Fire monitoring method and system combining lora wireless transmission and neural network algorithm

Abstract

The embodiment of the present application discloses a fire monitoring method and system combining Lora wireless transmission and neural network algorithm, which belongs to the field of fire Internet of Things technology. The fire monitoring method combining Lora wireless transmission and neural network algorithm proposed in the embodiment of the present application realizes comprehensive and efficient monitoring of fire terminal equipment through the intelligent execution of the fire monitoring system. This method not only improves the accuracy and timeliness of data collection, but also realizes accurate identification and timely warning of fire hazards through the powerful analysis ability of the neural network algorithm. At the same time, by comprehensively considering the influence of multiple fire trend elements and dynamic adjustment strategies, it ensures that the fire terminal equipment can make the most appropriate response in different situations, thereby minimizing the losses caused by fire.

Description

Technical Field

[0001] The embodiments of the present application belong to the field of fire protection Internet of Things technology, and specifically relate to a fire protection monitoring method and system that combines LoRa wireless transmission and a neural network algorithm. Background Art

[0002] With the acceleration of modern urbanization, fire safety issues have become increasingly prominent. Traditional fire monitoring technology often seems to be unable to cope with complex and ever-changing fire hazards. Traditional fire monitoring methods mostly rely on wired transmission and manual inspections, which are not only inefficient, but also difficult to achieve full coverage and real-time monitoring. Especially in some remote areas or large buildings, wired transmission has high wiring costs and difficult maintenance, while manual inspections have blind spots and time delays, making it difficult to detect and deal with potential fire hazards in a timely manner.

[0003] In recent years, with the rapid development of Internet of Things technology, the application of wireless transmission technology in the field of fire monitoring has gradually attracted attention. Among them, LoRa (Low-Power Wide-Area Network), as an emerging wireless communication technology, has shown great application potential in fire monitoring with its advantages of low power consumption, long-distance transmission and low cost. LoRa technology can realize remote and real-time data transmission of fire terminal equipment, greatly improve the timeliness and accuracy of data collection, and provide strong support for the intelligentization of fire monitoring.

[0004] However, relying solely on LoRa wireless transmission technology is not enough to completely solve the complex problems in fire monitoring. In practical applications, it also involves the technical problem of processing a large amount of data and accurately identifying potential fire hazards. Summary of the invention

[0005] The embodiments of the present application provide a fire monitoring method and system combining lora wireless transmission and neural network algorithm, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0006] The embodiment of the present application provides a fire monitoring method combining lora wireless transmission and neural network algorithm, the method is executed by a fire monitoring system, the fire monitoring system is configured with a fire IoT sensor monitoring algorithm, the fire IoT sensor monitoring algorithm includes a sensor information analysis network corresponding to each target fire trend element; the sensor information analysis network corresponding to the target fire trend element is used to output a local fire hazard discrimination result of a fire terminal device, the local fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal device under the target fire trend element belongs to a set state label; the method includes: obtaining the lora sensor corresponding to the fire terminal device to be monitored Transmit interactive information, the lora transmission interactive information includes the quantitative description value corresponding to the target fire trend element; for each target fire trend element of the fire terminal equipment to be monitored, through the sensor information analysis network corresponding to the target fire trend element, according to the quantitative description value of the target fire trend element, determine the local fire hazard discrimination result of the fire terminal equipment to be monitored; based on the local fire hazard discrimination result determined by the sensor information analysis network corresponding to each target fire trend element, determine the target fire hazard discrimination result of the fire terminal equipment to be monitored; according to the target fire hazard discrimination result, determine the label corresponding to the monitoring action response state of the fire terminal equipment to be monitored.

[0007] An embodiment of the present application provides a fire monitoring system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0008] An embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.

[0009] The fire monitoring method combining lora wireless transmission and neural network algorithm proposed in the embodiment of the present application realizes comprehensive and efficient monitoring of fire terminal equipment through the intelligent execution of the fire monitoring system. This method not only improves the accuracy and timeliness of data collection, but also realizes accurate identification and timely warning of fire hazards through the powerful analysis ability of the neural network algorithm. At the same time, by comprehensively considering the influence of multiple fire trend elements and dynamic adjustment strategies, it ensures that the fire terminal equipment can make the most appropriate response in different situations, thereby minimizing the losses caused by fire. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flowchart of a fire monitoring method combining lora wireless transmission and a neural network algorithm provided in an embodiment of the present application.

[0011] Figure 2 A schematic diagram of the structure of a fire monitoring system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of them. Based on the embodiments in the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.

[0013] The terms "first", "second", etc. in the embodiments of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. can be a class, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the embodiments of the present application, "and / or" represents at least one of the connected objects, and the character " / " generally represents that the objects associated before and after are in an "or" relationship.

[0014] Figure 1 A fire monitoring method combining lora wireless transmission and neural network algorithm is shown. The method is executed by a fire monitoring system. The fire monitoring system is configured with a fire IoT sensor monitoring algorithm. The fire IoT sensor monitoring algorithm includes a sensor information analysis network corresponding to each target fire trend element; the sensor information analysis network corresponding to the target fire trend element is used to output the local fire hazard discrimination result of the fire terminal equipment. The local fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal equipment under the target fire trend element belongs to the set state label. Based on this, the method includes the following steps 110-140.

[0015] Step 110: The fire monitoring system obtains the LoRa transmission interaction information corresponding to the fire terminal equipment to be monitored, and the LoRa transmission interaction information includes the quantitative description value corresponding to the target fire trend element.

[0016] The primary task of the fire monitoring system is to collect real-time data from various fire terminal devices, which are the basis for subsequent analysis and decision-making. In the embodiment of the present application, the fire monitoring system uses LoRa wireless transmission technology to obtain key interactive information from the fire terminal devices to be monitored. LoRa technology is known for its low power consumption, long-distance transmission and strong penetration ability, and is particularly suitable for scenarios such as fire monitoring that require wide coverage and have high requirements for battery life.

[0017] The acquired LoRa transmission interaction information contains quantitative description values ​​directly related to the target fire trend elements. These target fire trend elements may include temperature, smoke concentration, light intensity, gas composition and other parameters, which together constitute the key indicators for evaluating fire safety hazards. The stable transmission of LoRa technology ensures the accuracy and timeliness of these data, providing a reliable basis for subsequent hazard identification.

[0018] Step 120: The fire monitoring system determines the local fire hazard identification result of the fire terminal equipment to be monitored based on the quantitative description value of each target fire trend element of the fire terminal equipment to be monitored through the sensor information analysis network corresponding to the target fire trend element.

[0019] After collecting the LoRa transmission interaction information of the fire terminal equipment to be monitored, the fire monitoring system enters the data analysis stage. The key to this step is to use the pre-built sensor information analysis network to process each target fire trend element. The sensor information analysis network is designed based on the neural network algorithm, which can intelligently analyze and output the judgment results of local fire hazards based on the input quantitative description value.

[0020] The design of the sensor information analysis network fully considers the complexity and diversity of fire development. Through advanced technologies such as deep learning, it realizes the mining and learning of a large amount of historical data, thus forming the ability to accurately identify different fire trend elements. The local fire hazard identification result output by each network is actually a numerical value that represents the possibility that the fire terminal equipment monitoring action response status label belongs to the set status label under a specific fire trend element. This quantitative representation method provides a scientific basis for subsequent comprehensive judgment.

[0021] Step 130: The fire monitoring system determines the target fire hazard identification result of the fire terminal equipment to be monitored based on the local fire hazard identification result determined by the sensor information analysis network corresponding to each target fire trend element.

[0022] After obtaining the local fire hazard identification results corresponding to each target fire trend element, the fire monitoring system needs to further integrate this information to form a comprehensive assessment of the overall fire hazard. In this step, the system will conduct a comprehensive analysis of all local identification results based on the preset weights and rules, and finally determine the target fire hazard identification results of the fire terminal equipment to be monitored.

[0023] This process reflects the logical thinking from the local to the whole. By comprehensively considering the influence of multiple fire trend factors, the fire monitoring system can more accurately grasp the potential risk of fire. At the same time, this comprehensive analysis method also improves the system's ability to identify different types of fires, whether it is a fast-spreading open fire or a slowly accumulated smoldering fire, it can be monitored in a timely and effective manner.

[0024] Step 140: The fire monitoring system determines the label corresponding to the monitoring action response state of the fire terminal equipment to be monitored according to the target fire hazard identification result.

[0025] Finally, the fire monitoring system assigns corresponding labels to the monitoring action response status of the fire terminal equipment to be monitored based on the target fire hazard identification results. These labels represent the specific actions that the equipment should take under the current level of fire safety hazards, such as normal monitoring, early warning prompts, emergency alarms, etc. Through this step, the fire monitoring system realizes a complete closed loop from data collection, analysis to decision output, providing direct and effective guidance for fire safety management.

[0026] It can be understood that the determination of the label is not only based on the current hidden danger identification results, but also takes into account historical data, environmental factors and possible fire development trends. This dynamic adjustment strategy ensures that the fire terminal equipment can make the most appropriate response in different situations, thereby minimizing the losses caused by fire.

[0027] To facilitate understanding of the above technical solution, a complete application scenario example is used below for further introduction.

[0028] The primary task of the fire monitoring system is to collect accurate data from various fire terminal devices in real time. These data are the basis for subsequent analysis and decision-making, and are crucial to ensuring fire safety. In the embodiment of the present application, the fire monitoring system uses LoRa wireless transmission technology to obtain key interactive information from the fire terminal equipment to be monitored. LoRa technology is known for its low power consumption, long-distance transmission and strong penetration ability, and is particularly suitable for scenarios such as fire monitoring that require wide coverage and have high requirements for battery life. The LoRa module can transmit data stably, ensuring the accuracy and timeliness of the data even in complex urban environments. The obtained LoRa transmission interaction information contains quantitative description values ​​directly related to the target fire trend elements. These target fire trend elements may include temperature, smoke concentration, light intensity, gas composition (such as carbon monoxide, carbon dioxide, etc.) and other parameters. For example, a temperature sensor may provide real-time ambient temperature data, such as 25°C, 30°C, etc.; a smoke sensor may provide a quantitative value of smoke concentration, such as 0.1mg / m³, 0.5mg / m³, etc. These parameters together constitute the key indicators for assessing fire safety hazards.

[0029] After collecting the LoRa transmission interaction information of the fire terminal equipment to be monitored, the fire monitoring system enters the data analysis stage. The key to this step is: for each target fire trend element, use the pre-built sensor information analysis network for processing. The sensor information analysis network is designed based on the neural network algorithm, which can intelligently analyze and output the discrimination results of local fire hazards according to the input quantitative description value. These networks include input layer, hidden layer and output layer. The input layer receives the quantitative description value transmitted from LoRa, the hidden layer extracts useful features through complex calculations and processing, and the output layer gives the discrimination results of local fire hazards. Taking the temperature sensor information analysis network as an example, the network is set to adopt an exemplary three-layer structure: the input layer contains a neuron, corresponding to the quantitative description value of the temperature sensor; the hidden layer contains multiple neurons for extracting temperature features; the output layer contains a neuron, which outputs the discrimination results of local fire hazards, such as a value between 0 and 1, indicating the possibility of fire at the current temperature. In practical applications, the design of the sensor information analysis network will be more complex, and advanced technologies such as deep learning, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), may be used to mine and learn a large amount of historical data, thereby forming the ability to accurately identify different fire trend elements. The local fire hazard identification result output by each network is actually a numerical value that represents the possibility that the fire terminal equipment monitoring action response state label belongs to the set state label under a specific fire trend element. For example, for the temperature sensor information analysis network, an output value of 0.8 may indicate that at the current temperature, the possibility of a fire is high, and the fire terminal equipment should enter the early warning state.

