Underground space accident risk assessment method and system based on deep learning
By collecting data in real-time in underground space and establishing a risk assessment model using deep learning, the problems of low efficiency and insufficient accuracy of underground space accident risk assessment in the prior art are solved, and more efficient and accurate risk identification and early warning are achieved.
Patent Information
- Application Number
- CN202411375837.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2025-01-10
AI Technical Summary
The existing underground space accident risk assessment methods are inefficient and cannot guarantee the accuracy, mainly because they rely on manual observations and single data comparisons, and cannot effectively prevent and early warning.
Using a deep learning-based method, an accident risk assessment model is established by collecting data in underground space in real time, performing data annotation, normalization processing, feature extraction and model training, and using neural networks for risk assessment, and combining infrared sensors and speakers for Warning notice.
It improves the efficiency and accuracy of underground space accident risk assessment, can promptly identify potential risks and issue warnings to prevent accidents.
Smart Images
Figure CN119293687A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and system for assessing underground space accident risks based on deep learning. Background Art
[0002] At present, many industrial developments require operations in underground spaces, such as mining, subway development, and underground garage construction. Since operations in underground spaces are accompanied by greater risks, it is necessary to monitor relevant data of underground spaces to prevent safety accidents. However, most of the underground space data currently collected is manually observed, and the accident risk is judged by comparing the size of a single data. This method is inefficient and the accuracy cannot be guaranteed. Summary of the invention
[0003] The present invention provides an underground space accident risk assessment method and system based on deep learning, the main purpose of which is to solve the problem that the existing underground space accident risk assessment method is low in efficiency and cannot guarantee accuracy.
[0004] To achieve the above purpose, the present invention provides a method for assessing underground space accident risk based on deep learning, comprising: Relevant data of the underground space is acquired in real time by using a collection device preset in the underground space, corresponding accident types are marked for different types of data, and the underground space data set is classified according to the accident type to obtain an underground space classification data set; Normalizing the underground space classification data set to obtain a preprocessed classification data set, wherein the normalizing the underground space data set to obtain the preprocessed classification data set includes: The normalization process is expressed by the following formula: ; in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data; Using a feature extraction algorithm to extract features from the preprocessed classification data set to obtain a classified underground space feature data set; Using an accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type, and obtain an outlier; It is determined whether there is an accident risk in the underground space based on the abnormal value.
[0005] Optionally, the underground space data set is classified according to the accident type to obtain the underground space classification data set, including: Acquire the initial data collected by the collection device, and annotate the initial data according to the identification code of the collection device to obtain annotated data; Establish a corresponding number of blank data sets according to the number of predetermined accident types; According to the accident type standard of the standard data, the labeled data is placed in a blank data set of the corresponding classification to obtain an underground space classification data set; and the words in the word list are converted into vector form according to the position and frequency of occurrence in the vocabulary to obtain a word vector data set.
[0006] Optionally, an infrared sensor may be installed to detect the density of people in the underground space, and issue a warning when the density is too high to prevent stampede accidents.
[0007] Optionally, the extracting features of the preprocessed classified data set using a feature extraction algorithm to obtain a classified underground space feature data set includes: Processing missing values on the preprocessed classification data set to obtain a complete classification data set; Performing word segmentation processing on the complete classification data set to obtain a word list; constructing a vocabulary based on the list of words; Traversing the word list, converting the words in the word list into vector form according to the positions and frequencies of occurrence of the words in the vocabulary, to obtain a word vector data set; The word vector data set is subjected to dimensionality reduction processing to obtain a classified underground space feature data set.
[0008] Optionally, performing missing value processing on the preprocessed classification data set to obtain a complete classification data set includes: Performing missing value identification on the preprocessed classification data set to obtain a missing value identification result; Perform missing value statistics on the preprocessed classification data set according to the missing value identification result to obtain a statistical result; The missing value processing is performed on the cultivation environment data according to the statistical results by a preset missing value processing method to obtain a complete classification data set.
