Electrical fire prediction method and system based on Internet of Things

Abstract

The invention discloses an electrical fire prediction method and system based on the Internet of Things, and relates to the technical field of intelligent prediction, and the method comprises the following steps: setting a first parameter and a second parameter, fitting to obtain a first related equation, calculating an actual historical residual current, training a fire prediction model, and predicting a first probability. According to the method, the potential risk of the electrical fire can be recognized in advance by analyzing the historical residual current and the related parameters and combining the fire prediction model, the input data quality of the prediction model is improved by utilizing historical operation data and related coefficient analysis, the first related equation is fitted through the least square method, the calculation precision of the historical residual current is optimized, and the calculation accuracy of the electrical fire is improved. On the basis of the Internet of Things technology, operation data of the electrical equipment can be collected in real time and transmitted to the cloud or the local server for analysis, and the possibility of misjudgment of a single parameter is reduced by analyzing multiple parameters and relevance among the parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent prediction, and in particular to an electrical fire prediction method and system based on the Internet of Things. Background Art

[0002] In recent years, with the continuous development of technologies such as the Internet of Things, big data, and artificial intelligence, the scale of the smart fire protection market has expanded rapidly. Among them, electrical fire prediction technology, as an important part of smart fire protection, has also been widely used. Electrical fire prediction technology will develop in a more intelligent and automated direction. It can not only automatically collect and analyze data, but also automatically take corresponding measures based on the prediction results, such as automatically cutting off the power supply and starting fire extinguishing equipment.

[0003] At present, in a Chinese invention patent with publication number CN117172369A, an electrical fire prediction method, system, device and storage medium are disclosed. The method scores the safety level of the current building through the line information and the load information to obtain a first score; obtains historical fire data and weather data of the current building, wherein the historical fire data includes the number of fires and the causes of the fires; scores the current building according to the number of fires and the weather data to obtain a second score; combines the first score and the second score to determine the probability of electrical fires in the current building to obtain the risk probability; optimizes the risk probability according to the cause of the fire to obtain the final risk probability, but the related technology does not fuse and predict based on multiple sensor data at the front end of the Internet of Things, lacks the comprehensiveness of fire prediction, and does not compensate for the situation where the residual current changes due to changes in the load, etc., which is not conducive to the accuracy of the data source and the accuracy of the prediction. Summary of the invention

[0004] The technical problem solved by the present invention is that the related technology does not fuse and predict based on multiple sensor data at the front end of the Internet of Things, lacks comprehensiveness in fire prediction, and does not compensate for the situation where the residual current changes due to current changes caused by changes in load, etc., which is not conducive to the accuracy of the data source and the accuracy of the prediction.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, an electrical fire prediction method based on the Internet of Things comprises the following steps: Step S100, setting the historical residual current as the first parameter, calculating the correlation coefficient between the historical residual current and various parameters according to the historical operation data, and setting the second parameter according to the correlation coefficient; Step S200, fitting the first parameter and the second parameter according to the least square method to obtain a first correlation equation, and calculating the actual historical residual current according to the first correlation equation; Step S300, obtaining historical fire categories according to historical fire conditions, training a fire prediction model according to historical environmental data, actual historical residual current and historical fire categories, and obtaining first probabilities of various fire categories caused by historical environmental data and actual historical residual current according to the fire prediction model.

[0006] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, wherein: the historical operation data includes historical three-phase current and historical three-phase voltage, or historical single-phase current and historical single-phase voltage; The calculation method of correlation coefficient includes: Calculate the first average value of the historical residual current, calculate the second average value of any type of historical operating data, construct a correlation coefficient calculation expression, bring the first average value, the second average value, the historical residual current and the corresponding historical operating data into the correlation coefficient calculation expression, and obtain the correlation coefficient between the historical operating data and the historical residual current.

[0007] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, the calculation expression of the correlation coefficient is: ; Among them, r is the correlation coefficient, x i is the i-th historical residual current, y i is any one of the i-th historical operation data, n is the number of historical residual currents, x is the first average value, and y is the second average value.

[0008] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, the setting logic of the second parameter includes: The first value is set as the correlation coefficient threshold, and the first value is compared with each correlation coefficient. When the correlation coefficient is greater than or equal to the first value, the corresponding historical operation data is set to the second parameter. When the correlation coefficient is less than the first value, the corresponding historical operation data is deleted and jump to the next historical operation data.

