An intelligent monitoring system for electrical fire warning

Abstract

The present invention provides an intelligent monitoring system for electrical fire warning, including: obtaining video images, grouping them into multiple sub-video images, calculating feature vectors in each video image by using the change situation of pixels in the sub-video images, and then training a support vector machine through the feature vectors to realize the prediction of fire situations. The beneficial effects of the present invention: realizing the detection of fire situations in target video images, thereby giving early warnings according to the fire situations, being able to detect fires without open flames, improving the recognition rate of fires, and protecting the property safety of other devices.

Description

Technical Field

[0001] The present invention relates to the field of fire warning, and particularly to an intelligent monitoring system for electrical fire warning. Background Art

[0002] With the continuous development of the economic society, electrical appliances have gradually become an important part of families and factories. Since fires caused by electrical appliances can pose a serious threat to people's lives and property to a certain extent, it is necessary to monitor electrical fires.

[0003] Currently, the intelligent monitoring methods for electrical fires mainly involve placing temperature sensors at key positions or detecting firelight through video. However, the location of the fire ignition is uncertain. By the time the firelight burns to a position where the temperature sensor can sense it, an irreparable fire may have already occurred. Moreover, the method of detecting firelight through video cannot detect fires without visible flames, posing a great potential safety hazard. Summary of the Invention

[0004] The main object of the present invention is to provide an intelligent monitoring system for electrical fire warning, aiming to solve the problem that fires without visible flames cannot be detected by the method of detecting firelight through video.

[0005] The present invention provides an intelligent monitoring system for electrical fire warning, comprising:

[0006] An acquisition module for acquiring video images of the areas where multiple devices are located; wherein, the video images include at least two different light environments, which alternately change in sequence at a preset interval time;

[0007] A division module for dividing the video images into multiple groups of sub-video images according to the preset interval time; wherein, each group of sub-video images includes video frames in all light environments;

[0008] An extraction module for extracting one video frame in each light environment of each group of sub-video images and performing fusion according to a preset data fusion method to obtain a fused data frame of each group of sub-video images;

[0009] A segmentation module for segmenting each of the fused data frames into a preset number of equally sized blocks;

[0010] A first calculation module for calculating the difference value of the blocks at the same position of two adjacent fused data frames in chronological order; wherein, the difference value is an absolute value;

[0011] A second calculation module for calculating the sum of the difference values of all blocks of two adjacent fused data frames to obtain the difference sum value of the two adjacent fused data frames;

[0012] A third calculation module, configured to calculate the feature vectors of each video image according to the difference sum value , where represents the feature vector of the i-th video image, represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point;

[0013] A corresponding module, configured to correspond the feature vectors of each video image to the corresponding fire situation of the video image one by one to form a data set;

[0014] An allocation module, configured to divide the data set into a training data set and a test data set;

[0015] A first input module, configured to input the training data set into a preset support vector machine, and train the preset support vector machine according to optimal hyperparameters to obtain a temporary model;

[0016] A detection module, configured to detect the temporary model through the test data set, and obtain a target model when the detection result meets the training requirements of the model;

[0017] An acquisition module, configured to acquire a target video image to be detected, and calculate a target feature vector corresponding to the target video image;

[0018] A second input module, configured to input the target feature vector into the target model to obtain the fire situation of the target video image.

[0019] Further, the intelligent monitoring system for electrical fire warning further includes:

[0020] A difference sum value detection module, configured to detect whether the values of each of the difference sum values are all less than a preset sum value;

[0021] A determination module, configured to, if so, determine to execute the step of calculating the feature vectors of each video image according to the difference sum value.

[0022] Further, the extraction module includes:

[0023] A video frame extraction sub-module, configured to extract one video frame in each light environment of each group of sub-video images,

[0024] A pixel value acquisition sub-module, configured to acquire the pixel value of each pixel point from each of the video frames;

[0025] A summation sub-module, configured to sum and average the pixel points in each group of sub-videos to obtain a fused data frame corresponding to each group of sub-videos.

