Intelligent fire-fighting system and intelligent fire-fighting data optimization method

Abstract

The invention provides an intelligent fire-fighting system and an intelligent fire-fighting data optimization method, and particularly relates to the technical field of fire-fighting data processing, and the system comprises a data acquisition module, a data processing module, a data storage module, a data analysis module, an early warning decision module and a user interaction module. The system collects fire-fighting environment data in real time through the distributed sensor unit, and the fire-fighting environment data is stored in the distributed database and the block chain storage unit after being processed by edge computing and cloud computing. And performing multi-dimensional data analysis by utilizing a machine learning and deep learning model, and generating an early warning and executing a decision by an early warning decision module according to the analysis result. The user interaction module displays early warning information and decision results through a visual interface and a mobile terminal, the technical problem that a traditional fire fighting system is prone to paralysis due to data bloating, transmission delay, low storage efficiency and poor real-time performance is solved, the real-time performance, reliability and safety of the system are enhanced, and meanwhile user experience is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of fire fighting data processing, and more particularly, to an intelligent fire fighting system and an intelligent fire fighting data optimization method. Background Art

[0002] With the acceleration of the urbanization process and the continuous expansion of building scale, fire safety issues have become increasingly prominent, and intelligent fire fighting systems have gradually become the focus of industry development. In the prior art, traditional fire fighting systems usually adopt a centralized data collection and processing method, and data storage and analysis are carried out through a single data center. The traditional fire fighting system has the following problems:

[0003] First of all, in the process of data collection, transmission, storage and analysis of traditional fire fighting systems, there are generally problems such as data bloat, transmission delay, low storage efficiency, and insufficient analysis ability. Especially in large-scale fire fighting scenarios, traditional fire fighting systems are prone to system paralysis due to excessive data volume, unable to achieve real-time warning and rapid decision-making, seriously affecting the reliability and practicability of the fire fighting system.

[0004] Secondly, although the adoption of the centralized data collection and processing method simplifies the system architecture to a certain extent, due to the huge data volume and single transmission path, it is easy to cause data transmission bottlenecks and excessive system load.

[0005] Finally, traditional systems mostly use relational databases for data storage, and it is difficult to meet the needs of high-frequency data access and large-scale data synchronization. Although some systems introduce cloud computing technology to improve data processing ability, there are still problems such as insufficient data security and poor real-time performance, and the stubborn problems of data bloat and system paralysis cannot be fundamentally solved.

[0006] In summary, how to solve the technical problem that the traditional fire fighting system is prone to system paralysis due to data bloat, transmission delay, low storage efficiency and poor real-time performance is an urgent problem to be solved. Summary of the Invention

[0007] The main purpose of the present invention is to provide an intelligent fire fighting system and an intelligent fire fighting data optimization method to solve the technical problem that the traditional fire fighting system is prone to system paralysis due to data bloat, transmission delay, low storage efficiency and poor real-time performance, thereby significantly improving the efficiency of data collection, processing, storage and analysis, enhancing the real-time performance, reliability and security of the system, and optimizing the user experience at the same time.

[0008] To achieve the above object, the present invention provides an intelligent fire fighting system and an intelligent fire fighting data optimization method.

[0009] In the first aspect, the present invention provides an intelligent fire fighting system, and the system includes:

[0010] A data acquisition module, where the data acquisition module includes multiple distributed sensor units, and the distributed sensor units are used to collect fire environment data in real time;

[0011] A data processing module, where the data processing module includes an edge computing unit and a cloud computing unit. The edge computing unit and the cloud computing unit are respectively connected to the distributed sensor units through a wireless communication network. The edge computing unit and the cloud computing unit are connected through an encrypted communication protocol. The edge computing unit and the cloud computing unit are used to receive and process the fire environment data;

[0012] A data storage module, where the data storage module includes a distributed database unit and a blockchain storage unit. The distributed database unit and the blockchain storage unit are respectively connected to the data processing module. The distributed database unit and the blockchain storage unit are connected through a data synchronization protocol; the distributed database unit and the blockchain storage unit are used to store the processed fire environment data;

[0013] A data analysis module, where the data analysis module includes a machine learning model unit and a deep learning model unit. The machine learning model unit and the deep learning model unit are respectively connected to the data storage module. The machine learning model unit and the deep learning model unit are connected through a data interface. The machine learning model unit and the deep learning model unit are used to perform multi-dimensional analysis on the stored fire environment data;

[0014] An early warning decision-making module, where the early warning decision-making module includes an early warning generation unit and a decision execution unit. The early warning generation unit and the decision execution unit are respectively connected to the data analysis module. The early warning generation unit and the decision execution unit are connected through an instruction transmission protocol. The early warning generation unit and the decision execution unit are used to generate early warning information and execute decisions according to the output result after the multi-dimensional analysis;

[0015] A user interaction module, where the user interaction module includes a visualization interface unit and a mobile terminal unit. The visualization interface unit and the mobile terminal unit are respectively connected to the early warning decision-making module. The visualization interface unit and the mobile terminal unit are connected through a wireless communication network. The visualization interface unit and the mobile terminal unit are used to display early warning information and decision results to users.

[0016] Optionally, the distributed sensor units include a temperature sensor, a smoke sensor, a gas sensor, and an image sensor. The temperature sensor, the smoke sensor, the gas sensor, and the image sensor are connected to the data processing module through a wireless communication network.

[0017] Optionally, the edge computing unit is used to perform real-time preprocessing on the fire environment data, and the preprocessing includes data cleaning, data compression, and data encryption. The cloud computing unit is used to perform in-depth analysis and model training on the preprocessed fire environment data.

[0018] Optionally, the distributed database unit is used to store the fire environment data with an access frequency greater than or equal to N times per second, and the blockchain storage unit is used to store the fire environment data with an access frequency less than N times per second.

[0019] Optionally, the machine learning model unit is used to perform classification and regression analysis on the fire environment data, and the deep learning model unit is used to perform anomaly detection and trend prediction on the fire environment data.

[0020] Optionally, the warning generation unit is used to generate warning information according to the output result after the multi-dimensional analysis, and the decision execution unit is used to execute the control instruction of the fire-fighting equipment according to the warning information.

[0021] Optionally, the visualization interface unit is used to display the warning information and decision results in the form of charts, and the mobile terminal unit is used to receive and display the warning information and decision results to the user.

[0022] Optionally, both the machine learning model unit and the deep learning model unit adopt a distributed training framework, and the distributed training framework is connected to the data storage module through a data interface.

[0023] In a second aspect, the present invention provides a method for optimizing smart fire data, and the method is applied to the smart fire system described in the first aspect. The optimization method includes:

[0024] Real-time collect fire environment data;

[0025] Receive and process the fire environment data;

[0026] Store the processed fire environment data;

[0027] Perform multi-dimensional analysis on the stored fire environment data;

[0028] Generate warning information according to the output result after the multi-dimensional analysis and execute the decision;

[0029] Display the warning information and decision results to the user.

[0030] Optionally, the receiving and processing the fire environment data includes:

[0031] Perform real-time preprocessing on the fire environment data, and the preprocessing includes data cleaning, data compression, and data encryption;

[0032] Deeply analyze and model train the preprocessed fire environment data.

[0033] The intelligent fire protection system and intelligent fire protection data optimization method provided by this application. The system includes a data acquisition module, a data processing module, a data storage module, a data analysis module, an early warning decision module, and a user interaction module. The data acquisition module uses multiple distributed sensor units to obtain fire environment data in real time. The data processing module includes edge computing and cloud computing units, and efficiently processes data through wireless communication and encryption protocols. The data storage module combines a distributed database and a blockchain storage unit to ensure secure data synchronization. The data analysis module uses machine learning and deep learning technologies to perform multi-dimensional analysis on the stored data. The early warning decision module quickly generates an early warning and executes a decision based on the analysis results. The user interaction module uses a visual interface and a mobile terminal to display the early warning information and decision results in real time. This system solves the technical problem that the traditional fire protection system is prone to paralysis due to data bloat, transmission delay, low storage efficiency, and poor real-time performance, enhances the real-time performance, reliability, and security of the system, and optimizes the user experience at the same time. Description of the Drawings

[0034] The specification drawings forming a part of this application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0035] Figure 1 It is a schematic diagram of the intelligent fire protection system provided by this application;

[0036] Figure 2 It is a flowchart of the intelligent fire protection data optimization method provided by this application.

[0037] Through the above drawings, the clear embodiments of this application have been shown, and there will be more detailed descriptions later. These drawings and text descriptions are not intended to limit the scope of the concept of this application in any way, but to illustrate the concept of this application to those skilled in the art by referring to specific embodiments. Detailed Embodiments

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the protection scope of this application.

[0039] In the description and claims of the present invention and the above-mentioned drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order different from those illustrated or described here.

[0040] In the present invention, words such as "exemplary" or "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0041] The intelligent fire protection system and the intelligent fire protection data optimization method provided by this application. The system includes data acquisition, processing, storage, analysis, early warning decision-making and user interaction modules. The system collects data in real time through distributed sensors, processes it through the cooperation of the edge and the cloud, and stores it in a distributed database and a blockchain unit. Machine learning and deep learning models are used to analyze the data. The early warning decision-making module responds quickly, and the results are displayed through a visualization interface and a mobile terminal. This system solves the technical problem that the traditional fire protection system is prone to paralysis due to data bloat, transmission delay, low storage efficiency and poor real-time performance, enhances the real-time performance, reliability and security of the system, and optimizes the user experience at the same time.

