A state warning method and system for safety protection device for building construction

Abstract

The present invention discloses a state warning method and system for safety protection devices for construction, which relates to the field of construction technology. The method includes: obtaining multiple state sensor data sets of multiple safety protection devices; performing correlation analysis to generate multiple state companion networks; performing abnormal identification on multiple state sensor data sets to obtain multiple abnormal state features; analyzing multiple abnormal state features according to multiple state companion networks to obtain N companion detection groups; establishing corresponding N Siam‑Net twin recognition networks, making predictions based on the twin recognition results, and obtaining state warning signals. The present invention solves the technical problem that the traditional safety protection device warning in the prior art is a separate warning, and the correlation effect between each state data is not considered, resulting in false alarms and missed alarms, and achieves the technical effect of reducing the probability of false alarms and missed alarms through state information sharing and deep learning between multiple protection devices.

Description

Technical Field

[0001] The present invention relates to the technical field of building construction, and in particular to a state early warning method and system for a safety protection device used in building construction. Background Art

[0002] Construction sites often involve many high-altitude operations, mechanical equipment operations, and complex working environments, so the use of safety protection devices becomes particularly important. As an important part of modern construction safety management, status warning technology aims to timely discover potential safety hazards and prevent safety accidents through real-time monitoring and prediction of the status of safety protection devices. However, the current safety protection device warnings are mostly carried out separately, which is easy to form islands and ignore the correlation between status data, resulting in false alarms and missed alarms, which poses a greater risk to construction. Summary of the invention

[0003] The present application provides a status warning method and system for safety protection devices for construction, which is used to solve the technical problem that the traditional safety protection device warning in the prior art is a separate warning without considering the correlation between the various status data, resulting in false alarms and missed alarms.

[0004] In a first aspect of the present application, a state warning method for a safety protection device for construction is provided, the method comprising: obtaining multiple state sensor data sets for multiple safety protection devices; analyzing the correlations between the multiple safety protection devices to generate multiple state companion networks with the multiple safety protection devices as central nodes, respectively, wherein the correlations between other nodes constituting each state companion network and the central node are greater than a preset correlation; obtaining multiple abnormal state features by performing abnormality identification on the multiple state sensor data sets; analyzing the multiple abnormal state features according to the multiple state companion networks to obtain N companion detection groups corresponding to each state companion network, wherein each companion detection group includes two safety protection devices, and the two safety protection devices belong to the same state companion network; establishing N Siam-Net twin recognition networks corresponding to the N companion detection groups, performing predictions based on the twin recognition results of the N Siam-Net twin recognition networks, and obtaining a state warning signal.

[0005] In a second aspect of the present application, a state warning system for a safety protection device for building construction is provided, the system comprising: a state sensing data set acquisition module, the state sensing data set acquisition module is used to acquire multiple state sensing data sets of multiple safety protection devices; a state accompanying network generation module, the state accompanying network generation module is used to analyze the correlation between the multiple safety protection devices, and generate multiple state accompanying networks with the multiple safety protection devices as central nodes, wherein the correlation between other nodes constituting each state accompanying network and the central node is greater than a preset correlation; an abnormal state feature recognition module, the abnormal state feature recognition module is used to analyze the correlation between the multiple safety protection devices, and generate multiple state accompanying networks with the multiple safety protection devices as central nodes. A state sensing data set is used to perform anomaly identification and obtain multiple abnormal state features; an accompanying detection group acquisition module is used to analyze the multiple abnormal state features according to the multiple state accompanying networks, and obtain N accompanying detection groups corresponding to each state accompanying network, wherein each accompanying detection group includes two safety protection devices, and the two safety protection devices belong to the same state accompanying network; a state warning signal acquisition module is used to establish N Siam-Net twin recognition networks corresponding to the N accompanying detection groups, and make predictions based on the twin recognition results of the N Siam-Net twin recognition networks to obtain a state warning signal.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The present application provides a state warning method for a safety protection device for construction, which relates to the technical field of construction. The method performs abnormal identification on state sensor data of multiple safety protection devices to obtain multiple abnormal state features, performs correlation analysis on multiple safety protection devices, generates multiple state companion networks, analyzes multiple abnormal state features according to the multiple state companion networks, establishes N Siam-Net twin recognition networks, performs prediction according to the twin recognition results, and obtains state warning signals. The method solves the technical problem that the traditional safety protection device warning in the prior art is a separate warning, does not consider the correlation influence between each state data, and causes false alarms and missed alarms, and achieves the technical effect of reducing the probability of false alarms and missed alarms through state information sharing and deep learning among multiple protection devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A schematic diagram of a state warning method for a safety protection device for construction provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of a process for obtaining twin recognition results in a state warning method for a safety protection device for construction provided in an embodiment of the present application;