[0009] Optionally, before performing accident risk assessment on the classified underground space feature data set according to the accident type using the accident risk assessment model trained based on deep learning to obtain the abnormal value, the method further includes: Collecting historical data of the underground space through a preset collection device to obtain historical data of the underground space; Extracting features from the underground space historical data to obtain standard feature data of the underground space; Dividing part of the underground space standard characteristic data into training data and the remaining data into testing data; Using a neural network model to perform accident risk assessment on the training data to obtain training abnormality parameters; Calculating a training difference value between the training abnormality parameter and a standard abnormality parameter preset in the training data; If the training difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the step of performing accident risk assessment on the training data using the neural network model to obtain training abnormal parameters is performed again; If the training difference value is less than a preset difference value threshold, the neural network model is used to perform an accident risk assessment on the test data to obtain a test abnormality parameter; Calculating a test difference value between the test abnormality parameter and a preset standard abnormality parameter of the training data; If the test difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the above step of using the neural network model to perform accident risk assessment on the test data to obtain test abnormal parameters is performed again; If the test difference value is less than the preset difference value threshold, the model training is confirmed to be completed and the accident risk assessment model is obtained.
[0010] Optionally, the calculating a training difference value between the training abnormality parameter and a standard abnormality parameter preset in the training data includes: The training difference value is calculated using the following formula: ; in, Indicates the The training difference value between the training anomaly parameters in the group training data and the preset standard anomaly parameters of the training data, Indicates The number of samples of the training data set, For the said Training anomaly parameters for group training data ， For the said The preset standard anomaly parameters for the group training data.
[0011] Optionally, the accident risk assessment model trained based on deep learning is used to perform accident risk assessment on the classified underground space feature data set according to the accident type to obtain an outlier value, including: Selecting an accident type that needs to be risk assessed, obtaining underground feature data corresponding to the accident type in the classified underground space feature data set, and performing data conversion processing on the underground feature data using the accident risk assessment model to obtain a feature matrix; Using the accident risk assessment model to perform a convolution operation on the feature matrix through a convolution layer to obtain convolution features; Using the accident risk assessment model, performing a maximum pooling operation on the convolutional features through a pooling layer to obtain a pooling feature; The accident risk assessment model is used to perform accident risk assessment of the underground space corresponding to the accident type through a fully connected layer according to the pooling characteristics to obtain an outlier.
[0012] Optionally, when it is determined that there is an accident risk in the underground space, a warning is issued through a loudspeaker preset in the underground space to notify the staff to evacuate.
[0013] In order to solve the above problems, the present invention also provides an underground space accident risk assessment system based on deep learning, the system comprising: A data acquisition module is used to acquire relevant data of the underground space in real time through a collection device preset in the underground space, mark corresponding accident types for different types of data, and classify the underground space data set according to the accident type to obtain an underground space classification data set; The data processing module is used to perform normalization processing on the underground space classification data set to obtain a pre-processed classification data set, wherein the normalization processing on the underground space data set to obtain the pre-processed classification data set includes: The normalization process is expressed by the following formula: ; in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data; A feature extraction module, used to extract features from the preprocessed classification data set using a feature extraction algorithm to obtain a classified underground space feature data set; The risk assessment module is used to use the accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type, obtain outliers, and determine whether there is an accident risk in the underground space based on the outliers.
[0014] The embodiment of the present invention obtains a classified underground space data set by collecting underground space data and labeling the data with corresponding accident types, and then normalizes the classified underground space data set and extracts features to obtain a classified underground space feature data set, and uses an accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type to obtain an outlier, and judges whether there is an accident risk in the underground space based on the outlier. Therefore, the underground space accident risk assessment method and system based on deep learning proposed by the present invention can solve the problem that the existing underground space accident risk assessment method is low in efficiency and cannot guarantee accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 A schematic diagram of a process flow of an underground space accident risk assessment method based on deep learning provided by an embodiment of the present invention; Figure 2 A schematic diagram of a feature extraction process provided by an embodiment of the present invention; Figure 3 A schematic diagram of a process of performing accident risk assessment on a classified underground space feature data set using an accident risk assessment model provided by an embodiment of the present invention; Figure 4 A functional module diagram of an underground space accident risk assessment system based on deep learning provided in one embodiment of the present invention.