[0026] Further, the intelligent monitoring system for electrical fire warning further includes:

[0027] An initial support vector machine acquisition module, configured to acquire initial support vector machines with multiple different hyperparameter combinations;

[0028] A prediction module, configured to input the training data set into each of the initial support vector machines for prediction to obtain respective corresponding prediction results;

[0029] A loss value calculation module, configured to calculate the loss value of each initial support vector machine according to the prediction results;

[0030] A support vector machine selection module, configured to select, based on the loss value, the initial support vector machine with the smallest loss value as the preset support vector machine.

[0031] Further, the intelligent monitoring system for electrical fire warning further includes:

[0032] A fire situation judgment module, configured to judge whether the fire situation has reached the alarm level;

[0033] A calling-for-help module, configured to, if so, call for help from the fire department.

[0034] The present invention further provides an intelligent monitoring method for electrical fire warning, including:

[0035] Collecting video images of areas where multiple devices are located; wherein, the video images include at least two different light environments, which alternately change in sequence at a preset interval time;

[0036] Dividing the video images into multiple groups of sub-video images according to the preset interval time; wherein, each group of sub-video images includes video frames under all light environments;

[0037] Extracting one video frame under each light environment in each group of sub-video images and performing fusion according to a preset data fusion method to obtain a fusion data frame of each group of sub-video images;

[0038] Dividing each of the fusion data frames into preset numbers of equally-sized blocks;

[0039] Calculating the difference value of blocks at the same position of adjacent two fusion data frames in chronological order; wherein, the difference value is an absolute value;

[0040] Calculating the sum value of the difference values of all blocks of adjacent two fusion data frames to obtain the difference sum value of the adjacent two fusion data frames;

[0041] Calculating the feature vector of each video image according to the difference sum value , wherein, denote the feature vector of the i-th video image denote the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point

[0042] One-to-one correspondence is established between the feature vectors of each video image and the fire situation corresponding to the video image to form a data set

[0043] Divide the data set into a training data set and a test data set

[0044] Input the training data set into a preset support vector machine, and train the preset support vector machine according to the optimal hyperparameters to obtain a temporary model

[0045] Detect the temporary model through the test data set, and when the detection result meets the training requirements of the model, obtain a target model

[0046] Obtain a target video image to be detected, and calculate the target feature vector corresponding to the target video image

[0047] Input the target feature vector into the target model to obtain the fire situation of the target video image

[0048] Further, before the step of calculating the feature vectors of each video image according to the difference sum value, the following steps are also included

[0049] Detect whether the values of each difference sum value are all less than a preset sum value

[0050] If so, it is determined to execute the step of calculating the feature vectors of each video image according to the difference sum value

[0051] Further, the step of extracting one video frame in each light environment of each group of sub-video images and performing fusion according to a preset data fusion method to obtain a fusion data frame of each group of sub-video images includes

[0052] Extract one video frame in each light environment of each group of sub-video images

[0053] Obtain the pixel value of each pixel point from each of the video frames

[0054] Sum and average the pixel points in each group of sub-videos to obtain a fusion data frame corresponding to each group of sub-videos

[0055] Further, before the step of inputting the training data set into a preset support vector machine and training the preset support vector machine according to the optimal hyperparameters to obtain a temporary model, the following steps are also included

[0056] Obtain initial support vector machines with multiple different combinations of hyperparameters;

[0057] Input the training data set into each of the initial support vector machines for prediction to obtain respective corresponding prediction results;

[0058] Calculate the loss value of each initial support vector machine according to the prediction results;

[0059] Select the initial support vector machine with the smallest loss value based on the loss value as the preset support vector machine.

[0060] Further, after the step of inputting the target feature vector into the target model to obtain the fire situation of the target video image, the following steps are further included:

[0061] Judge whether the fire situation has reached the alarm level;

[0062] If so, call for help from the fire department.

[0063] Advantages of the present invention: By acquiring video images and grouping them into multiple sub-video images, calculating the feature vectors in each video image using the change situation of pixels in the sub-video images, and then training a support vector machine through the feature vectors to achieve the prediction of the fire situation, thereby realizing the detection of the fire situation of the target video image, and then giving an early warning according to the fire situation, it is possible to detect a fire without open flames, greatly improving the recognition rate of fires and protecting the property safety of other devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a schematic block diagram of the structure of an intelligent monitoring system for electrical fire early warning according to an embodiment of the present invention;

[0065] Figure 2 is a schematic flowchart of an intelligent monitoring method for electrical fire early warning according to an embodiment of the present invention;

[0066] Figure 3 is a schematic block diagram of the structure of a computer device according to an embodiment of the present application.