[0009] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, the calculation expression of the first correlation equation is: x i =F(y i ); Wherein, F is the operation rule for converting the second parameter into the first parameter.

[0010] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, the fitting method for the first correlation equation includes: Calculate the first error between the value of the dependent variable of the first association equation and the corresponding historical residual current value, traverse each first error, calculate the sum of the squares of the first errors, adjust the value of the coefficient in the operation rule so that the value of the sum of the squares of the first errors is minimized, and set the coefficient at this time as the final coefficient of the first association equation.

[0011] As a preferred solution of the electrical fire prediction method based on the Internet of Things described in the present invention, the fitting method for the first correlation equation includes: Calculate the first error between the value of the dependent variable of the first association equation and the corresponding historical residual current value, traverse each first error, calculate the sum of the squares of the first errors, adjust the value of the coefficient in the operation rule so that the value of the sum of the squares of the first errors is minimized, and set the coefficient at this time as the final coefficient of the first association equation.

[0012] As a preferred embodiment of the electrical fire prediction method based on the Internet of Things described in the present invention, the smoldering fire setting method includes: Retrieving infrared temperature data from the front end of the Internet of Things, converting the infrared temperature data into an infrared temperature image, extracting the grayscale value of each pixel of the infrared temperature image, setting the second value and the third value as the grayscale value threshold, and comparing the grayscale value of each pixel with the grayscale value threshold respectively; When there are pixels whose grayscale values ​​are distributed between the second value and the third value, the fire category is set to smoldering fire; when the grayscale values ​​of all pixels are not distributed between the second value and the third value, the method of setting an open fire is jumped; Open flame setup methods include: Retrieving a monitoring image of the industrial camera, extracting each RGB value of the monitoring image, setting the fifth value and the sixth value as the RGB value threshold, and comparing the RGB value of each pixel with the RGB value threshold respectively; When there are pixels whose RGB values ​​are distributed between the fourth value and the fifth value, the fire category is set to open fire; when all the pixels whose RGB values ​​are not distributed between the fourth value and the fifth value, the method of setting no fire is jumped; The no-fire setting method includes: when jumping to the no-fire setting method, setting the fire category to no-fire and ending the fire category setting.

[0013] As a preferred embodiment of the electrical fire prediction method based on the Internet of Things described in the present invention, the historical environmental data includes carbon monoxide concentration, ambient temperature and smoke concentration; The training methods of the fire prediction model include: Setting neural network parameters, including the number of neural network layers, the number of neurons in the input layer, the number of neurons in the hidden layer, the number of neurons in the output layer, the number of iterations, the activation function, and the minimum error of the training target. The number of neural network layers is set to 3, the number of neurons in the input layer is set to 4, the number of neurons in the output layer is set to 4, the number of neurons in the hidden layer is set to 4, the number of iterations is set to 10000, the activation function is set to the sigmond function, and the minimum error of the training target is set to 0.02; Taking historical environmental data and actual historical residual current as input and historical fire categories as output, a neural network is trained to obtain a fire prediction model. By inputting environmental data and residual current into the fire prediction model, a first probability of each fire type is obtained, and the first probability includes a smoldering fire probability, an open flame probability and a no fire probability.

[0014] Second, the electrical fire prediction system based on the Internet of Things includes a calculation module, a construction module and a prediction module; The calculation module sets the historical residual current as a first parameter, calculates the correlation coefficient between the historical residual current and various parameters according to the historical operation data, and sets the second parameter according to the correlation coefficient; The construction module fits the first parameter and the second parameter according to the least square method to obtain a first correlation equation, and calculates the actual historical residual current according to the first correlation equation; The prediction module classifies fires according to historical fire conditions, trains a fire prediction model according to historical environmental data, actual historical residual current and historical fire categories, and obtains first probabilities of various fire categories caused by historical environmental data and actual historical residual current according to the fire prediction model.

