Method and system for detecting electrical fire risk hidden danger

Abstract

The invention relates to the technical field of electrical fire early warning detection, and discloses a method and a system for detecting electrical fire risk hidden dangers. The method comprises the following steps: collecting current, voltage and temperature signals through a sensor, and generating a multi-dimensional original data set; extracting spectrum features and mutation features; inputting the feature vector into a trained model for risk trend prediction; and smoke features and flame features are detected and analyzed in combination with a video image recognition algorithm. Compared with the prior art that fire hazards are judged only depending on a fixed threshold value, and particularly, accurate recognition and real-time early warning of the fire hazards cannot be achieved in a complex electrical environment, the method adopts a cooperative mode of multi-source data acquisition, intelligent algorithm analysis and video image recognition; high-precision prediction and real-time alarm of the electrical fire risk are realized, so that the problem of false alarm or missing alarm is avoided, and the real-time performance and the stability of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of electrical fire early warning detection, and in particular relates to a detection method and system for electrical fire risk hazards. Background Art

[0002] At present, the accuracy, real-time and intelligent level of electrical fire hazard detection technology in complex environments are still greatly insufficient. For example, traditional fire detection methods mainly rely on fixed threshold alarms, usually based on a single parameter such as current and voltage for simple judgment, and it is difficult to effectively distinguish normal fluctuations from potential risks. This method shows a high false alarm rate and missed alarm rate in complex environments such as industrial facilities, data centers, and large public buildings. Especially in the case of aging equipment and frequent load fluctuations, it is difficult for existing technologies to achieve accurate early warning, resulting in the failure to timely discover and deal with fire hazards. Existing technologies also have obvious limitations in real-time monitoring and multi-source information fusion to deal with electrical fire hazards. The setting of fixed thresholds is easily affected by environmental noise and signal fluctuations, and cannot dynamically adapt to changes in the electrical system, resulting in a lack of sufficient flexibility in the system. In addition, some current systems lack the integration of video image information, and cannot use image recognition technology to further confirm fire characteristics such as smoke and flames, resulting in low accuracy of early warning signals. Therefore, there is an urgent need for an intelligent electrical fire hazard detection method that can achieve high-precision prediction and real-time early warning of fire risks in complex electrical environments through multi-source data acquisition and intelligent analysis, combined with technologies such as dynamic threshold adjustment and video image recognition. This method should be able to improve the fault tolerance and adaptability of the system, enhance the accuracy and response speed of fire warning, reduce the problem of false alarms and missed alarms, ensure the safety of personnel and property, and improve the overall effectiveness of the electrical fire warning system. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to propose a method and system for detecting electrical fire risk hazards, aiming to solve the technical problem that the prior art only relies on fixed thresholds to judge fire hazards, especially in complex electrical environments, and cannot achieve accurate identification and real-time warning of fire hazards.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a method for detecting electrical fire risk hazards, comprising: Step S10: Collect multi-source signals in the electrical system through current transformers, voltage transformers and thermistor sensors, respectively. The multi-source signals include current signals , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; Step S20: Perform Fourier transform on the de-noised electrical parameter data set D to obtain electrical spectrum feature Y, and apply wavelet transform on the electrical spectrum feature to extract electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; Step S30: Convert the electrical multidimensional feature vector Perform standardization to obtain the standardized feature vector ; Step S40: normalize the feature vector Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; Step S50: After the fire warning signal is generated, the video surveillance linkage module is automatically activated to retrieve the real-time monitoring image in the warning area and extract smoke features using the CNN model. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

[0005] Preferably, in step S10, the electrical parameter data set D is obtained by denoising the multidimensional original data set through moving average filtering, using the formula: ; Among them, i is the center position of the current moving average calculation, j is the index variable in the window, is the ith electrical parameter data in the electrical parameter data set after denoising, is the size of the moving average window, is the jth data point in the multidimensional original data set.

[0006] Preferably, in step S20, wavelet transform is applied to the electrical spectrum feature to extract the electrical mutation feature Z, wherein the wavelet transform is expressed as: ; in, is the electrical mutation characteristic after wavelet transformation, is the electrical spectrum characteristic, is the mother wavelet function, a is the scale parameter, b is the location parameter, is the time variable, is the integration of the time variable t.

