Photovoltaic power generation fire early warning and disposal method and system based on multi-modal perception

Abstract

The application discloses a photovoltaic power generation fire warning and disposal method and system based on multi-modal perception, relates to the technical field of fire warning, and is used for solving the problems of untimely fire hazard identification and insufficient disposal precision, calling the circuit loss level of each discharge device, detecting the comprehensive reverse power loss and discharge noise frequency of each discharge device to generate the fire warning coefficient of each discharge device, correcting the circuit loss level based on the fire warning coefficient, screening and marking the discharge device according to the corrected circuit loss level, calling the device number and ground potential data of the marked discharge device to evaluate the insulation degradation trend of the marked discharge device, detecting the hydrogen concentration data in the marked discharge device to calculate the insulation oil decomposition rate, combining the insulation degradation trend to determine whether the device number of the marked discharge device is stored into a fire warning disposal table, and improving the operation and maintenance efficiency and the active fire prevention and control capability.

Description

Technical Field

[0001] This invention relates to the field of fire early warning technology, and more specifically, to a method and system for early warning and handling of photovoltaic power generation fires based on multimodal sensing. Background Technology

[0002] As an important component of renewable energy, photovoltaic power generation systems have advantages such as being green, environmentally friendly, and highly efficient, and have been widely used in residential buildings, industrial parks, and large-scale photovoltaic power plants. A photovoltaic system mainly consists of photovoltaic modules, combiner boxes, transformer substations, inverters, control systems, and electrical wiring. Its operation involves high-voltage direct current, complex electrical connections, and various power conversion devices.

[0003] The existing technology has the following shortcomings:

[0004] Currently, existing technologies typically rely on a single electrical parameter for fire anomaly detection, making it difficult to capture insulation degradation and hydrogen accumulation characteristics in the early stages of discharge. This results in problems such as delayed early warning and high false alarm rates. Furthermore, the lack of a multimodal fusion analysis mechanism for electrical, acoustic, and gas data leads to untimely identification and insufficient accuracy in handling fire hazards in photovoltaic power generation systems, significantly reducing fire safety and operation and maintenance efficiency. Therefore, this paper proposes a method and system for early warning and handling of photovoltaic power generation fires based on multimodal perception.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for early warning and handling of photovoltaic power generation fires based on multimodal perception. This method and system solve the problems mentioned in the background art by employing a multimodal perception and dynamic risk assessment algorithm that integrates electrical parameters, acoustic signals, and gas characteristics.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for early warning and handling of photovoltaic power generation fires based on multimodal sensing, comprising the following steps:

[0008] Step S1: After the photovoltaic panel outputs current, call the circuit loss level of each discharge device, detect the output current data of each discharge device, and calculate the reverse power loss of each discharge device.

[0009] Step S2: Collect discharge acoustic data of each discharge device and count the discharge noise frequency. Generate the fire warning coefficient of each discharge device by combining the reverse power loss and the discharge noise frequency. Correct the circuit loss level based on the fire warning coefficient.

[0010] Step S3: Screen and mark the discharge equipment according to the corrected circuit loss level, retrieve the equipment number and grounding potential data of the marked discharge equipment through the photovoltaic power generation data log, and evaluate the insulation degradation trend of the marked discharge equipment based on the grounding potential data;

[0011] Step S4: Detect the hydrogen concentration data inside the marking discharge device, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the device number of the marking discharge device in the fire early warning and handling table based on the insulation degradation trend.

[0012] In a preferred embodiment, in step S1, after the photovoltaic module starts to perform photoelectric conversion and output current, the circuit loss level of each discharge device is called.

[0013] The circuit loss level is a set parameter determined by each discharge device during the operation calibration stage based on the device's rated power, conductor resistance characteristics, and long-term heat loss characteristics.

[0014] Output current data is collected in real time by current sensors and voltage sensors configured at the output terminals of each discharge device. The output current data includes instantaneous current intensity and output potential difference.

[0015] Instantaneous current intensity refers to the rate of charge flow across the conductor cross-section at the output end of a discharge device; its physical definition is the amount of charge passing through the conductor cross-section per unit time.

