A distributed photovoltaic fire warning method and system
Through the combination of sensor monitoring and image recognition technology, the multi-dimensional information comprehensive monitoring and analysis of photovoltaic panels is achieved, the accuracy of fire risk prediction and target detection is improved, and targeted measures are taken to solve the single monitoring of traditional photovoltaic fire early warning methods. and insufficient identification have significantly improved the accuracy and response speed of fire warnings.
Patent Information
- Application Number
- CN202411067243.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-06
AI Technical Summary
The existing photovoltaic fire warning methods have shortcomings in the single monitoring parameters, insufficient accuracy of target identification, and integrated optimization of prediction and measures, resulting in insufficient accuracy and timeliness of fire warnings.
The distributed photovoltaic fire warning method is adopted, and the multi-dimensional operation parameters of the photovoltaic panel are collected by sensors to input the fire warning model for risk degree prediction, and the target detection is carried out in combination with the pre-trained convolutional neural network image recognition model, and the on-site image is collected in real time for fire characteristics. Identify and take targeted measures based on the prediction results.
It improves the accuracy and response speed of fire warnings, enhances the safety guarantee of photovoltaic power generation systems, and reduces the safety risks brought by fires.
Smart Images

Figure CN118587829B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fire warning, and in particular to a distributed photovoltaic fire warning method and system. Background Art
[0002] With the widespread application of renewable energy, photovoltaic power generation plays an increasingly important role in the energy field as one of the clean energy sources. However, photovoltaic power generation systems may face various safety hazards during long-term operation, among which fire risk is an important issue that cannot be ignored. Fires in photovoltaic panels may cause serious losses to the power generation system and even threaten personal safety. Therefore, it is particularly important to develop an efficient and reliable photovoltaic fire warning method.
[0003] However, the existing photovoltaic fire warning methods still have some shortcomings. First, the traditional sensor-based monitoring methods can only monitor a certain parameter, lacking comprehensiveness and accuracy. Secondly, the image recognition-based methods have certain limitations in fire target detection, such as the ability to accurately identify targets such as flames and smoke needs to be improved. In addition, the existing methods lack effective integration and optimization in predicting the degree of fire risk and taking targeted measures.
[0004] In response to the above problems, we have proposed a distributed photovoltaic fire warning method. This method makes full use of sensors to monitor the operating parameters of photovoltaic panels, and combines image recognition technology to achieve fire target detection, thereby improving the accuracy and timeliness of fire warnings. Specifically, we collect the operating parameters of photovoltaic panels through sensors, and input them into the fire warning model to predict the degree of fire risk. At the same time, we collect on-site images of photovoltaic panels in real time, and use pre-trained image recognition models for target detection. Finally, targeted measures are taken according to the prediction results. Compared with the existing technology, our invention has the following advantages: First, it makes full use of sensors and image recognition technology to achieve comprehensive monitoring and analysis of multi-dimensional information; second, it uses pre-trained image recognition models to improve the accuracy and stability of target detection; third, based on the prediction results, targeted measures are taken to respond to fire risks in a timely manner and effectively protect the safety of photovoltaic power generation systems.
[0005] In summary, the distributed photovoltaic fire early warning method proposed in the present invention has certain innovations and advantages in overcoming the limitations of traditional methods, and is expected to play an important role in the safety assurance of photovoltaic power generation systems. Summary of the invention
[0006] In view of the above-mentioned problems, the present invention is proposed.
[0007] Therefore, the technical problem solved by the present invention is: the technical problem of insufficient accuracy of single monitoring parameters and target recognition in traditional methods for fire early warning of photovoltaic power generation systems.
[0008] In order to solve the above technical problems, the present invention provides the following technical solutions: a distributed photovoltaic fire early warning method, which comprises the following steps:
[0009] Collect the operating parameters of photovoltaic panels through sensors; input the operating parameters of photovoltaic panels into the fire warning model to predict the degree of fire risk; pre-train the image recognition model through convolutional neural network; collect on-site images of photovoltaic panels in real time, and input the on-site images into the image recognition model; use the pre-trained image recognition model to perform target detection on the real-time collected on-site images; combine the fire risk degree prediction results with target detection, and take targeted measures.