[0016] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0017] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.
[0018] The embodiment of the present application provides an underground space accident risk assessment method based on deep learning. The execution subject of the underground space accident risk assessment method based on deep learning includes but is not limited to at least one of the electronic devices such as the server, the terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the underground space accident risk assessment method based on deep learning can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.
[0019] Reference Figure 1 FIG. 1 is a flow chart of an underground space accident risk assessment method based on deep learning provided by an embodiment of the present invention. In this embodiment, the underground space accident risk assessment method based on deep learning includes: S1. Obtain relevant data of the underground space in real time through a collection device preset in the underground space, label different types of data with corresponding accident types, classify the underground space data set according to the accident type, and obtain an underground space classification data set.
[0020] In the embodiment of the present invention, the collection device preset in the underground space includes a temperature sensor, a pressure sensor, an oxygen detector, a PID photoionization detector, and the like.
[0021] In the embodiment of the present invention, the real-time acquisition of relevant data of the underground space by a collection device preset in the underground space includes: Using the temperature sensor to obtain underground space temperature data; Using the pressure sensor to obtain structural pressure data of the underground space; Using the oxygen detector to obtain oxygen concentration data of the underground space; The PID photoionization detector is used to obtain toxic gas concentration data and combustible gas concentration data in the underground space.
[0022] In the embodiment of the present invention, the PID photoionization detector is an instrument for detecting low-concentration gases. Its working principle is to use photoionization technology to ionize organic molecules into positive and negative ions (ionization) that can be detected by the detector, and then measure the charge of the ionized gas through the detector and convert it into a current signal. Finally, the detector measures the current intensity to reflect the content of the substance. The PID photoionization detector has the advantages of a wide measurement range, high accuracy, and fast response speed. In underground space detection, the PID photoionization detector can be used to detect some combustible gases, toxic gases, etc. in underground space, and is a very effective detection method.
[0023] In the embodiment of the present invention, the accident types corresponding to different types of data annotations refer to an accident type corresponding to one or more related data of the underground space being abnormal. For example, when the temperature of the underground space is too high and the flammable gas is too high, an explosion accident is likely to occur; when the oxygen concentration of the underground space is too low, a suffocation accident is likely to occur; when the structural pressure data of the underground space is abnormal, a collapse accident is likely to occur; when the concentration of toxic gases in the underground space is too high, a poisoning accident is likely to occur.
[0024] In the embodiment of the present invention, the underground space data set is classified according to the accident type to obtain the underground space classification data set, including: Acquire the initial data collected by the collection device, and annotate the initial data according to the identification code of the collection device to obtain annotated data; Establish a corresponding number of blank data sets according to the number of predetermined accident types; According to the accident type standard of the standard data, the labeled data is placed in a blank data set of the corresponding classification to obtain an underground space classification data set; and the words in the word list are converted into vector form according to the position and frequency of occurrence in the vocabulary to obtain a word vector data set.
[0025] In the embodiment of the present invention, an infrared sensor can be provided to detect the density of people in the underground space, and a warning can be issued when the density of people is too high, thereby preventing stampede accidents from occurring.
[0026] In an embodiment of the present invention, relevant data of the underground space is acquired in real time by a collection device preset in the underground space, corresponding accident types are marked for different types of data, and the underground space data set is classified according to the accident type to obtain an underground space classification data set, which provides necessary data for subsequent processing and facilitates subsequent targeted analysis based on the classification of the data, thereby improving the accuracy of the results of subsequent data analysis.
[0027] S2. Normalize the underground space classification data set to obtain a preprocessed classification data set.
[0028] In the embodiment of the present invention, the normalization process is a dimensionless processing method, which converts the absolute value of the physical system numerical value into a relative value relationship. It is an effective way to simplify calculations and reduce the value, and helps to improve the comparability and analyzability of data.
[0029] In the embodiment of the present invention, the normalization processing of the underground space dataset to obtain the preprocessed classification dataset includes: The normalization process is expressed by the following formula: ; in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data.