[0042] The technical solutions of this application and how the technical solutions of this application solve the above technical problems will be described in detail below with specific embodiments. These several specific embodiments below can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this application will be described below with reference to the drawings.

[0043] Figure 1 It is a schematic diagram of the intelligent fire protection system provided by this application, as Figure 1 shown, the intelligent fire protection system provided in this embodiment, the system includes:

[0044] A data acquisition module, the data acquisition module includes a plurality of distributed sensor units, and the distributed sensor units are used to collect fire protection environment data in real time;

[0045] A data processing module, the data processing module includes an edge computing unit and a cloud computing unit, the edge computing unit and the cloud computing unit are respectively connected to the distributed sensor units through a wireless communication network, the edge computing unit and the cloud computing unit are connected through an encrypted communication protocol, and the edge computing unit and the cloud computing unit are used to receive and process the fire protection environment data;

[0046] A data storage module, which includes a distributed database unit and a blockchain storage unit. The distributed database unit and the blockchain storage unit are respectively connected to the data processing module, and the distributed database unit and the blockchain storage unit are connected through a data synchronization protocol; the distributed database unit and the blockchain storage unit are used to store the processed fire environment data;

[0047] A data analysis module, which includes a machine learning model unit and a deep learning model unit. The machine learning model unit and the deep learning model unit are respectively connected to the data storage module, the machine learning model unit and the deep learning model unit are connected through a data interface, and the machine learning model unit and the deep learning model unit are used to perform multi-dimensional analysis on the stored fire environment data;

[0048] An early warning decision-making module, which includes an early warning generation unit and a decision execution unit. The early warning generation unit and the decision execution unit are respectively connected to the data analysis module, the early warning generation unit and the decision execution unit are connected through an instruction transmission protocol, and the early warning generation unit and the decision execution unit are used to generate early warning information and execute decisions according to the output result of the multi-dimensional analysis;

[0049] A user interaction module, which includes a visualization interface unit and a mobile terminal unit. The visualization interface unit and the mobile terminal unit are respectively connected to the early warning decision-making module, the visualization interface unit and the mobile terminal unit are connected through a wireless communication network, and the visualization interface unit and the mobile terminal unit are used to display early warning information and decision results to users.

[0050] Optionally, the distributed sensor unit includes a temperature sensor, a smoke sensor, a gas sensor, and an image sensor, and the temperature sensor, the smoke sensor, the gas sensor, and the image sensor are connected to the data processing module through a wireless communication network.

[0051] Optionally, the edge computing unit is used to perform real-time preprocessing on the fire environment data, and the preprocessing includes data cleaning, data compression, and data encryption. The cloud computing unit is used to perform in-depth analysis and model training on the preprocessed fire environment data.

[0052] Optionally, the distributed database unit is used to store the fire environment data with an access frequency greater than or equal to N times per second, and the blockchain storage unit is used to store the fire environment data with an access frequency less than N times per second.

[0053] Optionally, the machine learning model unit is used to classify and perform regression analysis on fire environment data, and the deep learning model unit is used to detect anomalies and predict trends in fire environment data.

[0054] Optionally, the warning generation unit is used to generate warning information based on the output result after multi-dimensional analysis, and the decision execution unit is used to execute the control instruction of the fire-fighting equipment according to the warning information.

[0055] Optionally, the visualization interface unit is used to display warning information and decision results in the form of charts, and the mobile terminal unit is used to receive and display warning information and decision results to users.

[0056] Optionally, both the machine learning model unit and the deep learning model unit adopt a distributed training framework, and the distributed training framework is connected to the data storage module through a data interface.

[0057] The system includes the following modules and units, and the specific implementation is as follows:

[0058] 1. Data acquisition module

[0059] The data acquisition module includes multiple distributed sensor units for real-time acquisition of fire environment data. The distributed sensor units include temperature sensors, smoke sensors, gas sensors, and image sensors. The temperature sensors are used to collect ambient temperature data, the smoke sensors are used to detect smoke concentration, the gas sensors are used to detect harmful gas concentration, and the image sensors are used to collect image data of the fire scene. All sensors are connected to the data processing module through a wireless communication network (such as Wi-Fi or LoRa).

[0060] 2. Data processing module

[0061] The data processing module includes an edge computing unit and a cloud computing unit. The edge computing unit is connected to the distributed sensor units through a wireless communication network and is used to perform real-time preprocessing on fire environment data. The preprocessing includes data cleaning, data compression, and data encryption. Data cleaning is used to remove noise data, data compression is used to reduce the amount of data transmission, and data encryption is used to ensure the security of data transmission. The cloud computing unit is connected to the edge computing unit through an encrypted communication protocol (such as TLS) and is used to perform in-depth analysis and model training on the preprocessed fire environment data. The data processing module includes an edge computing unit and a cloud computing unit, and the specific implementation is as follows:

[0062] 2.1 Edge computing unit

[0063] The edge computing unit is connected to the distributed sensor unit via a wireless communication network (such as Wi-Fi or LoRa) and is used for real-time preprocessing of fire environment data. The specific implementation of the edge computing unit includes the following steps:

[0064] 1. Data cleaning:

[0065] The edge computing unit uses the Median Filtering algorithm to remove the noise data collected by the sensors. The Median Filtering algorithm eliminates outliers by calculating the median of the data sequence to ensure the accuracy of the data.

[0066] 2. Data compression:

[0067] The edge computing unit uses Huffman Coding to compress the fire environment data and reduce the data transmission volume. Huffman Coding generates an optimal binary coding table by statistically analyzing the frequency distribution of the data to achieve efficient compression.

[0068] 3. Data encryption:

[0069] The edge computing unit uses the AES-256 encryption algorithm to encrypt the compressed data to ensure the security of data transmission. The AES-256 encryption algorithm converts the data into ciphertext through symmetric key encryption technology to prevent the data from being stolen or tampered with during transmission.

[0070] The edge computing unit transmits the preprocessed data to the cloud computing unit via the wireless communication network. During the transmission process, the TLS (Transport Layer Security) protocol is used for encrypted communication to ensure the integrity and security of data transmission.

[0071] 2.2 Cloud computing unit

[0072] The cloud computing unit is connected to the edge computing unit via an encrypted communication protocol (such as TLS) and is used for in-depth analysis and model training of the preprocessed fire environment data. The specific implementation of the cloud computing unit includes the following steps:

[0073] 1. In-depth analysis:

[0074] The cloud computing unit uses the K-means clustering algorithm to perform clustering analysis on the fire environment data to identify potential patterns in the data. The K-means clustering algorithm helps to discover outliers or trends in the data by dividing the data into multiple clusters.

[0075] The cloud computing unit uses the Principal Component Analysis (PCA) algorithm to perform dimensionality reduction on high-dimensional data, extract the main features, and reduce the data complexity.

[0076] 2. Model Training:

[0077] The cloud computing unit uses the Random Forest algorithm to train the classification model for fire environment data. The Random Forest algorithm improves the accuracy and robustness of classification by constructing multiple decision trees and performing ensemble learning.

[0078] The cloud computing unit uses the Long Short-Term Memory network (LSTM) to train the time series prediction model for fire environment data. LSTM can capture long-term dependencies in the data through memory units and gating mechanisms, and is suitable for fire trend prediction.

[0079] The cloud computing unit deploys the trained models to the edge computing unit. The edge computing unit uses these models to perform local analysis and decision-making on real-time data, reducing the dependence on the cloud and improving the real-time performance of the system.

[0080] 2.3 Connection Modes of the Data Processing Module

[0081] The edge computing unit is connected to the distributed sensor unit through a wireless communication network (such as Wi-Fi or LoRa), and specifically uses the MQTT protocol for data transmission. The MQTT protocol is a lightweight publish / subscribe message transmission protocol, suitable for low-bandwidth and unstable network environments.

[0082] The edge computing unit communicates with the cloud computing unit through the TLS protocol to ensure the security of data transmission. The TLS protocol provides end-to-end data encryption and authentication through digital certificates and key exchange mechanisms.

[0083] Technical Effects:

[0084] The edge computing unit reduces the data transmission volume and improves the data security through real-time preprocessing (data cleaning, data compression, and data encryption).

[0085] The cloud computing unit provides high-precision data analysis and prediction capabilities through in-depth analysis and model training.

[0086] The data processing module realizes the efficient processing and analysis of fire environment data through the collaborative work of edge computing and cloud computing, enhancing the real-time performance and reliability of the system.

[0087] 3. Data Storage Module

[0088] The data storage module includes a distributed database unit and a blockchain storage unit. The distributed database unit is used to store fire environment data with an access frequency greater than or equal to N times per second, and the blockchain storage unit is used to store fire environment data with an access frequency less than N times per second. The distributed database unit and the blockchain storage unit are connected through a data synchronization protocol (such as the Paxos protocol) to ensure data consistency and security. The data storage module includes a distributed database unit and a blockchain storage unit, and the specific implementation is as follows:

[0089] 3.1 Distributed Database Unit

[0090] The distributed database unit is used to store fire environment data with an access frequency greater than or equal to N times per second. The distributed database unit adopts the Cassandra database. Cassandra is a highly scalable distributed NoSQL database, suitable for high-frequency data access scenarios. The specific implementation includes the following steps:

[0091] 1. Data Partitioning:

[0092] The distributed database unit uses the Consistent Hashing algorithm to partition and store the fire environment data. The Consistent Hashing algorithm ensures high availability and load balancing of data by evenly distributing the data across multiple nodes.