[0067] The realization, functional characteristics and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0068] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0069] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly. The connection described may be a direct connection or an indirect connection.

[0070] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and B may represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0071] In addition, in the present invention, descriptions such as "first" and "second" are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions between various embodiments may be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0072] Referring to Figure 1 , the present invention proposes an intelligent monitoring system for electrical fire warning, including:

[0073] An acquisition module 10 for acquiring video images of areas where multiple devices are located; wherein, the video images include at least two different light environments and alternately change in sequence at preset intervals.

[0074] A division module 20 for dividing the video images into multiple groups of sub-video images according to the preset intervals; wherein, each group of sub-video images includes video frames in all light environments.

[0075] An extraction module 30 for extracting one video frame in each light environment of each group of sub-video images and fusing them according to a preset data fusion method to obtain a fused data frame of each group of sub-video images.

[0076] A splitting module 40 for splitting each of the fused data frames into a preset number of equally sized chunks;

[0077] A first calculation module 50 for calculating the difference value between chunks at the same position of adjacent two fused data frames in chronological order; wherein, the difference value is an absolute value;

[0078] A second calculation module 60 for calculating the sum of the difference values of all chunks of adjacent two fused data frames to obtain the difference sum value of the adjacent two fused data frames;

[0079] A third calculation module 70 for calculating the feature vector of each video image according to the difference sum value , where represents the feature vector of the i-th video image, represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point;

[0080] A corresponding module 80 for corresponding the feature vectors of each video image with the corresponding fire situation of the video image one by one to form a data set;

[0081] An allocation module 90 for dividing the data set into a training data set and a test data set;

[0082] A first input module 100 for inputting the training data set into a preset support vector machine and training the preset support vector machine according to the optimal hyperparameters to obtain a temporary model;

[0083] A detection module 110 for detecting the temporary model through the test data set, and obtaining a target model when the detection result meets the training requirements of the model;

[0084] An acquisition module 120 for acquiring a target video image to be detected and calculating the target feature vector corresponding to the target video image;

[0085] A second input module 130 for inputting the target feature vector into the target model to obtain the fire situation of the target video image.

[0086] As described in the acquisition module 10, video images of the areas where multiple devices are located are acquired. During the acquisition process of the video images, they will be alternately changed at preset intervals. For example, first irradiated with red light for 2s, and then irradiated with yellow light for 2s. This is because the combustion materials of the devices may be different. Therefore, under constant lighting conditions, there may be situations where it is impossible to observe. Therefore, by alternately changing them, the deformation of the devices and the detection of some smoke can be obtained. That is, after a fire occurs, even if there is no open fire, there will be some smoke or the devices will be deformed. Therefore, detection can be carried out through different light environments. The specific number of light environments is not limited. The more the number, the more accurate the detection, but it will bring problems such as higher costs for the acquisition devices and increased computational complexity. The fewer the number, the accuracy of detection may decrease. Therefore, at least two different light environments are required for shooting to obtain accurate data.

[0087] As described in the division module 20, the video images are divided into multiple groups of sub-video images according to the preset interval time. Among them, the preset interval time is the time for the alternate change of the light environment. Therefore, the division time period can be set according to the preset interval time, and the length of this time period needs to be greater than the time period of the light environment change, so as to obtain multiple groups of sub-video images.

[0088] As described in the extraction module 30, one video frame in each light environment of each group of sub-video images is extracted and fused according to a preset data fusion method to obtain the fused data frame of each group of sub-video images. Extracting the video frames in each light environment can be used for data fusion. It should be noted that since a group of sub-video images may contain multiple video frames in the same light environment, any one of them can be selected for calculation.