[0030] After obtaining the local fire hazard discrimination results corresponding to each target fire trend element, the fire monitoring system needs to further integrate this information to form a comprehensive assessment of the overall fire hazard. In this step, the system will conduct a comprehensive analysis of all local discrimination results based on the preset weights and rules. The setting of weights and rules is based on in-depth analysis of historical fire data and expert experience. For example, temperature may be regarded as the most important fire trend element, so its corresponding local discrimination result may be given the highest weight; while light intensity may be regarded as a secondary fire trend element, and its weight may be relatively low. The comprehensive analysis process may adopt statistical methods such as weighted summation and logistic regression. Taking weighted summation as an example, the system will multiply each local discrimination result by its corresponding weight, and then add all the results to obtain the final target fire hazard discrimination result. This result is a value between 0 and 1, indicating the possibility that the fire terminal equipment to be monitored currently has a fire hazard. By comprehensively considering the influence of multiple fire trend elements, the fire monitoring system can more accurately grasp the potential risk of fire. At the same time, this comprehensive analysis method also improves the system's ability to identify different types of fires. Whether it is a fast-spreading open flame or a slowly accumulating smoldering flame, it can be monitored promptly and effectively.

[0031] Finally, the fire monitoring system assigns corresponding labels to the monitoring action response status of the fire terminal equipment to be monitored according to the target fire hazard identification results. These labels represent the specific actions that the equipment should take under the current level of fire safety hazards. The setting of labels is based on a deep understanding of fire safety management needs. For example, when the target fire hazard identification result is lower than a certain threshold, the system may set the label to "normal monitoring", indicating that the current fire safety situation is good and no special measures are required; when the identification result is higher than a certain threshold, the system may set the label to "early warning prompt", indicating that there are potential fire safety hazards and that monitoring needs to be strengthened; when the identification result is extremely high, the system may set the label to "emergency alarm", indicating that a fire has occurred or is about to occur and that immediate countermeasures need to be taken. The determination of labels is not only based on the current hazard identification results, but also takes into account historical data, environmental factors, and possible fire development trends. For example, if a certain area has experienced many fires in history, or the current environmental humidity is low and the wind speed is high, the system may set labels more sensitively to ensure fire safety. Through this step, the fire monitoring system realizes a complete closed loop from data collection, analysis to decision output. This provides direct and effective guidance for fire safety management and helps reduce the occurrence of fire accidents.

[0032] In order to more specifically illustrate the method of the embodiment of the present application, it is assumed that the fire terminal equipment to be monitored includes a temperature sensor, a smoke sensor, and a light sensor. At a certain moment, the quantitative description value provided by the temperature sensor is 35°C (the normal range is 20°C-30°C), the quantitative description value provided by the smoke sensor is 0.3mg / m³ (the normal range is 0mg / m³-0.1mg / m³), and the quantitative description value provided by the light sensor is 100lx (the normal range is 50lx-200lx).

[0033] According to the preset weights and rules (temperature weight is set to 0.6, smoke weight is set to 0.3, and light weight is set to 0.1), the system calculates the target fire hazard identification results as follows:

[0034] Target fire hazard identification result = (35-30) × 0.6 + (0.3-0.1) × 0.3 + (100-50) × 0.1 × (-0.1) = 3.0 + 0.06-0.5 = 2.56 (Note: The light intensity here is within the normal range, so it has a negative impact on the hazard identification result)

[0035] Since 2.56 is higher than the preset warning threshold (such as 2.0), the system sets the monitoring action response status label of the fire terminal equipment to be monitored to "warning prompt".

[0036] In addition, taking the temperature sensor information analysis network as an example, an exemplary three-layer neural network structure is set up, the input layer contains one neuron (corresponding to the quantitative description value of the temperature sensor), the hidden layer contains three neurons (used to extract temperature features), and the output layer contains one neuron (outputs the judgment result of local fire hazards). During the network training process, historical temperature data and corresponding fire occurrence conditions are used as training samples. Through the back propagation algorithm and gradient descent optimization method, the network weights and biases are continuously adjusted so that the network can accurately output the possibility of fire at a given temperature. In practical applications, when the temperature sensor provides a new quantitative description value, the network can quickly output the corresponding local fire hazard judgment result, providing a basis for subsequent comprehensive analysis.

[0037] In summary, the fire monitoring method combining lora wireless transmission and neural network algorithm proposed in the embodiment of the present application realizes comprehensive and efficient monitoring of fire terminal equipment through the intelligent execution of the fire monitoring system. This method not only improves the accuracy and timeliness of data collection, but also realizes accurate identification and timely warning of fire hazards through the powerful analysis ability of the neural network algorithm. At the same time, by comprehensively considering the influence of multiple fire trend elements and dynamic adjustment strategies, it ensures that the fire terminal equipment can make the most appropriate response in different situations, thereby minimizing the losses caused by fire.

[0038] In some optional embodiments, the sensor information analysis network corresponding to the target fire trend element determines the local fire hazard identification result of the fire terminal equipment to be monitored according to the quantitative description value of the target fire trend element, including: determining the target quantitative description range to which the quantitative description value of the target fire trend element belongs from each quantitative description range of the target fire trend element through the sensor information analysis network corresponding to the target fire trend element; and determining the local fire hazard identification result of the fire terminal equipment to be monitored according to the confidence evaluation degree corresponding to the target quantitative description range.

[0039] It can be understood that the process of determining the local fire hazard identification results of the fire terminal equipment to be monitored through the sensor information analysis network corresponding to the target fire trend elements in the fire monitoring system is a process that is more intuitively demonstrated in combination with specific examples below.

[0040] First, define the target fire trend factors, which may include temperature, humidity, smoke concentration, light intensity, etc. Taking temperature as an example, three quantitative description ranges are set: normal (20℃-30℃), warning (30℃-40℃), and dangerous (>40℃). These ranges are set based on a comprehensive consideration of historical data, fire safety standards, and environmental factors.

[0041] Next, a sensor information analysis network is constructed, which consists of multiple sensor nodes, data processing units and intelligent analysis algorithms. The sensor nodes are responsible for collecting the values ​​of the target fire trend elements in real time and transmitting these data to the data processing unit. The data processing unit pre-processes these data, such as denoising and normalization, for subsequent intelligent analysis.

[0042] The intelligent analysis algorithm is the core of the sensor information analysis network. It uses neural networks or machine learning algorithms, such as support vector machines (SVM), decision trees, random forests, etc., to conduct in-depth analysis of pre-processed data. Taking temperature as an example, when the temperature value collected by a sensor node is 35°C, the intelligent analysis algorithm will first determine which quantitative description range this value belongs to. Obviously, 35°C belongs to the warning range.

[0043] However, simply judging the range to which the temperature value belongs is not enough to determine the existence of fire hazards. It is also necessary to make a more accurate judgment based on the confidence rating corresponding to the target quantitative description range. The calculation of the confidence rating can be based on multiple factors, such as the matching degree of historical data, the stability of current data, and the changing trend of environmental factors.

[0044] For the purpose of specific explanation, a confidence evaluation calculation formula is introduced:

[0045] Confidence evaluation degree = (matching degree between current data and historical data) * (stability weight of current data) * (weight of changing trend of environmental factors);

[0046] Among them, the matching degree between current data and historical data is calculated by comparing the difference between the current temperature value and the historical temperature value of the same period; the stability weight of the current data is determined based on the fluctuation of the temperature value over a period of time; the weight of the changing trend of environmental factors considers the influence of factors such as wind speed, wind direction, and weather conditions on the speed of fire spread.

[0047] For example, at a certain moment, the temperature sensor reading is 35°C, and the matching degree of this value with the historical data is 0.8, the stability weight of the current data is 0.9, and the weight of the environmental factor change trend is 0.7. According to the calculation formula of the confidence evaluation degree, we can get:

[0048] Confidence rating = 0.8*0.9*0.7 = 0.504;

[0049] Since the confidence evaluation is high (close to 0.5, for example, the threshold is 0.5), it can be judged that the warning range to which the current temperature value of 35°C belongs is credible, that is, there is a fire hazard. At this time, the sensor information analysis network will output a comprehensive hazard identification result, including the existence of the hazard, the severity of the hazard (such as the warning level), the possible location of the hazard, and the recommended response measures.

[0050] In addition to temperature, the sensor information analysis network also performs similar analysis and identification on other fire trend factors. For example, for smoke density, different quantitative description ranges and confidence evaluation calculation formulas can be set to achieve accurate monitoring of smoke density and hidden danger identification.

[0051] In addition, the sensor information analysis network also supports the function of dynamically adjusting the quantitative description range and confidence evaluation. In practical applications, it may be found that some quantitative description ranges or confidence evaluation calculation formulas are not completely applicable to the current monitoring environment or fire trend. In this case, the quantitative description range and confidence evaluation can be fine-tuned based on the real-time feedback data and results to improve the accuracy and reliability of hidden danger identification.

[0052] Finally, it should be emphasized that the sensor information analysis network is only a component of the fire monitoring system. In practical applications, the sensor information analysis network needs to be closely coordinated with other fire monitoring equipment, alarm devices, fire fighting equipment, etc. to form a complete and efficient fire monitoring system. Through this system, comprehensive perception, intelligent analysis and rapid response of fire trends can be achieved, providing a strong guarantee for fire safety.

[0053] It can be seen that based on the above content, it is demonstrated how to determine the local fire hazard identification results of the fire terminal equipment to be monitored through the sensor information analysis network corresponding to the target fire trend elements. This process not only relies on advanced sensor technology and data processing capabilities, but also requires intelligent analysis algorithms and dynamic adjustment mechanisms to ensure the accuracy and reliability of the identification.

[0054] In some other optional embodiments, after determining the label corresponding to the monitoring action response state of the fire terminal equipment to be monitored based on the target fire hazard identification result, it also includes: if the label corresponding to the monitoring action response state of the fire terminal equipment to be monitored belongs to the target state label, determining the target early warning situation keyword of the fire terminal equipment to be monitored based on the target fire hazard identification result; and performing a fire emergency response to the monitoring action response state of the fire terminal equipment to be monitored based on the fire emergency response strategy corresponding to the target early warning situation keyword.

[0055] In detail, for the subsequent processing flow of determining the corresponding label of the monitoring action response state of the fire terminal equipment to be monitored based on the target fire hazard identification result in the fire monitoring system, this embodiment further refines and expands the relevant functions, aiming to achieve a more accurate and efficient fire emergency response. This process not only involves the detailed classification and identification of the monitoring action response state, but also covers the early warning situation analysis based on specific labels and the corresponding emergency response strategy execution, thereby ensuring the intelligence and automation level of the fire monitoring system.

[0056] First, when the monitoring action response status of the fire terminal equipment to be monitored is assigned a specific label, the label becomes a bridge between hidden danger identification and subsequent emergency response. Among these labels, some are defined as "target status labels", which represent fire hazard states that require immediate action or special attention. The setting of these target status labels is based on the analysis of historical fire cases, the requirements of fire safety regulations, and the needs of actual monitoring scenarios. They can accurately reflect the urgency and severity of current fire safety.

[0057] Once the tag corresponding to the monitoring action response state of the fire terminal device to be monitored is identified as the target state tag, the system will immediately start a series of subsequent processing processes. The first task is to determine the target warning situation keywords of the terminal device based on the target fire hazard identification results. This step is to extract information related to the current fire trend, hazard type, possible impact range, etc. through in-depth analysis of the hazard identification results, and then condense them into representative and directional keywords. These keywords can not only help fire management personnel quickly understand the core issues of the current fire hazards, but also provide an important basis for the selection of subsequent emergency response strategies.