[0093] 2. Data Replication:

[0094] The distributed database unit adopts a multi-copy mechanism. Each data partition stores copies on multiple nodes, and the number of copies is set to 3. Through the multi-copy mechanism, it is ensured that when a certain node fails, the data can still be accessed from other nodes, improving the fault tolerance of the system.

[0095] 3. Data Access:

[0096] The distributed database unit provides a data access interface through CQL (Cassandra Query Language), supporting fast read and write operations for high-frequency data. CQL is a query language similar to SQL, suitable for data operations in distributed databases.

[0097] 3.2 Blockchain Storage Unit

[0098] The blockchain storage unit is used to store fire environment data with an access frequency less than N times per second. The blockchain storage unit adopts the Hyperledger Fabric framework. Hyperledger Fabric is an enterprise-level blockchain platform, suitable for data storage scenarios with low frequency but high security and immutability. The specific implementation includes the following steps:

[0099] 1. Data uploading to the blockchain:

[0100] The blockchain storage unit uses a smart contract to upload fire environment data to the blockchain. The smart contract is implemented through chaincode, which is written in the Go language and defines the rules and processes for data uploading to the blockchain.

[0101] 2. Data verification:

[0102] The blockchain storage unit uses the PBFT (Practical Byzantine Fault Tolerance) consensus algorithm to verify data. The PBFT consensus algorithm ensures data consistency and immutability through a multi-node voting mechanism.

[0103] 3. Data storage:

[0104] The blockchain storage unit stores fire environment data in the distributed ledger in the form of blocks. Each block contains the hash value of the data, a timestamp, and the hash value of the previous block, ensuring data integrity and traceability.

[0105] 3.3 Data synchronization protocol

[0106] The distributed database unit and the blockchain storage unit perform data synchronization through the Paxos protocol. The Paxos protocol is a distributed consensus algorithm used to ensure data consistency among multiple nodes. The specific implementation includes the following steps:

[0107] 1. Proposal stage:

[0108] The distributed database unit sends a data synchronization proposal to the blockchain storage unit. The proposal contains the fire environment data to be synchronized.

[0109] 2. Voting stage:

[0110] Multiple nodes of the blockchain storage unit vote on the proposal to ensure the legality and consistency of the proposal.

[0111] 3. Confirmation stage:

[0112] When the proposal is agreed upon by a majority of nodes, the distributed database unit synchronizes the data to the blockchain storage unit to ensure data consistency and security.

[0113] 3.4 Connection method of the data storage module

[0114] The distributed database unit and the blockchain storage unit perform data synchronization through the Paxos protocol, and specifically use the gRPC framework to implement communication between nodes. gRPC is a high-performance remote procedure call framework that supports multiple programming languages and is suitable for data synchronization in distributed systems.

[0115] The distributed database unit is connected to the data processing module through the CQL interface, providing read and write services for high-frequency data.

[0116] The blockchain storage unit is connected to the data processing module through the REST API interface, providing storage and query services for low-frequency data.

[0117] 3.5 Specific technical means of the data storage module

[0118] Distributed database unit: Uses the Cassandra database, adopts the consistent hashing algorithm for data partitioning, and the multi-copy mechanism ensures high data availability. The CQL interface provides data access.

[0119] Blockchain storage unit: Uses the Hyperledger Fabric framework, smart contracts implement data uploading to the chain, the PBFT consensus algorithm ensures data consistency, and the distributed ledger stores data.

[0120] Data synchronization protocol: Uses the Paxos protocol for data synchronization, and the gRPC framework realizes communication between nodes.

[0121] 3.6 Data interface design

[0122] The data analysis module obtains data from the storage module through the data interface (gRPC protocol). The interface definition is as follows:

[0123] service DataQuery{

[0124] rpc GetFireData(DataRequest)returns(stream DataResponse){}

[0125] }

[0126] message DataRequest{

[0127] string sensor_id = 1;

[0128] int64 start_timestamp = 2;

[0129] int64 end_timestamp = 3;

[0130] }

[0131] message DataResponse{

[0132] float temperature = 1;

[0133] float smoke_level = 2;

[0134] bytes image_data = 3; / / Compressed image in JPEG format

[0135] }

[0136] High-frequency data (access frequency ≥ 10 times / second) is directly read from the distributed database (Cassandra), and low-frequency data is queried from the blockchain (Hyperledger Fabric) through smart contracts.

[0137] Technical effects:

[0138] The distributed database unit realizes the efficient storage and access of high-frequency data through the Cassandra database and the consistent hashing algorithm.

[0139] The blockchain storage unit ensures the security and immutability of low-frequency data through Hyperledger Fabric and the PBFT consensus algorithm.

[0140] The data storage module realizes data synchronization between the distributed database unit and the blockchain storage unit through the Paxos protocol and the gRPC framework, ensuring data consistency and security.

[0141] 4. Data analysis module

[0142] The data analysis module includes a machine learning model unit and a deep learning model unit. The machine learning model unit is used for classification and regression analysis of fire environment data. Specifically, the support vector machine (SVM) and random forest (RandomForest) algorithms are used for classification, and the linear regression (Linear Regression) algorithm is used for regression analysis. The deep learning model unit is used for anomaly detection and trend prediction of fire environment data. Specifically, the long short-term memory network (LSTM) is used for anomaly detection, and the convolutional neural network (CNN) is used for trend prediction. Both the machine learning model unit and the deep learning model unit adopt a distributed training framework (such as TensorFlow or PyTorch), and the distributed training framework is connected to the data storage module through a data interface. The data analysis module includes a machine learning model unit and a deep learning model unit, and the specific implementation method is as follows:

[0143] 4.1 Machine learning model unit

[0144] The machine learning model unit is used for classification and regression analysis of fire environment data. The specific implementation includes the following steps:

[0145] 1. Classification analysis:

[0146] The machine learning model unit classifies the fire environment data using the Support Vector Machine (SVM) algorithm. SVM divides the data into different categories by constructing a hyperplane and is suitable for the classification task of fire risk levels. SVM uses the Radial Basis Function (RBF) as the kernel function and optimizes the classification effect by adjusting the penalty parameter C and the kernel function parameter γ.

[0147] The machine learning model unit classifies the fire environment data using the Random Forest algorithm. Random Forest improves the classification accuracy and robustness by constructing multiple decision trees and performing ensemble learning. The parameter settings of Random Forest include the number of trees (n_estimators) and the maximum depth of each tree (max_depth).

[0148] 2. Regression analysis:

[0149] The machine learning model unit performs regression analysis on the fire environment data using the Linear Regression algorithm. Linear Regression predicts continuous values by fitting a linear equation and is suitable for the prediction task of fire spread speed. Linear Regression uses the Least Squares method for parameter estimation to minimize the prediction error.

[0150] The machine learning model unit is connected to the data storage module through a data interface, obtains the fire environment data from the distributed database unit and the blockchain storage unit, and stores the analysis results back to the data storage module.

[0151] 4.2 Deep learning model unit

[0152] The deep learning model unit is used for anomaly detection and trend prediction of the fire environment data. The specific implementation includes the following steps:

[0153] 1. Anomaly detection:

[0154] The deep learning model unit uses the Long Short-Term Memory (LSTM) network to detect anomalies in the fire environment data. LSTM can capture long-term dependencies in the data through memory units and gating mechanisms and is suitable for the detection task of fire abnormal behaviors. The network structure of LSTM includes an input layer, a hidden layer, and an output layer. The number of units in the hidden layer is set to 128, and the activation function uses ReLU (Rectified Linear Unit).

[0155] 2. Trend prediction:

[0156] The deep learning model unit uses a convolutional neural network (CNN) to predict the trend of fire environment data. The CNN extracts the spatial features of the data through convolutional layers and pooling layers, and is suitable for the prediction task of fire spread trend. The network structure of the CNN includes convolutional layers, pooling layers and fully connected layers. The number of filters in the convolutional layer is set to 64, and the pooling layer uses the max pooling operation.

[0157] The deep learning model unit is connected to the data storage module through a data interface, obtains fire environment data from the distributed database unit and the blockchain storage unit, and stores the analysis results back to the data storage module.

[0158] 4.3 Distributed Training Framework

[0159] Both the machine learning model unit and the deep learning model unit adopt the TensorFlow distributed training framework. TensorFlow distributes the training tasks to multiple computing nodes through data parallelism and model parallelism to improve the training efficiency. The specific implementation includes the following steps:

[0160] 1. Data Parallelism:

[0161] TensorFlow divides the training data into multiple mini-batches. Each computing node is responsible for processing one mini-batch of data and updates the model parameters through gradient aggregation.

[0162] 2. Model Parallelism:

[0163] TensorFlow distributes different layers of the model to different computing nodes. Each node is responsible for calculating the output of a part of the layers and passes the intermediate results through communication between nodes.

[0164] TensorFlow is connected to the data storage module through a data interface, obtains training data from the distributed database unit and the blockchain storage unit, and stores the trained model back to the data storage module.

[0165] 4.4 Connection Method of the Data Analysis Module

[0166] The machine learning model unit and the deep learning model unit are connected to the data storage module through a data interface, and specifically use the REST API interface for data transmission. The REST API interface provides data read and write services through the HTTP protocol to ensure efficient data transmission and access.

[0167] The machine learning model unit and the deep learning model unit are connected to the computing nodes through the TensorFlow distributed training framework, and the gRPC framework is specifically used to implement communication between nodes. gRPC ensures the efficient execution of training tasks through efficient binary encoding and streaming transmission.

[0168] 4.5 Specific technical means of the data analysis module

[0169] Classification analysis: Use the support vector machine (SVM) and random forest (Random Forest) algorithms for data classification.

