A method for identifying major risk sources in highway construction
Through the CNN neural network model and clustering algorithm, combined with the emergency repair log of highway construction project and real-time monitoring data, the automatic identification of major risk sources in highway construction is achieved, and the problem of poor real-time identification of risk sources in the existing technology is solved, and the identification is improved Efficiency and precision.
Patent Information
- Application Number
- CN202211082593.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-09-06
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-09-06
AI Technical Summary
In the prior art, all risk sources need to be judged one by one, resulting in poor real-time identification of risk sources in highway construction and affecting the effectiveness of construction risk management.
The CNN neural network model is used to build a risk source identification model based on the emergency repair log of the highway construction project. The correlation analysis is carried out through feature extraction and clustering algorithms, and the construction data is monitored in real time. The principal component analysis is used to improve the risk source identification efficiency and realize the automation of major risk sources. Identify.
There is no need to make judgments on all risk sources one by one, which improves the efficiency and accuracy of risk sources identification, ensures the real-timeness of identification, and minimizes the frequency of accidents.
Smart Images

Figure CN115481880B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of highway construction, and in particular to a method for identifying major risk sources in highway construction. Background Art
[0002] With the increasingly accelerated pace of economic development, all walks of life have achieved rapid development. In order to better promote the construction of my country's infrastructure, my country's highway construction projects have also been significantly enhanced compared with before. In all stages of highway construction, highway roadbed engineering is one of the most important contents. Its quality is directly related to the quality of the highway, providing the most reliable guarantee for people's safe travel.
[0003] Safety risks during the construction period are centered around personal safety. Anything that affects personal safety is called a risk source and is a potential factor that causes safety accidents. Risk identification is the first important task to be carried out when conducting risk analysis. Risk identification must at least be able to solve three problems: first, what risks should be considered, second, what are the main factors that cause these risks, and third, what is the severity of the consequences caused by these risks.
[0004] Among the major dangerous events that occur during highway construction, risk sources are the main cause of dangerous events. This is also the focus of highway construction safety management departments. Only by classifying and handling risk sources in a systematic analysis can we fundamentally improve the safety factor in highway construction and minimize the frequency of accidents. Therefore, there are many ways to identify risk sources in highway construction, but these methods require one-by-one judgment of all risk sources, resulting in poor real-time identification, which ultimately affects the effect of risk source identification and increases construction risks. Summary of the invention
[0005] The purpose of the present invention is to provide a method for identifying major risk sources in highway construction, so as to solve the technical problem in the prior art that all risk sources need to be judged one by one, resulting in poor real-time identification.
[0006] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:
[0007] A method for identifying major risk sources in highway construction comprises the following steps:
[0008] Step S1, using the highway construction project emergency repair log to construct an identification sample for risk source identification, and using a CNN neural network model to perform convolution training based on the identification sample to obtain a risk source identification model;
[0009] Step S2: extracting features from the identified samples of each risk source category to obtain risk source associated features, and using a clustering algorithm to perform association analysis on each risk source category based on the risk source associated features to obtain associated risk source categories for each risk source category;
[0010] Step S3: monitor the construction data of the target highway in real time, and perform principal component analysis on the construction data to obtain real-time principal identification data; input the real-time principal identification data into the risk source identification model to obtain the real-time principal risk source category of the target highway; and then use the real-time principal risk source category and the associated risk source category of the real-time principal risk source category as the real-time risk source category of the target highway, so as to improve the efficiency of risk source identification by using association analysis.
[0011] As a preferred solution of the present invention, the method of using the highway construction engineering emergency repair log to construct an identification sample for risk source identification includes:
[0012] Extract the risk source categories at the emergency repair sequence and the construction data for risk source identification at the emergency repair sequence from the emergency repair log of the highway construction project, and organize the risk source categories and construction data into identification samples according to the emergency repair sequence;
[0013] The expression of the identification sample is: {(Y i ,X i )|i∈[1,N]}, where Y i Characterized as the i-th risk source category, X i It is represented as the construction data for identifying the i-th risk source category, N is the total number of risk source categories, and i is a measurement constant.