[0030] In the embodiment of the present invention, the preprocessed classified data set is obtained by normalizing the underground space classified data set, thereby improving the comparability and analyzability of the data and improving the efficiency of subsequent data processing.
[0031] S3. Use a feature extraction algorithm to extract features from the preprocessed classification data set to obtain a classified underground space feature data set.
[0032] In an embodiment of the present invention, feature extraction of the preprocessed classified data set can reduce the dimension of the data, reduce the complexity of the data, and simplify multiple features in the original data into a few representative features, thereby reducing the burden of calculation and storage, and making it easier for the neural network model to learn and understand the data.
[0033] In the embodiment of the present invention, the feature extraction algorithm may be a bag-of-words model method.
[0034] In the embodiment of the present invention, refer to Figure 2 As shown, a schematic diagram of a feature extraction process is provided in an embodiment of the present invention, wherein the feature extraction algorithm is used to extract features from the preprocessed classification data set to obtain a classified underground space feature data set, including: S21, performing missing value processing on the preprocessed classification data set to obtain a complete classification data set; S22, performing word segmentation processing on the complete classification data set to obtain a word list; S23, constructing a vocabulary according to the word list; S24, traversing the word list, converting the words in the word list into vector form according to the positions and frequencies of occurrence of the words in the vocabulary, to obtain a word vector data set; S25. Perform dimensionality reduction processing on the word vector dataset to obtain a classified underground space feature dataset.
[0035] In the embodiment of the present invention, the step of performing missing value processing on the preprocessed classification data set to obtain a complete classification data set includes: Performing missing value identification on the preprocessed classification data set to obtain a missing value identification result; Perform missing value statistics on the preprocessed classification data set according to the missing value identification result to obtain a statistical result; The missing value processing is performed on the cultivation environment data according to the statistical results by a preset missing value processing method to obtain a complete classification data set.
[0036] In an embodiment of the present invention, the preset missing value processing method includes deleting missing values and filling missing values.
[0037] Furthermore, the filling of missing values includes mean filling, median filling and mode filling.
[0038] In the embodiment of the present invention, the word segmentation processing is a basic task in Chinese natural language processing, which aims to segment the text into words or phrases.
[0039] In the embodiment of the present invention, by using a feature extraction algorithm to extract features from the preprocessed classification data set, a classified underground space feature data set is obtained, which facilitates the neural network model to learn and understand the data and improves the efficiency of subsequent data processing.
[0040] S4. Use the accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type to obtain outliers.
[0041] In an embodiment of the present invention, before performing accident risk assessment on the classified underground space feature data set according to the accident type using the accident risk assessment model trained based on deep learning and obtaining the abnormal value, the method further includes: Collecting historical data of the underground space through a preset collection device to obtain historical data of the underground space; Extracting features from the underground space historical data to obtain standard feature data of the underground space; Dividing part of the underground space standard characteristic data into training data and the remaining data into testing data; Using a neural network model to perform accident risk assessment on the training data to obtain training abnormality parameters; Calculating a training difference value between the training abnormality parameter and a standard abnormality parameter preset in the training data; If the training difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the step of performing accident risk assessment on the training data using the neural network model to obtain training abnormal parameters is performed again; If the training difference value is less than a preset difference value threshold, the neural network model is used to perform an accident risk assessment on the test data to obtain a test abnormality parameter; Calculating a test difference value between the test abnormality parameter and a preset standard abnormality parameter of the training data; If the test difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the above step of using the neural network model to perform accident risk assessment on the test data to obtain test abnormal parameters is performed again; If the test difference value is less than the preset difference value threshold, the model training is confirmed to be completed and the accident risk assessment model is obtained.
[0042] In an embodiment of the present invention, the back propagation algorithm is a commonly used algorithm for training a neural network model. The training set data is input into the input layer of the neural network model, passes through the hidden layer, and finally reaches the output layer and outputs the result. Since there is an error between the output result of the ANN and the actual result, the error between the estimated value and the actual value is calculated, and the error is back-propagated from the output layer to the hidden layer until it is propagated to the input layer. During the back propagation process, the values of various parameters are adjusted according to the error, and the above process is continuously iterated until the parameters of the model tend to stabilize.