[0015] The beneficial effects of the present invention are as follows: by analyzing historical residual current and related parameters and combining with a fire prediction model, the potential risk of electrical fires can be identified in advance, an early warning can be issued before a fire occurs, and time can be gained for taking preventive measures. Historical operating data and correlation coefficient analysis are used to screen out parameters with a high correlation with fire as the second parameter, thereby improving the input data quality of the prediction model. The calculation accuracy of the historical residual current is further optimized by fitting the first correlation equation using the least squares method. Model training is performed in combination with historical environmental data and fire categories, which can more accurately predict the probability of fire under different environmental conditions. Based on the Internet of Things technology, the operating data of electrical equipment can be collected in real time and transmitted to the cloud or local server for analysis. By analyzing multiple parameters and the correlation between them, the possibility of misjudgment of a single parameter is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1A schematic diagram of the basic flow of an electrical fire prediction method based on the Internet of Things provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0018] Example, see Figure 1 , as an embodiment of the present invention, provides an electrical fire prediction method based on the Internet of Things, comprising the following steps: Step S100, setting the historical residual current as the first parameter, calculating the correlation coefficient between the historical residual current and various parameters according to the historical operation data, and setting the second parameter according to the correlation coefficient; Step S200, fitting the first parameter and the second parameter according to the least square method to obtain a first correlation equation, and calculating the actual historical residual current according to the first correlation equation; Step S300, obtaining historical fire categories according to historical fire conditions, training a fire prediction model according to historical environmental data, actual historical residual current and historical fire categories, and obtaining first probabilities of various fire categories caused by historical environmental data and actual historical residual current according to the fire prediction model.

[0019] The present invention analyzes historical residual current and related parameters and combines them with a fire prediction model to identify the potential risks of electrical fires in advance, issue a warning before a fire occurs, and buy time for taking preventive measures. Historical operating data and correlation coefficient analysis are used to screen out parameters with a high correlation with fire as the second parameter, thereby improving the input data quality of the prediction model. The first correlation equation is fitted by the least squares method to further optimize the calculation accuracy of the historical residual current. Model training is performed in combination with historical environmental data and fire categories, so that the probability of fire under different environmental conditions can be more accurately predicted. Based on the Internet of Things technology, the operating data of electrical equipment can be collected in real time and transmitted to the cloud or local server for analysis. By analyzing multiple parameters and the correlation between them, the possibility of misjudgment of a single parameter is reduced.

[0020] The historical operation data includes the historical three-phase current and the historical three-phase voltage, or the historical single-phase current and the historical single-phase voltage; The calculation method of correlation coefficient includes: Calculate the first average value of the historical residual current, calculate the second average value of any type of historical operating data, construct a correlation coefficient calculation expression, bring the first average value, the second average value, the historical residual current and the corresponding historical operating data into the correlation coefficient calculation expression, and obtain the correlation coefficient between the historical operating data and the historical residual current.

[0021] The calculation expression of the correlation coefficient is: ; Among them, r is the correlation coefficient, x i is the i-th historical residual current, y i is any one of the i-th historical operation data, n is the number of historical residual currents, x is the first average value, and y is the second average value.

[0022] In the specific implementation, by calculating the correlation coefficient, it is clear which operating parameters (such as three-phase current and three-phase voltage) have a significant correlation with the historical residual current. This helps to screen out the key parameters that are really helpful for fire prediction from many parameters, avoid incorporating parameters that have little to do with the occurrence of fire into the model, thereby improving the efficiency and accuracy of the model. Parameters with high correlation coefficients are usually more valuable for fire prediction. By screening these parameters as model inputs, the prediction ability of the model is enhanced, enabling it to more accurately identify fire risks, reducing unnecessary parameter inputs, reducing the computational complexity in the model training and prediction process, and improving the response speed of the system. The calculation of the correlation coefficient helps understand the causal relationship between the historical residual current and the operating parameters. For example, if the correlation coefficient between a phase current and the residual current is very high, it means that the change of the phase current may have a significant impact on the residual current, thereby providing a clearer basis for fire prediction. As the operating environment and equipment status of the electrical system change, the correlation between certain parameters and the residual current may change. By regularly calculating the correlation coefficient, the model input parameters are dynamically adjusted to keep the model at its best performance.

[0023] The setting logic of the second parameter includes: The first value is set as the correlation coefficient threshold, and the first value is compared with each correlation coefficient. When the correlation coefficient is greater than or equal to the first value, the corresponding historical operation data is set to the second parameter. When the correlation coefficient is less than the first value, the corresponding historical operation data is deleted and jump to the next historical operation data.