[0014] As a preferred solution of the present invention, the risk source identification model is obtained by performing convolution training based on identification samples using a CNN neural network model, including:
[0015] Using the construction data in the identification sample as an input item of the CNN neural network, using the risk source category in the identification sample as an output item of the CNN neural network, and using the CNN neural network model to perform convolution training based on the input item and the output item to obtain the risk source identification model;
[0016] The model expression of the risk source identification model is:
[0017] Y = CNN(X);
[0018] Where Y is the model identifier of the risk source category, X is the model identifier of the construction data, and CNN is the model identifier of the CNN neural network model.
[0019] As a preferred solution of the present invention, the step of extracting features from the identification samples of each risk source category to obtain risk source associated features includes:
[0020] The construction data acquisition time sequence in the identification sample of each risk source category is sequentially changed from the emergency repair time sequence to the preset time sequence to obtain the preset time sequence, and the construction data sequence of each risk source category is obtained between the preset time sequence and the emergency repair time sequence;
[0021] The risk source association feature is obtained by extracting the data change rate feature and the data change trend feature of each construction data component in the construction data sequence in turn. The data change rate feature and the data change trend feature are obtained by analyzing the construction data components by plotting them as two-dimensional curves in time series.
[0022] As a preferred solution of the present invention, the method of using a clustering algorithm to perform association analysis on each risk source category based on risk source association characteristics to obtain an associated risk source category for each risk source category includes:
[0023] Taking each risk source category as a cluster center in turn, and using a clustering algorithm to cluster each risk source category based on the risk source association characteristics to obtain a category set for each risk source category;
[0024] In the category set of each risk source category, the intra-set distance between each non-cluster center risk source category and the cluster center is calculated in turn, and the risk source category of the non-cluster center in the category set whose intra-set distance is less than the preset distance is taken as the associated risk source category of the risk source category in the category set, so as to obtain an associated query table containing each risk source category and the associated risk source category of the risk source category.
[0025] As a preferred solution of the present invention, the step of performing principal component analysis on the construction data to obtain real-time principal identification data includes:
[0026] The acquisition time sequence of the construction data of the target highway is obtained by presetting the real-time time sequence by a preset time length to obtain the analysis time sequence, and the construction data sequence of the target highway is obtained between the analysis time sequence and the real-time time sequence;
[0027] The construction data sequence of the target highway is subjected to principal component analysis to obtain the data contribution value of each construction data component in the construction data, and the construction data components whose data contribution values are higher than the preset threshold of data contribution are taken as the main identification data components, and the data values of the main identification data components at the real-time time series are combined into real-time main identification data.
[0028] As a preferred solution of the present invention, the step of inputting the real-time main identification data into the risk source identification model to obtain the real-time main risk source category of the target highway includes:
[0029] The real-time main identification data is input into the risk source identification model, and the risk source identification model outputs the real-time main risk source category of the target highway.
[0030] As a preferred solution of the present invention, the associated risk source category of the real-time main risk source category is obtained by querying the associated query table.
[0031] As a preferred solution of the present invention, the construction data is normalized after being acquired, and data calculation is performed after the normalization process.
[0032] As a preferred solution of the present invention, the calculation formula of the distance within the set is:
[0033]
[0034] Where D k is the distance within the set of the kth risk source category in the category set, X k is the risk source association feature of the kth risk source category at the non-cluster center in the category set, X center is the risk source association characteristic of the risk source category at the cluster center within the category set, and k is a measurement constant.
[0035] Compared with the prior art, the present invention has the following beneficial effects:
[0036] The present invention constructs a risk source identification model to realize the automatic identification of construction risk sources, and uses a clustering algorithm to perform association analysis on each risk source category based on the risk source association characteristics to obtain the associated risk source category of each risk source category, so as to improve the efficiency of risk source identification by association analysis, without having to judge all risk sources one by one, and ensuring the real-time performance of identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.