[0010] As a preferred solution of the distributed photovoltaic fire early warning method described in the present invention, the sensors include a temperature sensor, a smoke sensor, a voltage sensor, a current sensor and an infrared thermal imaging sensor.
[0011] The operating parameters are converted into electrical signals by the fire detectors and transmitted to the fire warning model.
[0012] The infrared thermal imaging sensor identifies whether there are hot spots on the photovoltaic panels on site.
[0013] The operating parameters include temperature data, smoke data, current, and voltage data.
[0014] As a preferred solution of the distributed photovoltaic fire warning method described in the present invention, the fire warning model obtains the fire risk value according to the operating parameters of the photovoltaic panel, including the smoke data Perform smoothing, perform Hilbert transform on the smoothed smoke data, obtain the analytical signal, and extract the instantaneous amplitude from the analytical signal .
[0015] Temperature data Perform a Fourier transform to convert the temperature signal into the frequency domain and find the frequency component with the maximum amplitude in the spectrum.
[0016] Current data Perform Hilbert transform to obtain the instantaneous amplitude and extract the average value of the instantaneous amplitude As the instantaneous characteristics of current data.
[0017] Voltage data Perform wavelet transform to decompose the voltage data into wavelet coefficients of different scales, and select the amplitude with the highest wavelet coefficient energy. As the time-frequency characteristics of voltage data.
[0018] The fire risk level value is expressed as,
[0019] ,
[0020] Among them, R represents the fire risk value, is the adjustment factor, represents the maximum amplitude in the temperature data, represents the instantaneous amplitude of smoke data, represents the average value of the instantaneous amplitude of the current data, represents the time-frequency characteristics of voltage data, Represents the baseline risk value.
[0021] When the fire risk level value R is greater than 0 and less than or equal to the first-level fire risk threshold, it means that the photovoltaic area is at low risk, recorded as B1. When the fire risk level value is greater than the first-level fire risk threshold and less than or equal to the second-level fire risk threshold, it means that the photovoltaic area is at medium risk, recorded as B2. When the fire risk level value is greater than the second-level fire risk threshold, it means that there is a suspected fire in the photovoltaic area, recorded as B3.
[0022] As a preferred solution of the distributed photovoltaic fire warning method described in the present invention, the pre-trained image recognition model includes preparing an image data set containing normal and abnormal states of photovoltaic panels, training the image recognition model through the image data set, performing data enhancement during the training process, and evaluating the accuracy of the image recognition model through cross-validation technology after the training is completed.
[0023] As a preferred solution of the distributed photovoltaic fire warning method described in the present invention, the collecting of on-site images of photovoltaic panels includes adjusting the image size, brightness, and contrast of the collected images, using filters to remove image noise, and enhancing the images.
[0024] As a preferred solution of the distributed photovoltaic fire warning method described in the present invention, the target detection includes applying an image recognition model to identify abnormal features in the image, dividing the image into a normal state A1 and an abnormal state A2 according to the recognition detection result, and marking the abnormal area;
[0025] The abnormal state includes the appearance of flame shape or color features in the scene image, the appearance of smoke gathering or diffusion features in the scene image, and the identification of temperature abnormality using an infrared thermal imaging sensor.
[0026] As a preferred solution of the distributed photovoltaic fire warning method described in the present invention, the taking of targeted measures includes that when the photovoltaic area is at a low risk B1, there is no need to identify the target detection results and continue to perform regular equipment and safety inspections.
[0027] When the photovoltaic area is at medium risk B2, the identification and detection results are further combined. When it is at B2A1, the monitoring frequency of on-site images is increased, the frequency of on-site image target detection is increased, and the data collected by the sensors is regularly analyzed and trended.
[0028] When in B2A2, strengthen real-time monitoring of the photovoltaic area, pay close attention to changes in on-site data, increase the frequency and scope of on-site inspections and patrols, improve the sensitivity of the early warning system, increase on-site monitoring points and coverage, and prepare for emergency response work, including inspecting and maintaining fire-fighting equipment.
[0029] When it is suspected fire B3, further judgment is made based on the image information and the smoke data and temperature data in the operating parameters.
[0030] The further judgment is to judge whether there is a hot spot in the image of the abnormal state. If there is no hot spot in the scene image information, it is in B3A1. When there is a hot spot in the scene image, it is in B3A2, and further detection is performed by obtaining the evaluation index value.