[0011] Figure 3 A schematic diagram of a process for obtaining a status warning signal in a status warning method for a safety protection device for construction provided in an embodiment of the present application;

[0012] Figure 4 A schematic diagram of the structure of a status warning system for a safety protection device for construction provided in an embodiment of the present application.

[0013] Explanation of the accompanying symbols: state sensing data set acquisition module 11, state accompanying network generation module 12, abnormal state feature recognition module 13, accompanying detection group acquisition module 14, state warning signal acquisition module 15. DETAILED DESCRIPTION

[0014] The present application provides a status warning method for a safety protection device for construction, which is used to solve the technical problem that the traditional safety protection device warning in the prior art is a separate warning without considering the correlation between the status data, resulting in false alarms and missed alarms.

[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0016] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices.

[0017] Embodiment 1

[0018] like Figure 1 As shown, the present application provides a state warning method for a safety protection device for construction, the method comprising:

[0019] P10: Acquire multiple state sensor data sets of multiple safety protection devices;

[0020] Specifically, according to the specific needs of the construction site and the results of the safety risk assessment, appropriate types of safety protection devices are selected, such as gas protection devices, smart wearable protection devices, construction fire protection, etc. Furthermore, through multiple safety protection devices, various types of status data are collected in real time, including gas concentration, temperature, humidity, personnel location, physiological indicators and other information, and the collected status sensor data are preliminarily processed and cleaned to eliminate noise and outliers, and multiple status sensor data sets are obtained to provide a data basis for subsequent early warning analysis.

[0021] P20: Analyze the correlation between the multiple safety protection devices to generate multiple state accompanying networks with the multiple safety protection devices as central nodes, wherein the correlation between other nodes constituting each state accompanying network and the central node is greater than a preset correlation;

[0022] Optionally, the correlation between multiple real-time safety protection devices is deeply analyzed. For example, by obtaining the location, function, working status and historical working data of the multiple safety protection devices, and performing similarity quantification calculations according to the data category, for example, selecting a suitable correlation measurement method, such as Pearson correlation coefficient, Spearman rank correlation coefficient or mutual information, to calculate the correlation between different devices and obtain multiple correlation coefficients. Further, based on the calculated correlation coefficient, a correlation matrix is ​​constructed, and each element in the correlation matrix represents the correlation between two devices. Each safety protection device is taken as a central node. For each central node, the correlation matrix is ​​traversed to find other nodes with a correlation greater than a preset correlation, and these nodes are connected to the central node to form a state accompanying network, wherein the setting of the preset correlation needs to be set according to the actual application scenario, to ensure the connectivity of the network, and to avoid introducing too many noise nodes. By establishing multiple state accompanying networks, the potential associations and mutual influences between safety protection devices are determined, providing strong support for the accurate generation of early warning signals.

[0023] P30: Acquire multiple abnormal state features by performing abnormality identification on the multiple state sensor data sets;

[0024] It should be understood that the abnormality recognition of the multiple state sensor data sets can be performed by establishing an abnormality recognition model, and using the abnormality recognition model to perform abnormality recognition of the multiple state sensor data sets. Specifically, the construction process of the abnormality recognition model can be: based on big data, historical safety protection data is collected, and data differentiation and marking are performed to obtain a normal state data set and an abnormal state data set, and the normal state data set and the abnormal state data set are used as training data, combined with machine learning, such as a neural network model, to perform supervised training until the output data reaches convergence, and the abnormality recognition model is obtained. The multiple state sensor data sets are input into the abnormality recognition model for abnormal feature recognition to obtain multiple abnormal state features, which may include the number of abnormal values, the distribution of abnormal values, the duration of abnormal values, etc., which can reflect the abnormal performance of the safety protection device in different aspects.

