Hazardous chemical station safety and credible early warning method, system, equipment and medium

Abstract

The invention provides a dangerous chemical station safety and credible early warning method, system and device and a medium. The dangerous chemical station safety and credible early warning method comprises the steps that multi-modal information of a dangerous chemical station is acquired, the multi-modal information is converted into data to be processed, and the multi-modal information comprises at least one of sensor information, image information, environment and weather information and geography and traffic information; fusing the to-be-processed data, performing formatting processing, and outputting data in a standard format; and inputting the data in the standard format into the multi-layer perceptron model, evaluating the safety of the hazardous chemical station, and carrying out alarm pushing on the data exceeding a risk threshold according to a safety evaluation result. By acquiring the multi-modal information and performing fusion processing on the multi-modal information, the actual condition of the hazardous chemical station can be reflected more comprehensively, and the accuracy and credibility of early warning can be remarkably improved. And a multi-layer sensor model is adopted, and alarm pushing is carried out, so that the credibility and reliability of safety management of the hazardous chemical station are improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology for the Internet of Things, and more specifically to a safe and reliable early warning method, system, equipment and medium for hazardous chemical sites. Background Art

[0002] As an important part of the chemical industry, the safety of hazardous chemical sites is directly related to the safety of life, property and environment. However, hazardous chemical sites often face complex and changeable operating environments, including the storage, transportation and processing of various chemicals, all of which may pose safety hazards. Traditional safety monitoring methods often rely on a single data source, such as temperature sensors and pressure sensors. Although these data can provide certain safety information, they are often one-sided and limited, and it is difficult to fully reflect the actual safety status of hazardous chemical sites.

[0003] Therefore, it is necessary to provide a safe and reliable early warning method, system, equipment and medium for hazardous chemical sites to at least partially solve the above problems. Summary of the invention

[0004] A series of simplified concepts are introduced in the Summary of the Invention section, which will be further described in detail in the Detailed Description of the Invention section. The Summary of the Invention section of this application does not mean to attempt to define the key features and essential technical features of the claimed technical solution, nor does it mean to attempt to determine the scope of protection of the claimed technical solution.

[0005] In order to at least partially solve the above problems, the first aspect of the present application provides a safe and reliable early warning method for hazardous chemicals stations, comprising the following steps:

[0006] Acquire multimodal information of a hazardous chemical station and convert the multimodal information into data to be processed, wherein the multimodal information includes at least one of sensor information, image information, environmental and meteorological information, and geographical and traffic information;

[0007] The data to be processed are integrated and formatted, and the data in standard format are output;

[0008] The data in standard format is input into the multi-layer perceptron model for deep neural network learning to evaluate the safety of hazardous chemical sites. Based on the safety assessment results, alarms are pushed for data that exceeds the risk threshold.

[0009] According to the first aspect of the present application, the hazardous chemical site safety and reliable early warning method can more comprehensively reflect the actual situation of the hazardous chemical site by acquiring multimodal information and fusing it, and can significantly improve the accuracy and credibility of the early warning. The multi-layer perceptron model is used to continuously learn and thus improve the accuracy of the prediction. By pushing alarms, the credibility and reliability of the safety management of hazardous chemical sites are improved.

[0010] Optionally, the deep neural network learning step specifically includes:

[0011] Data preprocessing: normalizing the data in the standard format to the range of [0-1], and converting the normalized data into the data format required by the deep learning model;

[0012] Feature extraction: extract features from preprocessed data;

[0013] Model training: The extracted features are used as training data, and a deep learning model is used for model training to output a prediction model.

[0014] Safety prediction: obtain the data in the new standard format as the input of the prediction model, and use the output of the prediction model as the safety assessment result of the hazardous chemicals site.

[0015] Optionally, the multilayer perceptron model consists of an input layer, multiple hidden layers and an output layer; wherein,

[0016] The ReLU activation function is used in each hidden layer to handle nonlinear relationships, and training is performed using a back-propagation algorithm.

[0017] Optionally, the deep learning model is one of PyTorch, Paddle, TensorFlow, and Caffe.

[0018] Optionally, the deep learning model uses BCEloss as a loss function, and uses an Adam optimizer to update the weights of the deep learning model. The BCELoss loss function is specifically:

[0019] BCELoss(y,y^)=-ylog(y^)-(1-y)log(1-y^)

[0020] in:

[0021] y is the true label, which takes the value of 0 or 1;

[0022] y^ is the output value of the model.