[0031] When in B3A1, the monitoring frequency of on-site images is increased, on-site inspections are continued and the density of monitoring is increased. Through historical data and trend analysis, it is observed whether there is a trend of hot spots emerging. When there is a trend, it is detected in advance by obtaining the evaluation index value.
[0032] The evaluation index value is obtained as follows:
[0033] ,
[0034] Among them, a, o, and b represent adjustment factors.
[0035] When the comprehensive evaluation index value F is greater than the index value threshold Indicates that there is actually a fire. When the comprehensive index value F is less than or equal to the index value threshold Indicates a misjudgment.
[0036] When a fire actually occurs, emergency measures will be taken, including automatically notifying relevant personnel and fire departments via text messages and alarm systems, implementing automatic evacuation guidance, and guiding people in nearby areas to evacuate to designated safe areas.
[0037] When there is a misjudgment, a notification is sent to the relevant personnel to prompt the existence of a false alarm. The relevant personnel confirm and verify the false alarm and record the time and relevant data of the false alarm event.
[0038] Another object of the present invention is to provide a distributed photovoltaic fire warning system, which can solve the problems of the prior art in terms of real-time monitoring accuracy and fire warning response speed by integrating sensor data monitoring and advanced image recognition technology.
[0039] In order to solve the above technical problems, the present invention provides the following technical solutions: a distributed photovoltaic fire warning system, including a sensor module, a fire warning module, a real-time image acquisition module and a targeted measures module.
[0040] The sensor module is responsible for collecting the operating parameters of the photovoltaic panel.
[0041] The fire warning module is used to receive photovoltaic panel operating parameters collected by sensors and predict the degree of fire risk.
[0042] The real-time image acquisition module is used to acquire on-site images of photovoltaic panels in real time as input of the image recognition model.
[0043] The targeted measures module is used to combine the fire risk degree prediction result with the target detection and take targeted measures.
[0044] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned distributed photovoltaic fire early warning method when executing the computer program.
[0045] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned distributed photovoltaic fire warning method.
[0046] Beneficial effects of the present invention: The present invention makes comprehensive use of sensor monitoring and image recognition technology, collects the operating parameters of photovoltaic panels and inputs them into the fire warning model, realizes accurate prediction of the degree of fire risk, and simultaneously collects on-site images in real time and performs target detection through a pre-trained image recognition model, effectively identifying fire characteristics. In addition, targeted measures are taken according to the prediction results, including increasing the monitoring frequency, improving the sensitivity of the early warning system, etc., which ultimately greatly improves the accuracy and response speed of fire warnings, ensures the safe and stable operation of photovoltaic power generation systems, and reduces the safety risks caused by fires. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0048] Figure 1 This is an overall flow chart of a distributed photovoltaic fire warning method provided by the first embodiment of the present invention.
[0049] Figure 2 This is an overall framework diagram of a distributed photovoltaic fire warning system provided for the second embodiment of the present invention. DETAILED DESCRIPTION
[0050] In order to make the above-mentioned purposes, 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 of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0051] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a distributed photovoltaic fire warning method, characterized in that.
[0052] S1: Collect the operating parameters of the photovoltaic panels through sensors.
[0053] The sensors include temperature sensors, smoke sensors, voltage sensors, current sensors and infrared thermal imaging sensors.
[0054] The operating parameters are converted into electrical signals through fire detectors and transmitted to the fire warning model.
[0055] Infrared thermal imaging sensors identify if there are hot spots on the photovoltaic panels on site.
[0056] The operating parameters include temperature data, smoke data, current and voltage data.
[0057] To further supplement, the temperature sensor monitors the temperature changes of the photovoltaic panels and provides real-time data on the temperature changes.
[0058] The smoke sensor detects the smoke concentration in the area of the photovoltaic panels.
[0059] Voltage sensors monitor the current output of the photovoltaic panel system.
[0060] The voltage sensor detects the voltage output of the photovoltaic panel system.
[0061] It is further added that the infrared thermal imaging sensors are arranged around the perimeter of the PV panels to cover the entire PV panel area, and the area covered by each sensor overlaps to ensure that hot spots on the surface of the PV panels are not missed.