[0025] P40: Analyze the multiple abnormal state features according to the multiple state accompanying networks to obtain N accompanying detection groups corresponding to each state accompanying network, wherein each accompanying detection group includes two safety protection devices, and the two safety protection devices belong to the same state accompanying network;

[0026] In a possible embodiment of the present application, the multiple abnormal state features are deeply analyzed according to the multiple state accompanying networks to identify the combination of safety protection devices closely related to the abnormal state features to form an accompanying detection group. Specifically, by analyzing the correlation between the abnormal state features and each node in the state accompanying network, the identified multiple abnormal state features are mapped to the corresponding state accompanying network, and in each state accompanying network, according to the mapping result of the abnormal state features, the nodes most related to the abnormal state features are selected to form a combination of safety protection devices to form an accompanying detection group. In addition, each accompanying detection group contains two safety protection devices belonging to the same state accompanying network.

[0027] Furthermore, for each state accompanying network, the corresponding N accompanying detection groups are obtained, and the specific value of N can be determined according to the actual warning accuracy and efficiency requirements. By obtaining a combination of safety protection devices closely related to the abnormal state characteristics, that is, N accompanying detection groups corresponding to each state accompanying network, an influence association between multiple safety protection devices is established, multiple associated detection relationships are formed, and the comprehensiveness of detection is improved.

[0028] P50: Establish N Siam-Net twin recognition networks corresponding to the N accompanying detection groups, make predictions based on the twin recognition results of the N Siam-Net twin recognition networks, and obtain a status warning signal.

[0029] Specifically, for the N accompanying detection groups, a Siam-Net twin recognition network is established for each accompanying detection group. The Siam-Net is a deep learning network structure commonly used for target tracking and similarity comparison. By training N Siam-Net twin recognition networks, the state changes between different safety protection devices can be compared and identified, thereby predicting possible safety risks.

[0030] The specific process may be to construct a Siam-Net twin recognition network for each accompanying detection group, and the Siam-Net twin recognition network generally includes a feature extraction layer, a similarity comparison layer, and an output layer. The feature extraction layer is used to extract meaningful features from the input data, the similarity comparison layer calculates the similarity or difference between these features, and the output layer gives the final recognition result. The Siam-Net twin recognition network can be trained using historical data and known security events, and combined with real-time status data to predict device status and obtain status warning signals.

[0031] Further, such as Figure 2 As shown, step P50 of the embodiment of the present application also includes:

[0032] P51: Obtain the twin recognition result, wherein the twin recognition result includes an identification companion detection group;

[0033] P52: performing node coincidence identification on the identification accompanying detection group to obtain a coincidence accompanying detection group, wherein the coincidence accompanying detection group is composed of two accompanying detection groups including a coincidence safety protection device;

[0034] P53: Collecting the abnormal state feature group corresponding to the coincidence detection group;

[0035] P54: Obtaining a security risk indicator by predicting the abnormal state feature group;

[0036] P55: When the safety risk index is greater than the preset safety risk index, a status warning signal corresponding to the overlap companion detection group is generated.

[0037] It should be understood that by using N Siam-Net twin recognition networks, the N accompanying detection groups are tested for the similarity of abnormal features to obtain twin recognition results, which are composed of multiple identified accompanying detection groups. The identified accompanying detection groups refer to the accompanying detection groups that are judged to have similar abnormal features in the abnormal feature similarity detection.

[0038] Furthermore, node overlap identification is performed on the identified accompanying detection groups to check whether different accompanying detection groups contain the same safety protection devices. If two accompanying detection groups contain the same safety protection devices, they are determined to be overlapping accompanying detection groups. Overlapping accompanying detection groups may point to the same potential safety risk or failure mode, which is crucial for subsequent safety risk analysis.

[0039] Furthermore, the abnormal state feature group corresponding to the overlapped accompanying detection group is collected. The abnormal state feature group is a set of state features identified as abnormal in the overlapped accompanying detection group, which may include sensor readings, equipment status, environmental changes and other information.

[0040] Furthermore, by predicting the abnormal state feature group, a safety risk index is obtained. The prediction process may process and analyze the abnormal state feature group by using a machine learning model, statistical analysis or other prediction algorithms, thereby obtaining a quantitative safety risk index. When the safety risk index is greater than a preset safety risk index, a state warning signal corresponding to the coincidence detection group is generated. The preset safety risk index can be set according to actual application scenarios and requirements. The generation of the state warning signal can trigger an alarm, notify relevant personnel or automatically take safety measures to ensure timely response and handling of potential safety risks.