[0038] Figure 1 A flow chart of a method for identifying major risk sources in highway construction provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0040] like Figure 1 As shown, the present invention provides a method for identifying major risk sources in highway construction, comprising the following steps:
[0041] Step S1, using the highway construction project emergency repair log to construct an identification sample for risk source identification, and using a CNN neural network model to perform convolution training based on the identification sample to obtain a risk source identification model;
[0042] The highway construction repair log is used to construct identification samples for risk source identification, including:
[0043] Extract the risk source categories at the emergency repair sequence and the construction data for risk source identification at the emergency repair sequence from the emergency repair log of the highway construction project, and organize the risk source categories and construction data into identification samples according to the emergency repair sequence;
[0044] The expression of the identification sample is: {(Y i ,X i )|i∈[1,N]}, where Y i Characterized as the i-th risk source category, X i It is represented as the construction data for identifying the i-th risk source category, N is the total number of risk source categories, and i is a measurement constant.
[0045] The CNN neural network model is used to perform convolution training based on identification samples to obtain a risk source identification model, including:
[0046] The construction data in the identification sample is used as the input item of the CNN neural network, and the risk source category in the identification sample is used as the output item of the CNN neural network. The CNN neural network model is used to perform convolution training based on the input item and the output item to obtain the risk source identification model.
[0047] The model expression of the risk source identification model is:
[0048] Y = CNN(X);
[0049] Where Y is the model identifier of the risk source category, X is the model identifier of the construction data, and CNN is the model identifier of the CNN neural network model.
[0050] A risk source identification model is constructed, which can identify the risk sources that occur during the construction process through construction data, thereby replacing manual data comparison and analysis to complete risk source identification. The risk source identification model can avoid the randomness of manual identification and improve the efficiency and accuracy of risk source identification.
[0051] Step S2: extracting features from the identified samples of each risk source category to obtain risk source associated features, and using a clustering algorithm to perform association analysis on each risk source category based on the risk source associated features to obtain associated risk source categories for each risk source category;
[0052] The feature extraction of each risk source category identification sample obtains the risk source associated features, including:
[0053] The construction data acquisition time sequence in the identification sample of each risk source category is sequentially changed from the emergency repair time sequence to the preset time sequence to obtain the preset time sequence, and the construction data sequence of each risk source category is obtained between the preset time sequence and the emergency repair time sequence;
[0054] In the construction data sequence, the data change rate characteristics and data change trend characteristics are extracted for each construction data component in turn to obtain the risk source association characteristics. The risk source association characteristics can be added, deleted or modified as needed during actual use. The data change rate characteristics and data change trend characteristics are obtained by analyzing the construction data components by plotting them as two-dimensional curves in time series.
[0055] The clustering algorithm is used to perform association analysis on each risk source category based on the risk source association characteristics to obtain the associated risk source category of each risk source category, including:
[0056] Each risk source category is taken as the cluster center in turn, and the clustering algorithm is used to cluster each risk source category based on the risk source association characteristics to obtain the category set of each risk source category. Each risk source category in the same category set has similar risk source association characteristics, that is, the same data change rate characteristics and data change trend characteristics. Similar data change rate characteristics and data change trend characteristics indicate that the risk source categories have similar response characteristics, showing the correlation between risk source categories. For example, the construction data in risk source category A and risk source category B both change from time series 1 to time series 2, and the change rate state and change trend state are similar, which means that risk source category A and risk source category B show a certain degree of correlation, that is, the occurrence of risk source category A will be accompanied by risk source category B, or the occurrence of risk source category B will be accompanied by risk source category A. Therefore, it is only necessary to identify one of the risk source categories A or B. After identifying the risk source category A or B, it is known that the risk source category B or A will occur. A can be used as the associated risk source category of B, or B can be used as the associated risk source category of A.
