) i ) represents the Euclidean distance function; x is the normalized data object; C i Let x represent the i-th cluster center, 1 ≤ i ≤ k, where k is the number of cluster centers; n is the number of samples in the dataset; j Represents the data object of the j-th variable in the dataset; C ij This represents the cluster center of the j-th variable.

6. A monitoring system for a confined space monitoring robot based on 5G communication according to claim 1, characterized in that: In step S32, the construction of the conditional probability table involves: constructing a conditional probability table for each variable based on the parameter values ​​obtained from parameter estimation; for each variable, given its parent node, calculating its joint probability across multiple parent nodes, using the following formula: ， In the formula, P(U) represents a set of nodes U = B1, B2, ..., B n The joint probability distribution of Pa(B) i ) represents node B in a Bayesian network. i The set of parent nodes; P(B) i ) represents the target node B i The prior probability; n is the number of nodes.

7. A monitoring system for a confined space monitoring robot based on 5G communication according to claim 1, characterized in that: In step S33, the model structure learning method is the Bayesian Information Criterion (BIC), which enumerates all network structures and selects the network structure with the smallest BIC value as the optimal model structure.

Images