[0007] Preferably, in step S40, the loss function of the LSTM network model training is the mean square error, and the formula is: ; in, is the mean square error loss function, The number of electrical parameter dataset samples taken for the training process, is the true risk trend value of the Lth sample, is the risk trend value predicted by the LSTM model for the Lth sample.

[0008] Preferably, in step S50, the smoke features and fire features in the picture are analyzed by a video image recognition algorithm, a dual-stream model is added to analyze static image features and dynamic video changes, and a risk probability score is calculated: ; in, and is the weight coefficient, is the activation function, Score the risk probability.

[0009] Preferably, when When the risk score is greater than or equal to the preset risk score threshold, the fire confirmation signal P is set to 1; when When it is less than the preset risk score threshold, the fire confirmation signal P is set to 0.

[0010] Preferably, in step S30, the normalization formula is: ; in, is the standardized feature vector, is the electrical multidimensional eigenvector, is the minimum value of the characteristic in the electrical multidimensional characteristic vector, It is the maximum value of the features in the electrical multidimensional feature vector.

[0011] Preferably, in step S40, the calculation formula of the risk threshold K is: ; in, is the regulating factor, is the standard deviation of the risk trend value, is the mean of the risk trend values.

[0012] The present invention also provides a detection system for electrical fire risk hazards, comprising: The signal acquisition module is used to collect multi-source signals in the electrical system through current transformers, voltage transformers and thermistor sensors. The multi-source signals include current signals , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; The feature extraction module is used to perform Fourier transform on the de-noised electrical parameter data set D to obtain the electrical spectrum feature Y, and apply wavelet transform to the electrical spectrum feature to extract the electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; Standardization processing module is used to convert electrical multidimensional feature vector Perform standardization to obtain the standardized feature vector ; Risk trend prediction module is used to standardize feature vectors Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; The video linkage module is used to automatically activate the video surveillance linkage module after the fire warning signal is generated, retrieve the real-time monitoring screen in the warning area, and use the CNN model to extract smoke features. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

[0013] The beneficial effect of the present invention is that compared with the prior art which only relies on fixed thresholds to judge fire hazards, especially in complex electrical environments, the technical problem of being unable to achieve accurate identification and real-time warning of fire hazards, the present invention achieves high-precision prediction and real-time alarm of electrical fire risks by adopting a collaborative approach of multi-source data acquisition, intelligent algorithm analysis, and video image recognition, thereby avoiding problems due to false alarms or missed alarms and improving the real-time and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 A schematic flow chart of a first embodiment of a method for detecting electrical fire risk hazards provided by the present invention.

[0016] Figure 2 A system schematic diagram of a method for detecting electrical fire risk hazards provided by the present invention.

[0017] Figure 3 A schematic diagram of equipment for a method for detecting electrical fire risk hazards provided by the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Embodiment 1

[0019] like Figure 1 , which is a flow chart of a first embodiment of a method for detecting electrical fire risk hazards of the present invention, provides a first embodiment of a method for detecting electrical fire risk hazards of the present invention.

[0020] In a first embodiment, a method for detecting electrical fire risk hazards includes: Step S10: Collect multi-source signals in the electrical system through current transformers, voltage transformers and thermistor sensors, respectively. The multi-source signals include current signals , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; It should be noted that current signals, voltage signals and temperature signals are the main physical quantities in the electrical system, which can reflect the working status and environmental conditions of electrical equipment. By comprehensively analyzing these signals, the health status of the electrical system can be judged more comprehensively.

[0021] It is understandable that since electrical signals may contain various noises (such as electromagnetic interference, environmental fluctuations, etc.), directly using the original signal for analysis will lead to errors. Therefore, after data collection, denoising through moving average filtering helps to eliminate random fluctuations, improve signal stability, and provide a more reliable data basis for subsequent feature extraction and analysis.

[0022] It should be understood that the window size of the moving average filter needs to be set according to the dynamic characteristics of the electrical system. A window that is too large will increase the smoothness of the signal but lose details, while a window that is too small may not be able to effectively denoise. Therefore, the window size should be optimized according to the actual application scenario to ensure that the electrical parameter data set after denoising retains valid information.