[0089] As described in the above-mentioned segmentation module 40 and the first calculation module 50, each of the fused data frames is segmented into preset and equal-sized blocks. In order to facilitate distinguishing the location of the fire and the background, they can be segmented into blocks with equal numbers and sizes. In the actual process, most devices may not change, or some devices only move mechanically. Therefore, within a block, there is no change. Therefore, calculate the difference value of the blocks at the same position of adjacent two fused data frames in chronological order, and calculate according to the difference value of each block, so as to avoid the problem of inaccurate data caused by the small overall change of the data.

[0090] As described in the above-mentioned second calculation module 60 and the third calculation module 70, calculate the sum of the difference values of all blocks of adjacent two fused data frames to obtain the difference sum value of these two adjacent fused data frames, and calculate the feature vectors of each video image according to the difference sum value. , where, represents the feature vector of the i-th video image, represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point. Since the sum value of each difference value can reflect the change amount of pixels, and the feature vector calculated according to the difference sum value can reflect the fire situation, that is, if a fire occurs in the initial stage, that is, when there is no big fire, the corresponding feature vector will be very small because the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point will become smaller. And if the fire is very serious, then the change of pixel values will be very fast. Therefore, the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point will become larger. Therefore, the feature vector can reflect the fire situation.

[0091] As described in the corresponding module 80 - distribution module 90, the data set can be divided into a training data set and a test data set to facilitate the subsequent training of the model and improve the training accuracy of the model.

[0092] As described in the first input module 100 and the detection module 110, the support vector machine is a most popular machine learning technology, which is an approximate representation of minimizing structural risk. It seeks the best combination point between the model generalization performance and the fitting performance, rather than the traditional empirical risk minimization, and cleverly solves the dimension problem and can process the standardized data. Therefore, the prediction of the fire situation can be realized through the support vector machine. Specifically, the fire situation can be multiple preset levels, and the levels of each video image are manually set in advance by relevant personnel. The method of training the support vector machine is the same as the prior art, that is, training the support vector machine through the optimal difference parameter, which will not be elaborated in this application.

[0093] As described in the acquisition module 120 and the second input module 130, the target video image to be detected is acquired, and the target feature vector corresponding to the target video image is calculated for inputting the target feature vector into the target model to obtain the fire situation of the target video image. It should be noted that the method of obtaining the target feature vector is the same as that of the above-mentioned feature vector, which will not be elaborated here. After inputting into the target model, the fire situation of the target video image can be obtained. The detection of the fire situation of the target video image is realized, and thus early warning can be carried out according to the fire situation, and a fire without open flames can be discovered, greatly improving the recognition rate of fires and protecting the property safety of other equipment.

[0094] In one embodiment, the intelligent monitoring system for electrical fire early warning further includes:

[0095] A difference sum detection module, configured to detect whether the values of all the difference sums are less than a preset sum value;

[0096] A determination module, configured to, if so, determine to execute the step of calculating the feature vectors of each video image according to the difference sums.

[0097] As described in the above difference sum detection module and determination module, in actual situations, people will approach various devices to turn them off or check them, or even just pass by, which may cause pixel changes and lead to misjudgment. However, when people approach the devices, the change situation is very large. Therefore, by detecting whether the values of all the difference sums are less than the preset sum value, when there is a value greater than this value, it can be considered that human factors cause the pixel to become larger, so that the deletion of incorrect observed data can be realized. It should be noted that during the test process, people generally do not adjust the light environment when passing by, that is, there is generally only one light environment when people pass by. Therefore, it can be determined that it is not the detection time, and the light environment will change during the detection time. Therefore, this item of detection can also be omitted. This embodiment is only set to prevent people from forgetting to switch the light environment.

[0098] In one embodiment, the extraction module 30 includes:

[0099] A video frame extraction sub-module, configured to extract one video frame in each light environment of each group of sub-video images;

[0100] A pixel value acquisition sub-module, configured to acquire the pixel value of each pixel point from each of the video frames;

[0101] A summation sub-module, configured to sum and average the pixel points in each group of sub-videos to obtain a fusion data frame corresponding to each group of sub-videos.

[0102] As described in the above video frame extraction sub-module, pixel value acquisition sub-module, and summation sub-module, a method for fusing data frames is provided. In some embodiments, the fusion of data frames can also be reduced by the light reflection intensity. This application fuses the values of pixel points. Specifically, it can be the fusion of R, G, and B, and then the weighted average method is used to obtain the fusion data frame.