[0023] Optionally, the sensor information includes at least one of pressure information, temperature information, gas leakage information and distance information;

[0024] The environment and meteorological information includes at least one of strong wind information, heavy rain information, heavy fog information, low temperature information and high temperature information;

[0025] The geographic and traffic information includes at least one of road condition information, mountain information, river information and slope information;

[0026] The image information is processed by GPU computing power.

[0027] The second aspect of the present application provides a reliable safety warning system for hazardous chemicals stations, which is used to implement the reliable safety warning method for hazardous chemicals stations, including:

[0028] An Internet of Things terminal, the Internet of Things terminal comprising a plurality of collection devices, the collection devices being used to collect multimodal information of hazardous chemical sites and convert the multimodal information into data to be processed;

[0029] An edge computing platform, the edge computing platform is used to obtain the data to be processed from different acquisition devices, merge the data to be processed, format the data, and output the data in a standard format;

[0030] A safe and reliable early warning system is based on a multi-layer perceptron model, which analyzes the received standard format and outputs the safety score of the hazardous chemicals site. If the safety score exceeds the risk threshold, an alarm is pushed.

[0031] The hazardous chemicals station safety and reliable early warning system according to the second aspect of the present application has similar technical effects as the above method. This solution is simple to implement, does not require changes to existing equipment at hazardous chemicals stations, and does not require changes to the site. The algorithm is simple and easy to implement.

[0032] Optionally, the Internet of Things terminal is communicatively connected to the edge computing platform, and the communication connection mode includes wired communication mode and / or wireless communication mode; the wired communication mode includes Ethernet communication, serial port communication, HDMI communication and optical fiber communication mode; the wireless communication mode includes WiFi, Bluetooth, Lora and 4G / 5G mobile network communication mode.

[0033] A third aspect of the present application provides an electronic device, the device comprising:

[0034] a memory storing executable program code; and

[0035] at least one processor coupled to the memory; wherein,

[0036] The processor calls the executable program code stored in the memory to execute the steps of the above-mentioned hazardous chemicals site safety and reliable early warning method.

[0037] A fourth aspect of the present application provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the steps of the hazardous chemicals site safety and reliable early warning method as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The following drawings of the embodiments of the present application are hereby used as part of the present application for understanding the present application. The drawings show the embodiments of the present application and their descriptions, and are used to explain the principles of the present application. In the drawings,

[0039] Figure 1 This is a flow chart of the safe and reliable early warning process of hazardous chemicals stations in the embodiment of this application;

[0040] Figure 2 A schematic diagram of the process of deep neural network learning in an embodiment of the present application;

[0041] Figure 3 This is a schematic diagram of a safe and reliable early warning system for hazardous chemicals sites in an embodiment of the present application. DETAILED DESCRIPTION

[0042] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present application. However, it is obvious to those skilled in the art that the present application embodiments can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present application embodiments, some technical features well known in the art are not described.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should also be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of the features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0044] Unless otherwise stated, the numerical ranges herein include not only the entire range within its two endpoints but also include several sub-ranges contained therein.

[0045] Now, exemplary embodiments according to the present application will be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in a variety of different forms and should not be interpreted as being limited to the embodiments described herein. It should be understood that these embodiments are provided to make the disclosure of the present application thorough and complete, and to fully convey the concepts of these exemplary embodiments to those of ordinary skill in the art.

[0046] Reference Figures 1 to 3 The present application provides a safe and reliable early warning method for hazardous chemicals sites, comprising the following steps:

[0047] S100: Acquire multimodal information of hazardous chemicals sites and convert the multimodal information into data to be processed.

[0048] Multimodal information is obtained through IoT terminals, which include various IoT modules, sensors, and image and video acquisition devices.

[0049] The multi-module information includes at least one of sensor information, image information, environment and meteorological information, and geographic and traffic information.

[0050] The sensor information includes at least one of pressure information, temperature information, gas leakage information and distance information. Specifically, the pressure information is obtained by a pressure sensor. The temperature information is obtained by a temperature sensor. The gas leakage information is obtained by a gas leakage sensor. The distance information is obtained by a distance measuring sensor.

[0051] The environmental and meteorological information includes at least one of strong wind information, heavy rain information, heavy fog information, low temperature information and high temperature information.

[0052] The geographic and traffic information includes at least one of road condition information, mountain information, river information, and slope information.

[0053] The image information is processed by GPU computing power. Specifically, the image information is acquired through video image acquisition module data. Optionally, the video image acquisition module data is various explosion-proof cameras of hazardous chemical sites.

[0054] Optionally, this step also includes: before converting the multimodal information into the data to be processed, cleaning the multimodal information to remove abnormal values ​​or duplicate data. Specifically, for sensor information, it is necessary to check the integrity and consistency of the data, for example, to check whether there are missing values, abnormal values ​​or data jumps. Ensuring the quality and accuracy of the input data provides a reliable foundation for subsequent deep learning and analysis.