[0043] In the embodiment of the present invention, the preset difference value threshold is studied and set according to the model accuracy requirement.
[0044] In the embodiment of the present invention, the step of calculating the test difference value between the test abnormality parameter and the preset standard abnormality parameter of the training data is the same as the step of calculating the training difference value, which will not be repeated here.
[0045] In the embodiment of the present invention, the step of calculating the training difference value between the training abnormality parameter and the standard abnormality parameter preset in the training data includes: The training difference value is calculated using the following formula: ; in, Indicates the The training difference value between the training anomaly parameters in the group training data and the preset standard anomaly parameters of the training data, Indicates The number of samples of the training data set, For the said Training anomaly parameters for group training data ， For the said The preset standard anomaly parameters for the group training data.
[0046] In the embodiment of the present invention, refer to Figure 3 As shown, an embodiment of the present invention provides a flow chart of an accident risk assessment model for performing accident risk assessment on a classified underground space feature data set, wherein the accident risk assessment model trained based on deep learning is used to perform accident risk assessment on the classified underground space feature data set according to the accident type to obtain an abnormal value, including: S31, selecting an accident type that needs to be risk assessed, obtaining underground feature data corresponding to the accident type in the classified underground space feature data set, and performing data conversion processing on the underground feature data using the accident risk assessment model to obtain a feature matrix; S32, using the accident risk assessment model to perform a convolution operation on the feature matrix through a convolution layer to obtain a convolution feature; S33, using the accident risk assessment model to perform a maximum pooling operation on the convolutional features through a pooling layer to obtain pooling features; S34. Using the accident risk assessment model, a fully connected layer is used to perform an accident risk assessment on the underground space corresponding to the accident type according to the pooling feature to obtain an outlier value.
[0047] In the embodiment of the present invention, the data conversion process refers to representing the underground feature data in the form of word vectors, and then concatenating the word vectors to form a matrix.
[0048] In the embodiment of the present invention, the convolution operation refers to performing a convolution operation on the feature matrix and according to the convolution kernel parameters to extract some features in the feature matrix.
[0049] In the embodiment of the present invention, the convolution kernel parameter is an important concept in convolutional neural networks in deep learning. In convolutional neural networks, the convolution kernel is a parameter used to slide on the input data and perform convolution operations. Each convolution kernel has one or more weights, which are learned during the training process and used to determine the mode of convolution operation.
[0050] In the embodiment of the present invention, the maximum pooling operation can effectively reduce the dimension of the data while retaining the most important feature information. Since the maximum pooling operation only focuses on the maximum value in each pooling window and ignores other values, it can effectively reduce noise and redundant information and improve the generalization ability of the model.
[0051] In an embodiment of the present invention, the accident risk assessment model has the ability to learn implicit features. During the training process, the accident risk assessment model continuously updates the parameters of the model through the back propagation algorithm, so that the model can automatically learn the implicit features in the data. In the problem of accident risk assessment for underground space, these implicit features may be abstract features related to the environment, such as a certain waveform shape, spectrum distribution, etc. Through deep neural networks, higher-level and more expressive features can be gradually extracted, thereby better assessing the probability of accident risk in underground space.
[0052] In detail, the performance of the accident risk assessment model depends largely on the quality of available data and the representativeness of the sample. Therefore, collecting sufficient, representative and accurately labeled underground space characteristic data is the key to ensuring the accuracy of the model. In addition, timely updating and maintenance of the model is also an important factor to ensure the effectiveness of the model under different operating environments and data changes.
[0053] In an embodiment of the present invention, an accident risk assessment model trained based on deep learning is used to perform accident risk assessment on the classified underground space feature data set according to the accident type, and outliers are obtained, which facilitates subsequent targeted analysis based on the outliers.
[0054] S5. Determine whether there is an accident risk in the underground space based on the abnormal value.
[0055] In the embodiment of the present invention, it is necessary to set an abnormal value threshold value, and determine whether there is an accident risk in the underground space by determining the magnitude relationship between the abnormal value and the abnormal value threshold value.