[0170] Regression analysis: Use the linear regression (Linear Regression) algorithm for data regression.

[0171] Anomaly detection: Use the long short-term memory network (LSTM) for data anomaly detection.

[0172] Trend prediction: Use the convolutional neural network (CNN) for data trend prediction.

[0173] Distributed training: Use the TensorFlow distributed training framework for model training.

[0174] 4.6 Model structure and hierarchical relationship

[0175] The machine learning model unit adopts the random forest (Random Forest) algorithm, which contains 100 decision trees. The input layer receives the standardized fire environment data (temperature, smoke concentration, gas concentration, image feature vector) from the data storage module, and the output layer generates the fire risk level classification results (low, medium, high) through the Softmax function and linearly regresses to predict the fire spread rate.

[0176] The deep learning model unit adopts the long short-term memory network (LSTM) structure, which contains 3 hidden layers (each layer has 128 neurons). The input layer receives the sequential fire data (time window is 60 seconds), and the output layer generates the anomaly detection probability value (0 - 1) through the Sigmoid function and the trend prediction result based on the Attention mechanism.

[0177] Model connection method: The classification result of the machine learning model is used as the auxiliary input of the deep model, and parameter transfer between models is realized through the data interface (REST API), and the input / output data is encapsulated in JSON format.

[0178] 4.7 Data input / output process

[0179] Input data processing: Extract real-time fire environment data (in JSON format) from the distributed database unit, including temperature (unit: °C), smoke concentration (ppm), CO concentration (ppm), and the proportion of flame pixels extracted by the image sensor (0 - 100%).

[0180] Stream the data to the machine learning model unit and the deep learning model unit through the data interface (Apache Kafka message queue). The input data needs to be standardized (Z-Score normalization).

[0181] Output data definition:

[0182] Output of the machine learning model: Fire risk level (classification label) and predicted value of the spread rate (m 2 / s).

[0183] Output of the deep learning model: Abnormal detection confidence level (warning is triggered when the threshold ≥ 0.85) and fire trend in the next 5 minutes (rising / steady / falling).

[0184] Model update mechanism:

[0185] The model adopts an offline batch update strategy. Every week, historical data (accounting for 30%) is extracted from the blockchain storage unit, and incremental training is carried out through the TensorFlow distributed training framework (parameter server architecture). After the model version is verified by the hash value, it is deployed to the production environment.

[0186] Technical effects:

[0187] The machine learning model unit realizes the classification and regression analysis of fire environment data through support vector machine, random forest, and linear regression algorithms, improving the prediction accuracy of fire risk level and spread speed.

[0188] The deep learning model unit realizes the abnormal detection and trend prediction of fire environment data through long short-term memory network and convolutional neural network, improving the detection ability of fire abnormal behaviors and spread trends.

[0189] The data analysis module realizes the efficient training and deployment of the model through the TensorFlow distributed training framework, improving the real-time performance and reliability of the system.

[0190] 5. Early warning decision-making module

[0191] The early warning decision-making module includes an early warning generation unit and a decision execution unit. The early warning generation unit is used to generate early warning information based on the multi-dimensional analysis results of the data analysis module, and specifically uses a rule-based early warning generation algorithm (such as the IF-THEN rule) to generate early warning information. The decision execution unit is used to execute the control instructions of the fire-fighting equipment according to the early warning information, and specifically controls the start or shutdown of the fire-fighting equipment through the Modbus protocol. The early warning decision-making module includes an early warning generation unit and a decision execution unit, and the specific implementation method is as follows:

[0192] 5.1 Early warning generation unit

[0193] The early warning generation unit is used to generate early warning information based on the multi-dimensional analysis results of the data analysis module, and the specific implementation includes the following steps:

[0194] 1. Rule-based early warning generation algorithm:

[0195] The early warning generation unit uses the IF-THEN rule to generate early warning information. The IF-THEN rule generates corresponding early warning information through condition judgment and logical reasoning. The specific rules are as follows:

[0196] Rule 1: If the temperature value detected by the temperature sensor is greater than 60 degrees Celsius, then generate the "high temperature warning" information.

[0197] Rule 2: If the smoke concentration detected by the smoke sensor is greater than 50 ppm, then generate the "smoke warning" information.

[0198] Rule 3: If the concentration of harmful gases detected by the gas sensor is greater than 20 ppm, then generate the "harmful gas warning" information.

[0199] Rule 4: If the flame area detected by the image sensor is greater than 10 square centimeters, then generate the "flame warning" information.

[0200] 2. Early warning information generation:

[0201] The early warning generation unit generates early warning information according to the IF-THEN rule and stores the early warning information in the distributed database unit. The early warning information includes the early warning type, early warning level, early warning time, and early warning location.

[0202] The early warning generation unit is connected to the data analysis module through a data interface, obtains the multi-dimensional analysis results from the data analysis module, and generates early warning information according to the analysis results.

[0203] 5.2 Decision execution unit

[0204] The decision execution unit is used to execute the control instructions of the fire-fighting equipment according to the early warning information, and the specific implementation includes the following steps:

[0205] 1. Control instruction generation:

[0206] The decision execution unit generates control instructions according to the warning information. The specific instructions are as follows:

[0207] Instruction 1: If the "high temperature warning" information is generated, generate the "start sprinkler system" instruction.

[0208] Instruction 2: If the "smoke warning" information is generated, generate the "start smoke exhaust system" instruction.

[0209] Instruction 3: If the "hazardous gas warning" information is generated, generate the "start ventilation system" instruction.

[0210] Instruction 4: If the "flame warning" information is generated, generate the "start fire extinguishing system" instruction.

[0211] 2. Control instruction execution:

[0212] The decision execution unit sends the control instructions to the fire protection equipment through the Modbus protocol. The Modbus protocol is connected to the fire protection equipment through the RS-485 serial communication interface to ensure the accurate transmission and execution of the instructions. The specific implementation is as follows:

[0213] Sprinkler system: Send a start instruction through the Modbus protocol to control the solenoid valve of the sprinkler system to open and start the sprinkler.

[0214] Smoke exhaust system: Send a start instruction through the Modbus protocol to control the fan of the smoke exhaust system to start and conduct smoke exhaust.

[0215] Ventilation system: Send a start instruction through the Modbus protocol to control the fan of the ventilation system to start and conduct ventilation.

[0216] Fire extinguishing system: Send a start instruction through the Modbus protocol to control the release of the fire extinguishing agent of the fire extinguishing system to conduct fire extinguishing.

[0217] The decision execution unit is connected to the warning generation unit through the data interface, obtains the warning information from the warning generation unit, and generates and executes control instructions according to the warning information.

[0218] 5.3 Connection method of the warning decision module

[0219] The warning generation unit is connected to the data analysis module through the data interface, and specifically uses the REST API interface for data transmission. The REST API interface provides data reading and writing services through the HTTP protocol to ensure the efficient generation and storage of warning information.

[0220] The decision execution unit is connected to the fire-fighting equipment through the Modbus protocol, and specifically uses the RS-485 serial communication interface for instruction transmission. The RS-485 interface transmits through differential signals to ensure the accuracy and reliability of control instructions.

[0221] 5.4 Specific technical means of the early warning decision module

[0222] Early warning generation: Use IF-THEN rules to generate early warning information. The specific rules include high-temperature warning, smoke warning, harmful gas warning, and flame warning.

[0223] Decision execution: Use the Modbus protocol to execute control instructions. The specific instructions include starting the sprinkler system, starting the smoke exhaust system, starting the ventilation system, and starting the fire extinguishing system.

[0224] 5.5 Model output and decision logic

[0225] Decision logic rules:

[0226] If the machine learning model outputs a fire risk level of "high" and the confidence level of anomaly detection by the deep learning model ≥ 0.9, trigger a first-level early warning (red alert), and the decision execution unit automatically starts the sprinkler system and unlocks the escape route.

[0227] If the deep learning model predicts that the fire trend in the next 5 minutes is "rising" and the current risk level is "medium", trigger a second-level early warning (orange alert), and the decision execution unit closes the ventilation system and notifies the fire station.

[0228] Fault tolerance mechanism:

[0229] When the model output results conflict (such as the machine learning classification is "low" but the deep learning detects an anomaly), start the manual review process, and at the same time push the review request to the administrator through the mobile terminal unit.

[0230] Technical effects:

[0231] The early warning generation unit generates early warning information through IF-THEN rules to ensure the accuracy and timeliness of early warning information.

[0232] The decision execution unit executes control instructions through the Modbus protocol to ensure the rapid response and effective control of fire-fighting equipment.

[0233] Through the coordinated work of early warning generation and decision execution, the early warning decision module realizes the automation of fire early warning and emergency handling, and improves the real-time performance and reliability of the system.

[0234] 6. User interaction module

[0235] The user interaction module includes a visualization interface unit and a mobile terminal unit. The visualization interface unit is used to display warning information and decision results in the form of charts, and specifically uses the ECharts library to generate bar charts, line charts, and pie charts. The mobile terminal unit is connected to the warning decision module through a wireless communication network (such as 4G or 5G), and is used to receive and display warning information and decision results to users, and is specifically implemented through a mobile application (such as an Android or iOS application). The user interaction module includes a visualization interface unit and a mobile terminal unit, and the specific implementation method is as follows:

[0236] 6.1 Visualization Interface Unit

[0237] The visualization interface unit is used to display warning information and decision results in the form of charts, and the specific implementation includes the following steps:

[0238] 1. Chart Generation:

[0239] The visualization interface unit uses the ECharts library to generate bar charts, line charts, and pie charts. ECharts is an open-source visualization library based on JavaScript and is suitable for generating interactive charts. The specific implementation is as follows:

[0240] Bar Chart: Used to display the number of fire warnings in different time periods. The X-axis of the bar chart represents time, the Y-axis represents the number of warnings, and each bar represents the number of warnings in a time period.