[0016] Output potential difference refers to the potential difference between the output terminal of a discharge device and the reference ground, which reflects the voltage level at the output terminal of the device during energy transmission.

[0017] In a preferred embodiment, in step S1, phase synchronization analysis is performed on the instantaneous current intensity and output potential difference to extract the phase difference between voltage and current in each discharge device;

[0018] Based on the output current data, the reverse power loss of each discharge device is calculated according to the power transmission theory formula. The specific calculation process is as follows: using the instantaneous current intensity, output potential difference, and phase difference within a preset sampling period as inputs, the reverse power loss is calculated, as shown in the following expression:

[0019] ;

[0020] in, For reverse power loss, These are the start and end times of the preset sampling period, respectively. For the output potential difference, Instantaneous current intensity This represents the phase difference.

[0021] In a preferred embodiment, in step S2, acoustic wave signals generated by the discharge device in the discharge state are collected by an acoustic sensor array configured at a fixed position inside the discharge device housing, and discharge acoustic wave data of the discharge device is obtained.

[0022] The frequency domain energy spectrum is obtained by extracting the energy distribution characteristics of the acoustic signal in the discharge acoustic wave data in different frequency ranges using fast Fourier transform.

[0023] The frequency domain energy spectrum is screened to identify high-energy discharge pulse signals generated during the discharge process:

[0024] If the amplitude of the sound wave of each frequency component in the frequency domain energy spectrum exceeds the preset noise threshold, it is determined to be a valid discharge pulse event.

[0025] The total number of high-energy discharge pulse events is counted and defined as the discharge noise frequency.

[0026] In a preferred embodiment, in step S2, the reverse power loss and discharge noise frequency are standardized to obtain the power loss factor and noise frequency factor.

[0027] The product of reverse power loss and discharge noise frequency is used as the fire warning coefficient.

[0028] The circuit loss level of the discharge equipment is corrected based on the fire early warning coefficient:

[0029] ;

[0030] in, This is the corrected circuit loss level. This refers to the circuit loss level of the discharge equipment during the initial operation phase. Fire early warning coefficient, This is the preset correction factor.

[0031] In a preferred embodiment, in step S3, the corrected circuit loss level is compared with a preset screening threshold:

[0032] When the corrected circuit loss level is greater than or equal to the preset screening threshold, the discharge device is screened and marked.

[0033] If the corrected circuit loss level is less than the preset screening threshold, it will not be screened or marked.

[0034] In a preferred embodiment, in step S3, the device number and grounding potential data of the marked discharge device are retrieved through the photovoltaic power generation data log;

[0035] The equipment number serves as a unique identifier for each discharge device;

[0036] The grounding potential data is the potential value of the output terminal of the discharge device relative to the reference ground;

[0037] Calculate the relative standard deviation of the grounding potential data time series, which is the ratio of the standard deviation of the grounding potential data to the mean, and use the relative standard deviation as the insulation degradation trend.

[0038] In a preferred embodiment, in step S4, the gas sensor uses the infrared spectroscopy detection principle to measure the volume fraction of hydrogen in the internal operating medium of the marking discharge device, thereby obtaining hydrogen concentration data inside the marking discharge device.

[0039] The increment of hydrogen concentration is obtained by subtracting the minimum value from the maximum value of the hydrogen concentration data;

[0040] The decomposition rate of insulating oil in the marked discharge device was calculated based on the increase in hydrogen concentration and the volume of insulating oil. ;

[0041] in, This represents the increase in hydrogen concentration. This is the preset solubility coefficient of hydrogen in insulating oil. This represents the volume of the insulating oil.

[0042] In a preferred embodiment, in step S4, the insulating oil decomposition rate and the insulation degradation trend are standardized to obtain the insulating oil decomposition factor and the insulation degradation factor.

[0043] The fire risk index of the marked discharge equipment is obtained by weighted summation of the insulating oil decomposition factor and the insulation degradation factor.

[0044] When the fire risk index is greater than or equal to the preset fire risk threshold, the device number of the marked discharge device is stored in the fire early warning and handling table.

[0045] When the fire risk index is less than the preset fire risk threshold, it is determined that there is no need to store the equipment number of the marked discharge device in the fire early warning and handling table.