[0056] In detail, judging whether there is an accident risk in the underground space according to the abnormal value includes: If the abnormal value is greater than or equal to the preset abnormal value threshold, it is determined that there is an accident risk in the underground space; If the abnormal value is smaller than the abnormal value threshold, it is determined that there is no accident risk in the underground space.
[0057] In the embodiment of the present invention, when it is determined that there is an accident risk in the underground space, a warning is issued through a loudspeaker preset in the underground space to notify the staff to evacuate.
[0058] like Figure 4The figure shows a functional module diagram of an underground space accident risk assessment system based on deep learning provided by one embodiment of the present invention.
[0059] The underground space accident risk assessment system 100 based on deep learning of the present invention can be installed in an electronic device. According to the functions to be implemented, the underground space accident risk assessment system 100 based on deep learning can include a data acquisition module 101, a data processing module 102, a feature extraction module 103 and a risk assessment module 104. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.
[0060] In this embodiment, the functions of each module / unit are as follows: The data acquisition module 101 is used to acquire relevant data of the underground space in real time through a collection device preset in the underground space, mark corresponding accident types for different types of data, and classify the underground space data set according to the accident type to obtain an underground space classification data set; The data processing module 102 is used to perform normalization processing on the underground space classification data set to obtain a pre-processed classification data set, wherein the normalization processing on the underground space data set to obtain the pre-processed classification data set includes: The normalization process is expressed by the following formula: ; in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data; The feature extraction module 103 is used to extract features from the pre-processed classification data set using a feature extraction algorithm to obtain a classified underground space feature data set; The risk assessment module 104 is used to perform accident risk assessment on the classified underground space feature data set according to the accident type using an accident risk assessment model trained based on deep learning, obtain outliers, and determine whether there is an accident risk in the underground space based on the outliers.
[0061] In detail, each module described in the underground space accident risk assessment system 100 based on deep learning in the embodiment of the present invention is used in the same manner as described above. Figures 1 to 3 The underground space accident risk assessment method based on deep learning described in the article has the same technical means and can produce the same technical effects, so I will not go into details here.
[0062] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0063] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0064] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0065] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0066] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.
[0067] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0068] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.
[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.
Claims
1. A method for assessing underground space accident risk based on deep learning, characterized in that: The method comprises: Relevant data of the underground space is acquired in real time by using a collection device preset in the underground space, corresponding accident types are marked for different types of data, and the underground space data set is classified according to the accident type to obtain an underground space classification data set; Normalizing the underground space classification data set to obtain a preprocessed classification data set, wherein the normalizing the underground space data set to obtain the preprocessed classification data set includes: The normalization process is expressed by the following formula: ； in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data; Using a feature extraction algorithm to extract features from the preprocessed classification data set to obtain a classified underground space feature data set; Using an accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type, and obtain an outlier; It is determined whether there is an accident risk in the underground space based on the abnormal value.
2. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: The underground space data set is classified according to the accident type to obtain the underground space classification data set, including: Acquire the initial data collected by the collection device, and annotate the initial data according to the identification code of the collection device to obtain annotated data; Establish a corresponding number of blank data sets according to the number of predetermined accident types; According to the accident type standard of the standard data, the labeled data is placed in a blank data set of the corresponding classification to obtain an underground space classification data set; and the words in the word list are converted into vector form according to the position and frequency of occurrence in the vocabulary to obtain a word vector data set.
3. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: Infrared sensors can be installed to detect the density of people in underground spaces, and issue warnings when the density is too high to prevent stampede accidents.
4. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: The method of extracting features from the preprocessed classified data set using a feature extraction algorithm to obtain a classified underground space feature data set includes: Processing missing values on the preprocessed classification data set to obtain a complete classification data set; Performing word segmentation processing on the complete classification data set to obtain a word list; constructing a vocabulary based on the list of words; Traversing the word list, converting the words in the word list into vector form according to the positions and frequencies of occurrence of the words in the vocabulary, to obtain a word vector data set; The word vector data set is subjected to dimensionality reduction processing to obtain a classified underground space feature data set.
5. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: The step of performing missing value processing on the preprocessed classification data set to obtain a complete classification data set includes: Performing missing value identification on the preprocessed classification data set to obtain a missing value identification result; Perform missing value statistics on the preprocessed classification data set according to the missing value identification result to obtain a statistical result; The missing value processing is performed on the cultivation environment data according to the statistical results by a preset missing value processing method to obtain a complete classification data set.
6. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: Before performing accident risk assessment on the classified underground space feature data set according to the accident type using the accident risk assessment model trained based on deep learning to obtain the abnormal value, the method further includes: Collecting historical data of the underground space through a preset collection device to obtain historical data of the underground space; Extracting features from the underground space historical data to obtain standard feature data of the underground space; Dividing part of the underground space standard characteristic data into training data and the remaining data into testing data; Using a neural network model to perform accident risk assessment on the training data to obtain training abnormality parameters; Calculating a training difference value between the training abnormality parameter and a standard abnormality parameter preset in the training data; If the training difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the step of performing accident risk assessment on the training data using the neural network model to obtain training abnormal parameters is performed again; If the training difference value is less than a preset difference value threshold, the neural network model is used to perform an accident risk assessment on the test data to obtain a test abnormality parameter; Calculating a test difference value between the test abnormality parameter and a preset standard abnormality parameter of the training data; If the test difference value is greater than or equal to a preset difference value threshold, the weight parameters and bias items of the neural network model are adjusted by a back propagation algorithm, and then the above step of using the neural network model to perform accident risk assessment on the test data to obtain test abnormal parameters is performed again; If the test difference value is less than the preset difference value threshold, the model training is confirmed to be completed and the accident risk assessment model is obtained.
7. The underground space accident risk assessment method based on deep learning according to claim 6, characterized in that: The calculating of the training difference value between the training abnormality parameter and the standard abnormality parameter preset in the training data includes: The training difference value is calculated using the following formula: ； in, Indicates the The training difference value between the training anomaly parameters in the group training data and the preset standard anomaly parameters of the training data, Indicates The number of samples of the training data set, For the said Training anomaly parameters for group training data ， For the said Preset standard anomaly parameters for group training data.
8. The underground space accident risk assessment method based on deep learning according to claim 6, characterized in that: The accident risk assessment model trained based on deep learning is used to perform accident risk assessment on the classified underground space feature data set according to the accident type to obtain abnormal values, including: Selecting an accident type that needs to be risk assessed, obtaining underground feature data corresponding to the accident type in the classified underground space feature data set, and performing data conversion processing on the underground feature data using the accident risk assessment model to obtain a feature matrix; Using the accident risk assessment model to perform a convolution operation on the feature matrix through a convolution layer to obtain convolution features; Using the accident risk assessment model, performing a maximum pooling operation on the convolutional features through a pooling layer to obtain a pooling feature; The accident risk assessment model is used to perform accident risk assessment of the underground space corresponding to the accident type through a fully connected layer according to the pooling characteristics to obtain an outlier.
9. The underground space accident risk assessment method based on deep learning according to claim 1, characterized in that: When it is determined that there is a risk of accident in the underground space, a warning will be issued through a loudspeaker preset in the underground space to notify the staff to evacuate.
10. The underground space accident risk assessment system based on deep learning is characterized by: The system comprises: A data acquisition module is used to acquire relevant data of the underground space in real time through a collection device preset in the underground space, mark corresponding accident types for different types of data, and classify the underground space data set according to the accident type to obtain an underground space classification data set; The data processing module is used to perform normalization processing on the underground space classification data set to obtain a pre-processed classification data set, wherein the normalization processing on the underground space data set to obtain the pre-processed classification data set includes: The normalization process is expressed by the following formula: ； in, is the first Group data, is the first Group data, is the first The average of the group data, is the first The number of samples for group data; A feature extraction module, used to extract features from the preprocessed classification data set using a feature extraction algorithm to obtain a classified underground space feature data set; The risk assessment module is used to use the accident risk assessment model trained based on deep learning to perform accident risk assessment on the classified underground space feature data set according to the accident type, obtain outliers, and determine whether there is an accident risk in the underground space based on the outliers.