[0041] Further, such as Figure 3 As shown, step P51 of the embodiment of the present application also includes:

[0042] P51-1: Using the N Siam-Net twin recognition networks, perform accompanying detection of abnormal feature similarity on the N accompanying detection groups to obtain N accompanying similarities;

[0043] P51-2: Judge the N accompanying similarities, identify the accompanying detection group that is within the preset similarity threshold, output the identified accompanying detection group, and output the identified accompanying detection group as the twin recognition result.

[0044] Furthermore, step P51-1 of the embodiment of the present application also includes:

[0045] P51-11: Setting a feature mapping space, wherein the feature mapping space includes a plurality of mapping indicators, and the plurality of mapping indicators include abnormal target similarity, feature change similarity, and protection function similarity;

[0046] P51-12: The Siam-Net twin recognition network uses the feature mapping space to perform high-dimensional mapping on the abnormal features of each accompanying detection group, and outputs target similarity, change similarity and functional similarity;

[0047] P51-13: Perform weight calculation based on the target similarity, change similarity and function similarity, and output the accompanying similarity.

[0048] It should be understood that the Siam-Net twin recognition network is used to perform a companion detection of the abnormal feature similarity of the accompanying detection group, and output the identification of the accompanying detection group as the twin recognition result. Specifically, first, a feature mapping space is set, and the feature mapping space includes multiple mapping indicators for comprehensively evaluating and quantifying the similarity of abnormal features in the accompanying detection group. And the multiple mapping indicators include abnormal target similarity, feature change similarity, and protection function similarity. The abnormal target similarity is used to evaluate the detection consistency of abnormal targets by different devices in the accompanying detection group. The feature change similarity is used to analyze the similarity of the changing trends of the features of different devices in the accompanying detection group over time or other conditions. The protection function similarity is used to measure the similarity or complementarity of the security protection functions of different devices in the accompanying detection group.

[0049] Furthermore, the Siam-Net twin recognition network is used to map the abnormal features of each accompanying detection group into the feature mapping space. The mapping process is a high-dimensional transformation that can convert the original abnormal feature data into a metric that can reflect similarity, and obtain the target similarity, change similarity and functional similarity of each accompanying detection group, and provide a quantitative description of the similarity of abnormal features between devices in each accompanying detection group. Furthermore, according to the importance of different mapping indicators in the overall evaluation, a corresponding weight is assigned to each mapping indicator, and the target similarity, change similarity and functional similarity are weightedly calculated to obtain the accompanying similarity, which can fully reflect the similarity of abnormal features in multiple dimensions.

[0050] Furthermore, the N accompanying similarities are judged by comparing with a preset similarity threshold. When any accompanying similarity exceeds the preset similarity threshold, it is considered that there is a high abnormal feature similarity between the accompanying detection groups, which may point to the same potential safety risk or failure mode. The accompanying detection group is then identified and output as an identified accompanying detection group, and the identified accompanying detection group is output as the twin recognition result for subsequent node overlap identification and safety risk prediction.

[0051] Furthermore, step P54 of the embodiment of the present application also includes:

[0052] P54-1: Outputting a plurality of learned abnormal state features by performing federated learning on the abnormal state feature group;

[0053] P54-2: Send multiple learned abnormal state features to the safety protection devices corresponding to the overlapping detection group, perform predictions according to their corresponding safety risk prediction modules, and obtain safety risk indicators corresponding to each safety protection device.

[0054] Optionally, security risk indicators are obtained by predicting feature groups that identify abnormal states. First, federated learning is used to process feature groups that identify abnormal states. Federated learning is a machine learning framework that allows multiple parties to jointly train a model without sharing original data, which not only protects data privacy but also can improve model performance by using data from multiple parties.

[0055] Specifically, the abnormal state feature data from different safety protection devices are collected, including sensor readings, device status, log information, etc., to form a data set for federated learning. Further, a suitable algorithm model is selected, such as a neural network, decision tree or support vector machine, etc. In the federated learning framework, each safety protection device is taken as a participant, and the abnormal state feature data corresponding to each participant is used for training to obtain the abnormal recognition model of each safety protection device. A global update model is obtained through model aggregation, and the aggregated global model update is distributed to each participant. After receiving the update, the participant applies it to the local model and continues the next round of training. Through multiple iterations, until the model reaches the preset performance requirements or meets the stop condition.