[0046] The photovoltaic power generation fire early warning and response system based on multimodal perception includes a loss calculation module, an acoustic correction module, an insulation assessment module, and a maintenance judgment module. The functions of each module are as follows:

[0047] The loss calculation module is used to call the circuit loss level of each discharge device after the photovoltaic panel outputs current, detect the output current data of each discharge device and calculate the reverse power loss of each discharge device.

[0048] The acoustic correction module is used to collect the discharge acoustic data of each discharge device and count the discharge noise frequency. It combines the reverse power loss and the discharge noise frequency to generate the fire warning coefficient of each discharge device, and corrects the circuit loss level based on the fire warning coefficient.

[0049] The insulation assessment module is used to screen and mark discharge equipment according to the corrected circuit loss level. It retrieves the equipment number and grounding potential data of the marked discharge equipment from the photovoltaic power generation data log, and assesses the insulation degradation trend of the marked discharge equipment based on the grounding potential data.

[0050] The maintenance judgment module is used to detect the hydrogen concentration data inside the marked discharge equipment, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the equipment number of the marked discharge equipment in the fire early warning and handling table based on the insulation degradation trend.

[0051] The technical effects and advantages of this invention are as follows:

[0052] This invention generates a fire warning coefficient for each discharge device by calling the circuit loss level of each discharge device, detecting the comprehensive reverse power loss and discharge noise frequency of each discharge device, and correcting the circuit loss level based on the fire warning coefficient. The discharge devices are then screened and marked according to the corrected circuit loss level. The device number and grounding potential data of the marked discharge devices are retrieved to assess the insulation degradation trend of the marked discharge devices. The hydrogen concentration data inside the marked discharge devices is detected to calculate the insulating oil decomposition rate. Based on the insulation degradation trend, it is determined whether to store the device number of the marked discharge device in the fire warning response table. By integrating electrical, acoustic, and gas multimodal sensing data, a dynamic fire risk assessment system covering the entire chain from circuit anomalies and insulation degradation to insulating oil decomposition is constructed. This enables early and accurate warning of potential fire hazards in photovoltaic power generation systems, improving operation and maintenance efficiency and proactive fire prevention capabilities. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the implementation of the photovoltaic power generation fire early warning and response method based on multimodal sensing according to the present invention.

[0054] Figure 2 This is a module framework diagram of the photovoltaic power generation fire early warning and response system based on multimodal perception of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] This invention generates a fire warning coefficient for each discharge device by calling the circuit loss level of each discharge device, detecting the comprehensive reverse power loss and discharge noise frequency of each discharge device, and correcting the circuit loss level based on the fire warning coefficient. The discharge devices are then screened and marked according to the corrected circuit loss level. The device number and grounding potential data of the marked discharge devices are retrieved to assess the insulation degradation trend of the marked discharge devices. The hydrogen concentration data inside the marked discharge devices is detected to calculate the insulating oil decomposition rate. Based on the insulation degradation trend, it is determined whether to store the device number of the marked discharge device in the fire warning response table. By integrating electrical, acoustic, and gas multimodal sensing data, a dynamic fire risk assessment system covering the entire chain from circuit anomalies and insulation degradation to insulating oil decomposition is constructed, achieving early and accurate warnings of potential fire hazards in photovoltaic power generation systems.

[0057] Example 1

[0058] Please see Figure 1 A method for early warning and handling of photovoltaic power generation fires based on multimodal sensing includes the following steps:

[0059] Step S1: After the photovoltaic panel outputs current, call the circuit loss level of each discharge device, detect the output current data of each discharge device, and calculate the reverse power loss of each discharge device.

[0060] Step S2: Collect discharge acoustic data of each discharge device and count the discharge noise frequency. Generate the fire warning coefficient of each discharge device by combining the reverse power loss and the discharge noise frequency. Correct the circuit loss level based on the fire warning coefficient.

[0061] Step S3: Screen and mark the discharge equipment according to the corrected circuit loss level, retrieve the equipment number and grounding potential data of the marked discharge equipment through the photovoltaic power generation data log, and evaluate the insulation degradation trend of the marked discharge equipment based on the grounding potential data;

[0062] Step S4: Detect the hydrogen concentration data inside the marking discharge device, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the device number of the marking discharge device in the fire early warning and handling table based on the insulation degradation trend.