[0057] In the category set of each risk source category, the intra-set distance between each non-cluster center risk source category and the cluster center is calculated in turn, and the risk source category of the non-cluster center whose intra-set distance in the category set is less than the preset distance is taken as the associated risk source category of the risk source category in the category set, so as to obtain an associated query table containing each risk source category and the associated risk source category of the risk source category.
[0058] The calculation formula for the distance within a set is:
[0059]
[0060] Where D k is the distance within the set of the kth risk source category in the category set, X k is the risk source association feature of the kth risk source category at the non-cluster center in the category set, X center is the risk source association characteristic of the risk source category at the cluster center within the category set, and k is a measurement constant.
[0061] By conducting correlation analysis on each risk source category, the associated risk source category of each risk source category can be identified. Then, when subsequently identifying the risk source, the main risk source category can be mined, and after association, more risk source categories can be mined in depth to increase the breadth of risk source category identification, achieve multiple identifications with a single identification, and improve identification efficiency.
[0062] Step S3: monitor the construction data of the target highway in real time, and perform principal component analysis on the construction data to obtain real-time principal identification data; input the real-time principal identification data into the risk source identification model to obtain the real-time principal risk source category of the target highway; and then use the real-time principal risk source category and the associated risk source category of the real-time principal risk source category as the real-time risk source category of the target highway, so as to improve the efficiency of risk source identification by using association analysis.
[0063] The construction data is subjected to principal component analysis to obtain real-time principal identification data, including:
[0064] The acquisition time sequence of the construction data of the target highway is obtained by presetting the real-time time sequence by a preset time length to obtain the analysis time sequence, and the construction data sequence of the target highway is obtained between the analysis time sequence and the real-time time sequence;
[0065] The construction data sequence of the target highway is subjected to principal component analysis to obtain the data contribution value of each construction data component in the construction data, and the construction data components whose data contribution values are higher than the preset threshold of data contribution are taken as the main identification data components, and the data values of the main identification data components at the real-time time series are combined into real-time main identification data.
[0066] Input the real-time main identification data into the risk source identification model to obtain the real-time main risk source categories of the target highway, including:
[0067] The real-time main identification data is input into the risk source identification model, and the risk source identification model outputs the real-time main risk source category of the target highway.
[0068] The associated risk source categories of the real-time main risk source categories are obtained by querying the associated query table.
[0069] The construction data are normalized after acquisition, and data calculation is performed after normalization.
[0070] The present invention constructs a risk source identification model to realize the automatic identification of construction risk sources, and uses a clustering algorithm to perform association analysis on each risk source category based on the risk source association characteristics to obtain the associated risk source category of each risk source category, so as to improve the efficiency of risk source identification by association analysis, without having to judge all risk sources one by one, and ensuring the real-time performance of identification.
[0071] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.
Claims
1. A method for identifying major risk sources in highway construction, characterized in that: The following steps are involved: Step S1, using the highway construction project emergency repair log to construct an identification sample for risk source identification, and using a CNN neural network model to perform convolution training based on the identification sample to obtain a risk source identification model; Step S2: extracting features from the identified samples of each risk source category to obtain risk source associated features, and using a clustering algorithm to perform association analysis on each risk source category based on the risk source associated features to obtain associated risk source categories for each risk source category; Step S3, real-time monitoring of the construction data of the target highway, and performing principal component analysis on the construction data to obtain real-time principal identification data, inputting the real-time principal identification data into the risk source identification model to obtain the real-time principal risk source category of the target highway, and then using the real-time principal risk source category and the associated risk source category of the real-time principal risk source category as the real-time risk source category of the target highway, so as to improve the risk source identification efficiency by using the association analysis; The identification samples for risk source identification are constructed by using the highway construction engineering emergency repair log, including: Extract the risk source categories at the emergency repair sequence and the construction data for risk source identification at the emergency repair sequence from the emergency repair log of the highway construction