[0056] After multiple rounds of federated learning training, an optimized machine learning model is obtained, and the optimized machine learning model is used to identify and analyze abnormal state features, and multiple learned abnormal state features are output. Further, the multiple learned abnormal state features are sent to the safety protection device corresponding to the overlapped companion detection group, and each safety protection device is equipped with a safety risk prediction module. These modules can be models based on machine learning or other prediction algorithms. According to the corresponding safety risk prediction modules, the received abnormal state features are predicted to obtain the safety risk index corresponding to each safety protection device, and the safety risk index is the degree of safety risk faced by the safety protection device.

[0057] Furthermore, step P55 of the embodiment of the present application also includes:

[0058] P55-1: Acquire each safety protection device included in the coincidence detection group;

[0059] P55-2: Determine the address where the warning signal of each safety protection device is sent;

[0060] P55-3: If the warning signal sending addresses of various safety protection devices are the same, a fusion status warning signal is generated for unified warning.

[0061] Specifically, when the security risk index is greater than the preset security risk index, a status warning signal needs to be generated to remind relevant personnel to take corresponding measures. First, the various security protection devices included in the overlap detection group are obtained. These devices may be distributed in different locations and are responsible for monitoring and protecting different key assets or areas. The identity and location information of these devices are obtained by searching the database or querying relevant configuration information.

[0062] Furthermore, the warning signal sending address of each of the safety protection devices is determined. The warning signal sending address may be an email address, a text message receiving number, a specific monitoring platform interface, etc. If the warning signal sending addresses of each of the safety protection devices are the same address, that is, the warning signal needs to be sent to the same receiving point, a fusion state warning signal is generated. The fusion state warning signal is a summary and integration of the security risk information of multiple devices, including the security risk information of all relevant devices in the overlapping accompanying detection group, which is used for unified warning. The generation of a fusion state warning signal can centrally display the security risk information of multiple devices, so that relevant personnel can have a more comprehensive understanding of the current security risk status, and thus make more accurate judgments and countermeasures. At the same time, through unified warning, information redundancy and confusion caused by multiple devices sending warning signals separately can also be avoided.

[0063] Furthermore, step P55-3 of the embodiment of the present application also includes:

[0064] P55-3a: If the warning signal sending addresses of the various safety protection devices are different addresses, corresponding status warning signals are generated according to the various safety protection devices, and multiple status warning signals are output;

[0065] P55-3b: Issue warnings separately according to the target addresses corresponding to the multiple status warning signals.

[0066] Exemplarily, if the warning signal sending addresses of each safety protection device in the overlapping detection group are different, that is, the warning signal needs to be sent to different receiving points, a corresponding status warning signal is generated according to the specific situation of each safety protection device, including the corresponding risk type, level, location, etc., and separate warnings are issued according to the target addresses corresponding to the multiple status warning signals. The warning signal of each safety protection device will be sent to the designated recipient to ensure that relevant personnel can receive the corresponding warning information in a timely manner.

[0067] In summary, the embodiments of the present application have at least the following technical effects:

[0068] The present application performs abnormality identification on the state sensor data of multiple safety protection devices, obtains multiple abnormal state characteristics, analyzes the correlation of multiple safety protection devices, generates multiple state companion networks, analyzes multiple abnormal state characteristics based on the multiple state companion networks, establishes N Siam-Net twin recognition networks, makes predictions based on the twin recognition results, and obtains state warning signals.

[0069] The technical effect of reducing the probability of false alarms and missed alarms is achieved through status information sharing and deep learning among multiple protective devices.

[0070] Embodiment 2

[0071] Based on the same inventive concept as the state warning method of a safety protection device for construction in the aforementioned embodiment, Figure 4 As shown, the present application provides a state warning system for safety protection devices for construction, and the system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0072] A state sensing data set acquisition module 11, wherein the state sensing data set acquisition module 11 is used to acquire multiple state sensing data sets of multiple safety protection devices;

[0073] A state accompanying network generation module 12, wherein the state accompanying network generation module 12 is used to analyze the correlation between the multiple safety protection devices, and generate multiple state accompanying networks with the multiple safety protection devices as central nodes, wherein the correlation between other nodes constituting each state accompanying network and the central node is greater than a preset correlation;