[0058] After determining the target early warning situation keywords, the system will respond to the monitoring action response status of the monitored fire terminal equipment according to the fire emergency response strategies corresponding to these keywords. This process is an important manifestation of the intelligence of the fire monitoring system. It requires the system to automatically match and execute the corresponding emergency response strategies according to different types of early warning situations, so as to achieve rapid and accurate handling of fire situations.

[0059] Specifically, the fire emergency response strategy may include a variety of measures, such as activating the alarm device, adjusting the operating status of the fire-fighting equipment, notifying relevant personnel to go to the site for inspection, etc. For different types of early warning situations, the system will select the most appropriate response measures for execution based on the preset strategy library. For example, when the target early warning situation keyword is "high temperature warning", the system may automatically start the sprinkler system to reduce the temperature on site, and at the same time trigger the alarm device to notify relevant personnel; when the keyword is "smoke concentration exceeds the standard", the system may turn on the smoke exhaust fan to accelerate the exhaust of smoke and also trigger the alarm.

[0060] During the fire emergency response process, the system will also continuously monitor the status changes of the fire terminal equipment to be monitored, and adjust and optimize the response strategy based on the real-time feedback data. This dynamic adjustment mechanism is the key to ensuring the effectiveness and adaptability of emergency response. It can flexibly adjust the response measures according to the real-time changes in the fire trend, thereby minimizing the losses caused by the fire.

[0061] In addition, in order to improve the accuracy and efficiency of fire emergency response, the system also supports historical data analysis and pattern recognition related to target warning situation keywords. By learning from historical fire cases and hidden danger identification results, the system can gradually accumulate rich knowledge and experience, and then when faced with new warning situations, it can more accurately judge their types and severity and choose the most appropriate emergency response strategy.

[0062] It is worth noting that although the automated execution of fire emergency response strategies can significantly improve response speed and efficiency, the human factor is still indispensable in fire monitoring. Therefore, when designing the system, the needs of human-computer interaction are fully considered, providing fire management personnel with a convenient monitoring interface and operation tools so that they can manually intervene in the emergency response process when necessary to ensure the comprehensive control of fire safety.

[0063] In addition to the above emergency response measures, this embodiment also emphasizes the importance of prevention and early warning. By continuously monitoring the status changes of the fire terminal equipment to be monitored and conducting a comprehensive analysis based on historical data and environmental factors, the system can detect potential fire hazards in advance and issue early warning signals. This early warning mechanism can not only provide fire management personnel with a valuable time window so that they can take preventive measures to avoid fires, but also improve the initiative and predictability of the entire fire monitoring system.

[0064] To achieve this goal, the system uses advanced sensing technology and data processing algorithms to ensure accurate perception and prediction of fire trends. At the same time, the system also supports seamless connection with other components of the fire monitoring system, such as sensor networks, data transmission systems, alarm devices, etc., to form a complete and efficient fire monitoring system. Through this system, comprehensive perception, intelligent analysis and rapid response to fire hazards can be achieved, providing a strong guarantee for fire safety.

[0065] In the actual application of the system, we also focus on in-depth understanding and satisfaction of user needs. Through full communication and exchanges with users, we understand their pain points and needs in fire monitoring, and then optimize and improve the system in a targeted manner. For example, we provide flexible quantitative description range settings and confidence evaluation calculation methods so that users can customize configurations according to their own scenarios and needs; we also provide a variety of alarm methods and response strategy options so that users can flexibly adjust according to actual conditions.

[0066] It can be seen that through the implementation of this embodiment, accurate identification, efficient disposal and effective prevention of fire hazards can be achieved, providing strong technical support for fire safety.

[0067] In the embodiment of the present application, the fire-fighting IoT sensor monitoring algorithm is one of the core invention points, and its debugging and training process is crucial. Based on this, in an innovative design idea, the debugging steps of the fire-fighting IoT sensor monitoring algorithm include the following contents:

[0068] (1) Obtaining LoRa transmission interaction information corresponding to at least one reference fire terminal device and an authentication tag of the monitoring action response status of the reference fire terminal device, wherein the LoRa transmission interaction information includes a quantitative description value corresponding to at least one reference fire trend element;

[0069] (2) determining the proportion of fire terminal devices with set status labels corresponding to each quantitative description range of the reference fire trend element according to the quantitative description value of the reference fire terminal device under the reference fire trend element, so as to obtain the confidence evaluation degree corresponding to each quantitative description range;

[0070] (3) selecting at least one target fire trend element from the reference fire trend elements based on the confidence evaluation corresponding to each quantitative description range of the reference fire trend element;

[0071] (4) performing a relationship statistical evaluation on each target fire trend element based on the authentication tag of the monitoring action response status of the reference fire terminal equipment to obtain an attention value corresponding to each target fire trend element;

[0072] (5) Based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, generate the fire IoT sensor monitoring algorithm for the fire terminal equipment.

[0073] In the field of fire safety, efficient and accurate fire monitoring is the key to preventing and reducing fire losses. In this context, the performance optimization and debugging of the fire IoT sensor monitoring algorithm as a core component of the system is particularly important. Based on this, the embodiment of the present application proposes an innovative debugging and training method for the fire IoT sensor monitoring algorithm, which aims to improve the recognition accuracy and response speed of the algorithm in complex fire environments through refined step design.

[0074] First, in order to build a reliable training foundation, it is necessary to obtain the LoRa (Long Range) transmission interaction information corresponding to at least one reference fire terminal device, as well as the authentication labels of the monitoring action response status of these reference fire terminal devices. LoRa transmission interaction information plays an important role in the fire IoT sensor network due to its low power consumption and long-distance transmission characteristics. This information contains quantitative description values ​​corresponding to at least one reference fire trend element. These fire trend elements may include but are not limited to smoke concentration, temperature, humidity, light intensity, etc., which are the direct basis for evaluating the possibility of fire. The authentication label is an authoritative authentication of whether the fire terminal equipment can correctly perform monitoring actions (such as alarm, start fire extinguishing equipment, etc.) under specific fire trend elements, which ensures the authenticity and validity of the training data.

[0075] Secondly, after obtaining the preliminary transmission interaction information and authentication tags, the algorithm debugging enters the in-depth analysis stage. Based on the quantitative description value of the reference fire terminal equipment under the reference fire trend element, it is necessary to determine the proportion of fire terminal equipment with set status labels corresponding to each quantitative description range. The essence of this step is to classify and count the data, and calculate the proportion of terminal equipment that can respond correctly within the quantitative range of different fire trend elements, so as to obtain the confidence evaluation degree corresponding to each quantitative description range. The level of confidence evaluation reflects the reliability of the algorithm in judging the fire trend within the quantitative range, and is an important basis for the subsequent screening of target fire trend elements.

[0076] Next, based on the confidence rating obtained in the previous step, the algorithm debugging enters the key target fire trend factor screening stage. Since multiple fire trend factors may be involved in the fire monitoring scenario, and not all factors are equally important for fire judgment, it is necessary to screen out at least one target fire trend factor from the reference fire trend factors. The screening is based on the confidence rating corresponding to the quantitative description range of each factor. Factors with high confidence ratings, i.e., higher judgment accuracy, can be preferentially selected as target fire trend factors. This step helps to simplify the algorithm model, improve the algorithm's operating efficiency and accuracy, and reduce unnecessary computational overhead.

[0077] After determining the target fire trend elements, the algorithm debugging enters the relationship statistical evaluation stage. Based on the authentication label of the monitoring action response status of the reference fire terminal equipment, an in-depth relationship statistical evaluation is performed on each target fire trend element. The purpose of this step is to explore the intrinsic connection between each target fire trend element and the monitoring action response of the fire terminal equipment, and to quantify the degree of influence of each element on the response state of the terminal equipment through statistical analysis methods (such as correlation analysis, regression analysis, etc.), and then obtain the attention value corresponding to each target fire trend element. The attention value reflects the degree of attention that the algorithm should pay when processing specific fire trend elements, and is the basis for the algorithm weight allocation and decision logic design.

[0078] Finally, based on the attention values ​​corresponding to the target fire trend elements and the confidence ratings corresponding to the quantitative description ranges of these elements, the algorithm debugging enters the final model generation stage. In this stage, by comprehensively using machine learning algorithms (such as neural networks, support vector machines, etc.) or deep learning frameworks, the attention values ​​and confidence ratings are used as feature inputs to construct a fire IoT sensor monitoring algorithm for fire terminal equipment. The algorithm can quickly and accurately identify fire trends based on the real-time LoRa transmission interaction information, and prioritize the data that is most critical to fire judgment based on the allocation of attention values, thereby achieving efficient and accurate fire warning and response.

[0079] It is worth noting that the entire debugging and training process is not completed overnight, but an iterative optimization cycle. In practical applications, it may be necessary to continuously adjust the screening criteria of the target fire trend elements, optimize the calculation method of the attention value, and even redesign the algorithm architecture according to the actual performance of the algorithm to achieve the best monitoring effect. In addition, with the addition of new equipment and the elimination of old equipment in the fire IoT sensor network, as well as changes in the fire monitoring environment, the algorithm's training data set also needs to be updated regularly to ensure that the algorithm can continue to adapt to new monitoring needs.

[0080] It can be seen that the above debugging and training method for the fire IoT sensor monitoring algorithm, through refined step design and scientific data analysis, has achieved effective screening of fire trend elements and reasonable allocation of attention values, thereby generating a high-performance fire IoT sensor monitoring algorithm. This algorithm can not only improve the accuracy and response speed of fire monitoring, but also maintain stable monitoring performance in complex and changeable fire environments, providing strong technical support for fire safety.

[0081] In order to facilitate the clear and complete implementation of the above technical solution, an exemplary implementation is further described below.

[0082] 1. Data Acquisition and Preprocessing

[0083] 1.1 Data Acquisition

[0084] In order to build a solid data foundation, it is necessary to obtain LoRa transmission interaction information from at least one reference fire terminal device. This information includes quantitative description values ​​of multiple reference fire trend elements, such as smoke concentration (unit: ppm), temperature (unit: ℃), humidity (unit: %) and light intensity (unit: lx). At the same time, it is also necessary to obtain the authentication tags of the monitoring action response status of these terminal devices under specific fire trend elements to verify whether the terminal devices can correctly perform monitoring actions.

[0085] For example, there is a dataset containing 100 reference fire terminal devices, each of which provides the following information:

[0086] Smoke concentration (ppm): range 0-1000, step 10;

[0087] Temperature (℃): range -20-80, step 1;

[0088] Humidity (%): range 0-100, step 5;

[0089] Light intensity (lx): range 0-10000, step 100.

[0090] and an authentication tag for each terminal device’s monitoring action response status under a specific combination of fire trend elements.

[0091] 1.2 Data Preprocessing

[0092] After obtaining the raw data, data preprocessing is required to ensure the accuracy and consistency of the data. This includes steps such as data cleaning, data normalization, and data standardization.

[0093] For example, for smoke concentration data, the following normalization formula can be used:

[0094] Norm_Smoke=(Smoke-Smoke_min) / (Smoke_max-Smoke_min);

[0095] Among them, Smoke is the original smoke concentration value, Smoke_min and Smoke_max are the minimum and maximum smoke concentration values ​​in the data set respectively. Through normalization, the smoke concentration value can be converted to the range of 0-1, which is convenient for subsequent processing.

[0096] 2. Analysis and screening of fire trend factors

[0097] 2.1. Quantitative description range and confidence evaluation calculation

[0098] After the data preprocessing is completed, it is necessary to determine the proportion of fire terminal equipment with set status labels corresponding to each quantitative description range based on the quantitative description value of the reference fire terminal equipment under the reference fire trend element. This step is essentially to classify and count the data.