[0063] The specific implementation is as follows:

[0064] In step S1, after the photovoltaic module starts to output current through photoelectric conversion, the circuit loss level of each discharge device is called. The circuit loss level is a set parameter determined by each discharge device during the operation calibration stage based on the rated power of the device design, conductor resistance characteristics and long-term operation heat loss law. It is used to characterize the power consumption benchmark of the discharge device under rated operation state and reflect the degree of abnormal heat generation of the device and the potential fire risk level.

[0065] Output current data is collected in real time by current sensors and voltage sensors configured at the output terminals of each discharge device. The output current data includes instantaneous current intensity and output potential difference.

[0066] Instantaneous current intensity refers to the rate of charge flow across the conductor cross-section at the output end of a discharge device; its physical definition is the amount of charge passing through the conductor cross-section per unit time.

[0067] Output potential difference refers to the potential difference between the output terminal of the discharge device and the reference ground, which is used to reflect the voltage level at the output terminal of the device during energy transmission.

[0068] Phase synchronization analysis is performed on instantaneous current intensity and output potential difference to extract the phase difference between voltage and current in each discharge device. The phase difference is the relative phase offset angle between the voltage waveform and the current waveform, which characterizes the power factor change and energy transmission direction of the discharge device during operation.

[0069] It should be noted that a current sensor is a quantity measuring element used to detect the magnitude and variation of current in the conductor at the output end of a discharge device. Its function is to convert the current signal passing through the conductor into a proportional digital signal. A voltage sensor is a quantity detection device used to measure the potential difference between the output end of the discharge device and the system reference ground. Its function is to convert the measured voltage signal into a proportional digital signal. Phase synchronization analysis refers to the process of calculating the relative phase offset angle of the instantaneous current intensity and output potential difference at the same frequency component after acquiring the time series of the instantaneous current intensity and output potential difference at the output end of the discharge device, through time alignment, waveform filtering, and frequency domain transformation.

[0070] Based on the output current data, the reverse power loss of each discharge device is calculated according to the power transmission theory formula. The specific calculation process is as follows: using the instantaneous current intensity, output potential difference, and phase difference within a preset sampling period as inputs, the reverse power loss is calculated, as shown in the following expression:

[0071] ;

[0072] in, For reverse power loss, These are the start and end times of the sampling period, respectively. For the output potential difference, Instantaneous current intensity For phase difference, This is a power factor correction term used to extract the effective active power components of current and voltage in the opposite energy flow direction. When near hour, A value close to 1 indicates a significant reverse flow of electrical energy, meaning that some of the output power is fed back to the next stage circuit or the photovoltaic bus, leading to a decrease in the overall energy efficiency of the system and the risk of localized overheating.

[0073] Reverse power loss reflects the active power of energy flowing backward from the output terminal to the upper circuit during the operation of the discharge equipment. The larger the value, the more obvious the reverse energy flow and the greater the circuit power loss, which may lead to local overheating, conductor aging and stress accumulation in the insulating medium, thus causing fire hazards. The smaller the value, the more stable the output energy transmission direction of the equipment, the normal circuit operation, the power loss maintained within the design range, and the low fire risk.

[0074] In step S2, discharge acoustic wave data during the operation of each discharge device is collected. Specifically, the acoustic wave signals generated by the discharge device in the discharge state are collected in real time by an acoustic sensor array configured at a fixed position inside the discharge device housing, thereby obtaining the discharge acoustic wave data of the discharge device.

[0075] It should be noted that an acoustic sensor array is a multi-point acoustic wave sensing structure composed of multiple acoustic sensing units arranged in a spatial distribution pattern; the acoustic sensor is based on the principle of miniature capacitive sound pickup, which converts the mechanical acoustic vibration generated during the discharge process into a quantifiable electrical signal, thereby realizing the digital recording of acoustic signals; the principle of miniature capacitive sound pickup is a physical measurement mechanism based on capacitance change to realize sound wave sensing. It generates an electrical signal proportional to the sound wave vibration by causing a change in charge distribution due to the change in the spacing of the capacitor structure under the action of sound pressure.