[0074] An abnormal state feature recognition module 13, the abnormal state feature recognition module 13 is used to obtain multiple abnormal state features by performing abnormality recognition on the multiple state sensor data sets;

[0075] An accompanying detection group acquisition module 14, the accompanying detection group acquisition module 14 is used to analyze the multiple abnormal state characteristics according to the multiple state accompanying networks, and obtain N accompanying detection groups corresponding to each state accompanying network, wherein each accompanying detection group includes two safety protection devices, and the two safety protection devices belong to the same state accompanying network;

[0076] The status warning signal acquisition module 15 is used to establish N Siam-Net twin recognition networks corresponding to the N accompanying detection groups, make predictions based on the twin recognition results of the N Siam-Net twin recognition networks, and obtain status warning signals.

[0077] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0078] Using the N Siam-Net twin recognition networks, perform accompanying detection of abnormal feature similarities on the N accompanying detection groups to obtain N accompanying similarities;

[0079] The N accompanying similarities are judged, the accompanying detection group within the preset similarity threshold is identified, the identified accompanying detection group is output, and the identified accompanying detection group is output as the twin recognition result.

[0080] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0081] Obtaining the twin recognition result, wherein the twin recognition result includes an identification companion detection group;

[0082] Perform node coincidence identification on the identification accompanying detection group to obtain a coincidence accompanying detection group, wherein the coincidence accompanying detection group is composed of two accompanying detection groups including a coincidence safety protection device;

[0083] Collecting an abnormal state feature group corresponding to the coincidence detection group;

[0084] Obtaining a security risk indicator by predicting the abnormal state feature group;

[0085] When the safety risk index is greater than a preset safety risk index, a status warning signal corresponding to the overlap companion detection group is generated.

[0086] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0087] Outputting a plurality of learned abnormal state features by performing federated learning on the abnormal state feature group;

[0088] The multiple learned abnormal state features are sent to the safety protection devices corresponding to the overlapped accompanying detection group, and predictions are performed according to the corresponding safety risk prediction modules to obtain the safety risk indicators corresponding to each safety protection device.

[0089] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0090] Acquire each safety protection device included in the coincidence detection group;

[0091] Determine the address where the warning signal of each safety protection device is sent;

[0092] If the warning signal sending addresses of various safety protection devices are the same address, a fusion state warning signal is generated for unified warning.

[0093] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0094] If the warning signal sending addresses of the various safety protection devices are different addresses, corresponding status warning signals are generated according to the various safety protection devices, and multiple status warning signals are output;

[0095] Warnings are issued separately according to the target addresses respectively corresponding to the multiple status warning signals.

[0096] Furthermore, the status warning signal acquisition module 15 is also used to perform the following steps:

[0097] Setting a feature mapping space, wherein the feature mapping space includes a plurality of mapping indicators, and the plurality of mapping indicators include abnormal target similarity, feature change similarity, and protection function similarity;

[0098] The Siam-Net twin recognition network uses the feature mapping space to perform high-dimensional mapping on the abnormal features of each accompanying detection group, and outputs target similarity, change similarity and functional similarity;

[0099] A weight calculation is performed according to the target similarity, the change similarity and the function similarity, and the accompanying similarity is output.