[0099] Taking smoke concentration as an example, it can be divided into 10 quantitative description ranges (each 100 ppm is a range), and then the proportion of terminal devices that can correctly respond (i.e. alarm) in each range is calculated. This proportion is the confidence evaluation corresponding to the quantitative description range, which reflects the reliability of the algorithm in judging the fire trend within the quantitative range.

[0100] 2.2. Screening of target fire trend elements

[0101] Based on the confidence rating obtained in the previous step, at least one target fire trend element needs to be selected from the reference fire trend elements. The selection is based on the confidence rating corresponding to the quantitative description range of each element. Yes, elements with high confidence rating and higher judgment accuracy will be preferred as target fire trend elements.

[0102] For example, if it is found that the confidence evaluation of smoke concentration and temperature in most quantitative description ranges is high (such as greater than 85%), while the confidence evaluation of humidity and light intensity is relatively low, then smoke concentration and temperature can be selected as target fire trend elements.

[0103] 3. Relationship Statistical Evaluation and Attention Value Calculation

[0104] 3.1 Relationship Statistical Evaluation

[0105] After determining the target fire trend elements, it is necessary to conduct an in-depth statistical evaluation of the relationship between each target fire trend element. The purpose of this step is to explore the inherent relationship between each target fire trend element and the monitoring action response of the fire terminal equipment.

[0106] Statistical methods such as correlation analysis and regression analysis can be used to quantify the degree of influence of each factor on the response state of the terminal equipment. For example, the correlation coefficient between smoke concentration and the alarm probability of the terminal equipment can be calculated to evaluate the influence of smoke concentration on the alarm probability.

[0107] 3.2. Attention value calculation

[0108] Based on the results of the relationship statistical evaluation, the attention value corresponding to each target fire trend element can be obtained. The attention value reflects the degree of attention that the algorithm should pay when processing a specific fire trend element.

[0109] A simple way to calculate the attention value is to determine it based on the product of the confidence rating and correlation coefficient of each factor. For example, if the confidence rating of smoke concentration is 0.9 and its correlation coefficient with alarm probability is 0.8, then the attention value of smoke concentration can be 0.9*0.8=0.72.

[0110] 4. Algorithm Generation and Optimization

[0111] 4.1 Algorithm Generation

[0112] After obtaining the attention values ​​corresponding to the target fire trend elements, a fire IoT sensor monitoring algorithm for fire terminal equipment can be generated based on these values ​​and confidence ratings.

[0113] One possible algorithm is based on machine learning methods, such as support vector machines (SVM) or neural networks (NN). The attention value can be used as the feature weight, and the quantitative description value of each target fire trend element can be used as the input feature to train a classification model to predict the monitoring action response state of the terminal device.

[0114] Taking neural network as an example, a three-layer neural network consisting of input layer, hidden layer and output layer can be constructed. The input layer receives the quantitative description value of each target fire trend element; the hidden layer extracts features through nonlinear transformation; the output layer outputs the prediction result (such as "alarm" or "no alarm"). During the training process, the cross entropy loss function and gradient descent algorithm can be used to optimize the network parameters.

[0115] 4.2 Algorithm Optimization

[0116] In practical applications, it may be necessary to continuously adjust the screening criteria for target fire trend elements, optimize the calculation method of attention value, or even redesign the algorithm architecture according to the actual performance of the algorithm to achieve the best monitoring effect.

[0117] One of the optimization methods is to use ensemble learning methods, such as RandomForest or Gradient Boosting Decision Tree (GBDT). These methods can build a strong classifier by combining multiple weak classifiers, thereby improving the accuracy and generalization ability of the algorithm.

[0118] Another optimization method is to use deep learning methods, such as convolutional neural network (CNN) or recurrent neural network (RNN). These methods can automatically extract features and handle complex nonlinear relationships, thereby further improving the performance of the algorithm.

[0119] This design achieves effective screening of fire trend elements and reasonable allocation of attention values. Experimental results show that this method can significantly improve the accuracy and response speed of fire monitoring, providing strong technical support for fire safety.

[0120] In some exemplary embodiments, the proportion of fire terminal equipment with set state labels corresponding to each quantitative description range of the reference fire trend element is determined based on the quantitative description value of the reference fire terminal equipment under the reference fire trend element to obtain the confidence evaluation degree corresponding to each quantitative description range, including: determining the number of fire terminal equipment whose labels of monitoring action response states belong to set state labels based on the quantitative description value of the reference fire terminal equipment under the reference fire trend element and at least one quantitative description range of the reference fire trend element to obtain the proportion of fire terminal equipment with set state labels corresponding to each quantitative description range of the reference fire trend element; for each quantitative description range, mapping processing is performed on the proportion of fire terminal equipment with each set state label corresponding to the quantitative description range to obtain the confidence evaluation degree corresponding to the quantitative description range.

[0121] The core of this embodiment is to quantify the confidence rating under different fire conditions by analyzing and evaluating the performance of fire terminal equipment under specific fire trend elements. Specifically, the system first matches these values ​​with the preset quantitative description range based on the quantitative description values ​​of the fire terminal equipment under a certain reference fire trend element (such as smoke concentration, temperature, etc.). Subsequently, the system counts the proportion of fire terminal equipment in a specific state (such as set state labels such as "normal", "warning", and "alarm") within each quantitative description range. Finally, by mapping and processing these proportions, the system can generate a confidence rating for each quantitative description range, which reflects the system's confidence level in the fire judgment within the range.

[0122] From another perspective, this embodiment is building a data-driven fire situation awareness model. By comparing the actual measurement values ​​of the fire terminal equipment with the preset standard range and combining the response status of the equipment, the model can evaluate the reliability of the system in making correct judgments under different fire conditions. This not only helps to improve the accuracy of fire warnings, but also provides a more scientific basis for firefighting decisions.

[0123] For example, there are 100 fire terminal devices, which monitor the reference fire trend element based on smoke density. The quantitative description range of smoke density is divided into four levels: [0, 10], (10, 30], (30, 50], (50, +∞), corresponding to the four set status labels of "safe", "mild danger", "moderate danger", and "severe danger" respectively.

[0124] In the range of [0, 10], 80 devices are in a “safe” state, 5 are in a “slightly dangerous” state, and the remaining 15 are not responding due to failures.

[0125] In the (10, 30] range, 60 devices are in “mildly dangerous” state, 20 are in “safe” state, 10 are in “medium dangerous” state, and 10 are unresponsive.

[0126] In the range (30, 50], 30 devices are in "medium danger" state, 40 are in "slight danger" state, 20 are in "severe danger" state, and 10 are unresponsive.

[0127] In the range of (50, +∞), there are 90 devices in the “severe danger” state, 5 in the “moderate danger” state, and 5 unresponsive.

[0128] Furthermore, for the range [0, 10], the proportion of the "safe" state is 80 / 100 = 0.8, the proportion of the "slightly dangerous" state is 0.05, and the total proportion of the remaining states is 0.15. Similarly, the proportions of other ranges are calculated.

[0129] As for the mapping process, a mapping function, such as a linear mapping or a nonlinear mapping (such as a sigmoid function), is designed to convert the proportion of each state into a confidence rating. For example, a weighted average method can be used, in which the weight is assigned according to the importance of the state. The weight of the "safe" state is set to 1, the weight of "mild danger" is 2, the weight of "moderate danger" is 3, the weight of "severe danger" is 4, and non-response is not included in the total score, then:

[0130] The confidence rating of the range [0, 10] = (0.81 + 0.052) / (0.8 + 0.05) = 1.1 / 0.85 ≈ 1.294 (rounded to three decimal places); similarly, the confidence ratings of other ranges are calculated.

[0131] Finally, the confidence rating corresponding to each quantitative description range is obtained, such as the range of [0, 10] is 1.294, the range of (10, 30] is 1.8 (the result after setting the calculation), and so on.

[0132] Through the above process, the system can conduct confidence evaluation on the quantitative description range under different fire trend elements based on the actual response of the fire terminal equipment, thereby providing strong support for fire fighting decision-making.

[0133] In some other embodiments, the number of fire terminal devices whose labels of monitoring action response status belong to set status labels is determined based on the quantitative description value of the reference fire terminal device under the reference fire trend element and at least one quantitative description range of the reference fire trend element, and the proportion of fire terminal devices with set status labels corresponding to each quantitative description range of the reference fire trend element is obtained, including: recording the reference fire terminal devices whose labels of monitoring action response status belong to set status labels to obtain a first number; for each quantitative description range of the reference fire trend element, counting the reference fire terminal devices whose quantitative description values ​​of the reference fire trend element are within the quantitative description range to obtain the target fire terminal devices corresponding to the quantitative description range; determining a second number of the target fire terminal devices whose labels of monitoring action response status belong to set status labels; based on the first number and the second number, determining the proportion of fire terminal devices with set status labels corresponding to each quantitative description range of the reference fire trend element.

[0134] In the fire monitoring system, accurate judgment of the fire trend is crucial. In order to achieve this goal, the fire monitoring system needs to continuously collect and analyze the quantitative description values ​​from various fire terminal devices. These values ​​reflect various elements of the fire trend, such as smoke concentration, temperature, flame brightness, etc. However, it is not enough to just collect these data. The fire monitoring system also needs to be able to judge the response status of the fire terminal equipment based on these data, and then evaluate the effectiveness of the entire fire monitoring system.

[0135] The method provided in this embodiment is to achieve an accurate assessment of the response status of fire terminal equipment. Specifically, it determines whether the label of the monitoring action response status belongs to the set status label based on the quantitative description value of the reference fire terminal equipment under the reference fire trend factor and the quantitative description range of these factors. Through this process, the fire monitoring system can count the proportion of fire terminal equipment in a specific response state within the quantitative description range of different fire trend factors, thereby providing strong data support for fire decision-making.

[0136] This not only improves the intelligence level of the fire monitoring system, but also enables firefighters to more accurately understand the development trend of the fire and the response of firefighting equipment, so as to make more scientific and reasonable decisions.

[0137] In other words, the method described in this embodiment reveals the relationship between the fire trend and the response status of the fire equipment by statistically analyzing the quantitative description values ​​of the fire terminal equipment. In this method, the reference fire terminal equipment is the basis for evaluation, and their quantitative description values ​​reflect the actual situation of the fire trend. The quantitative description range of the reference fire trend elements is the benchmark for evaluation, which divides the fire trend into different intervals, which is convenient for the fire monitoring system to analyze and compare.

[0138] By counting the fire terminal devices in different quantitative description ranges and determining whether the labels of the monitoring action response status in these devices belong to the set status labels, the fire monitoring system can obtain the proportion of fire terminal devices in a specific response status within each quantitative description range. This proportion not only reflects the response of fire equipment under different fire trends, but also provides an important basis for firefighters to evaluate the effectiveness of the fire monitoring system.

[0139] For example, in a fire monitoring system, there are 100 reference fire terminal devices, which monitor the smoke density and fire trend elements respectively. The quantitative description values ​​of these devices are between 0 and 100, indicating different levels of smoke density. At the same time, the fire monitoring system sets three quantitative description ranges: [0, 30], (30, 70], (70, 100], corresponding to the three response status labels of "safety", "warning" and "alarm" respectively.

[0140] Now, the fire monitoring system needs to count the proportion of fire terminal equipment in a specific response state within each quantitative description range. The specific steps are as follows:

[0141] (1) Record the reference fire terminal equipment whose tag of monitoring action response status belongs to the set status tag. In the initial state, all equipment is in the "safe" state, so the first number is 100;

[0142] (2) For each quantitative description range of smoke concentration, the reference fire terminal equipment whose quantitative description value is within this range is counted. Suppose that after a period of monitoring, the fire monitoring system obtains the following data:

[0143] There are 80 devices in the range [0, 30];

[0144] There are 15 devices in the range of (30,70];

[0145] There are 5 devices within the range of (70, 100].