[0241] Line Chart: Used to display the fire spread trend. The X-axis of the line chart represents time, the Y-axis represents the fire spread speed, and the line represents the change trend of the fire spread.

[0242] Pie Chart: Used to display the proportion of different warning types. Each sector area of the pie chart represents a warning type, and the size of the area represents the proportion of this type of warning.

[0243] 2. Data Display:

[0244] The visualization interface unit obtains warning information and decision results from the warning decision module through the REST API interface, and passes the data to the ECharts library to generate charts. The charts are embedded in the web page through HTML5 and CSS3 technologies, and users can access the visualization interface through a browser.

[0245] The visualization interface unit is connected to the warning decision module through the REST API interface, and specifically uses the HTTP protocol for data transmission. The REST API interface obtains warning information and decision results through a GET request to ensure the efficient transmission and display of data.

[0246] 6.2 Mobile Terminal Unit

[0247] The mobile terminal unit is connected to the early warning decision-making module through a wireless communication network (such as 4G or 5G), and is used to receive and display early warning information and decision-making results to users. The specific implementation includes the following steps:

[0248] 1. Mobile application development:

[0249] The mobile terminal unit realizes the display of early warning information and decision-making results through a mobile application (such as an Android or iOS application). The mobile application is developed using the React Native framework. React Native is an open-source framework based on JavaScript and is suitable for developing cross-platform mobile applications.

[0250] 2. Data reception and display:

[0251] The mobile application obtains early warning information and decision-making results from the early warning decision-making module through the REST API interface and displays the data on the user interface. The specific display methods are as follows:

[0252] Early warning information display: The early warning information is displayed in the form of a list. Each piece of early warning information includes the early warning type, early warning level, early warning time, and early warning location.

[0253] Decision result display: The decision result is displayed in the form of a notification. Each notification includes the decision type, execution time, and execution status.

[0254] 3. User interaction:

[0255] The mobile application provides user interaction functions. Users can view detailed information by clicking on the early warning information or view the execution status by clicking on the decision result. Users can also set the early warning threshold and the way of receiving notifications through the application.

[0256] The mobile terminal unit is connected to the early warning decision-making module through a wireless communication network (such as 4G or 5G). The HTTP protocol is specifically used for data transmission. The REST API interface obtains early warning information and decision-making results through a GET request to ensure the efficient transmission and display of data.

[0257] 6.3 Connection method of the user interaction module

[0258] The visualization interface unit is connected to the early warning decision-making module through the REST API interface. The HTTP protocol is specifically used for data transmission. The REST API interface obtains early warning information and decision-making results through a GET request to ensure the efficient transmission and display of data.

[0259] The mobile terminal unit is connected to the early warning decision-making module through a wireless communication network (such as 4G or 5G), and specifically uses the HTTP protocol for data transmission. The REST API interface obtains early warning information and decision-making results through GET requests to ensure efficient data transmission and display.

[0260] 6.4 Specific technical means of the user interaction module

[0261] Chart generation: Use the ECharts library to generate bar charts, line charts, and pie charts. Specific charts include the bar chart of the number of early warnings, the line chart of the fire spread trend, and the pie chart of the proportion of early warning types.

[0262] Mobile application: Develop Android and iOS applications using the React Native framework. Specific functions include the display of early warning information, the display of decision-making results, and user interaction.

[0263] Data transmission: Use the REST API interface to perform data transmission through the HTTP protocol to ensure efficient data transmission and display.

[0264] Technical effects:

[0265] The visualization interface unit generates charts through the ECharts library, realizing the visual display of early warning information and decision-making results, and improving the readability and comprehensibility of data.

[0266] The mobile terminal unit develops a mobile application through the React Native framework, realizing the real-time display of early warning information and decision-making results and user interaction, and improving the convenience of the system and the user experience.

[0267] The user interaction module realizes the multi-channel display of early warning information and decision-making results through the collaborative work of the visualization interface and the mobile terminal, enhancing the real-time performance and reliability of the system.

[0268] Specific implementation steps of the system:

[0269] 1. Real-time collection of fire environment data:

[0270] Temperature sensors, smoke sensors, gas sensors, and image sensors collect fire environment data in real time through a wireless communication network and transmit the data to the data processing module.

[0271] 2. Receive and process fire environment data:

[0272] The edge computing unit performs real-time preprocessing on the received fire environment data, including data cleaning, data compression, and data encryption. The preprocessed data is transmitted to the cloud computing unit through an encrypted communication protocol for in-depth analysis and model training.

[0273] 3. Store the processed fire environment data:

[0274] The processed fire environment data is stored in the distributed database unit and the blockchain storage unit respectively according to the access frequency. The distributed database unit stores the high-frequency access data, and the blockchain storage unit stores the low-frequency access data.

[0275] 4. Conduct multi-dimensional analysis on the stored fire environment data:

[0276] The machine learning model unit uses the support vector machine and random forest algorithms to conduct classification and regression analysis on the fire environment data. The deep learning model unit uses the long short-term memory network and convolutional neural network to conduct anomaly detection and trend prediction on the fire environment data.

[0277] 5. Generate warning information according to the multi-dimensional analysis results and execute decisions:

[0278] The warning generation unit generates warning information according to the multi-dimensional analysis results, and the decision execution unit controls the startup or shutdown of the fire-fighting equipment through the Modbus protocol according to the warning information.

[0279] 6. Display the warning information and decision results to the user:

[0280] The visualization interface unit displays the warning information and decision results in the form of charts, and the mobile terminal unit displays the warning information and decision results to the user through the mobile application.

[0281] The intelligent fire-fighting system provided by the present invention collects fire environment data in real time through the distributed sensor unit, processes it through the cooperation of edge computing and cloud computing, and stores it in the distributed database and the blockchain storage unit. Utilize machine learning and deep learning models for multi-dimensional data analysis, and the warning decision module responds quickly, and the results are displayed through the visualization interface and the mobile terminal. This system solves the technical problems of the traditional fire-fighting system being prone to paralysis due to data bloat, transmission delay, low storage efficiency and poor real-time performance, enhances the real-time performance, reliability and security of the system, and optimizes the user experience at the same time.

[0282] Figure 2 It is a schematic flow chart of the intelligent fire-fighting data optimization method provided by this application, and the intelligent fire-fighting data optimization method is described in detail. As Figure 2 shown, the intelligent fire-fighting data optimization method provided by this embodiment includes:

[0283] S201: Collect fire environment data in real time.

[0284] Among them, in this step, fire environment data is collected in real time through multiple distributed sensors. The distributed sensor unit includes a temperature sensor, a smoke sensor, a gas sensor, and an image sensor. The temperature sensor is used to collect ambient temperature data, the smoke sensor is used to detect the smoke concentration, the gas sensor is used to detect the concentration of harmful gases, and the image sensor is used to collect image data of the fire scene.

[0285] In this step, fire environment data is collected in real time through multiple distributed sensors. The specific implementation method is as follows:

[0286] 1. Distributed sensor unit

[0287] The distributed sensor unit includes a temperature sensor, a smoke sensor, a gas sensor, and an image sensor. These sensors are connected to the data processing module through a wireless communication network (such as Wi-Fi or LoRa) to collect fire environment data in real time.

[0288] 2. Temperature sensor

[0289] The temperature sensor is used to collect ambient temperature data. The DS18B20 digital temperature sensor is adopted for the temperature sensor. This sensor communicates with the data processing module through the 1-Wire protocol. The temperature measurement range of the DS18B20 sensor is from -55°C to +125°C, and the accuracy is ±0.5°C. The temperature sensor collects ambient temperature data once per second and transmits the data to the data processing module through the wireless communication network.

[0290] 3. Smoke sensor

[0291] The smoke sensor is used to detect the smoke concentration. The MQ-2 gas sensor is adopted for the smoke sensor. This sensor outputs the smoke concentration value through an analog signal. The detection range of the MQ-2 sensor is from 300 ppm to 10000 ppm, and the response time is less than 10 seconds. The smoke sensor collects smoke concentration data once per second and transmits the data to the data processing module through the wireless communication network.

[0292] 4. Gas sensor

[0293] The gas sensor is used to detect the concentration of harmful gases. The MQ-135 gas sensor is adopted for the gas sensor. This sensor outputs the concentration value of harmful gases through an analog signal. The detection range of the MQ-135 sensor is from 10 ppm to 1000 ppm, and the response time is less than 30 seconds. The gas sensor collects the concentration data of harmful gases once per second and transmits the data to the data processing module through the wireless communication network.

[0294] 5. Image sensor

[0295] The image sensor is used to collect image data of the fire scene. The OV2640 image sensor is adopted for the image sensor, and this sensor communicates with the data processing module through the I2C protocol. The resolution of the OV2640 sensor is 1600x1200 pixels, and the frame rate is 30fps. The image sensor collects the image data of the fire scene once every second and transmits the data to the data processing module through the wireless communication network.

[0296] 6. Data Acquisition and Transmission

[0297] The temperature sensor, smoke sensor, gas sensor, and image sensor are connected to the data processing module through the wireless communication network (such as Wi-Fi or LoRa). The wireless communication network uses the MQTT protocol for data transmission. The MQTT protocol is a lightweight publish / subscribe message transmission protocol, which is suitable for low-bandwidth and unstable network environments.