[0100] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0101] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0102] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A state warning method for a safety protection device for construction, characterized in that: The method comprises: Acquire multiple state sensor data sets of multiple safety protection devices, and collect various types of state data in real time through the multiple safety protection devices, including gas concentration, temperature, humidity, personnel position, physiological indicators and other information; Analyze the correlations between the multiple safety protection devices, generate multiple state accompanying networks with the multiple safety protection devices as central nodes, obtain the locations, functions, working states and historical working data of the multiple safety protection devices, perform similarity quantification calculations according to data categories, select a correlation measurement method to calculate the correlations between different devices, and obtain multiple correlation coefficients, wherein the correlations between other nodes constituting each state accompanying network and the central node are greater than a preset correlation; Acquire multiple abnormal state features by performing abnormality identification on the multiple state sensor data sets; Analyze the multiple abnormal state features according to the multiple state accompanying networks to obtain N accompanying detection groups corresponding to each state accompanying network, wherein each accompanying detection group includes two safety protection devices, and the two safety protection devices belong to the same state accompanying network; Establishing N Siam-Net twin recognition networks corresponding to the N accompanying detection groups, making predictions based on the twin recognition results of the N Siam-Net twin recognition networks, and obtaining a status warning signal; Wherein, the acquisition status warning signal includes: Obtaining the twin recognition result, wherein the twin recognition result includes an identification companion detection group; Perform node coincidence identification on the identification accompanying detection group to obtain a coincidence accompanying detection group, wherein the coincidence accompanying detection group is composed of two accompanying detection groups including a coincidence safety protection device; Collecting an abnormal state feature group corresponding to the coincidence detection group; Obtaining a security risk indicator by predicting the abnormal state feature group; When the safety risk index is greater than a preset safety risk index, a state warning signal corresponding to the coincidence detection group is generated; The method of obtaining the security risk index by predicting the abnormal state feature group includes: Outputting a plurality of learned abnormal state features by performing federated learning on the abnormal state feature group; Sending a plurality of learned abnormal state features to the safety protection devices corresponding to the overlapped accompanying detection group, performing predictions according to the corresponding safety risk prediction modules, and obtaining safety risk indicators corresponding to each safety protection device; Wherein, when the safety risk index is greater than a preset safety risk index, a state warning signal corresponding to the coincidence detection group is generated, and the method further comprises: Acquire each safety protection device included in the coincidence detection group; Determine the address where the warning signal of each safety protection device is sent; If the warning signal sending addresses of various safety protection devices are the same, a fusion state warning signal is generated for unified warning; Wherein, if the warning signal sending addresses of each safety protection device are different addresses, a corresponding status warning signal is generated according to each safety protection device, and multiple status warning signals are output; Warnings are issued separately according to the target addresses respectively corresponding to the multiple status warning signals.

2. The method according to claim 1, characterized in that Prediction is performed based on the twin recognition results of the N Siam-Net twin recognition networks, where the calculation of the twin recognition results includes: Using the N Siam-Net twin recognition networks, perform accompanying detection of abnormal feature similarities on the N accompanying detection groups to obtain N accompanying similarities; The N accompanying similarities are judged, the accompanying detection group within the preset similarity threshold is identified, the identified accompanying detection group is output, and the identified accompanying detection group is output as the twin recognition result.

3. The method according to claim 2, characterized in that Using the N Siam-Net twin recognition networks to perform an accompanying detection of abnormal feature similarity on the N accompanying detection groups includes: Setting a feature mapping space, wherein the feature mapping space includes a plurality of mapping indicators, and the plurality of mapping indicators include abnormal target similarity, feature change similarity, and protection function similarity; The Siam-Net twin recognition network uses the feature mapping space to perform high-dimensional mapping on the abnormal features of each accompanying detection group, and outputs target similarity, change similarity and functional similarity; A weight calculation is performed according to the target similarity, the change similarity and the function similarity, and the accompanying similarity is output.

4. A status warning system for safety protection devices for construction, characterized in that: For executing the method according to any one of claims 1 to 3, the system comprises: A state sensing data set acquisition module, which is used to acquire multiple state sensing data sets of multiple safety protection devices, and collect various types of state data in real time through the multiple safety protection devices, including gas concentration, temperature, humidity, personnel position, physiological indicators and other information; A state accompanying network generation module, wherein the state accompanying network generation module is used to analyze the correlation between the multiple safety protection devices, generate multiple state accompanying networks with the multiple safety protection devices as central nodes, obtain the positions, functions, working states and historical working data of the multiple safety protection devices, perform similarity quantification calculations according to data categories, select a correlation measurement method to calculate the correlation between different devices, and obtain multiple correlation coefficients, wherein the correlation between other nodes constituting each state accompanying network and the central node is greater than a preset correlation; An abnormal state feature recognition module, the abnormal state feature recognition module is used to obtain multiple abnormal state features by performing abnormality recognition on the multiple state sensor data sets; An accompanying detection group acquisition module, the accompanying detection group acquisition module is used to analyze the multiple abnormal state features according to the multiple state accompanying networks, and obtain N accompanying detection groups corresponding to each state accompanying network, wherein each accompanying detection group includes two safety protection devices, and the two safety protection devices belong to the same state accompanying network; A status warning signal acquisition module, wherein the status warning signal acquisition module is used to establish N Siam-Net twin recognition networks corresponding to the N accompanying detection groups, perform predictions based on the twin recognition results of the N Siam-Net twin recognition networks, and acquire status warning signals.

Images