[0024] In the specific implementation, by setting the correlation coefficient threshold, only the historical operating data with a strong correlation with the historical residual current is retained as the second parameter. These data contribute more to fire prediction, thereby improving the prediction accuracy of the model. The second parameter screened by the correlation coefficient threshold can more clearly reflect which operating data is directly related to the fire risk. For example, if the correlation coefficient between a phase current and the residual current is higher than the threshold, it means that the change in the phase current may be closely related to the fire risk. This method makes the input parameters of the model more logical and explainable, making it easier for technicians and managers to understand the decision-making basis of the model.

[0025] The calculation expression of the first correlation equation is: x i =F(y i ); Wherein, F is the operation rule for converting the second parameter into the first parameter.

[0026] The fitting method for the first correlation equation includes: Calculate the first error between the value of the dependent variable of the first association equation and the corresponding historical residual current value, traverse each first error, calculate the sum of the squares of the first errors, adjust the value of the coefficient in the operation rule so that the value of the sum of the squares of the first errors is minimized, and set the coefficient at this time as the final coefficient of the first association equation.

[0027] In specific implementation, by adjusting the coefficients to minimize the sum of squared errors, it is possible to ensure that the first correlation equation is as close as possible to the actual historical residual current value, thereby improving the accuracy and reliability of the fitting. This method can automatically find the optimal coefficient combination so that the model can better reflect the laws of historical data and provide more accurate input for subsequent fire predictions. The least squares fitting method can effectively reduce the deviation between the model predicted value and the actual value, and improve the consistency of the model's performance on different data sets. By minimizing the sum of squared errors, the model has stronger robustness to data fluctuations and outliers, and thus performs more stably in practical applications. At the same time, it eliminates the interference of false residual current (such as the current caused by load input) on the real residual current, thereby improving the accuracy of the data source. Compared with other complex optimization methods, the calculation process of the least squares method is relatively simple and suitable for processing large-scale data.

[0028] Historical fire categories include smoldering fire, open fire, and no fire; The setting methods of historical fire categories include smoldering fire setting method, open fire setting method and no fire setting method; The historical fire categories are set in the order of smoldering fire, open fire, and no fire.

[0029] Smoldering fire setting methods include: Retrieving infrared temperature data from the front end of the Internet of Things, converting the infrared temperature data into an infrared temperature image, extracting the grayscale value of each pixel of the infrared temperature image, setting the second value and the third value as the grayscale value threshold, and comparing the grayscale value of each pixel with the grayscale value threshold respectively; When there are pixels whose grayscale values ​​are distributed between the second value and the third value, the fire category is set to smoldering fire; when the grayscale values ​​of all pixels are not distributed between the second value and the third value, the method of setting an open fire is jumped; Open flame setup methods include: Retrieving a monitoring image of the industrial camera, extracting each RGB value of the monitoring image, setting the fifth value and the sixth value as the RGB value threshold, and comparing the RGB value of each pixel with the RGB value threshold respectively; When there are pixels whose RGB values ​​are distributed between the fourth value and the fifth value, the fire category is set to open fire; when all the pixels whose RGB values ​​are not distributed between the fourth value and the fifth value, the method of setting no fire is jumped; The no-fire setting method includes: when jumping to the no-fire setting method, setting the fire category to no-fire and ending the fire category setting.

[0030] In the specific implementation, by retrieving infrared temperature data and converting it into an infrared temperature image, extracting the gray value of the pixel point, and comparing it with the preset gray value threshold, the smoldering fire can be accurately identified. This method effectively distinguishes smoldering fire from open fire and avoids misjudgment. Smoldering fire usually occurs in the early stage of a fire, with a low burning temperature and a slow spread. Through infrared temperature image monitoring, smoldering fire can be discovered in time in the early stage of a fire, which buys time for taking preventive measures. Early detection of smoldering fire prolongs the warning time and provides more time for personnel evacuation and fire fighting. The monitoring method of infrared temperature image and gray value threshold is combined with other fire monitoring parameters (such as smoke concentration, temperature, etc.) to form a multi-parameter comprehensive monitoring system. This makes fire monitoring more flexible and can adapt to different fire scenes. Through the Internet of Things technology, infrared temperature data is transmitted to the monitoring system in real time to realize automatic monitoring and early warning. This method reduces manual intervention and improves monitoring efficiency.