[0146] (3) Determine whether the tag of the monitoring action response status in the target fire terminal device belongs to the second number of the set status tag. For example, during statistics, the fire monitoring system finds:

[0147] Of the 80 devices in the [0, 30] range, 75 are still in the "safe" state, and 5 are in the "warning" state due to false alarms;

[0148] Of the 15 devices in the (30, 70] range, 10 were in the "warning" state, and 5 were still in the "safe" state due to missed reports;

[0149] Of the five devices in the range of [70, 100], four are in "alarm" state and one is not responding due to a fault.

[0150] (4) Based on the first and second numbers, determine the proportion of fire terminal equipment with set status labels corresponding to each quantitative description range of the reference fire trend element. The specific calculation is as follows:

[0151] In the range [0, 30], the proportion of devices in the "safe" state is 75 / 80=0.9375, and the proportion of devices in the "warning" state is 5 / 80=0.0625;

[0152] (In the range of 30, 70], the proportion of devices in the "warning" state is 10 / 15=0.6667, and the proportion in the "safe" state is 5 / 15=0.3333 (Note: the "safe" state here is actually a missed report and should be treated as an abnormal situation);

[0153] (70, 100] range, the proportion of devices in the "alarm" state is 4 / 5=0.8, and the proportion of devices in the "unresponsive" state is 1 / 5=0.2 (Note: the "unresponsive" state here should be treated as a fault condition).

[0154] Furthermore, in order to achieve the above functions, the following algorithm can be designed:

[0155] Python

[0156] # Set the data of fire terminal equipment to be stored in list form, each element is a dictionary containing device ID, quantitative description value and response status label

[0157] # Sample data: terminals = [{'id': 1, 'value': 25, 'status': 'Safe'}, {'id': 2, 'value': 55, 'status': 'Warning'}, ...]

[0158] # Define the quantitative description range and its corresponding response status label

[0159] ranges = {

[0160] 'safe': (0, 30),

[0161] 'warning': (30, 70),

[0162] 'alarm': (70, 100)

[0163] }

[0164] # Initialize the counter

[0165] counters = {

[0166] 'safe': {'safe': 0, 'warning': 0, 'alarm': 0, 'unresponsive': 0},

[0167] 'warning': {'safe': 0, 'warning': 0, 'alarm': 0, 'unresponsive': 0},

[0168] 'alarm': {'safe': 0, 'warning': 0, 'alarm': 0, 'unresponsive': 0}

[0169] }

[0170] # Traverse the terminal device data and count the number of devices with each response status label within each quantitative description range

[0171] for terminal in terminals:

[0172] value = terminal['value']

[0173] status = terminal['status']

[0174] # Determine which range the quantitative description value belongs to

[0175] if ranges['safe'][0]<= value<= ranges['safe'][1]:

[0176] range_key = 'safe'

[0177] elif ranges['warning'][0] <value<= ranges['warning'][1]:

[0178] range_key = 'warning'

[0179] elif ranges['alarm'][0] <value<= ranges['alarm'][1]:

[0180] range_key = 'alarm'

[0181] else:

[0182] # Handle the situation where the quantitative description value exceeds the preset range (here it is set as an abnormal situation and is not counted in the statistics)

[0183] continue

[0184] # Update counter

[0185] counters[range_key][status]+= 1

[0186] # Calculate the proportion of devices with each response status label within each quantitative description range

[0187] proportions = {}

[0188] for range_key, range_counters in counters.items():

[0189] total = sum(range_counters.values())

[0190] if total == 0:

[0191] # Handle the situation where no device is within the quantitative description range (here the ratio is set to 0)

[0192] proportions[range_key] = {status: 0 for status in range_counters}

[0193] else:

[0194] proportions[range_key] = {status: count / total for status, count in range_counters.items()}

[0195] # Output results

[0196] print(proportions)

[0197] In the above algorithm, terminals is a list storing fire terminal equipment data, ranges is a dictionary defining the quantitative description range and its corresponding response status label, counters is a counter used to count the number of devices with each response status label within each quantitative description range, and proportions is a dictionary used to store the proportion of devices with each response status label within each quantitative description range.

[0198] By running this algorithm, the fire monitoring system can obtain the proportion of fire terminal equipment in a specific response state within each quantitative description range, thereby achieving an accurate evaluation of the effectiveness of the fire monitoring system.

[0199] In summary, by statistically analyzing the quantitative description values ​​of fire terminal equipment, the relationship between the fire trend and the response status of fire equipment is revealed. This not only improves the intelligence level of the fire monitoring system, but also provides firefighters with a more accurate and scientific basis for decision-making. At the same time, this method also has certain versatility and scalability, and can be applied to other similar fire monitoring systems.

[0200] In some alternative embodiments, the method of selecting at least one target fire trend element from the reference fire trend elements based on the confidence ratings corresponding to the respective quantitative description ranges of the reference fire trend elements comprises: selecting original fire trend elements from the reference fire trend elements based on the confidence ratings corresponding to the respective quantitative description ranges of the reference fire trend elements; determining the spatiotemporal impact characteristics of the original fire trend elements based on the proportion of fire terminal equipment with set status labels corresponding to the respective quantitative description ranges of the original fire trend elements and the confidence ratings of the respective quantitative description ranges to obtain a spatiotemporal impact feature set corresponding to the original fire trend elements; and selecting at least one target fire trend element from the original fire trend elements based on the spatiotemporal impact feature set.

[0201] Furthermore, based on the proportion of fire terminal equipment with set status labels corresponding to each quantitative description range of the original fire trend element and the confidence evaluation degree of each quantitative description range, the spatiotemporal impact characteristics of the original fire trend element are determined to obtain the spatiotemporal impact feature set corresponding to the original fire trend element, including: for each quantitative description range of the original fire trend element, the proportion of fire terminal equipment with set status labels corresponding to the quantitative description range and the confidence evaluation degree of the quantitative description range are combined based on the spatiotemporal domain to obtain the sub-spatiotemporal impact feature set corresponding to the quantitative description range; the sub-spatiotemporal impact feature sets corresponding to each quantitative description range of the original fire trend element are combined based on the spatiotemporal domain to obtain the spatiotemporal impact feature set corresponding to the original fire trend element.

[0202] In practical applications, accurate identification and analysis of fire trend elements are crucial for developing effective fire fighting and rescue strategies. This embodiment aims to select the most representative target fire trend elements from multiple reference fire trend elements through a systematic method, so as to provide a scientific basis for fire fighting decision-making.

[0203] On the one hand, the core function of this embodiment is to select at least one target fire trend element from a set of reference fire trend elements. The process is divided into three main steps: selecting original fire trend elements, determining spatiotemporal impact characteristics, and selecting target fire trend elements. Each step relies on specific data analysis and processing methods to ensure that the target fire trend elements finally selected are highly reliable and practical.

[0204] (1) Screening of original fire trend elements:

[0205] Input: Reference fire trend elements and the confidence ratings corresponding to their various quantitative description ranges.

[0206] Processing: Based on the confidence evaluation, select the original fire trend elements with higher confidence from all reference fire trend elements.

[0207] Output: Filtered original fire trend elements.

[0208] (2) Determine the spatiotemporal impact characteristics:

[0209] Input: original fire trend elements, proportion of fire terminal equipment with set status labels corresponding to each quantitative description range, and confidence evaluation degree.

[0210] deal with:

[0211] For each quantitative description range, the proportion of fire terminal equipment and the confidence evaluation degree are combined based on the spatiotemporal domain to generate a sub-spatiotemporal impact feature set.

[0212] All sub-spatiotemporal impact feature sets are combined based on the spatiotemporal domain to generate the spatiotemporal impact feature set corresponding to the original fire trend elements.

[0213] Output: The spatiotemporal impact feature set corresponding to the original fire trend elements.

[0214] (3) Screening target fire trend factors:

[0215] Input: The spatiotemporal impact feature set corresponding to the original fire trend elements.

[0216] Processing: Based on the spatiotemporal impact feature set, appropriate evaluation methods are used to screen out target fire trend elements from the original fire trend elements.

[0217] Output: Filtered target fire trend elements.

[0218] In other words, the core of this embodiment is to gradually narrow down the scope of fire trend elements through multi-level screening and feature extraction, and finally lock in the target elements that have the greatest impact on fire fighting decisions. This process not only considers the confidence of fire trend elements, but also deeply analyzes their impact characteristics in the time and space domain, thereby ensuring the comprehensiveness and accuracy of the screening results.

[0219] Confidence evaluation is an important indicator to measure the reliability of fire trend elements. By screening high-confidence original fire trend elements, it can ensure that the basic data for subsequent analysis has a high degree of credibility.

[0220] The spatiotemporal influence characteristics reflect the distribution and change characteristics of fire trend elements in time and space. By combining the proportion of fire terminal equipment and confidence evaluation degree in the spatiotemporal domain, the spatiotemporal influence laws of fire trend elements can be fully revealed.

[0221] Based on the spatiotemporal impact feature set, the importance of the original fire trend elements can be further evaluated, and those target fire trend elements that are most instructive for firefighting decisions can be screened out.

[0222] In order to explain this embodiment more intuitively, supplementary explanation will be given below in conjunction with specific examples.

[0223] (1) Screening of original fire trend elements:

[0224] For example, there are five reference fire trend elements, recorded as A, B, C, D, and E respectively.

[0225] The quantitative description range of each factor and its corresponding confidence evaluation degree are as follows:

[0226] A: [0, 10], confidence level 0.8;

[0227] B: [10, 20], confidence level 0.75;

[0228] C: [20, 30], confidence level 0.9;

[0229] D: [30, 40], confidence level 0.6;

[0230] E: [40, 50], confidence level 0.85.

[0231] According to the confidence evaluation, the original fire trend elements with confidence greater than 0.75, namely A, C, and E, are screened out.

[0232] (2) Determine the spatiotemporal impact characteristics:

[0233] The proportion of fire terminal equipment with set status labels corresponding to each quantitative description range of the original fire trend elements A, C, and E is as follows:

[0234] A: [0, 10], scale 0.6;

[0235] C: [20, 30], scale 0.7;

[0236] E: [40, 50], scale 0.8.

[0237] For each quantitative description range, the proportion of fire terminal equipment and the confidence evaluation degree are combined based on the spatiotemporal domain to generate a sub-spatiotemporal impact feature set. For example, for A: [0, 10], its sub-spatiotemporal impact feature set can be expressed as {(0, 0.6, 0.8), (10, 0.6, 0.8)}, where the first element represents the lower limit of the quantitative description range, the second element represents the proportion of fire terminal equipment, and the third element represents the confidence evaluation degree.

[0238] All sub-spatiotemporal impact feature sets are combined based on the spatiotemporal domain to generate the spatiotemporal impact feature set corresponding to the original fire trend elements. For example, for A, C, and E, their spatiotemporal impact feature set can be expressed as {A's sub-spatiotemporal impact feature set, C's sub-spatiotemporal impact feature set, and E's sub-spatiotemporal impact feature set}.

[0239] (3) Screening target fire trend factors:

[0240] For example, an evaluation method based on a spatiotemporal impact feature set, such as a weighted summation method, is used to score the original fire trend elements. The higher the score, the greater the impact of the element on firefighting decisions.

[0241] For example, the scoring results are as follows: A: 75 points; C: 85 points; E: 90 points.

[0242] According to the scoring results, the target fire trend factor with the highest score, namely E, is screened out.

[0243] It can be seen that this embodiment uses a systematic method to select the most representative target fire trend elements from multiple reference fire trend elements. This process not only considers the confidence of the fire trend elements, but also deeply analyzes their impact characteristics in the time and space domain, thereby ensuring the comprehensiveness and accuracy of the screening results.