[0062] Temperature sensors are evenly distributed on the surface of the photovoltaic panels to capture the temperature changes in the entire photovoltaic panel area. The temperature data collected by the temperature sensor in S6 is combined with the hot spot information obtained by the infrared thermal imaging sensor and the smoke data to obtain the evaluation index value and realize the final fire prediction.
[0063] Smoke sensors are arranged around and at the bottom of the photovoltaic panels to detect the generation of smoke in time.
[0064] Voltage sensors and current sensors are distributed at key locations of the photovoltaic panel to monitor the voltage and current changes of the photovoltaic panel circuit.
[0065] S2: Input the operating parameters of the photovoltaic panels into the fire warning model to predict the degree of fire risk.
[0066] The fire warning model calculates the fire risk level based on the operating parameters of the photovoltaic panels, including smoke data. Perform smoothing, perform Hilbert transform on the smoothed smoke data, obtain the analytical signal, and extract the instantaneous amplitude from the analytical signal .
[0067] Temperature data Perform a Fourier transform to convert the temperature signal into the frequency domain and find the frequency component with the maximum amplitude in the spectrum.
[0068] Current data Perform Hilbert transform to obtain the instantaneous amplitude and extract the average value of the instantaneous amplitude As the instantaneous characteristics of current data.
[0069] Voltage data Perform wavelet transform to decompose the voltage data into wavelet coefficients of different scales, and select the amplitude with the highest wavelet coefficient energy. As the time-frequency characteristics of voltage data.
[0070] It is further supplemented that, according to the expected rate of temperature change of the photovoltaic panel and the distribution of frequency components, an appropriate window size is selected to balance the time resolution and frequency resolution, and a Hamming window is used to reduce spectrum leakage and edge effects and improve the identification accuracy of frequency components.
[0071] Before applying the Hilbert transform, the smoke data and current data are pre-filtered to remove noise and non-correlated frequency components to enhance the effectiveness of the transform, while the construction of the analytical signal ensures that all key time domain information is retained in the process of constructing the analytical signal to accurately calculate the instantaneous amplitude.
[0072] According to the characteristics of voltage data, appropriate wavelet basis functions are selected to capture mutations and non-periodic fluctuations in voltage data. According to the time-frequency characteristics of voltage signals, appropriate decomposition levels are selected to ensure sufficient resolution to analyze key time-frequency features.
[0073] The fire risk level is expressed as,
[0074] ,
[0075] Among them, R represents the fire risk value, is the adjustment factor, represents the maximum amplitude in the temperature data, represents the instantaneous amplitude of smoke data, represents the average value of the instantaneous amplitude of the current data, represents the time-frequency characteristics of voltage data, Represents the baseline risk value, which is set based on historical data.
[0076] When the fire risk level value R is greater than 0 and less than or equal to the first-level fire risk threshold, it means that the photovoltaic area is at low risk, recorded as B1. When the fire risk level value is greater than the first-level fire risk threshold and less than or equal to the second-level fire risk threshold, it means that the photovoltaic area is at medium risk, recorded as B2. When the fire risk level value is greater than the second-level fire risk threshold, it means that there is a suspected fire in the photovoltaic area, recorded as B3.
[0077] Furthermore, the output results of the fire warning model are combined with the anomaly detection results of the image recognition model to achieve a multi-dimensional assessment of fire risks.
[0078] S3: Pre-trained image recognition model via convolutional neural network.
[0079] The pre-trained image recognition model includes preparing an image dataset containing normal and abnormal states of photovoltaic panels, training the image recognition model through the image dataset, performing data enhancement during the training process, and evaluating the accuracy of the image recognition model through cross-validation technology after the training is completed.
[0080] Among them, normal state images are collected during the normal operation of photovoltaic panels, while abnormal state images are obtained by simulating fire scenes or from existing fire cases. The dataset contains images taken under various lighting conditions, at different angles and distances, which gives the model good generalization ability.
[0081] During the training process, corresponding labels are added to fire-related abnormal features such as flame shape and smoke accumulation through manual annotation. In the training of the image recognition model, these labels are learned as key information, so that the image recognition model can accurately identify fire-related abnormal features.
[0082] S4: Collect on-site images of photovoltaic panels in real time and input the on-site images into the image recognition model.
[0083] Collecting on-site images of photovoltaic panels includes adjusting the size, brightness, and contrast of the collected images, using filters to remove image noise, and enhancing the images.