[0031] Historical environmental data include carbon monoxide concentrations, ambient temperature, and smoke concentrations; The training methods of the fire prediction model include: Set the neural network parameters, which include the number of neural network layers, the number of neurons in the input layer, the number of neurons in the hidden layer, the number of neurons in the output layer, the number of iterations, the activation function, and the minimum error of the training target. Set the number of neural network layers to 3, the number of neurons in the input layer to 4, the number of neurons in the output layer to 4, the number of neurons in the hidden layer to 4, the number of iterations to 10000, the activation function to the sigmond function, and the minimum error of the training target to 0.02; Taking historical environmental data and actual historical residual current as input and historical fire categories as output, the neural network is trained to obtain a fire prediction model. By inputting environmental data and residual current into the fire prediction model, the first probability of each fire type is obtained, which includes the probability of smoldering fire, the probability of open flame and the probability of no fire.

[0032] In the specific implementation, historical environmental data (carbon monoxide concentration, ambient temperature, smoke concentration) and historical residual current are used as input, which can more comprehensively reflect the various characteristics before the fire occurs. This multi-dimensional data input can improve the model's ability to identify fire types. Through the training of the neural network, the model can learn the complex relationship between different fire types (smoldering fire, open fire, no fire) and input data, so as to more accurately predict the fire type. Setting a high number of iterations (10,000 times) and a reasonable training target minimum error (0.02) can ensure that the model is fully trained on a large amount of historical data, thereby improving the generalization ability of the model. The fully trained neural network model can adapt to different environmental conditions and fire scenes, and has strong robustness. By comprehensively considering multiple environmental parameters and electrical parameters, it reduces false alarms caused by single parameter abnormalities.

[0033] The present invention analyzes historical residual current and related parameters and combines them with a fire prediction model to identify the potential risks of electrical fires in advance, issue a warning before a fire occurs, and buy time for taking preventive measures. Historical operating data and correlation coefficient analysis are used to screen out parameters with a high correlation with fire as the second parameter, thereby improving the input data quality of the prediction model. The first correlation equation is fitted by the least squares method to further optimize the calculation accuracy of the historical residual current. Model training is performed in combination with historical environmental data and fire categories, so that the probability of fire under different environmental conditions can be more accurately predicted. Based on the Internet of Things technology, the operating data of electrical equipment can be collected in real time and transmitted to the cloud or local server for analysis. By analyzing multiple parameters and the correlation between them, the possibility of misjudgment of a single parameter is reduced.

[0034] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Wherein, the storage medium is implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0035] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions of the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention should be included in the scope of the claims of the present invention.

Claims

1. An electrical fire prediction method based on the Internet of Things, characterized in that: The following steps are involved: Step S100, setting the historical residual current as the first parameter, calculating the correlation coefficient between the historical residual current and various parameters according to the historical operation data, and setting the second parameter according to the correlation coefficient; Step S200, fitting the first parameter and the second parameter according to the least square method to obtain a first correlation equation, and calculating the actual historical residual current according to the first correlation equation; Step S300, obtaining historical fire categories according to historical fire conditions, training a fire prediction model according to historical environmental data, actual historical residual current and historical fire categories, and obtaining first probabilities of various fire categories caused by historical environmental data and actual historical residual current according to the fire prediction model.

2. The electrical fire prediction method based on the Internet of Things according to claim 1, characterized in that: The historical operation data includes the historical three-phase current and the historical three-phase voltage, or the historical single-phase current and the historical single-phase voltage; The calculation method of correlation coefficient includes: Calculate the first average value of the historical residual current, calculate the second average value of any type of historical operating data, construct a correlation coefficient calculation expression, bring the first average value, the second average value, the historical residual current and the corresponding historical operating data into the correlation coefficient calculation expression, and obtain the correlation coefficient between the historical operating data and the historical residual current.

3. The electrical fire prediction method based on the Internet of Things as claimed in claim 2, characterized in that: The calculation expression of the correlation coefficient is: ； Among them, r is the correlation coefficient, x i is the i-th historical residual current, y i is any one of the i-th historical operation data, n is the number of historical residual currents, x is the first average value, and y is the second average value.