[0103] In one embodiment, the intelligent monitoring system for electrical fire warning further includes:

[0104] An initial support vector machine acquisition module, configured to acquire initial support vector machines with multiple different combinations of hyperparameters;

[0105] A prediction module, configured to input the training data set into each of the initial support vector machines for prediction to obtain respective corresponding prediction results;

[0106] A loss value calculation module, configured to calculate the loss value of each initial support vector machine according to the prediction result;

[0107] A support vector machine selection module, configured to select the initial support vector machine with the smallest loss value as the preset support vector machine based on the loss value.

[0108] As described in the above module, the generalization ability of the SVM model depends to a large extent on the internal parameters. For example, if the penalty coefficient is too large, the model training is difficult and overfitting occurs. If the penalty coefficient is too small, the model is prone to underfitting. The empirical risk ratio affects the complexity of the distribution of samples in the high-dimensional feature space. Therefore, multiple initial support vector machines with different hyperparameter combinations can be selected, and the initial support vector machine with the smallest loss value is selected as the preset support vector machine, so as to further improve the prediction accuracy of the support vector machine. The calculation method of the loss value is the ratio of the difference between the prediction result and the actual result to the actual result as the loss value.

[0109] In one embodiment, the intelligent monitoring system for electrical fire warning further includes:

[0110] A fire situation judgment module, configured to judge whether the fire situation has reached the alarm level;

[0111] A calling module, configured to, if so, call the fire department for help.

[0112] As described in the above module, when the fire situation reaches the alarm level, the fire department can be called for help, that is, when the fire situation reaches the preset fire level, the alarm device can be automatically triggered to give an alarm.

[0113] Refer to Figure 2 , the present invention further provides an intelligent monitoring method for electrical fire warning, including:

[0114] S1: Collect video images of multiple device locations; wherein, the video images include at least two different light environments, which alternately change in sequence at a preset interval time;

[0115] S2: Divide the video images into multiple groups of sub-video images according to the preset interval time; wherein, each group of sub-video images includes video frames in all light environments;

[0116] S3: Extract one video frame in each light environment of each group of sub-video images, and fuse them according to a preset data fusion method to obtain a fused data frame of each group of sub-video images;

[0117] S4: Divide each of the fused data frames into preset numbers of equal-sized blocks;

[0118] S5: Calculate the difference values of the blocks at the same positions in two adjacent fused data frames in chronological order; wherein, the difference values are absolute values.

[0119] S6: Calculate the sum of the difference values of all the blocks in two adjacent fused data frames to obtain the difference sum value of the two adjacent fused data frames.

[0120] S7: Calculate the feature vectors of each video image according to the difference sum value , where represents the feature vector of the i-th video image, represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point.

[0121] S8: One-to-one correspond the feature vectors of each video image with the fire situation corresponding to the video image to form a data set.

[0122] S9: Divide the data set into a training data set and a test data set.

[0123] S10: Input the training data set into a preset support vector machine, and train the preset support vector machine according to the optimal hyperparameters to obtain a temporary model.

[0124] S11: Detect the temporary model through the test data set. When the detection result meets the training requirements of the model, obtain the target model.

[0125] S12: Obtain the target video image to be detected, and calculate the target feature vector corresponding to the target video image.

[0126] S13: Input the target feature vector into the target model to obtain the fire situation of the target video image.

[0127] In one embodiment, before step S7 of calculating the feature vectors of each video image according to the difference sum value, it further includes:

[0128] S601: Detect whether the values of all the difference sum values are less than a preset sum value;

[0129] S602: If so, determine to execute the step of calculating the feature vectors of each video image according to the difference sum value.

[0130] In one embodiment, step S3 of extracting one video frame in each light environment of each group of sub-video images and performing fusion according to a preset data fusion method to obtain the fused data frame of each group of sub-video images includes:

[0131] S301: Extract one video frame in each light environment from each group of sub-video images.

[0132] S302: Obtain the pixel values of each pixel point from each of the said video frames.

[0133] S303: Sum and average the pixel points in each group of sub-videos to obtain the fusion data frame corresponding to each group of sub-videos.