[0244] Under some preferred design ideas, the fire Internet of Things sensor monitoring algorithm for fire terminal equipment is generated based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, including: based on the attention value corresponding to the target fire trend element, a decision tree network between the fire trend element and the cross possibility is generated, and the cross possibility is the cross possibility that the label of the monitoring action response state of the fire terminal equipment belongs to the set state label; based on the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element and the decision tree network, a fire Internet of Things sensor monitoring algorithm for fire terminal equipment is generated.

[0245] It can be understood that the core of this embodiment is to generate a fire IoT sensor monitoring algorithm for fire terminal equipment. The algorithm aims to achieve effective monitoring and response to fire conditions by comprehensively considering the attention value of the target fire trend element and the confidence evaluation of each quantitative description range. Specifically, the functions of the algorithm include:

[0246] 1) Construct a decision tree network based on attention value: The attention value reflects the importance of the target fire trend factor in the overall fire assessment. Using the attention value, a decision tree network is constructed between the fire trend factor and the fire terminal equipment monitoring action response state (i.e., cross probability). This network can learn and predict the probability distribution of the fire terminal equipment response state under different fire trend factors;

[0247] 2) Generate monitoring algorithm by combining confidence evaluation: Confidence evaluation indicates the degree of trust in the quantitative description range of the target fire trend element. The confidence evaluation is combined with the decision tree network to generate the final fire IoT sensor monitoring algorithm. This algorithm can comprehensively consider the importance of the fire trend element (attention value) and the reliability of the description range (confidence evaluation), so as to more accurately trigger the monitoring action of the fire terminal equipment.

[0248] In other words, the fire IoT sensor monitoring algorithm proposed in this embodiment is a data-driven intelligent decision-making method. It conducts in-depth analysis of fire trend elements, digs out the potential correlation between the fire terminal equipment response status, and then realizes accurate monitoring. The core of this method is:

[0249] 1) Data fusion and mining: effectively fuse two different types of information, attention value and confidence evaluation, to discover the hidden laws between fire trends and equipment responses;

[0250] 2) Intelligent decision-making: Using the decision tree network, a machine learning model, intelligent mapping of fire trends to equipment response status is achieved, improving the accuracy and efficiency of monitoring.

[0251] For example, the target fire trend factor is "smoke density", and its corresponding attention value is 0.8 (indicating that smoke density is a relatively important factor in the overall fire assessment). Furthermore, the quantitative description range of smoke density is set to be divided into three intervals: [0, 10], [10, 50], and [50, +∞], and the corresponding confidence evaluation degrees are 0.7, 0.9, and 0.6, respectively.

[0252] Construct a decision tree network based on attention value: Based on the attention value of 0.8, a decision tree network with smoke concentration as input feature is constructed. Each node in the network represents a decision point, and the probability distribution of the response state of the fire terminal equipment is predicted according to the different value ranges of smoke concentration.

[0253] Generate monitoring algorithm based on confidence rating: For each quantitative description range of smoke concentration, the output of the decision tree network is weighted and adjusted based on its confidence rating. For example, for the interval [10, 50] (confidence rating is 0.9), the decision result of this interval will have a larger weight in the final output.

[0254] The resulting fire IoT sensor monitoring algorithm can accurately trigger the monitoring action of the fire terminal equipment based on the real-time smoke concentration value, taking into account the attention value and confidence evaluation. For example, when the smoke concentration reaches 30 (in the interval [10, 50]), the algorithm will trigger monitoring actions such as alarm or sprinkler with a higher confidence.

[0255] It can be seen that the fire IoT sensor monitoring algorithm proposed in this embodiment realizes effective monitoring and response to the fire situation by comprehensively considering the attention value and confidence evaluation degree of the target fire trend elements.

[0256] In the next step, the fire IoT sensor monitoring algorithm for the fire terminal equipment is generated based on the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element and the decision tree network, including: based on the decision tree network and the set algorithm construction index, the initial fire hazard discrimination weight for the fire terminal equipment is obtained; according to the initial fire hazard discrimination weight, the attention value corresponding to the target fire trend element, and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, the fire IoT sensor monitoring algorithm is generated, the fire IoT sensor monitoring algorithm includes a feature association diagram between the target fire hazard discrimination result of the fire terminal equipment and the quantitative description value of the target fire trend element, the target fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal equipment belongs to the set state label.

[0257] It can be understood that the core of this embodiment is to generate a fire IoT sensor monitoring algorithm for fire terminal equipment. The functions of this algorithm mainly include two aspects:

[0258] 1) Acquisition of initial fire hazard discrimination weights: Based on the decision tree network and the set algorithm construction indicators, the algorithm can calculate the initial fire hazard discrimination weights for fire terminal equipment. This weight reflects the sensitivity and response priority of fire terminal equipment to potential fire hazards under specific fire trend factors;

[0259] 2) Generation of fire protection IoT sensor monitoring algorithm: Using the initial fire hazard discrimination weight, the attention value of the target fire trend element, and the confidence evaluation of each quantitative description range, the algorithm can generate a comprehensive fire protection IoT sensor monitoring solution. This solution not only includes the target fire hazard discrimination results of the fire terminal equipment, but also shows the relationship between the target fire hazard discrimination results and the quantitative description value of the target fire trend element through the feature association diagram.

[0260] In other words, the fire IoT sensor monitoring algorithm proposed in this embodiment is an intelligent decision-making system that integrates machine learning (such as decision tree network) and expert knowledge (such as setting algorithm construction indicators, attention value, and confidence evaluation degree). It achieves accurate prediction and visual display of the response status of the monitoring action of the fire terminal equipment through a comprehensive analysis of the fire trend elements.

[0261] 1) Machine learning: As a representative of machine learning, the decision tree network can automatically learn the complex relationship between fire trend elements and the response status of fire terminal equipment.

[0262] 2) Expert knowledge: Setting algorithm construction indicators, attention values, confidence ratings, etc. reflects the knowledge and experience of domain experts, providing a strong guarantee for the accuracy and reliability of the algorithm.

[0263] For example, the target fire trend factor is "smoke density", and its corresponding attention value is 0.85 (indicating that smoke density is of high importance in fire assessment). The quantitative description range of smoke density is divided into three intervals: [0, 10], [10, 50], and [50, +∞], and the corresponding confidence evaluation degrees are 0.7, 0.8, and 0.9, respectively.

[0264] 1) Obtaining the initial fire hazard discrimination weight: Based on the decision tree network and the set algorithm construction indicators (such as accuracy, recall rate, etc.), calculate the initial fire hazard discrimination weight for the fire terminal equipment. In this example, the weight value is set to 0.6 (indicating that when smoke density is used as a fire trend factor, the fire terminal equipment has a 60% ability to discriminate potential hazards).

[0265] 2) Generation of fire IoT sensor monitoring algorithm: The fire IoT sensor monitoring algorithm is generated using the initial fire hazard discrimination weight (0.6), the attention value of smoke concentration (0.85), and the confidence evaluation degree of each quantitative description range (0.7, 0.8, 0.9). The feature correlation diagram output by the algorithm shows the relationship between the discrimination results of the fire terminal equipment for the target fire hazard (such as "hazard" or "no hazard") and the smoke concentration value in different smoke concentration ranges. For example, in the smoke concentration range [10, 50], if the confidence evaluation degree is 0.8, the fire terminal equipment is more likely to be judged as "hazardous". The target fire hazard discrimination result represents the possibility that the monitoring action response state of the fire terminal equipment (such as alarm, sprinkler, etc.) belongs to the set state label (such as "fire"). In this example, if the smoke concentration reaches or exceeds a certain threshold (such as 30), the possibility of the fire terminal equipment judging it as a "fire" state will increase significantly.

[0266] It can be seen that the above-mentioned fire IoT sensor monitoring algorithm realizes the accurate prediction and visual display of the monitoring action response status of fire terminal equipment by combining machine learning and expert knowledge.

[0267] In a further embodiment, the fire IoT sensor monitoring algorithm is generated based on the initial fire hazard discrimination weight, the attention value corresponding to the target fire trend element, and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, including: for each target fire trend element, based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, a sensor information analysis network is generated for the target fire trend element, the sensor information analysis network is used to output a local fire hazard discrimination result of the fire terminal equipment, the local fire hazard discrimination result characterizes the possibility that the label of the monitoring action response state of the fire terminal equipment under the target fire trend element belongs to the set state label; based on the initial fire hazard discrimination weight and the sensor information analysis network corresponding to each target fire trend element, a fire IoT sensor monitoring algorithm is obtained.

[0268] In detail, this embodiment further refines the generation process of the fire IoT sensor monitoring algorithm, especially the processing method for each target fire trend element:

[0269] (1) Generation of sensor information analysis network: For each target fire trend element, the algorithm generates a dedicated sensor information analysis network based on its corresponding attention value and the confidence evaluation corresponding to each quantitative description range. The function of this network is to output the local fire hazard identification result of the fire terminal equipment, that is, the possibility that the monitoring action response state of the fire terminal equipment belongs to the set state label (such as "fire") under a specific fire trend element;

[0270] (2) Acquisition of the fire IoT sensor monitoring algorithm: After generating the sensor information analysis network of all target fire trend elements, the algorithm will integrate the initial fire hazard discrimination weights and these networks to obtain the final fire IoT sensor monitoring algorithm. This algorithm comprehensively considers the impact of all fire trend elements and can more comprehensively evaluate the monitoring action response status of the fire terminal equipment.

[0271] It can be seen that the fire protection IoT sensor monitoring algorithm generation method proposed in this embodiment is a divide-and-conquer strategy. It first decomposes the complex fire assessment problem into multiple smaller sub-problems targeting a single fire trend element, and then generates a dedicated sensor information analysis network for each sub-problem. Finally, by integrating the results of these networks, the overall fire hazard identification result is obtained.

[0272] Divide and conquer: By generating an independent sensor information analysis network for each target fire trend factor, the algorithm can more finely handle the impact of different fire trend factors on the response status of fire terminal equipment.

[0273] Comprehensive evaluation: When integrating the results of various sensor information analysis networks, the algorithm takes into account the initial fire hazard discrimination weights, thereby ensuring the accuracy and reliability of the overall evaluation.

[0274] For example, the target fire trend elements include "smoke density" and "temperature", whose attention values ​​are 0.8 and 0.7 respectively. The quantitative description range and corresponding confidence evaluation are as follows:

[0275] Smoke density: [0, 10] (confidence rating 0.6), [10, 50] (confidence rating 0.8), [50, +∞] (confidence rating 0.9);

[0276] Temperature: [0, 40] (confidence rating 0.7), [40, 80] (confidence rating 0.5), [80, +∞] (confidence rating 0.95).

[0277] Generation of sensor information analysis network: For smoke density, a sensor information analysis network based on an attention value of 0.8 and a confidence evaluation degree (0.6, 0.8, 0.9) is generated to output the local fire hazard discrimination result. For temperature, a sensor information analysis network based on an attention value of 0.7 and a confidence evaluation degree (0.7, 0.5, 0.95) is generated to output the local fire hazard discrimination result.

[0278] Acquisition of the fire protection IoT sensor monitoring algorithm: The initial fire hazard discrimination weight is set to 0.65 (indicates that the discrimination ability after comprehensive consideration of all fire trend factors is 65%). Based on the initial fire hazard discrimination weight and the results of the two sensor information analysis networks, the final fire protection IoT sensor monitoring algorithm is integrated.

[0279] For example, when the smoke concentration reaches 30 (belongs to the interval [10, 50], with a confidence rating of 0.8) and the temperature reaches 60 (belongs to the interval [40, 80], but with a lower confidence rating of 0.5), the algorithm will comprehensively consider these two factors as well as their attention values ​​and confidence ratings to come up with a comprehensive fire hazard identification result. If this result exceeds a certain set threshold (such as 0.5), the corresponding monitoring action of the fire terminal equipment is triggered.