[0084] S5: Use the pre-trained image recognition model to perform object detection on the scene images collected in real time.
[0085] Target detection includes applying an image recognition model to identify abnormal features in an image, dividing the image into a normal state A1 and an abnormal state A2 based on the recognition and detection results, and marking the abnormal area.
[0086] Abnormal conditions include the appearance of flame shape or color features in the scene image, the appearance of smoke accumulation or diffusion features in the scene image, and the identification of temperature abnormalities using infrared thermal imaging sensors.
[0087] Furthermore, image recognition models provide another level of monitoring, increasing the accuracy and reliability of fire detection.
[0088] S6: Combine the fire risk prediction results with target detection and take targeted measures.
[0089] Targeted measures include continuing to conduct regular equipment and safety inspections without identifying target detection results when the PV area is at low risk B1.
[0090] When the photovoltaic area is at medium risk B2, the identification and detection results are further combined. When it is at B2A1, the monitoring frequency of on-site images is increased, the frequency of on-site image target detection is increased, and the data collected by the sensors is regularly analyzed and trended.
[0091] When in B2A2, strengthen real-time monitoring of the photovoltaic area, pay close attention to changes in on-site data, increase the frequency and scope of on-site inspections and patrols, improve the sensitivity of the early warning system, increase on-site monitoring points and coverage, and prepare for emergency response work, including inspecting and maintaining fire-fighting equipment.
[0092] When it is suspected fire B3, further judgment is made based on the image information and the smoke data and temperature data in the operating parameters.
[0093] Further judgment is made to determine whether there is a hot spot in the image of the abnormal state. If there is no hot spot in the scene image information, it is in B3A1. When there is a hot spot in the scene image, it is in B3A2, and further detection is performed by obtaining the evaluation index value.
[0094] It is further added that not all hot spots indicate the presence of a fire. Hot spots simply refer to higher temperature areas that exist in equipment or environments. They may be caused by a variety of reasons, including normal operating heat, friction, electrical failure, mechanical failure, etc. Although some hot spots may be potential fire risks, especially when they heat up rapidly or are located near flammable materials, not all hot spots are directly related to fire. Therefore, the present invention further determines whether a fire has occurred by obtaining an evaluation value. In addition, the image recognition results are combined with the sensor data to utilize the advantages of the two data sources to improve the accuracy and response speed of fire detection.
[0095] When in B3A1, the monitoring frequency of on-site images is increased, on-site inspections are continued and the density of monitoring is increased. Through historical data and trend analysis, it is observed whether there is a trend of hot spots emerging. When there is a trend, it is detected in advance by obtaining the evaluation index value.
[0096] It is further added that B3A1 focuses more on prevention and close monitoring, while B2A2 requires more direct response and possible intervention measures because specific abnormal conditions have been identified.
[0097] The evaluation index value is obtained as,
[0098] ,
[0099] Among them, a, o, and b represent adjustment factors.
[0100] When the comprehensive evaluation index value F is greater than the index value threshold Indicates that there is actually a fire. When the comprehensive index value F is less than or equal to the index value threshold Indicates a misjudgment.
[0101] When a fire actually occurs, emergency measures will be taken, including automatically notifying relevant personnel and fire departments via text messages and alarm systems, implementing automatic evacuation guidance, and guiding people in nearby areas to evacuate to designated safe areas.
[0102] When there is a misjudgment, a notification is sent to the relevant personnel to prompt the existence of a false alarm. The relevant personnel confirm and verify the false alarm and record the time and relevant data of the false alarm event.
[0103] Furthermore, decisions are made based on the prediction results and image recognition results, which realizes intelligent fire warning and response measures and improves the response speed and accuracy of the system.
[0104] Develop targeted measures based on prediction and image recognition results to minimize fire risks and losses.
[0105] Example 2, reference Figure 2 , which is an embodiment of the present invention, provides a system of a distributed photovoltaic fire warning method. A distributed photovoltaic fire warning system includes a sensor module, a fire warning module, a real-time image acquisition module and a targeted measures module.
[0106] The sensor module is responsible for collecting the operating parameters of the photovoltaic panels.
[0107] The fire warning module is used to receive the photovoltaic panel operating parameters collected by sensors and predict the degree of fire risk.