project, and organize the risk source categories and construction data into identification samples according to the emergency repair sequence; The expression of the identification sample is: {(Y i ,X i )|i∈[1,N]}, where Y i Characterized as the i-th risk source category, X i It is represented as the construction data for identifying the i-th risk source category, N is the total number of risk source categories, and i is a measurement constant; The CNN neural network model is used to perform convolution training based on identification samples to obtain a risk source identification model, including: Using the construction data in the identification sample as an input item of the CNN neural network, using the risk source category in the identification sample as an output item of the CNN neural network, and using the CNN neural network model to perform convolution training based on the input item and the output item to obtain the risk source identification model; The model expression of the risk source identification model is: Y = CNN(X); In the formula, Y is the model identifier of the risk source category, X is the model identifier of the construction data, and CNN is the model identifier of the CNN neural network model; The feature extraction of the identification samples of each risk source category to obtain risk source associated features includes: The construction data acquisition time sequence in the identification sample of each risk source category is sequentially changed from the emergency repair time sequence to the preset time sequence to obtain the preset time sequence, and the construction data sequence of each risk source category is obtained between the preset time sequence and the emergency repair time sequence; Extracting the data change rate feature and the data change trend feature of each construction data component in the construction data sequence in turn to obtain the risk source association feature, wherein the data change rate feature and the data change trend feature are obtained by analyzing the construction data components by plotting them as a two-dimensional curve in time series; The clustering algorithm is used to perform association analysis on each risk source category based on the risk source association characteristics to obtain the associated risk source category of each risk source category, including: Taking each risk source category as a cluster center in turn, and using a clustering algorithm to cluster each risk source category based on the risk source association characteristics to obtain a category set for each risk source category; In the category set of each risk source category, the intra-set distance between each non-cluster center risk source category and the cluster center is calculated in turn, and the risk source category of the non-cluster center in the category set whose intra-set distance is less than the preset distance is taken as the associated risk source category of the risk source category in the category set, so as to obtain an associated query table containing each risk source category and the associated risk source category of the risk source category.
2. A method for identifying major risk sources in highway construction according to claim 1, characterized in that: The method of performing principal component analysis on the construction data to obtain real-time principal identification data includes: The acquisition time sequence of the construction data of the target highway is obtained by presetting the real-time time sequence by a preset time length to obtain the analysis time sequence, and the construction data sequence of the target highway is obtained between the analysis time sequence and the real-time time sequence; The construction data sequence of the target highway is subjected to principal component analysis to obtain the data contribution value of each construction data component in the construction data, and the construction data components whose data contribution values are higher than the preset threshold of data contribution are taken as the main identification data components, and the data values of the main identification data components at the real-time time series are combined into real-time main identification data.
3. A method for identifying major risk sources in highway construction according to claim 2, characterized in that: The step of inputting the real-time main identification data into the risk source identification model to obtain the real-time main risk source category of the target highway includes: The real-time main identification data is input into the risk source identification model, and the risk source identification model outputs the real-time main risk source category of the target highway.
4. A method for identifying major risk sources in highway construction according to claim 3, characterized in that: The associated risk source category of the real-time main risk source category is obtained by querying the associated query table.
5. A method for identifying major risk sources in highway construction according to claim 4, characterized in that: The construction data is normalized after being acquired, and data calculation is performed after the normalization process.
6. A method for identifying major risk sources in highway construction according to claim 1, characterized in that: The calculation formula of the distance within the set is: Where D k is the distance within the set of the kth risk source category in the category set, X k is the risk source association feature of the kth risk source category at the non-cluster center in the category set, X center is the risk source association characteristic of the risk source category at the cluster center within the category set, and k is a measurement constant.
Citation Information
Patent Citations
Method of real-time safety warning of tunnel approaching construction based on data fusion
AU2021101951A4
Highway construction major risk source identification method
CN108665169A
Rolling bearing intelligent fault diagnosis method based on CNN and FCMC
CN111562108A
Tunnel construction site risk assessment method and system based on Internet of Things
CN112862273A
Event records in a log file
US20220156247A1