[0280] In this way, by generating independent sensor information analysis networks for each target fire trend element and comprehensively considering the initial fire hazard discrimination weights and the results of these networks, a more comprehensive and accurate fire IoT sensor monitoring algorithm is generated.

[0281] In some independent embodiments, the method also includes: obtaining lora transmission interaction information corresponding to at least one verification fire terminal device, and an authentication label of the monitoring action response status of the verification fire terminal device, the lora transmission interaction information including the quantitative description value corresponding to the target fire trend element; through the fire IoT sensor monitoring algorithm, according to the quantitative description value of the target fire trend element of the verification fire terminal device, making a state label decision for the verification fire terminal device, and obtaining the target fire hazard discrimination result of the verification fire terminal device; for each discrimination weight range set for the fire terminal device, according to the authentication label corresponding to the verification fire terminal device, determining the number of verification fire terminal devices whose target fire hazard discrimination result is within the set state label of the discrimination weight range; summing the number of devices with the set state label corresponding to each associated discrimination weight range of the discrimination weight range, and obtaining the device superposition ratio of the set state label corresponding to the discrimination weight range; based on the device superposition ratio, detecting the algorithm quality of the fire IoT sensor monitoring algorithm, and obtaining an algorithm quality detection point of view.

[0282] It is understandable that in the design and implementation of the fire IoT sensor monitoring system, it is crucial to ensure the accuracy and reliability of the algorithm. This embodiment proposes a method to verify and evaluate the fire IoT sensor monitoring algorithm by obtaining the authentication tag of the lora (LongRange) transmission interaction information and monitoring action response status of the fire terminal equipment. This method not only helps to improve the performance of the algorithm, but also provides strong quality assurance for practical applications.

[0283] On the one hand, the method of this embodiment mainly includes the following key steps:

[0284] 1) Information acquisition: Obtain the LoRa transmission interaction information corresponding to at least one verification fire terminal device, as well as the authentication tag of the monitoring action response status of the device. The LoRa transmission interaction information contains the quantitative description value corresponding to the target fire trend element, which is the basic data for subsequent algorithm verification;

[0285] 2) Status label decision: Through the fire IoT sensor monitoring algorithm, the status label decision of the device is made based on the quantitative description value of the target fire trend factor of the verification fire terminal device. The result of the decision is the target fire hazard identification result, which reflects the monitoring action response status of the verification fire terminal device under a specific fire trend factor;

[0286] 3) Equipment number statistics: For each discrimination weight range set for the fire terminal equipment, according to the certification label corresponding to the verification fire terminal equipment, determine the number of verification fire terminal equipment with the set status label of the target fire hazard discrimination result within the specific discrimination weight range. This step is to count the distribution of verification fire terminal equipment within different discrimination weight ranges;

[0287] 4) Calculation of equipment superposition ratio: sum up the number of equipment with set status labels corresponding to each associated discrimination weight range, and obtain the equipment superposition ratio of the set status labels corresponding to the discrimination weight range. The equipment superposition ratio reflects the accumulation of fire terminal equipment verification within different discrimination weight ranges, and is an important indicator for evaluating algorithm performance;

[0288] 5) Algorithm quality detection: According to the equipment superposition ratio, the algorithm quality of the fire IoT sensor monitoring algorithm is tested to obtain the algorithm quality detection point of view. Through this step, the accuracy and reliability of the algorithm in practical applications can be comprehensively evaluated, providing strong support for the optimization and improvement of the algorithm.

[0289] On the other hand, the method of this embodiment is a comparative analysis method based on the actual data of the fire terminal equipment and the algorithm decision results. It implements the evaluation of the algorithm quality through the following core concepts:

[0290] 1) Verify fire terminal equipment: As a sample for algorithm verification, the verification fire terminal equipment provides actual LoRa transmission interaction information and authentication tags for monitoring action response status. These data are the basis for algorithm verification and are used to evaluate the performance of the algorithm in practical applications;

[0291] 2) LoRa transmission interaction information: LoRa transmission interaction information contains the quantitative description values ​​corresponding to the target fire trend elements, which is an important basis for the algorithm to make decisions. By obtaining this information, the actual fire scene can be simulated and the algorithm can be fully tested;

[0292] 3) Status label decision: Status label decision is one of the core functions of the algorithm. It makes status label decisions for the verification fire terminal equipment based on the quantitative description value of the target fire trend element. The result of the decision reflects the algorithm's ability to identify and respond to actual fire scenarios;

[0293] 4) Discrimination weight range and equipment number statistics: The discrimination weight range is set to classify and count the decision results of the algorithm. By counting the number of fire terminal devices verified within different discrimination weight ranges, we can understand the performance of the algorithm in different scenarios.

[0294] 5) Equipment superposition ratio and algorithm quality detection: Equipment superposition ratio is one of the important indicators for evaluating algorithm performance. It reflects the cumulative situation of verifying fire terminal equipment within different discrimination weight ranges. By calculating the equipment superposition ratio and conducting a comprehensive quality test on the algorithm, the accuracy and reliability of the algorithm in practical applications can be evaluated;

[0295] For example, there is a fire IoT sensor monitoring system that contains 100 verified fire terminal devices. These devices transmit interactive information through lora, providing quantitative description values ​​corresponding to target fire trend elements (such as smoke concentration, temperature, etc.). At the same time, the authentication labels of the monitoring action response status of these devices (such as "normal", "warning", "fire", etc.) are also known.

[0296] 1) Information acquisition: Obtain the LoRa transmission interaction information and authentication tags of 100 verification fire terminal devices from the system. Set the LoRa transmission interaction information of one of the devices to include a smoke concentration value of 30 (in the [10, 50] interval) and a temperature value of 65 (in the [40, 80] interval), and the authentication tag is "warning".

[0297] 2) Status label decision: Through the fire IoT sensor monitoring algorithm, the status label decision is made based on the smoke concentration and temperature values ​​of the device. The result of the algorithm decision is set to "warning", which is consistent with the certification label of the device.

[0298] 3) Equipment number statistics: For each discrimination weight range set for fire terminal equipment (such as [0, 0.3], [0.3, 0.6], [0.6, 1], etc.), count the number of verification fire terminal equipment with set status labels in a specific discrimination weight range. In the [0.3, 0.6] discrimination weight range, 30 devices have a status label of "warning".

[0299] 4) Calculation of equipment superposition ratio: Sum the number of devices with the set status label "warning" corresponding to each associated judgment weight range of the judgment weight range [0.3, 0.6] (such as [0, 0.3] and [0.6, 1]). Assume that there are 10 "warning" devices in the range [0, 0.3] and 5 "warning" devices in the range [0.6, 1]. Then the superposition ratio of devices with the set status label "warning" corresponding to the judgment weight range [0.3, 0.6] is (30+10+5) / 100=45%.

[0300] 5) Algorithm quality detection: The algorithm quality of the fire IoT sensor monitoring algorithm is detected based on the equipment superposition ratio. A quality standard is set, that is, the equipment superposition ratio should not be less than 40%. In this case, the algorithm quality detection point of view is "qualified" because the equipment superposition ratio is 45%, which meets the quality standard.

[0301] To further illustrate the method of this embodiment, similar analysis can be performed on more verification fire terminal equipment, and the proportion of equipment superposition within different discrimination weight ranges can be counted. Through comprehensive data analysis and algorithm quality testing, the accuracy and reliability of the fire IoT sensor monitoring algorithm in practical applications can be evaluated, providing strong support for algorithm optimization and improvement.

[0302] In addition, we can also consider introducing more evaluation indicators and methods to comprehensively evaluate the performance of the algorithm. For example, we can calculate the accuracy, recall, F1 score and other indicators of the algorithm to more comprehensively reflect the performance of the algorithm in different scenarios. At the same time, we can also use cross-validation, leave-one-out method and other methods to verify the generalization ability and stability of the algorithm.

[0303] Therefore, by obtaining the LoRa transmission interaction information and authentication tags of the fire terminal equipment, the fire IoT sensor monitoring algorithm is verified and evaluated. This method not only helps to improve the performance of the algorithm, but also provides strong quality assurance for practical applications.

[0304] In other independent embodiments, the algorithm quality of the fire IoT sensor monitoring algorithm is detected based on the device superposition ratio to obtain an algorithm quality detection point of view, including: for each discrimination weight range, mapping processing is performed on the device superposition ratio of each set state label corresponding to the discrimination weight range; screening target mapping processing results from the mapping processing results corresponding to each discrimination weight range; based on the target mapping processing results, the algorithm quality of the fire IoT sensor monitoring algorithm is detected to obtain an algorithm quality detection point of view.

[0305] Furthermore, in the in-depth development of the fire IoT sensor monitoring system, accurate evaluation of algorithm quality is a key link to ensure system performance. This embodiment focuses on the core indicator of device superposition ratio to conduct a comprehensive and in-depth quality test of the fire IoT sensor monitoring algorithm.

[0306] On the one hand, this embodiment mainly covers the following core steps:

[0307] 1) Mapping of device overlay ratios: For each discrimination weight range, the device overlay ratios of each set state label within the range need to be calculated first. Subsequently, these device overlay ratios will undergo a mapping process to convert them into numerical or graphical representations that are easier to compare, analyze, and understand. Specific methods of mapping may include linear transformation, logarithmic transformation, normalization, etc., to ensure that the processed data can accurately reflect the performance of the algorithm within different discrimination weight ranges;

[0308] 2) Screening of target mapping processing results: After completing the mapping process, it is necessary to screen out the target mapping processing results from the mapping processing results corresponding to each discrimination weight range. The target mapping processing results can refer to those that can represent the key indicators of algorithm performance or have special significance. The screening process may be based on preset thresholds, rankings, cluster analysis and other methods to ensure that the selected results can fully and accurately reflect the actual performance of the algorithm;

[0309] 3) Detection and evaluation of algorithm quality: Based on the selected target mapping processing results, the algorithm quality of the fire IoT sensor monitoring algorithm can be detected and evaluated. This process may involve comprehensive consideration of multiple aspects such as algorithm accuracy, stability, and robustness. By comparing with the preset quality standards or historical data, the algorithm quality detection point of view can be obtained, providing strong support for further optimization and improvement of the algorithm.

[0310] On the other hand, the core of this embodiment is to link the actual performance of the fire IoT sensor monitoring algorithm with the preset quality standard or expected performance through the bridge of equipment superposition ratio. Specifically:

[0311] 1) The significance of the equipment superposition ratio: The equipment superposition ratio is an important indicator that reflects the performance of the algorithm within different discrimination weight ranges. It comprehensively considers the algorithm's ability to identify various fire trend elements, response speed, stability and other aspects. By calculating and analyzing the equipment superposition ratio, we can have a more comprehensive understanding of the performance of the algorithm in actual applications, providing a strong basis for subsequent optimization and improvement.

[0312] 2) The role of mapping: Mapping is a key step to convert the device stacking ratio into a form that is easier to analyze and understand. Through mapping, the original device stacking ratio data can be converted into numerical or graphical representations with clear physical or statistical meanings. This helps to more intuitively understand the performance differences of the algorithm within different discrimination weight ranges and the impact of these differences on the overall performance of the algorithm.

[0313] 3) Screening and evaluation of target mapping processing results: Screening of target mapping processing results is an important step to ensure the accuracy and representativeness of the evaluation results. By screening out representative or special target mapping processing results, the performance of the algorithm can be evaluated more accurately. At the same time, by comparing with the preset quality standards or historical data, the algorithm quality detection point of view can be obtained, providing a clear direction and motivation for the continuous improvement and optimization of the algorithm.