[0108] The real-time image acquisition module is used to collect on-site images of photovoltaic panels in real time as input to the image recognition model.
[0109] The targeted measures module is used to combine the fire risk level prediction results with target detection and take targeted measures.
[0110] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.
[0111] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0112] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.
[0113] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0114] Example 3. In this example, in order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. In the traditional photovoltaic fire early warning method, only basic sensors such as temperature and smoke sensors are generally relied on for monitoring. Although this method can provide certain early warnings, it has limitations in the accuracy and response speed of fire identification. Common problems include high false alarm rates, delayed response time, and inability to effectively visually confirm the fire scene. In order to overcome these shortcomings, the present invention proposes a distributed photovoltaic fire early warning method that combines operating parameter monitoring and image recognition, aiming to improve the accuracy and efficiency of the early warning system through more comprehensive data analysis and real-time image processing. The experimental results of the traditional invention and the present invention are shown in Table 1.
[0115] Table 1 Experimental effect comparison table
[0116]
[0117] By combining the operation parameter monitoring of photovoltaic panels and image recognition technology, the method of the present invention can analyze the on-site situation more comprehensively, identify key features such as flame shape and smoke diffusion, and greatly improve the accuracy of fire identification. By using the image recognition model pre-trained by the convolutional neural network to quickly process the real-time acquired images, the method of the present invention greatly shortens the time from detecting an abnormality to issuing an early warning, significantly improving the response speed of the early warning system. The method of the present invention combines multi-source data (such as temperature, smoke, current and voltage information) with image analysis, effectively reducing misunderstandings and false alarms that may be caused by relying on a single data source. In practical applications, the method of the present invention shows higher system stability, reduces operation and maintenance problems caused by system errors or instability, and improves system reliability.
[0118] 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 the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A distributed photovoltaic fire early warning method, characterized in that: include: Collect the operating parameters of the photovoltaic panels through sensors; Input the operating parameters of the photovoltaic panels into the fire warning model to predict the degree of fire risk; Pre-training image recognition models through convolutional neural networks; Collect on-site images of photovoltaic panels in real time and input the on-site images into the image recognition model; Use pre-trained image recognition models to detect objects in live images collected in real time; Combine the fire risk prediction results with target detection and take targeted measures; The fire warning model obtains the fire risk value according to the operating parameters of the photovoltaic panel, including smoothing the smoke data S(t), performing Hilbert transform on the smoothed smoke data, obtaining an analytical signal, and extracting the instantaneous amplitude A from the analytical signal. S ; Perform Fourier transform on the temperature data T(t), convert the temperature signal into the frequency domain, and find the frequency component with the largest amplitude in the spectrum; Perform Hilbert transform on the current data I(t) to obtain the instantaneous amplitude and extract the average value of the instantaneous amplitude As the instantaneous characteristics of current data; Perform wavelet transform on the voltage data V(t), decompose the voltage data into wavelet coefficients of different scales, and select the amplitude W with the highest wavelet coefficient energy. max (V) as the time-frequency characteristics of voltage data; The fire risk level value is expressed as, Among them, R represents the fire risk value, α, β, γ, δ, κ are adjustment factors, and F max (T) represents the maximum amplitude in the temperature data, A S represents the instantaneous amplitude of smoke data, Represents the average value of the instantaneous amplitude of the current data, W max (V) represents the time-frequency characteristics of voltage data, B s represents the baseline risk value; When the fire risk level value R is greater than 0 and less than or equal to the first-level fire risk threshold, it means that the photovoltaic area is at low risk, recorded as B1. When the fire risk level value is greater than the first-level fire risk threshold and less than or equal to the second-level fire risk threshold, it means that the photovoltaic area is at medium risk, recorded as B2. When the fire risk level value is greater than the second-level fire risk threshold, it means that there is a suspected fire in the photovoltaic area, recorded as B3.
2. A distributed photovoltaic fire early warning method as claimed in claim 1, characterized in that: The sensors include temperature sensors, smoke sensors, voltage sensors, current sensors and infrared thermal imaging sensors; The operating parameters are converted into electrical signals by fire detectors and transmitted to the fire warning model; The infrared thermal imaging sensor identifies whether there are hot spots on the photovoltaic panels at the site; The operating parameters include temperature data, smoke data, current, and voltage data.