[0076] Frequency domain analysis is performed on the discharge acoustic wave data. This involves extracting the energy distribution characteristics of the acoustic signal in different frequency ranges from the discharge acoustic wave data using Fast Fourier Transform to obtain the frequency domain energy spectrum. The frequency domain energy spectrum is then filtered to identify high-energy discharge pulse signals generated during the discharge process.

[0077] If the amplitude of the sound wave of each frequency component in the frequency domain energy spectrum exceeds the preset noise threshold, it is determined to be a valid discharge pulse event.

[0078] The total number of high-energy discharge pulse events is counted and defined as the discharge noise frequency.

[0079] The frequency of discharge noise reflects the activity level of micro-discharge inside the insulation structure or at the port connection of the discharge equipment. The higher the value, the more frequent the partial discharge inside the discharge equipment, the higher the risk of arc formation, and the greater the potential fire hazard.

[0080] It should be noted that Fast Fourier Transform (FFT) is a digital signal processing method that converts the time-domain signal of a discharge acoustic wave into a frequency-domain signal. The Discrete Fourier Transform (DFT) algorithm decomposes the amplitude and phase information of the continuous or discrete time-domain signal at different frequency components, thereby obtaining the distribution characteristics of the acoustic wave energy in the frequency domain. The noise threshold is set by collecting reference acoustic wave data under the condition that the photovoltaic power generation equipment is operating normally and there is no abnormal discharge, statistically analyzing its amplitude distribution and calculating the mean and standard deviation. The sum of the mean and standard deviation is used as the noise threshold.

[0081] The reverse power loss and discharge noise frequency are standardized separately to obtain the power loss factor and noise frequency factor.

[0082] The product of reverse power loss and discharge noise frequency is used as the fire warning coefficient.

[0083] The fire warning coefficient reflects the comprehensive characteristics of the abnormal electrical energy consumption and the degree of arc formation during the operation of the discharge equipment. The larger the value, the more significant the reverse power loss and the more frequent the discharge noise events, indicating that the internal partial discharge is active and the fire risk is high. The smaller the value, the more the electrical energy consumption of the discharge equipment is close to the rated level and the discharge activity is less, the circuit is operating stably and is in normal working condition.

[0084] It should be noted that standardization refers to the process of mapping raw data of different physical quantities or different dimensions to a uniform dimension, uniform numerical range or uniform statistical distribution through a specific mathematical transformation. Standardization methods include, but are not limited to, standard linear transformation based on interval scaling, Z-Score standardization based on statistics or normalization method based on nonlinear mapping function. The application methods of standardization will not be elaborated here.

[0085] The circuit loss level of the discharge equipment is corrected based on the fire early warning coefficient to form a circuit loss status that reflects the current actual operating state. The correction formula is as follows:

[0086] ;

[0087] in, This is the corrected circuit loss level. This refers to the circuit loss level of the discharge equipment during the initial operation phase. Fire early warning coefficient, This is a correction factor used to control the correction magnitude and ensure the stability of the level mapping.

[0088] It should be noted that the correction coefficient is a proportional factor used to moderately adjust the impact of the fire warning coefficient on the original circuit loss level. It is used to control the correction range and avoid excessive deviation of the loss level due to a single abnormal data, thereby ensuring that the corrected circuit loss level maps the actual operating state of the equipment within a reasonable range. Its setting is based on the statistical distribution of the original circuit loss level of the discharge equipment under rated operating conditions and the fire warning coefficient in historical operating data, calculating the change range and extreme value range, and selecting the correction coefficient based on the change range and extreme value range.

[0089] In step S3, the discharge devices are screened and marked according to the corrected circuit loss level;

[0090] The corrected circuit loss level is compared with the preset screening threshold:

[0091] When the corrected circuit loss level is greater than or equal to the preset screening threshold, it is determined that the discharge equipment has abnormal heating and fire risk, and the discharge equipment is screened and marked.

[0092] If the corrected circuit loss level is less than the preset screening threshold, the fire risk of the discharge equipment is determined to be controllable, and it will not be screened or marked.