[0314] For example, there is a fire IoT sensor monitoring system that contains 1,000 verified fire terminal devices. These devices transmit interactive information through LoRa, providing quantitative description values ​​corresponding to target fire trend elements (such as smoke concentration, temperature, etc.). At the same time, the authentication labels of the monitoring action response status of these devices are also known (such as "normal", "warning", "fire", etc.).

[0315] 1) Calculation of equipment superposition ratio: First, according to the preset discrimination weight range (such as [0, 0.2], [0.2, 0.4], [0.4, 0.6], [0.6, 0.8], [0.8, 1.0]) and the set state label (such as "normal", "warning", "fire"), the equipment superposition ratio of each set state label in each discrimination weight range is calculated. In the discrimination weight range of [0.2, 0.4], the equipment superposition ratio of the "normal" state label is 0.6, the equipment superposition ratio of the "warning" state label is 0.3, and the equipment superposition ratio of the "fire" state label is 0.1.

[0316] 2) Mapping of equipment overlay ratio: Next, these equipment overlay ratios are mapped. To simplify the description, the linear transformation method is used to map the equipment overlay ratio to the range of [0, 100]. After the mapping process, the equipment overlay ratio of the "normal" state label becomes 60, the equipment overlay ratio of the "warning" state label becomes 30, and the equipment overlay ratio of the "fire" state label becomes 10.

[0317] 3) Screening of target mapping processing results: After completing the mapping processing, the target mapping processing results are screened out from the mapping processing results corresponding to each discrimination weight range. The recognition ability of the attention algorithm for the "warning" state label is set, so the mapping processing results of the "warning" state label within the [0.2, 0.4] discrimination weight range (i.e. 30) are used as the target mapping processing results.

[0318] 4) Detection and evaluation of algorithm quality: Finally, the algorithm quality of the fire IoT sensor monitoring algorithm is detected and evaluated based on the selected target mapping processing results. The preset quality standard is set as follows: within the [0.2, 0.4] discrimination weight range, the device superposition ratio of the "warning" status label should not be less than 40 (the value after mapping processing). In this case, the actual performance of the algorithm (i.e., 30) is lower than the preset quality standard (i.e., 40), so it is believed that the performance of the algorithm within this discrimination weight range needs to be improved.

[0319] In summary, by introducing the core indicator of equipment overlay ratio and combining mapping processing, screening of target mapping processing results, and detection and evaluation of algorithm quality, a comprehensive and in-depth method is provided for the quality evaluation of fire IoT sensor monitoring algorithms.

[0320] Further, Figure 2 FIG. 2 is a schematic diagram of a fire monitoring system 200 provided in an embodiment of the present application. Figure 2 The fire monitoring system 200 shown includes a processor 210, and the processor 210 can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0321] Alternatively, if Figure 2 As shown, the fire monitoring system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.

[0322] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0323] Alternatively, if Figure 2 As shown, the fire monitoring system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0324] Optionally, the fire monitoring system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application, which will not be described in detail here for the sake of brevity.

[0325] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities.

[0326] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0327] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0328] It should be noted that, in this article, the term "includes", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be noted that the scope of the method and device in the embodiment of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0329] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), including a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the embodiment of the present application.

[0330] The embodiments of the embodiments of the present application are described above in conjunction with the accompanying drawings, but the embodiments of the present application are not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the embodiments of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the embodiments of the present application and the scope of protection of the embodiments of the present application, all of which are within the protection of the embodiments of the present application.

Claims

1. A fire monitoring method combining lora wireless transmission and neural network algorithm, It is characterized in that The method is performed by a fire monitoring system, wherein a fire IoT sensor monitoring algorithm is configured in the fire monitoring system, and the fire IoT sensor monitoring algorithm includes a sensor information analysis network corresponding to each target fire trend element; The sensor information analysis network corresponding to the target fire situation trend element is used to output a local fire hazard discrimination result of the fire terminal equipment, and the local fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal equipment under the target fire situation trend element belongs to the set state label; the method includes: Obtain the lora transmission interaction information corresponding to the fire terminal equipment to be monitored, wherein the lora transmission interaction information includes the quantitative description value corresponding to the target fire trend element; For each target fire trend element of the fire terminal equipment to be monitored, the local fire hazard identification result of the fire terminal equipment to be monitored is determined according to the quantitative description value of the target fire trend element through the sensor information analysis network corresponding to the target fire trend element; Determine the target fire hazard discrimination result of the fire terminal equipment to be monitored based on the local fire hazard discrimination result determined by the sensor information analysis network corresponding to each target fire trend element; Determine the label corresponding to the monitoring action response state of the fire terminal equipment to be monitored according to the target fire hazard identification result; The determination steps of the fire protection IoT sensor monitoring algorithm include: Obtain lora transmission interaction information corresponding to at least one reference fire terminal device and an authentication tag of a monitoring action response state of the reference fire terminal device, wherein the lora transmission interaction information includes a quantitative description value corresponding to at least one reference fire trend element; According to the quantitative description value of the reference fire terminal equipment under the reference fire trend element, determine the proportion of fire terminal equipment with set state labels corresponding to each quantitative description range of the reference fire trend element to obtain the confidence evaluation degree corresponding to each quantitative description range; Based on the confidence evaluations corresponding to the respective quantitative description ranges of the reference fire trend elements, selecting at least one target fire trend element from the reference fire trend elements; According to the authentication tag of the monitoring action response state of the reference fire terminal equipment, a relationship statistical evaluation is performed on each target fire trend element to obtain an attention value corresponding to each target fire trend element; Based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, the fire Internet of Things sensor monitoring algorithm for the fire terminal equipment is generated.

2. The method according to claim 1, It is characterized in that The method of determining the local fire hazard identification result of the fire terminal equipment to be monitored based on the quantitative description value of the target fire trend element through the sensor information analysis network corresponding to the target fire trend element includes: Determining, through the sensor information analysis network corresponding to the target fire trend element, from each quantitative description range of the target fire trend element, the target quantitative description range to which the quantitative description value of the target fire trend element belongs; According to the confidence evaluation degree corresponding to the target quantitative description range, the local fire hazard identification result of the fire terminal equipment to be monitored is determined.

3. The method according to claim 1, It is characterized in that After determining the label corresponding to the monitoring action response state of the fire terminal device to be monitored according to the target fire hazard identification result, the method further includes: If the tag corresponding to the monitoring action response state of the fire terminal device to be monitored belongs to the target state tag, determine the target warning situation keyword of the fire terminal device to be monitored according to the target fire hazard identification result; According to the fire emergency response strategy corresponding to the target warning situation keyword, a fire emergency response is performed on the monitoring action response status of the fire terminal equipment to be monitored.

4. The method according to claim 1, It is characterized in that The method of determining the proportion of fire terminal devices with set status labels corresponding to each quantitative description range of the reference fire trend element according to the quantitative description value of the reference fire terminal device under the reference fire trend element, so as to obtain the confidence evaluation degree corresponding to each quantitative description range, includes: According to the quantitative description value of the reference fire terminal equipment under the reference fire trend element and at least one quantitative description range of the reference fire trend element, determine the number of fire terminal equipment whose labels of the monitoring action response state belong to the set state label, and obtain the proportion of fire terminal equipment with the set state label corresponding to each quantitative description range of the reference fire trend element; For each quantitative description range, mapping processing is performed on the proportion of fire terminal equipment of each set state label corresponding to the quantitative description range to obtain the confidence evaluation degree corresponding to the quantitative description range.

5. The method according to claim 4, It is characterized in that The method of determining the number of fire terminal devices whose labels of monitoring action response states belong to set state labels based on the quantitative description value of the reference fire terminal device under the reference fire trend element and at least one quantitative description range of the reference fire trend element, and obtaining the proportion of fire terminal devices with set state labels corresponding to each quantitative description range of the reference fire trend element, includes: Record the reference fire terminal equipment whose tag of monitoring action response status belongs to the set status tag to obtain the first number; For each quantitative description range of the reference fire situation trend element, count the reference fire terminal equipment whose quantitative description value of the reference fire situation trend element is within the quantitative description range, and obtain the target fire terminal equipment corresponding to the quantitative description range; Determine that the tag of the monitoring action response state in the target fire terminal device belongs to the second number of the set state tag; Based on the first number and the second number, the proportion of fire terminal equipment with set status labels corresponding to each quantitative description range of the reference fire trend element is determined.

6. The method according to claim 1, It is characterized in that The step of selecting at least one target fire trend element from the reference fire trend elements based on the confidence evaluation corresponding to each quantitative description range of the reference fire trend element comprises: Filtering original fire trend elements from the reference fire trend elements according to the confidence evaluations corresponding to the respective quantitative description ranges of the reference fire trend elements; Determine the spatiotemporal impact characteristics of the original fire trend elements according to the proportion of fire terminal equipment with set state labels corresponding to each quantitative description range of the original fire trend elements and the confidence evaluation degree of each quantitative description range, and obtain the spatiotemporal impact feature set corresponding to the original fire trend elements; Based on the spatiotemporal impact feature set, at least one target fire trend element is screened from the original fire trend elements.

7. The method according to claim 6, It is characterized in that The method of determining the spatiotemporal impact characteristics of the original fire situation trend elements according to the proportion of fire terminal equipment with set state labels corresponding to each quantitative description range of the original fire situation trend elements and the confidence evaluation degree of each quantitative description range, and obtaining the spatiotemporal impact feature set corresponding to the original fire situation trend elements, includes: For each quantitative description range of the original fire situation trend element, the proportion of fire terminal equipment of each set state label corresponding to the quantitative description range and the confidence evaluation degree of the quantitative description range are combined based on the time and space domain to obtain a sub-time and space influence feature set corresponding to the quantitative description range; The sub-spatiotemporal impact feature sets corresponding to each quantitative description range of the original fire trend element are combined based on the spatiotemporal domain to obtain the spatiotemporal impact feature set corresponding to the original fire trend element.

8. The method according to claim 1, It is characterized in that The generating of a fire IoT sensor monitoring algorithm for fire terminal equipment based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element includes: Based on the attention value corresponding to the target fire trend element, a decision tree network is generated between the fire trend element and the cross possibility, wherein the cross possibility is the cross possibility that the label of the monitoring action response state of the fire terminal equipment belongs to the set state label; Generate a firefighting IoT sensor monitoring algorithm for firefighting terminal equipment based on the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element and the decision tree network; The method of generating a fire-fighting IoT sensor monitoring algorithm for fire-fighting terminal equipment based on the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element and the decision tree network includes: Based on the decision tree network and the set algorithm construction index, an initial fire hazard discrimination weight for the fire terminal equipment is obtained; According to the initial fire hazard discrimination weight, the attention value corresponding to the target fire trend element, and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, a fire IoT sensor monitoring algorithm is generated, wherein the fire IoT sensor monitoring algorithm includes a characteristic association diagram between the target fire hazard discrimination result of the fire terminal equipment and the quantitative description value of the target fire trend element, and the target fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal equipment belongs to the set state label; The method of generating a fire protection IoT sensor monitoring algorithm based on the initial fire hazard discrimination weight, the attention value corresponding to the target fire trend element, and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element includes: For each target fire trend element, based on the attention value corresponding to the target fire trend element and the confidence evaluation degree corresponding to each quantitative description range of the target fire trend element, a sensor information analysis network for the target fire trend element is generated, and the sensor information analysis network is used to output a local fire hazard discrimination result of the fire terminal equipment, and the local fire hazard discrimination result represents the possibility that the label of the monitoring action response state of the fire terminal equipment under the target fire trend element belongs to the set state label; According to the initial fire hazard discrimination weights and the sensor information analysis network corresponding to each target fire trend element, a fire protection Internet of Things sensor monitoring algorithm is obtained.

9. A fire monitoring system, It is characterized in that The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 8.

Images