[0134] In one embodiment, before the step S10 of inputting the training data set into a preset support vector machine and training the preset support vector machine according to the optimal hyperparameters to obtain a temporary model, it further includes:

[0135] S901: Obtain initial support vector machines with multiple different combinations of hyperparameters.

[0136] S902: Input the training data set into each of the initial support vector machines for prediction to obtain the corresponding prediction results respectively.

[0137] S903: Calculate the loss value of each initial support vector machine according to the prediction results.

[0138] S904: Select the initial support vector machine with the smallest loss value based on the loss value as the preset support vector machine.

[0139] In one embodiment, after the step S13 of inputting the target feature vector into the target model to obtain the fire situation of the target video image, it further includes:

[0140] S1401: Judge whether the fire situation has reached the alarm level.

[0141] S1402: If so, call for help from the fire department.

[0142] The beneficial effects of the present invention: By acquiring video images and grouping them into multiple sub-video images, calculating the feature vectors in each video image using the change situation of the pixels in the sub-video images, and then training a support vector machine through the feature vectors to realize the prediction of the fire situation, thereby realizing the detection of the fire situation of the target video image, and giving an early warning according to the fire situation, it is possible to discover fires without open flames, greatly improving the recognition rate of fires and protecting the property safety of other devices.

[0143] Refer to Figure 3 , in the embodiments of the present application, a computer device is further provided. This computer device can be a server, and its internal structure can be as Figure 3As shown. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store various video images, etc. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it can implement the intelligent monitoring system for electrical fire warning described in any of the above embodiments.

[0144] Those skilled in the art can understand that Figure 3 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.

[0145] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor, it can implement the intelligent monitoring system for electrical fire warning described in any of the above embodiments.

[0146] Those of ordinary skill in the art can understand that all or part of the processes in the above-described method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-described method embodiments. Among them, any reference to memory, storage, database, or other media provided in the present application and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, there are various forms of RAM, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0147] It should be noted that in this text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method comprising a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such a process, device, article or method. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of additional identical elements in the process, device, article or method comprising such an element.

[0148] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0149] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0150] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. An intelligent monitoring system for electrical fire warning, characterized in that, Including: A collection module for collecting video images of the areas where multiple devices are located; wherein, the video images include at least two different light environments, which alternately change in sequence at a preset interval time; A division module for dividing the video images into multiple groups of sub-video images according to the preset interval time; wherein, each group of sub-video images includes video frames under all light environments; An extraction module for extracting one video frame under each light environment in each group of sub-video images and fusing them according to a preset data fusion method to obtain a fused data frame for each group of sub-video images; A segmentation module for segmenting each of the fused data frames into a preset number of equally sized blocks; A first calculation module for calculating the difference value of the blocks at the same position of two adjacent fused data frames in chronological order; wherein, the difference value is an absolute value; A second calculation module for calculating the sum value of the difference values of all blocks of two adjacent fused data frames to obtain the difference sum value of the two adjacent fused data frames; A third calculation module, configured to calculate eigenvectors of respective video images according to the difference sum value , wherein represents the eigenvector of the i-th video image represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point; A corresponding module for corresponding each of the feature vectors of the video images to the corresponding fire situation of the video images one by one to form a data set; An allocation module for dividing the data set into a training data set and a test data set; A first input module for inputting the training data set into a preset support vector machine and training the preset support vector machine according to the optimal hyperparameters to obtain a temporary model; A detection module for detecting the temporary model through the test data set, and obtaining a target model when the detection result meets the training requirements of the model; An acquisition module for acquiring a target video image to be detected and calculating the target feature vector corresponding to the target video image; A second input module for inputting the target feature vector into the target model to obtain the fire situation of the target video image.

2. The intelligent monitoring system for electrical fire warning according to claim 1, characterized in that, The intelligent monitoring system for electrical fire warning further includes: A difference sum value detection module for detecting whether the values of each of the difference sum values are all less than a preset sum value; A determination module for, if so, determining to execute the step of calculating the feature vectors of each video image according to the difference sum value.