[0055] Through this solution, multimodal information is used as a data source, which can better predict safety than a single data source. The fusion processing of multimodal information can more comprehensively reflect the actual situation of hazardous chemical sites. Compared with a single information source, this fusion processing method evaluates safety from multiple dimensions, which can significantly improve the accuracy and credibility of early warnings. Using a multi-layer perceptron model to conduct deep learning and analysis of the fused data can more accurately evaluate the safety of hazardous chemical sites. Through continuous learning and optimization, the model can gradually improve the accuracy and reliability of early warnings.

[0056] Step S200: The data to be processed are merged and formatted, and the data in a standard format are output.

[0057] This step is performed through the edge computing platform. The edge computing platform is equipped with multi-source data parsing and processing middleware, which can process and analyze multi-source data collected by the IoT terminal in real time, including data classification analysis, data fusion, data formatting and standard data output. The edge computing platform can realize real-time processing and analysis of data, improve the response speed of the application, further enhance the credibility of the early warning system, and effectively ensure data security, so as to realize real-time monitoring and rapid response to safety information of hazardous chemical sites. The edge computing platform receives the data to be processed output in step S100 and processes it. The formats and protocols of the data obtained by the IoT terminal are different, and need to be fused.

[0058] Optionally, the fused data is formatted and output in a standard format, such as JSON. The formatting process includes determining the structure, type, and naming rules of the data to ensure that subsequent algorithms and models can correctly process and parse the data.

[0059] In this step, after the data formatting is completed, the generated standard format data set is output for further analysis and processing by subsequent security and reliable early warning algorithms, models or systems. The output data set has a clear data structure, consistent data types and naming rules, which facilitates the subsequent algorithms and models to correctly process and parse the data.

[0060] In this solution, the IoT terminal is connected to the edge computing platform. The communication connection methods include wired communication and wireless communication. Among them, the wired communication method includes Ethernet communication, serial communication, HDMI communication and optical fiber communication. Wireless communication methods include wifi, Bluetooth, Lora and 4G / 5G mobile network communication. Both wired and wireless communication methods can achieve efficient data transmission between the IoT terminal and the edge computing platform. The wired communication method has high-speed and stable transmission characteristics; while the wireless communication method has the advantages of wide coverage, high flexibility and easy deployment. In some embodiments, wired communication and wireless communication can be available at the same time to further improve the flexibility, reliability and resource utilization of the communication system.

[0061] However, data processing alone is not enough to achieve efficient, safe and reliable early warning. In order to more accurately assess the safety of hazardous chemical sites, advanced algorithms and models are needed to conduct deep learning and analysis on the processed data.

[0062] S300: Input the data in standard format into the multilayer perceptron model. As a deep learning model, the multilayer perceptron model has strong nonlinear fitting and generalization capabilities, and can gradually approach the real safety assessment function through training and learning of historical data. In this step, deep neural network learning is performed based on the multilayer perceptron model to evaluate the safety of hazardous chemical sites. According to the safety assessment results, an alarm is pushed for data that exceeds the risk threshold.

[0063] This step is carried out by the safe and reliable early warning system. Based on the multi-layer perceptron model, the safe and reliable early warning system analyzes the received standard format and outputs the safety score of the hazardous chemical site. If the safety score exceeds the risk threshold, an alarm is pushed. By setting a unified scoring standard and threshold, the safety of different hazardous chemical sites can be standardized for comparison and sorting. Among them, the closer the safety score value is to 1, the higher the safety. The early warning capability of the system is enhanced through alarm push, and the credibility and reliability of the safety management of hazardous chemical sites are improved.

[0064] The multilayer perceptron model in this step consists of an input layer, multiple hidden layers, and an output layer; in each hidden layer, the ReLU activation function is used to process nonlinear relationships and is trained using the back-propagation algorithm. The use of the ReLU activation function in the hidden layer can simplify calculations and improve training efficiency. When the input is positive, the output of the ReLU function is equal to the input. At the same time, the output of the ReLU function is zero when the input is negative, which helps to sparse the network and reduce the mutual dependence between parameters, thereby enhancing the robustness of the model. Using the back-propagation algorithm for training can efficiently update the model weights so that the model gradually approaches the true security assessment function. The simple calculation method of the ReLU activation function and the high efficiency of the back-propagation algorithm make the model training process faster and more efficient.

[0065] The formula for the ReLU activation function is as follows:

[0066] f(x)=max(0,x)

[0067] Where x is the input and f(x) is the output after the ReLU unit. This formula means that when the input x is greater than or equal to 0, the output is x; when the input x is less than 0, the output is 0.