[0023] Step S20: Perform Fourier transform on the de-noised electrical parameter data set D to obtain electrical spectrum feature Y, and apply wavelet transform on the electrical spectrum feature to extract electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; It should be noted that Fourier transform can convert electrical signals in the time domain into frequency domain features, allowing the system to identify the main frequency components and energy distribution in the signal. Spectral features can effectively reflect the operating status of the electrical system and facilitate the identification of abnormal changes in frequency.

[0024] It can be understood that wavelet transform is a mathematical tool suitable for detecting sudden signal changes. Compared with Fourier transform, wavelet transform can not only provide frequency information, but also locate the specific time when the mutation occurs. Therefore, by applying wavelet transform to the spectrum characteristics, the mutation characteristics can be extracted, thereby identifying possible abnormal fluctuations and fault characteristics in electrical signals.

[0025] It should be understood that the combination of Fourier transform and wavelet transform can more comprehensively analyze electrical signals. Fourier transform is used to extract stable frequency features, while wavelet transform is used to identify instantaneous mutation features. The combination of the two can generate a more informative multidimensional feature vector, providing a reliable input for subsequent risk assessment.

[0026] Step S30: Convert the electrical multidimensional feature vector Standardize the process; It should be noted that the standardized formula is: ; in, is the standardized feature vector, is the electrical multidimensional eigenvector, is the minimum value of the characteristic in the electrical multidimensional characteristic vector, is the maximum value of the feature in the electrical multidimensional feature vector; It should be noted that the purpose of normalization is to convert feature values ​​of different dimensions and ranges into a unified numerical range, usually [0,1], so as to eliminate the dimensional differences between the features and prevent some features from having too much or too little impact on the model input, thereby improving the stability of subsequent analysis.

[0027] It is understandable that the electrical multidimensional feature vector contains multiple features of the electrical system, such as spectrum features and mutation features, which may have different numerical ranges. Through standardization, it can be ensured that all features are in the same numerical range, which can improve training efficiency and prediction accuracy.

[0028] It should be understood that the maximum and minimum values ​​used in the normalization process are typically calculated based on historical data or dynamically updated in real-time data to accommodate changes in the electrical system.

[0029] Step S40: normalize the feature vector Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; It should be noted that the risk threshold K is dynamically calculated based on the risk trend value, and the calculation formula is:

[0030] ; in, is the regulating factor, is the standard deviation of the risk trend value, is the mean of the risk trend value; When the risk trend value When the risk threshold K is exceeded, a fire warning signal X3 is generated; It should be noted that LSTM (Long Short-Term Memory Network) is a neural network model suitable for processing time series data. It can learn the pattern of electrical system risk trends changing over time, thereby predicting current and future risk conditions. The risk trend value is the output of the LSTM model, reflecting the potential risk level of the electrical system at the current moment.

[0031] It is understandable that since the working state of the electrical system will fluctuate over time, it is not possible to rely solely on a fixed threshold for early warning, but the threshold K should be set dynamically based on the historical distribution of the risk trend value. and mean The dynamic threshold method can effectively reduce false alarms and missed alarms and enhance the accuracy of early warning.

[0032] It should be understood that the modulating factor It can be adjusted according to the different risk levels of the electrical environment. For example, for a high risk environment, a smaller value, so that an alert is triggered at a lower risk trend value; for a low-risk environment, a larger value can be selected value, thereby reducing unnecessary warnings.

[0033] For example, assuming that in a certain electrical environment, the mean of the risk trend value is 0.5 and the standard deviation is 0.1. If the adjustment factor is set to 2, the threshold K is calculated as: K=2×0.1+0.5=0.7. When the risk trend value exceeds 0.7, for example, reaching 0.8, the system will generate a fire warning signal to remind relevant personnel to check and handle it.

[0034] By dynamically adjusting the risk threshold, we can effectively adapt to changes in different environments and ensure the sensitivity and accuracy of the system at different risk levels.

[0035] Step S50: After the fire warning signal is generated, the video surveillance linkage module is automatically activated to retrieve the real-time monitoring image in the warning area and extract smoke features using the CNN model. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

[0036] It should be noted that CNN (convolutional neural network) is a deep learning model suitable for image recognition and feature extraction, which can effectively extract smoke and fire features in video images. By analyzing these features, the system can further determine whether the warning signal is accurate to reduce the false alarm rate.