3. The intelligent monitoring system for electrical fire warning according to claim 1, characterized in that, The extraction module includes: A video frame extraction sub-module for extracting one video frame under each light environment in each group of sub-video images, A pixel value acquisition sub-module for acquiring the pixel value of each pixel point from each of the video frames; A summation sub-module for summing and averaging the pixel points in each group of sub-videos to obtain the fused data frame corresponding to each group of sub-videos.

4. The intelligent monitoring system for electrical fire warning according to claim 1, characterized in that, The intelligent monitoring system for electrical fire warning further includes: An initial support vector machine acquisition module for acquiring initial support vector machines with multiple different hyperparameter combinations; A prediction module for inputting the training data set into each of the initial support vector machines for prediction to obtain the respective corresponding prediction results; A loss value calculation module for calculating the loss value of each initial support vector machine according to the prediction results; A support vector machine selection module for selecting the initial support vector machine with the smallest loss value based on the loss value as the preset support vector machine.

5. The intelligent monitoring system for electrical fire warning according to claim 1, characterized in that, The intelligent monitoring system for electrical fire warning further includes: A fire situation judgment module for judging whether the fire situation has reached the alarm level; A calling for help module for, if so, calling for help from the fire department.

6. An intelligent monitoring method for electrical fire warning, characterized in that, It includes: Collect video images of the areas where multiple devices are located; among them, the video images include at least two different light environments, which alternately change in sequence at a preset interval time; Divide the video images into multiple groups of sub-video images according to the preset interval time; among them, each group of sub-video images includes video frames in all light environments; Extract one video frame in each light environment of each group of sub-video images, and perform fusion according to a preset data fusion method to obtain a fusion data frame of each group of sub-video images; Divide each of the fusion data frames into preset numbers of equally sized blocks; Calculate the difference value of the blocks at the same position of adjacent two fusion data frames in chronological order; among them, the difference value is an absolute value; Calculate the sum of the difference values of all blocks of adjacent two fusion data frames to obtain the difference sum value of the adjacent two fusion data frames; Calculate the feature vectors of each video image according to the differences and values , where represents the feature vector of the i-th video image, represents the change amount of the difference sum value of the sub-video image corresponding to the q-th time point and the difference sum value of the sub-video image corresponding to the z-th time point; one-to-one correspondence is made between the feature vectors of each video image and the fire situation corresponding to the video image to form a data set; Divide the data set into a training data set and a test data set; Input the training data set into a preset support vector machine, and train the preset support vector machine according to the optimal hyperparameters to obtain a temporary model; Detect the temporary model through the test data set, and when the detection result meets the training requirements of the model, obtain a target model; Obtain a target video image to be detected, and calculate the target feature vector corresponding to the target video image; Input the target feature vector into the target model to obtain the fire situation of the target video image.

7. The intelligent monitoring method for electrical fire warning according to claim 6, wherein Before the step of calculating the feature vectors of each video image according to the difference sum value, it further includes: Detect whether the values of each of the difference sum values are all less than a preset sum value; If so, determine to execute the step of calculating the feature vectors of each video image according to the difference sum value.

8. The intelligent monitoring method for electrical fire warning according to claim 6, characterized in that, The step of extracting one video frame in each light environment of each group of sub-video images and performing fusion according to a preset data fusion method to obtain a fusion data frame of each group of sub-video images includes: Extract one video frame in each light environment of each group of sub-video images, Obtain the pixel values of each pixel point from each of the video frames; Sum and average the pixel points in each group of sub-videos to obtain the corresponding fusion data frame of each group of sub-videos.

9. The intelligent monitoring method for electrical fire warning according to claim 6, wherein Before the step of inputting the training data set into a preset support vector machine and training the preset support vector machine according to the optimal hyperparameters to obtain a temporary model, it further includes: Obtain multiple initial support vector machines with different hyperparameter combinations; Input the training data set into each of the initial support vector machines for prediction to obtain their respective corresponding prediction results; Calculate the loss value of each initial support vector machine according to the prediction results; Select the initial support vector machine with the smallest loss value based on the loss value as the preset support vector machine.

10. The intelligent monitoring method for electrical fire warning according to claim 6, characterized in that, After the step of inputting the target feature vector into the target model to obtain the fire situation of the target video image, it further includes: Judge whether the fire situation has reached the alarm level; If so, call for help from the fire department.

Images