[0068] Reference Figure 2 , step S300 moderate neural network learning step includes:

[0069] S301: Data preprocessing: Normalize the data in the standard format to the range of [0-1], and convert the normalized data into the data format required by the deep learning model.

[0070] Normalization ensures that all features are on the same scale, and the normalized data has a more stable distribution, which helps the model converge to the optimal solution faster during training, so as to efficiently train the model.

[0071] Before step S301, a collection step is also included to receive and store data from the edge computing platform.

[0072] S302: Feature extraction: Extract features from the preprocessed data. In the multi-layer perceptron model processing flow, feature extraction is used to extract key information that is helpful for model training, thereby helping the model to better fit the data and improve the accuracy of prediction.

[0073] In this step, for numerical data, since it has been expressed in numerical form and usually has a high information density, no additional feature extraction is required. These numerical values ​​can be directly used as input for model training. For non-numerical data such as images, feature extraction is required because the information it contains is more complex and difficult to use directly for model training. In this solution, feature extraction is performed on the image through a deep learning backbone network, the key information in the image is extracted, and it is converted into numerical form for subsequent model training. Optionally, the deep learning backbone network is Resnet18 or MobilenetV2.

[0074] S303: Model training: The extracted features are used as training data, a deep learning model is used for model training, and a prediction model is output.

[0075] In this step, the data is fitted based on deep learning. The training process is iterative until the model performance reaches the preset accuracy requirement or the preset number of iterations is reached.

[0076] The deep learning model is one of PyTorch, Paddle, TensorFlow, and Caffe. Different deep learning frameworks differ in terms of architecture design and computing performance. In specific implementation, the framework can be rotated according to actual needs to meet different scenarios.

[0077] The deep learning model uses BCEloss as the loss function and uses the Adam optimizer to update the deep learning model weights. During the training process, BCEloss is used as the loss function to measure the difference between the model prediction and the true label, and the weights are updated through the Adam optimizer. The Adam optimizer can efficiently update the model weights, which not only speeds up the training speed, but also improves the convergence performance of the model. The training process is iterative until the model performance reaches the preset accuracy requirements or the preset number of iterations is reached.

[0078] The BCELoss loss function is specifically:

[0079] BCELoss(y,y^)=-ylog(y^)-(1-y)log(1-y^)

[0080] in:

[0081] y is the true label, which takes the value of 0 or 1;

[0082] y^ is the output value of the model.

[0083] In this scheme, y^ is usually a probability value between 0 and 1. The BCELoss loss function calculates the cross entropy loss between the predicted probability and the true label. Specifically, when the true label y is 1, the loss function focuses on the probability y^ that the model predicts to be 1. When the true label y is 0, the loss function focuses on the probability 1-y^ that the model predicts to be 0.

[0084] S304: Safety prediction: Obtain data in a new standard format as the input of the prediction model, and use the output of the prediction model as the safety assessment result of the hazardous chemical site. By using the output of the prediction model as the safety assessment result, real-time monitoring and early warning of the safety of hazardous chemical sites can be achieved. This helps to timely discover potential safety hazards and risks, and take appropriate measures for intervention and management.

[0085] The embodiment of the present application also provides a safe and reliable early warning device for hazardous chemicals stations, including:

[0086] A memory storing executable program codes; and at least one processor coupled to the memory. The processor calls the executable program codes stored in the memory.

[0087] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the following operations:

[0088] Acquire multimodal information of a hazardous chemical station and convert the multimodal information into data to be processed, wherein the multimodal information includes at least one of sensor information, image information, environmental and meteorological information, and geographical and traffic information;

[0089] The data to be processed are integrated and formatted, and the data in standard format are output;

[0090] Data in a standard format is input into the multi-layer perceptron model to evaluate the safety of hazardous chemical sites. Based on the safety assessment results, alarms are pushed for data that exceeds the risk threshold.

[0091] The embodiment of the present application further provides a computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as follows:

[0092] Acquire multimodal information of a hazardous chemical station and convert the multimodal information into data to be processed, wherein the multimodal information includes at least one of sensor information, image information, environmental and meteorological information, and geographical and traffic information;

[0093] The data to be processed are integrated and formatted, and the data in standard format are output;

[0094] Data in a standard format is input into the multi-layer perceptron model to evaluate the safety of hazardous chemical sites. Based on the safety assessment results, alarms are pushed for data that exceeds the risk threshold.