[0298] Technical Effects:

[0299] The temperature sensor collects the ambient temperature data in real time through the DS18B20 digital temperature sensor to ensure the accuracy and real-time nature of the temperature data.

[0300] The smoke sensor detects the smoke concentration in real time through the MQ-2 gas sensor to ensure the accuracy and real-time nature of the smoke concentration data.

[0301] The gas sensor detects the concentration of harmful gases in real time through the MQ-135 gas sensor to ensure the accuracy and real-time nature of the harmful gas concentration data.

[0302] The image sensor collects the image data of the fire scene in real time through the OV2640 image sensor to ensure the clarity and real-time nature of the image data.

[0303] The distributed sensor unit is connected to the data processing module through the wireless communication network (such as Wi-Fi or LoRa) to ensure the efficient transmission and real-time nature of the data.

[0304] S202: Receive and process the fire environment data.

[0305] Furthermore, the receiving and processing of the fire environment data specifically include:

[0306] Perform real-time preprocessing on the fire environment data. The preprocessing includes data cleaning, data compression, and data encryption; perform in-depth analysis and model training on the preprocessed fire environment data.

[0307] Specifically, the edge computing unit performs real-time preprocessing on the received fire environment data, including data cleaning, data compression, and data encryption. The preprocessed fire environment data is transmitted to the cloud computing unit through an encrypted communication protocol for in-depth analysis and model training.

[0308] This step specifically includes performing real-time preprocessing on the fire environment data, where the preprocessing includes data cleaning, data compression, and data encryption; performing in-depth analysis and model training on the preprocessed fire environment data. The specific implementation methods are as follows:

[0309] 1. Real-time preprocessing of the edge computing unit

[0310] The edge computing unit performs real-time preprocessing on the received fire environment data, specifically including the following steps:

[0311] Data cleaning:

[0312] The edge computing unit uses the Median Filtering algorithm to remove the noise data collected by the sensors. The Median Filtering algorithm eliminates outliers by calculating the median of the data sequence to ensure the accuracy of the data.

[0313] Data compression:

[0314] The edge computing unit uses Huffman Coding to compress the fire environment data, reducing the data transmission volume. Huffman Coding generates an optimal binary coding table by statistically analyzing the frequency distribution of the data to achieve efficient compression.

[0315] Data encryption:

[0316] The edge computing unit uses the AES-256 encryption algorithm to encrypt the compressed data to ensure the security of data transmission. The AES-256 encryption algorithm converts the data into ciphertext through symmetric key encryption technology to prevent the data from being stolen or tampered with during transmission.

[0317] The edge computing unit is connected to the distributed sensor unit through a wireless communication network (such as Wi-Fi or LoRa), receives the fire environment data, and transmits the preprocessed data to the cloud computing unit through an encrypted communication protocol (such as TLS).

[0318] 2. In-depth analysis and model training of the cloud computing unit

[0319] The cloud computing unit performs in-depth analysis and model training on the preprocessed fire environment data, specifically including the following steps:

[0320] In-depth analysis:

[0321] The cloud computing unit uses the K-means clustering algorithm to perform clustering analysis on the fire environment data and identify potential patterns in the data. The K-means clustering algorithm helps discover outliers or trends in the data by dividing the data into multiple clusters.

[0322] The cloud computing unit uses principal component analysis (PCA) to reduce the dimensionality of high-dimensional data, extract the main features, and reduce the data complexity.

[0323] Model training:

[0324] The cloud computing unit uses the Random Forest algorithm to train a classification model for the fire environment data. Random Forest improves the accuracy and robustness of classification by constructing multiple decision trees and performing ensemble learning.

[0325] The cloud computing unit uses long short-term memory network (LSTM) to train a time series prediction model for the fire environment data. LSTM can capture long-term dependencies in the data through memory units and gating mechanisms, and is suitable for fire trend prediction.

[0326] The cloud computing unit connects to the edge computing unit through an encryption communication protocol (such as TLS), receives the preprocessed fire environment data, and deploys the trained model to the edge computing unit for real-time data analysis and decision-making.

[0327] Technical effects:

[0328] The edge computing unit realizes the real-time preprocessing of the fire environment data through median filtering algorithm, Huffman coding, and AES-256 encryption algorithm, ensuring the accuracy, efficiency, and security of the data.

[0329] The cloud computing unit realizes the in-depth analysis and model training of the fire environment data through the K-means clustering algorithm, principal component analysis, Random Forest, and long short-term memory network, improving the data analysis ability and prediction accuracy.

[0330] The data processing module realizes the efficient processing and analysis of the fire environment data through the collaborative work of edge computing and cloud computing, enhancing the real-time performance and reliability of the system.

[0331] S203: Store the preprocessed fire environment data.

[0332] Specifically, the preprocessed fire environment data is stored in the distributed database unit and the blockchain storage unit according to the access frequency. The distributed database unit stores the high-frequency access data, and the blockchain storage unit stores the low-frequency access data.

[0333] This step specifically includes storing the processed fire environment data in a distributed database unit and a blockchain storage unit respectively according to the access frequency. The distributed database unit stores high-frequency access data, and the blockchain storage unit stores low-frequency access data. The specific implementation is as follows:

[0334] 1. Distributed database unit

[0335] The distributed database unit is used to store fire environment data with an access frequency greater than or equal to N times per second. The distributed database unit uses the Cassandra database. Cassandra is a highly scalable distributed NoSQL database, suitable for high-frequency data access scenarios. The specific implementation includes the following steps:

[0336] Data partitioning:

[0337] The distributed database unit uses the Consistent Hashing algorithm to partition and store the fire environment data. The Consistent Hashing algorithm evenly distributes the data across multiple nodes to ensure high availability and load balancing of the data.

[0338] Data replication:

[0339] The distributed database unit adopts a multi-copy mechanism. Each data partition stores copies on multiple nodes, and the number of copies is set to 3. Through the multi-copy mechanism, it is ensured that when a certain node fails, the data can still be accessed from other nodes, improving the fault tolerance of the system.

[0340] Data access:

[0341] The distributed database unit provides a data access interface through CQL (Cassandra Query Language), supporting fast read and write operations for high-frequency data. CQL is a query language similar to SQL, suitable for data operations in distributed databases.

[0342] 2. Blockchain storage unit

[0343] The blockchain storage unit is used to store fire environment data with an access frequency less than N times per second. The blockchain storage unit uses the Hyperledger Fabric framework. Hyperledger Fabric is an enterprise-level blockchain platform, suitable for data storage scenarios with low frequency but high security and immutability. The specific implementation includes the following steps:

[0344] Data on-chain:

[0345] The blockchain storage unit uses a smart contract to upload fire environment data to the blockchain. The smart contract is implemented through chaincode, which is written in the Go language and defines the rules and processes for data upload to the blockchain.

[0346] Data verification:

[0347] The blockchain storage unit uses the PBFT (Practical Byzantine Fault Tolerance) consensus algorithm to verify data. The PBFT consensus algorithm ensures data consistency and immutability through a multi-node voting mechanism.

[0348] Data storage:

[0349] The blockchain storage unit stores fire environment data in the distributed ledger in the form of blocks. Each block contains the hash value of the data, a timestamp, and the hash value of the previous block, ensuring data integrity and traceability.

[0350] 3. Data synchronization protocol

[0351] The distributed database unit and the blockchain storage unit perform data synchronization through the Paxos protocol. The Paxos protocol is a distributed consensus algorithm used to ensure data consistency among multiple nodes. The specific implementation includes the following steps:

[0352] Proposal stage:

[0353] The distributed database unit sends a data synchronization proposal to the blockchain storage unit. The proposal contains the fire environment data to be synchronized.

[0354] Voting stage:

[0355] Multiple nodes of the blockchain storage unit vote on the proposal to ensure the legality and consistency of the proposal.

[0356] Confirmation stage:

[0357] When the proposal is approved by a majority of nodes, the distributed database unit synchronizes the data to the blockchain storage unit to ensure data consistency and security.

[0358] 4. Connection method of the data storage module

[0359] The distributed database unit and the blockchain storage unit perform data synchronization through the Paxos protocol, and specifically use the gRPC framework to implement communication between nodes. gRPC is a high-performance remote procedure call framework that supports multiple programming languages and is suitable for data synchronization in distributed systems.

[0360] The distributed database unit is connected to the data processing module through the CQL interface, providing read and write services for high-frequency data.

[0361] The blockchain storage unit is connected to the data processing module through the REST API interface, providing storage and query services for low-frequency data.

[0362] Technical effects:

[0363] The distributed database unit realizes the efficient storage and access of high-frequency data through the Cassandra database and the consistent hashing algorithm.

[0364] The blockchain storage unit ensures the security and immutability of low-frequency data through Hyperledger Fabric and the PBFT consensus algorithm.

[0365] The data storage module realizes data synchronization between the distributed database unit and the blockchain storage unit through the Paxos protocol and the gRPC framework, ensuring data consistency and security.

[0366] S204: Perform multi-dimensional analysis on the stored fire environment data.

[0367] Specifically, the machine learning model unit uses the support vector machine and random forest algorithms to perform classification and regression analysis on the fire environment data. The deep learning model unit uses the long short-term memory network and convolutional neural network to perform anomaly detection and trend prediction on the fire environment data.