4. The electrical fire prediction method based on the Internet of Things according to claim 1, characterized in that: The setting logic of the second parameter includes: The first value is set as the correlation coefficient threshold, and the first value is compared with each correlation coefficient. When the correlation coefficient is greater than or equal to the first value, the corresponding historical operation data is set to the second parameter. When the correlation coefficient is less than the first value, the corresponding historical operation data is deleted and jump to the next historical operation data.

5. The electrical fire prediction method based on the Internet of Things according to claim 1, characterized in that: The calculation expression of the first correlation equation is: x i =F(y i ); Wherein, F is the operation rule for converting the second parameter into the first parameter.

6. The electrical fire prediction method based on the Internet of Things according to claim 5, characterized in that: The fitting method for the first correlation equation includes: Calculate the first error between the value of the dependent variable of the first association equation and the corresponding historical residual current value, traverse each first error, calculate the sum of the squares of the first errors, adjust the value of the coefficient in the operation rule so that the value of the sum of the squares of the first errors is minimized, and set the coefficient at this time as the final coefficient of the first association equation.

7. The electrical fire prediction method based on the Internet of Things according to claim 1, characterized in that: Historical fire categories include smoldering fire, open fire, and no fire; The setting methods of historical fire categories include smoldering fire setting method, open fire setting method and no fire setting method; The historical fire categories are arranged in the order of smoldering fire, open fire and no fire.

8. The electrical fire prediction method based on the Internet of Things according to claim 7, characterized in that: Smoldering fire setting methods include: Retrieving infrared temperature data from the front end of the Internet of Things, converting the infrared temperature data into an infrared temperature image, extracting the grayscale value of each pixel of the infrared temperature image, setting the second value and the third value as the grayscale value threshold, and comparing the grayscale value of each pixel with the grayscale value threshold respectively; When there are pixels whose grayscale values ​​are distributed between the second value and the third value, the fire category is set to smoldering fire; when the grayscale values ​​of all pixels are not distributed between the second value and the third value, the method of setting an open fire is jumped; Open flame setup methods include: Retrieving a monitoring image of the industrial camera, extracting each RGB value of the monitoring image, setting the fifth value and the sixth value as the RGB value threshold, and comparing the RGB value of each pixel with the RGB value threshold respectively; When there are pixels whose RGB values ​​are distributed between the fourth value and the fifth value, the fire category is set to open fire; when all the pixels whose RGB values ​​are not distributed between the fourth value and the fifth value, the method of setting no fire is jumped; The no-fire setting method includes: when jumping to the no-fire setting method, setting the fire category to no-fire and ending the fire category setting.

9. The electrical fire prediction method based on the Internet of Things according to claim 1, characterized in that: Historical environmental data include carbon monoxide concentrations, ambient temperature, and smoke concentrations; The training methods of the fire prediction model include: Setting neural network parameters, including the number of neural network layers, the number of neurons in the input layer, the number of neurons in the hidden layer, the number of neurons in the output layer, the number of iterations, the activation function, and the minimum error of the training target. The number of neural network layers is set to 3, the number of neurons in the input layer is set to 4, the number of neurons in the output layer is set to 4, the number of neurons in the hidden layer is set to 4, the number of iterations is set to 10000, the activation function is set to the sigmond function, and the minimum error of the training target is set to 0.02; Taking historical environmental data and actual historical residual current as input and historical fire categories as output, a neural network is trained to obtain a fire prediction model. By inputting environmental data and residual current into the fire prediction model, a first probability of each fire type is obtained, and the first probability includes a smoldering fire probability, an open flame probability and a no fire probability.

10. An electrical fire prediction system based on the Internet of Things, the system being used to execute the electrical fire prediction method based on the Internet of Things as claimed in claim 1, characterized in that: It includes calculation module, construction module and prediction module; The calculation module sets the historical residual current as a first parameter, calculates the correlation coefficient between the historical residual current and various parameters according to the historical operation data, and sets the second parameter according to the correlation coefficient; The construction module fits the first parameter and the second parameter according to the least square method to obtain a first correlation equation, and calculates the actual historical residual current according to the first correlation equation; The prediction module classifies fires according to historical fire conditions, trains a fire prediction model according to historical environmental data, actual historical residual current and historical fire categories, and obtains first probabilities of various fire categories caused by historical environmental data and actual historical residual current according to the fire prediction model.

Images