3. A distributed photovoltaic fire early warning method as claimed in claim 2, characterized in that: The pre-trained image recognition model includes preparing an image data set containing normal and abnormal states of photovoltaic panels, training the image recognition model through the image data set, performing data enhancement during the training process, and evaluating the accuracy of the image recognition model through cross-validation technology after the training is completed.
4. A distributed photovoltaic fire early warning method as claimed in claim 3, characterized in that: The method of collecting on-site images of photovoltaic panels includes adjusting the size, brightness and contrast of the collected images, removing image noise using a filter, and enhancing the images.
5. A distributed photovoltaic fire early warning method as claimed in claim 4, characterized in that: The target detection includes applying an image recognition model to identify abnormal features in the image, dividing the image into a normal state A1 and an abnormal state A2 according to the recognition detection results, and marking the abnormal area; The abnormal state includes the appearance of flame shape or color features in the scene image, the appearance of smoke gathering or diffusion features in the scene image, and the identification of temperature abnormality using an infrared thermal imaging sensor.
6. A distributed photovoltaic fire early warning method as claimed in claim 5, characterized in that: The targeted measures include that when the photovoltaic area is at low risk B1, there is no need to identify the target detection results and continue to conduct regular equipment and safety inspections; When the photovoltaic area is at medium risk B2, the identification and detection results are further combined. When it is at B2A1, the monitoring frequency of the on-site images is increased, the frequency of target detection of the on-site images is increased, and the data collected by the sensors is regularly analyzed and trended. When in B2A2, strengthen real-time monitoring of the PV area, pay close attention to changes in on-site data, increase the frequency and scope of on-site inspections and patrols, improve the sensitivity of the early warning system, increase on-site monitoring points and coverage, and prepare for emergency response work, including inspection and maintenance of fire-fighting equipment; When it is suspected fire B3, further judgment is made by combining image information and smoke data and temperature data in the operating parameters; The further determination is to determine whether there is a hot spot in the image of the abnormal state, wherein there is no hot spot in the scene image information, then it is in B3A1, and when there is a hot spot in the scene image, then it is in B3A2, and further detection is performed by obtaining the evaluation index value; When in B3A1, the monitoring frequency of on-site images is increased, on-site inspections are continued, and the density of monitoring is increased. Through historical data and trend analysis, it is observed whether there is a trend of hot spots. When there is a trend, it is detected in advance by obtaining the evaluation index value; The evaluation index value is obtained as follows: Among them, a, o, and b represent adjustment factors; When the comprehensive evaluation index value F is greater than the index value threshold θ, it indicates that there is actually a fire. When the comprehensive index value F is less than or equal to the index value threshold θ, it indicates that there is a misjudgment. When a fire actually occurs, emergency measures will be taken, including automatically notifying relevant personnel and fire departments through text messages and alarm systems, and implementing automatic evacuation guidance to guide people in nearby areas to evacuate to a predetermined safe area; When there is a misjudgment, a notification is sent to the relevant personnel to prompt the existence of a false alarm. The relevant personnel confirm and verify the false alarm and record the time and relevant data of the false alarm event.
7. A system using a distributed photovoltaic fire early warning method as claimed in any one of claims 1 to 6, characterized in that: It includes sensor module, fire warning module, real-time image acquisition module and targeted measures module; The sensor module is responsible for collecting the operating parameters of the photovoltaic panel; The fire warning module is used to receive the photovoltaic panel operating parameters collected by the sensor and predict the degree of fire risk; The real-time image acquisition module is used to acquire on-site images of photovoltaic panels in real time as input to the image recognition model; The targeted measures module is used to combine the fire risk degree prediction result with the target detection and take targeted measures.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a distributed photovoltaic fire warning method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a distributed photovoltaic fire warning method as described in any one of claims 1 to 6 are implemented.
Citation Information
Patent Citations
Intelligent power distribution room fire early-stage monitoring method and system based on multi-source data fusion
CN116524674A
Fire safety monitoring system and method based on alarm area model
CN117095506A
Fire hazard camera shooting detection device, fire hazard detection alarm method and fire hazard detection alarm system
CN117253333A
Multi-scene electrical fire on-line monitoring method and system
CN117409529A
Dangerous and waste reservoir fire risk monitoring system based on deep learning
CN117475366A