[0368] This step specifically includes using the machine learning model unit and the deep learning model unit to perform multi-dimensional analysis on the stored fire environment data. The machine learning model unit uses the support vector machine and random forest algorithms to perform classification and regression analysis on the fire environment data, and the deep learning model unit uses the long short-term memory network and convolutional neural network to perform anomaly detection and trend prediction on the fire environment data. The specific implementation method is as follows:

[0369] 1. Machine learning model unit

[0370] The machine learning model unit is used to perform classification and regression analysis on the fire environment data. The specific implementation includes the following steps:

[0371] Classification analysis:

[0372] The machine learning model unit uses the support vector machine (SVM) algorithm to classify the fire environment data. The support vector machine divides the data into different categories by constructing a hyperplane, which is suitable for the classification task of the fire risk level. The support vector machine uses the radial basis function (RBF) as the kernel function and optimizes the classification effect by adjusting the penalty parameter C and the kernel function parameter γ.

[0373] The machine learning model unit classifies fire environment data using the Random Forest algorithm. Random Forest improves the accuracy and robustness of classification by constructing multiple decision trees and performing ensemble learning. The parameter settings of Random Forest include the number of trees (n_estimators) and the maximum depth of each tree (max_depth).

[0374] Regression analysis:

[0375] The machine learning model unit performs regression analysis on fire environment data using the Linear Regression algorithm. Linear Regression predicts continuous values by fitting a linear equation and is suitable for predicting the fire spread speed. Linear Regression uses the least squares method for parameter estimation to minimize the prediction error.

[0376] The machine learning model unit is connected to the data storage module through a data interface, obtains fire environment data from the distributed database unit and the blockchain storage unit, and stores the analysis results back to the data storage module.

[0377] 2. Deep learning model unit

[0378] The deep learning model unit is used for anomaly detection and trend prediction of fire environment data. The specific implementation includes the following steps:

[0379] Anomaly detection:

[0380] The deep learning model unit uses the Long Short-Term Memory network (LSTM) to perform anomaly detection on fire environment data. The Long Short-Term Memory network can capture long-term dependencies in the data through memory units and gating mechanisms and is suitable for detecting abnormal fire behaviors. The network structure of the Long Short-Term Memory network includes an input layer, a hidden layer, and an output layer. The number of units in the hidden layer is set to 128, and the activation function uses ReLU (Rectified Linear Unit).

[0381] Trend prediction:

[0382] The deep learning model unit uses a Convolutional Neural Network (CNN) to perform trend prediction on fire environment data. The Convolutional Neural Network extracts the spatial features of the data through convolutional layers and pooling layers and is suitable for predicting the fire spread trend. The network structure of the Convolutional Neural Network includes convolutional layers, pooling layers, and fully connected layers. The number of filters in the convolutional layer is set to 64, and the pooling layer uses the Max Pooling operation.

[0383] The deep learning model unit is connected to the data storage module through a data interface, obtains fire environment data from the distributed database unit and the blockchain storage unit, and stores the analysis results back to the data storage module.

[0384] 3. Distributed Training Framework

[0385] Both the machine learning model unit and the deep learning model unit adopt the TensorFlow distributed training framework. TensorFlow distributes the training tasks to multiple computing nodes through data parallelism and model parallelism to improve the training efficiency. The specific implementation includes the following steps:

[0386] Data Parallelism:

[0387] TensorFlow splits the training data into multiple small batches (Mini-batch), each computing node is responsible for processing one small batch of data, and updates the model parameters through gradient aggregation.

[0388] Model Parallelism:

[0389] TensorFlow distributes different layers of the model to different computing nodes, each node is responsible for calculating the output of a part of the layers, and passes the intermediate results through communication between nodes.

[0390] TensorFlow is connected to the data storage module through a data interface, obtains training data from the distributed database unit and the blockchain storage unit, and stores the trained model back to the data storage module.

[0391] 4. Connection Method of the Data Analysis Module

[0392] The machine learning model unit and the deep learning model unit are connected to the data storage module through a data interface, and specifically use the REST API interface for data transmission. The REST API interface provides data reading and writing services through the HTTP protocol to ensure efficient data transmission and access.

[0393] The machine learning model unit and the deep learning model unit are connected to the computing nodes through the TensorFlow distributed training framework, and specifically use the gRPC framework to implement communication between nodes. gRPC ensures the efficient execution of training tasks through efficient binary encoding and streaming transmission.

[0394] Technical Effects:

[0395] The machine learning model unit realizes the classification and regression analysis of fire environment data through support vector machine, random forest and linear regression algorithms, and improves the prediction accuracy of fire risk level and spread speed.

[0396] The deep learning model unit realizes the anomaly detection and trend prediction of fire environment data through long short-term memory networks and convolutional neural networks, improving the detection ability of fire abnormal behaviors and spreading trends.

[0397] The data analysis module realizes the efficient training and deployment of the model through the TensorFlow distributed training framework, improving the real-time performance and reliability of the system.

[0398] S205: Generate a warning message based on the output result after the multi-dimensional analysis and execute a decision.

[0399] Specifically, the warning generation unit generates a warning message according to the multi-dimensional analysis result, and the decision execution unit controls the startup or shutdown of fire-fighting equipment through the Modbus protocol according to the warning message.

[0400] This step specifically includes that the warning generation unit generates a warning message according to the multi-dimensional analysis result, and the decision execution unit controls the startup or shutdown of fire-fighting equipment through the Modbus protocol according to the warning message. The specific implementation method is as follows:

[0401] 1. Warning generation unit

[0402] The warning generation unit is used to generate a warning message according to the multi-dimensional analysis result of the data analysis module. The specific implementation includes the following steps:

[0403] Rule-based warning generation algorithm:

[0404] The warning generation unit uses IF-THEN rules to generate warning messages. The IF-THEN rules generate corresponding warning messages through conditional judgment and logical reasoning. The specific rules are as follows:

[0405] Rule 1: If the temperature value detected by the temperature sensor is greater than 60 degrees Celsius, then generate a "high temperature warning" message.

[0406] Rule 2: If the smoke concentration detected by the smoke sensor is greater than 50 ppm, then generate a "smoke warning" message.

[0407] Rule 3: If the concentration of harmful gases detected by the gas sensor is greater than 20 ppm, then generate a "harmful gas warning" message.

[0408] Rule 4: If the flame area detected by the image sensor is greater than 10 square centimeters, then generate a "flame warning" message.

[0409] Warning message generation:

[0410] The early warning generation unit generates early warning information according to the IF-THEN rules and stores the early warning information in the distributed database unit. The early warning information includes the early warning type, early warning level, early warning time, and early warning location.

[0411] The early warning generation unit is connected to the data analysis module through a data interface, obtains the multi-dimensional analysis results from the data analysis module, and generates early warning information according to the analysis results.

[0412] 2. Decision Execution Unit

[0413] The decision execution unit is used to execute the control instructions of the fire-fighting equipment according to the early warning information. The specific implementation includes the following steps:

[0414] Control Instruction Generation:

[0415] The decision execution unit generates control instructions according to the early warning information. The specific instructions are as follows:

[0416] Instruction 1: If the "high temperature warning" information is generated, then generate the "start sprinkler system" instruction.

[0417] Instruction 2: If the "smoke warning" information is generated, then generate the "start smoke exhaust system" instruction.

[0418] Instruction 3: If the "hazardous gas warning" information is generated, then generate the "start ventilation system" instruction.

[0419] Instruction 4: If the "flame warning" information is generated, then generate the "start fire extinguishing system" instruction.

[0420] Control Instruction Execution:

[0421] The decision execution unit sends the control instructions to the fire-fighting equipment through the Modbus protocol. The Modbus protocol is connected to the fire-fighting equipment through the RS-485 serial communication interface to ensure the accurate transmission and execution of the instructions. The specific implementation is as follows:

[0422] Sprinkler System: Send a start instruction through the Modbus protocol to control the solenoid valve of the sprinkler system to open and start the sprinkler.

[0423] Smoke Exhaust System: Send a start instruction through the Modbus protocol to control the fan of the smoke exhaust system to start and conduct smoke exhaust.

[0424] Ventilation System: Send a start instruction through the Modbus protocol to control the fan of the ventilation system to start and conduct ventilation.

[0425] Fire Extinguishing System: Send a start instruction through the Modbus protocol to control the release of the fire extinguishing agent of the fire extinguishing system to conduct fire extinguishing.

[0426] The decision execution unit is connected to the warning generation unit through a data interface, obtains warning information from the warning generation unit, and generates and executes control instructions based on the warning information.

[0427] 3. Connection method of the warning decision module

[0428] The warning generation unit is connected to the data analysis module through a data interface, and specifically uses the REST API interface for data transmission. The REST API interface provides data reading and writing services through the HTTP protocol to ensure the efficient generation and storage of warning information.

[0429] The decision execution unit is connected to the fire-fighting equipment through the Modbus protocol, and specifically uses the RS-485 serial communication interface for instruction transmission. The RS-485 interface transmits through differential signals to ensure the accuracy and reliability of control instructions.

[0430] 4. Specific technical means of the warning decision module

[0431] Warning generation: Use IF-THEN rules to generate warning information, and the specific rules include high-temperature warning, smoke warning, harmful gas warning, and flame warning.

[0432] Decision execution: Use the Modbus protocol to execute control instructions, and the specific instructions include starting the sprinkler system, starting the smoke exhaust system, starting the ventilation system, and starting the fire extinguishing system.

[0433] Technical effects:

[0434] The warning generation unit generates warning information through IF-THEN rules to ensure the accuracy and timeliness of warning information.

[0435] The decision execution unit executes control instructions through the Modbus protocol to ensure the quick response and effective control of fire-fighting equipment.