[0037] It is understandable that the fire confirmation signal P takes the value of 0 or 1 to indicate whether the fire is confirmed. When the video image recognition algorithm detects obvious smoke or fire features, it generates a signal of P=1 to confirm the fire and trigger further alarm linkage; when no fire features are detected, it maintains the state of P=0 to avoid false alarms.

[0038] It should be understood that the effective duration of the fire warning signal determines the duration of the system's video analysis after the initial warning signal is generated. This duration can be set according to the needs of the actual site to ensure that the system can fully monitor the warning area when the fire risk is high. If no fire characteristics are detected within the effective duration, the system can automatically cancel the warning signal to avoid unnecessary alarm interference.

[0039] For example, suppose the system detects a fire risk and generates an early warning signal, activates the video surveillance linkage module, and calls the real-time image in the early warning area. The smoke and fire features are extracted through the CNN model. If P=1 is obtained after analysis, the system confirms the fire and sends a notification text message to emergency personnel, including the time and location of the fire, to facilitate emergency response. If smoke or fire features are not detected within the effective duration of the fire warning signal, for example, within 10 minutes, the system maintains P=0, automatically cancels the fire warning signal, and returns to normal.

[0040] Through step S50, the system can confirm the accuracy of the fire warning in combination with video surveillance, greatly reduce the false alarm rate, and ensure that the formal alarm is triggered only when the fire risk is confirmed, thereby improving the reliability and response efficiency of the fire warning. Embodiment 2

[0041] In addition, the present invention also provides a detection system for electrical fire risk hazards, please refer to Figure 2 , the system includes: The signal acquisition module 10 is used to respectively acquire multi-source signals in the electrical system through a current transformer, a voltage transformer and a thermistor sensor. The multi-source signals include a current signal , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; The feature extraction module 20 is used to perform Fourier transform on the de-noised electrical parameter data set D to obtain the electrical spectrum feature Y, and apply wavelet transform to the electrical spectrum feature to extract the electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; The standardization processing module 30 is used to convert the electrical multi-dimensional feature vector Perform standardization to obtain the standardized feature vector ; The risk trend prediction module 40 is used to convert the standardized feature vector Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; The video linkage module 50 is used to automatically activate the video monitoring linkage module after the fire warning signal is generated, retrieve the real-time monitoring screen in the warning area, and use the CNN model to extract smoke features. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

[0042] The present invention provides a detection system for electrical fire risk hazards, which adopts a detection method for electrical fire risk hazards in the above embodiment, and can solve a technical problem of detecting electrical fire risk hazards. Compared with the prior art, the beneficial effects of the detection system for electrical fire risk hazards provided by the present invention are the same as the beneficial effects of the detection method for electrical fire risk hazards provided by the above embodiment, and other technical features of the detection system for electrical fire risk hazards are the same as the features disclosed in the above embodiment method, which will not be described in detail here. Embodiment 3

[0043] The present invention provides a detection device for electrical fire risk hazards, please refer to Figure 3, a detection device for electrical fire risk hazards includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the detection method for electrical fire risk hazards in the above-mentioned embodiment 1. A detection device for electrical fire risk hazards in the embodiment of the present invention may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. A detection device for electrical fire risk hazards is only an example and should not bring any limitations to the functions and scope of use of the embodiments of the present invention. A detection device for electrical fire risk hazards may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM) 1004. In RAM 1004, various programs and data required for the operation of a detection device for electrical fire risk hazards are also stored. The processing device 1001, ROM 1002, and RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems may be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 may allow a detection device for electrical fire risk hazards to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a detection device for electrical fire risk hazards with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have alternatively.

[0044] The present invention also provides a computer program product, including a computer program, which implements the steps of the above-mentioned method for detecting electrical fire risk hazards when the computer program is executed by a processor. The computer program product provided by the present invention can solve the technical problem of detecting electrical fire risk hazards. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as the beneficial effects of the method for detecting electrical fire risk hazards provided by the above-mentioned embodiment, which will not be described in detail here.

[0045] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present invention are executed.