[0095] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0096] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0097] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0098] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0101] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0102] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0103] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0104] Unless otherwise defined, the technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art in the technical field of this application. The terms used herein are only for describing specific implementation purposes and are not intended to limit this application. Terms such as "setting" appearing in this article can mean that one component is directly attached to another component, or that one component is attached to another component through an intermediate. Features described in this article in one embodiment may be applied to another embodiment alone or in combination with other features, unless the feature is not applicable in the other embodiment or otherwise specified.

[0105] The present application has been described through the above-mentioned embodiments, but it should be understood that the above-mentioned embodiments are only for the purpose of example and description, and are not intended to limit the present application to the described embodiments. It can be understood by those skilled in the art that more variations and modifications can be made according to the teachings of the present application, and these variations and modifications all fall within the scope of protection claimed in the present application.

Claims

1. A reliable method for safety early warning of hazardous chemicals sites, characterized in that: The following steps are involved: Acquire multimodal information of a hazardous chemical station and convert the multimodal information into data to be processed, wherein the multimodal information includes at least one of sensor information, image information, environmental and meteorological information, and geographical and traffic information; The data to be processed are integrated and formatted, and the data in standard format are output; The data in standard format is input into the multi-layer perceptron model for deep neural network learning to evaluate the safety of hazardous chemical sites. Based on the safety assessment results, alarms are pushed for data that exceeds the risk threshold.

2. The method for safe and reliable early warning of hazardous chemicals stations according to claim 1 is characterized in that: The deep neural network learning steps specifically include: Data preprocessing: normalizing the data in the standard format to the range of [0-1], and converting the normalized data into the data format required by the deep learning model; Feature extraction: extract features from preprocessed data; Model training: The extracted features are used as training data, and a deep learning model is used for model training to output a prediction model. Safety prediction: obtain the data in the new standard format as the input of the prediction model, and use the output of the prediction model as the safety assessment result of the hazardous chemicals site.

3. The method for safe and reliable early warning of hazardous chemicals stations according to claim 2 is characterized in that: The multi-layer perceptron model consists of an input layer, multiple hidden layers and an output layer; wherein, The ReLU activation function is used in each hidden layer to handle nonlinear relationships, and training is performed using a back-propagation algorithm.

4. The method for safe and reliable early warning of hazardous chemicals stations according to claim 2 is characterized in that: The deep learning model is one of PyTorch, Paddle, TensorFlow, and Caffe.

5. The method for safe and reliable early warning of hazardous chemicals stations according to claim 4 is characterized in that: The deep learning model uses BCEloss as the loss function and uses the Adam optimizer to update the weights of the deep learning model. The BCELoss loss function is specifically: BCELoss(y,y^)=-ylog(y^)-(1-y)log(1-y^) in: y is the true label, which takes the value of 0 or 1; y^ is the output value of the model.

6. The method for safe and reliable early warning of hazardous chemicals stations according to claim 1 is characterized in that: The sensor information includes at least one of pressure information, temperature information, gas leakage information and distance information; The environment and meteorological information includes at least one of strong wind information, heavy rain information, heavy fog information, low temperature information and high temperature information; The geographic and traffic information includes at least one of road condition information, mountain information, river information and slope information; The image information is processed by GPU computing power.

7. A reliable safety warning system for hazardous chemicals stations, used to implement the safety warning method for hazardous chemicals stations as described in any one of claims 1 to 6, characterized in that: include: An Internet of Things terminal, the Internet of Things terminal comprising a plurality of collection devices, the collection devices being used to collect multimodal information of hazardous chemical sites and convert the multimodal information into data to be processed; An edge computing platform, the edge computing platform is used to obtain the data to be processed from different acquisition devices, merge the data to be processed, format the data, and output the data in a standard format; The safety warning system is based on a multi-layer perceptron model, analyzes the received standard format, and outputs the safety score of the hazardous chemicals site. If the safety score exceeds the risk threshold, an alarm is pushed.

8. The hazardous chemicals station safety and reliable early warning system according to claim 7 is characterized in that: The Internet of Things terminal is communicatively connected to the edge computing platform, and the communication connection method includes wired communication method and / or wireless communication method; the wired communication method includes Ethernet communication, serial port communication, HDMI communication and optical fiber communication method; the wireless communication method includes WiFi, Bluetooth, Lora and 4G / 5G mobile network communication method.

9. An electronic device, characterized in that: The device comprises: a memory storing executable program code; and at least one processor coupled to the memory; wherein, The processor calls the executable program code stored in the memory to execute the steps of the hazardous chemicals site safety and reliable early warning method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which, when called, are used to execute the steps of the hazardous chemicals site safety and reliable early warning method as described in any one of claims 1 to 6.

Images