[0436] Through the collaborative work of warning generation and decision execution, the warning decision module realizes the automation of fire warning and emergency handling, and improves the real-time performance and reliability of the system.

[0437] S206: Display warning information and decision results to the user.

[0438] Specifically, the visualization interface unit displays warning information and decision results in the form of charts, and the mobile terminal unit displays warning information and decision results to the user through a mobile application.

[0439] This step specifically includes that the visualization interface unit displays warning information and decision results in the form of charts, and the mobile terminal unit displays warning information and decision results to the user through a mobile application. The specific implementation method is as follows:

[0440] 1. Visualization Interface Unit

[0441] The visualization interface unit is used to display warning information and decision results in the form of charts. The specific implementation includes the following steps:

[0442] Chart Generation:

[0443] The visualization interface unit uses the ECharts library to generate bar charts, line charts, and pie charts. ECharts is an open-source visualization library based on JavaScript, suitable for generating interactive charts. The specific implementation is as follows:

[0444] Bar Chart: Used to display the number of fire warnings in different time periods. The X-axis of the bar chart represents time, the Y-axis represents the number of warnings, and each bar represents the number of warnings in a time period.

[0445] Line Chart: Used to display the fire spread trend. The X-axis of the line chart represents time, the Y-axis represents the fire spread speed, and the line represents the changing trend of the fire spread.

[0446] Pie Chart: Used to display the proportion of different warning types. Each sector area of the pie chart represents a warning type, and the size of the area represents the proportion of that type of warning.

[0447] Data Display:

[0448] The visualization interface unit obtains warning information and decision results from the warning decision module through the REST API interface and passes the data to the ECharts library to generate charts. The charts are embedded in the web page through HTML5 and CSS3 technologies, and users can access the visualization interface through a browser.

[0449] The visualization interface unit is connected to the warning decision module through the REST API interface, and specifically uses the HTTP protocol for data transmission. The REST API interface obtains warning information and decision results through a GET request to ensure the efficient transmission and display of data.

[0450] 2. Mobile Terminal Unit

[0451] The mobile terminal unit is connected to the warning decision module through a wireless communication network (such as 4G or 5G), and is used to receive and display warning information and decision results to users. The specific implementation includes the following steps:

[0452] Mobile Application Development:

[0453] The mobile terminal unit realizes the display of early warning information and decision-making results through a mobile application (such as an Android or iOS application). The mobile application is developed using the React Native framework, which is an open-source framework based on JavaScript and is suitable for developing cross-platform mobile applications.

[0454] Data reception and display:

[0455] The mobile application obtains early warning information and decision-making results from the early warning decision-making module through the REST API interface and displays the data on the user interface. The specific display methods are as follows:

[0456] Display of early warning information: The early warning information is displayed in a list form, and each piece of early warning information includes the early warning type, early warning level, early warning time, and early warning location.

[0457] Display of decision-making results: The decision-making results are displayed in the form of notifications, and each notification includes the decision-making type, execution time, and execution status.

[0458] User interaction:

[0459] The mobile application provides user interaction functions. Users can view detailed information by clicking on the early warning information or view the execution status by clicking on the decision-making results. Users can also set the early warning threshold and the way of receiving notifications through the application.

[0460] The mobile terminal unit is connected to the early warning decision-making module through a wireless communication network (such as 4G or 5G), and the HTTP protocol is specifically used for data transmission. The REST API interface obtains early warning information and decision-making results through GET requests to ensure the efficient transmission and display of data.

[0461] 3. Connection method of the user interaction module

[0462] The visualization interface unit is connected to the early warning decision-making module through the REST API interface, and the HTTP protocol is specifically used for data transmission. The REST API interface obtains early warning information and decision-making results through GET requests to ensure the efficient transmission and display of data.

[0463] The mobile terminal unit is connected to the early warning decision-making module through a wireless communication network (such as 4G or 5G), and the HTTP protocol is specifically used for data transmission. The REST API interface obtains early warning information and decision-making results through GET requests to ensure the efficient transmission and display of data.

[0464] 4. Specific technical means of the user interaction module

[0465] Chart Generation: Use the ECharts library to generate bar charts, line charts, and pie charts. The specific charts include the bar chart of the number of early warnings, the line chart of the fire spread trend, and the pie chart of the proportion of early warning types.

[0466] Mobile Application: Develop Android and iOS applications using the React Native framework. The specific functions include the display of early warning information, the display of decision-making results, and user interaction.

[0467] Data Transmission: Use the REST API interface to transmit data via the HTTP protocol to ensure the efficient transmission and display of data.

[0468] Technical Effects:

[0469] The visualization interface unit generates charts through the ECharts library, realizing the visual display of early warning information and decision-making results, and improving the readability and understandability of data.

[0470] The mobile terminal unit develops mobile applications through the React Native framework, realizing the real-time display of early warning information and decision-making results and user interaction, and improving the convenience and user experience of the system.

[0471] The user interaction module realizes the multi-channel display of early warning information and decision-making results through the collaborative work of the visualization interface and the mobile terminal, enhancing the real-time performance and reliability of the system.

[0472] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices can be implemented as software, firmware, hardware, and their appropriate combinations.

[0473] After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily think of other embodiments of the present application. The present application is intended to cover any variations, uses, or adaptations of the present application, which follow the general principles of the present application and include the common general knowledge or conventional technical means in the technical field not disclosed in the present application. The specification and examples are only considered exemplary, and the true scope and spirit of the present application are pointed out by the following claims.

[0474] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present application is only limited by the appended claims.

Claims

1. An intelligent fire protection system, characterized in that, Comprising: A data acquisition module, the data acquisition module includes a plurality of distributed sensor units, and the distributed sensor units are used to collect fire environment data in real time; A data processing module, the data processing module includes an edge computing unit and a cloud computing unit, the edge computing unit and the cloud computing unit are respectively connected to the distributed sensor units through a wireless communication network, the edge computing unit and the cloud computing unit are connected through an encrypted communication protocol, and the edge computing unit and the cloud computing unit are used to receive and process the fire environment data; A data storage module, the data storage module includes a distributed database unit and a blockchain storage unit, the distributed database unit and the blockchain storage unit are respectively connected to the data processing module, and the distributed database unit and the blockchain storage unit are connected through a data synchronization protocol; the distributed database unit and the blockchain storage unit are used to store the processed fire environment data; A data analysis module, the data analysis module includes a machine learning model unit and a deep learning model unit, the machine learning model unit and the deep learning model unit are respectively connected to the data storage module, the machine learning model unit and the deep learning model unit are connected through a data interface, and the machine learning model unit and the deep learning model unit are used to perform multi-dimensional analysis on the stored fire environment data; An early warning decision-making module, the early warning decision-making module includes an early warning generation unit and a decision execution unit, the early warning generation unit and the decision execution unit are respectively connected to the data analysis module, the early warning generation unit and the decision execution unit are connected through an instruction transmission protocol, and the early warning generation unit and the decision execution unit are used to generate early warning information and execute decisions according to the output results after the multi-dimensional analysis; A user interaction module, the user interaction module includes a visualization interface unit and a mobile terminal unit, the visualization interface unit and the mobile terminal unit are respectively connected to the early warning decision-making module, the visualization interface unit and the mobile terminal unit are connected through a wireless communication network, and the visualization interface unit and the mobile terminal unit are used to display early warning information and decision results to users.

2. The intelligent fire protection system according to claim 1, characterized in that The distributed sensor units include temperature sensors, smoke sensors, gas sensors and image sensors, and the temperature sensors, smoke sensors, gas sensors and image sensors are connected to the data processing module through a wireless communication network.

3. The intelligent fire protection system according to claim 1, characterized in that The edge computing unit is used to perform real-time preprocessing on the fire environment data, and the preprocessing includes data cleaning, data compression and data encryption. The cloud computing unit is used to perform in-depth analysis and model training on the preprocessed fire environment data.

4. The intelligent fire protection system according to claim 1, characterized in that, The distributed database unit is used to store fire environment data with an access frequency greater than or equal to N times / second, and the blockchain storage unit is used to store fire environment data with an access frequency less than N times / second.

5. The intelligent fire protection system according to claim 1, characterized in that The machine learning model unit is used to classify and perform regression analysis on the fire environment data, and the deep learning model unit is used to detect anomalies and predict trends in the fire environment data.

6. The intelligent fire protection system according to claim 1, characterized in that The warning generation unit is used to generate warning information according to the output result after the multi-dimensional analysis, and the decision execution unit is used to execute the control instructions of the fire-fighting equipment according to the warning information.

7. The intelligent fire protection system according to claim 1, characterized in that, The visualization interface unit is used to display the warning information and decision results in the form of charts, and the mobile terminal unit is used to receive and display the warning information and decision results to the user.

8. The intelligent fire protection system according to claim 1, characterized in that, Both the machine learning model unit and the deep learning model unit adopt a distributed training framework, and the distributed training framework is connected to the data storage module through a data interface.

9. A method for optimizing smart fire data, the optimization method is applied to the fire-fighting system according to any one of claims 1-8, and the optimization method includes: Collect fire environment data in real time; Receive and process the fire environment data; Store the processed fire environment data; Perform multi-dimensional analysis on the stored fire environment data; Generate warning information and make decisions according to the output result after the multi-dimensional analysis; Display the warning information and decision results to the user.

10. The optimization method according to claim 9, characterized in that, The receiving and processing the fire environment data includes: Perform real-time preprocessing on the fire environment data, and the preprocessing includes data cleaning, data compression, and data encryption; Perform in-depth analysis and model training on the preprocessed fire environment data.

Images