[0046] It should be understood that the various parts disclosed in the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0047] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for detecting electrical fire risk hazards, characterized in that: include: Step S10: Collect multi-source signals in the electrical system through current transformers, voltage transformers and thermistor sensors, respectively. The multi-source signals include current signals , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; Step S20: Perform Fourier transform on the de-noised electrical parameter data set D to obtain electrical spectrum feature Y, and apply wavelet transform on the electrical spectrum feature to extract electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; Step S30: Convert the electrical multidimensional feature vector Perform standardization to obtain the standardized feature vector ; Step S40: normalize the feature vector Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; Step S50: After the fire warning signal is generated, the video surveillance linkage module is automatically activated to retrieve the real-time monitoring image in the warning area and extract smoke features using the CNN model. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

2. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S10, the multi-dimensional original data set is denoised by moving average filtering to obtain an electrical parameter data set D, using the formula: ； Among them, i is the center position of the current moving average calculation, j is the index variable in the window, is the ith electrical parameter data in the electrical parameter data set after denoising, is the size of the moving average window, is the jth data point in the multidimensional original data set.

3. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S20, wavelet transform is applied to the electrical spectrum feature to extract the electrical mutation feature Z, where the wavelet transform is expressed as: ； in, is the electrical mutation characteristic after wavelet transformation, is the electrical spectrum characteristic, is the mother wavelet function, a is the scale parameter, b is the location parameter, is the time variable, is the integration of the time variable t.

4. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S40, the loss function of the LSTM network model training is the mean square error, and the formula is: ； in, is the mean square error loss function, The number of electrical parameter dataset samples taken for the training process, is the true risk trend value of the Lth sample, is the risk trend value predicted by the LSTM model for the Lth sample.

5. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S50, the smoke and fire features in the picture are analyzed by the video image recognition algorithm, and the dual-stream model is added to analyze the static image features and dynamic video changes, and the risk probability score is calculated: ； in, and is the weight coefficient, is the activation function, Score the risk probability.

6. The method for detecting electrical fire risk hazards according to claim 5, characterized in that: when When the risk score is greater than or equal to the preset risk score threshold, the fire confirmation signal P is set to 1; when When it is less than the preset risk score threshold, the fire confirmation signal P is set to 0.

7. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S30, the normalization formula is: ； in, is the standardized feature vector, is the electrical multidimensional eigenvector, is the minimum value of the characteristic in the electrical multidimensional characteristic vector, It is the maximum value of the features in the electrical multidimensional feature vector.

8. The method for detecting electrical fire risk hazards according to claim 1, characterized in that: In step S40, the calculation formula of the risk threshold K is: ； in, is the regulating factor, is the standard deviation of the risk trend value, is the mean of the risk trend values.

9. A detection system for electrical fire risk hazards, applied to the detection method for electrical fire risk hazards according to any one of claims 1 to 8, characterized in that: The system comprises: The signal acquisition module is used to collect multi-source signals in the electrical system through current transformers, voltage transformers and thermistor sensors. The multi-source signals include current signals , voltage signal V and temperature signal T, generate a multi-dimensional original data set according to the collected multi-source signals, and perform denoising on the multi-dimensional original data set through moving average filtering to obtain an electrical parameter data set D; The feature extraction module is used to perform Fourier transform on the de-noised electrical parameter data set D to obtain the electrical spectrum feature Y, and apply wavelet transform to the electrical spectrum feature to extract the electrical mutation feature Z, and combine the obtained electrical spectrum feature Y and electrical mutation feature Z to form an electrical multi-dimensional feature vector ; The standardization processing module is used to convert the electrical multidimensional feature vector Perform standardization to obtain the standardized feature vector ; Risk trend prediction module is used to standardize feature vector Input into the trained LSTM network model, and the LSTM network model outputs the risk trend value at the current moment , dynamically calculate the risk threshold K according to the risk trend value. When the risk threshold K is exceeded, a fire warning signal X3 is generated; When the risk trend value When the risk threshold K is exceeded, a fire warning signal X3 is generated; The video linkage module is used to automatically activate the video surveillance linkage module after the fire warning signal is generated, retrieve the real-time monitoring screen in the warning area, and use the CNN model to extract smoke features. and fire characteristics Analyze the smoke and fire features in the picture through the video image recognition algorithm to generate a fire confirmation signal P; When P=1, the fire is confirmed, a notification SMS is immediately sent to emergency personnel, and the time and location of the fire are recorded; When P=0, the smoke and fire characteristics in the image are continuously analyzed during the effective duration of the fire warning signal.

Images