[0093] It should be noted that the method for setting the screening threshold is based on the dynamic distribution characteristics of circuit loss levels and the principle of quantile statistics. A time series sample set of circuit loss levels of discharge equipment is established in long-term operation data. According to the operation risk control strategy, a target quantile (e.g., 90% to 95%) is selected, and the circuit loss level corresponding to the target quantile is used as the screening threshold.

[0094] The equipment number and grounding potential data of the marked discharge equipment are retrieved from the photovoltaic power generation data log;

[0095] The device number serves as a unique identifier for each discharge device, used to enable data association, device tracking, and subsequent maintenance operations.

[0096] Grounding potential data is the potential value of the output terminal of the discharge equipment relative to the reference ground. It reflects the insulation integrity between the discharge equipment and the ground and the grounding continuity. If the grounding potential fluctuates or changes abruptly, it indicates that the insulation layer of the discharge equipment has failed, thus causing grounding imbalance.

[0097] Calculate the relative standard deviation of the grounding potential data time series, which is the ratio of the standard deviation of the grounding potential data to the mean, and use it as the insulation degradation trend;

[0098] The insulation degradation trend reflects the fluctuation state of the grounding potential. The larger the value, the more severe the fluctuation of the grounding potential of the discharge equipment, the more frequent the insulation deterioration or partial discharge activity, and the higher the fire risk. The smaller the value, the more stable the grounding potential, the better the insulation condition, and the lower the fire risk.

[0099] It should be noted that photovoltaic power generation data logs refer to a collection of records that are collected in real time, digitally processed, and stored in a time series for key parameters of photovoltaic power generation equipment during operation.

[0100] In step S4, the hydrogen concentration data in the operating medium inside the marking discharge device is detected. The combustible gas components generated during the operation of the discharge device are monitored in real time by a gas sensor installed inside the discharge device housing. The gas sensor uses the infrared spectroscopy detection principle to measure the volume fraction of hydrogen components in the mixed gas to obtain the hydrogen concentration data inside the marking discharge device.

[0101] Hydrogen concentration data reflects the degree of decomposition of the insulating oil inside the marked discharge equipment under the influence of heat, electrical breakdown, or partial discharge, characterizing the state of insulation degradation and potential fire risk.

[0102] It should be noted that a gas sensor is a detection device that can detect the concentration of a specific gas component in air or a confined space in real time and output a quantifiable electrical signal; the infrared spectroscopy detection principle is a detection method based on the absorption characteristics of gas molecules to specific wavelengths of infrared light. It uses a specific wavelength beam emitted by an infrared light source to pass through the gas to be measured. When the gas molecules absorb light energy within this wavelength range, the change in the intensity of the transmitted light is measured, and the concentration value of the gas is calculated through a calibration curve.

[0103] The increment of hydrogen concentration is obtained by subtracting the minimum value from the maximum value of the hydrogen concentration data;

[0104] The decomposition rate of the insulating oil in the marked discharge device is calculated based on the increase in hydrogen concentration and the volume of insulating oil, as expressed below:

[0105] ;

[0106] in, This represents the increase in hydrogen concentration. This is the solubility coefficient of hydrogen in insulating oil. This represents the volume of the insulating oil.

[0107] The insulating oil decomposition rate quantifies the rate of thermochemical decomposition of a unit volume of insulating oil during the operating cycle. It reflects the degree of damage to the insulating medium caused by partial discharge or thermal breakdown. The higher the value, the greater the fire risk.

[0108] The decomposition rate of insulating oil and the trend of insulation degradation were standardized to obtain the decomposition factor of insulating oil and the degradation factor of insulation.

[0109] The fire risk index of the marked discharge equipment is obtained by weighted summation of the insulating oil decomposition factor and the insulation degradation factor, as shown in the following expression:

[0110] ;

[0111] in, This is a fire risk index. This is the decomposition factor of insulating oil. For insulation degradation factor, and These are the weighting coefficients.

[0112] The fire risk index reflects the overall health status of the insulation system of the marked discharge equipment. The higher the value, the higher the degree of decomposition of the insulating oil inside the equipment and the significant fluctuation of the grounding potential, indicating that the insulation material has undergone obvious degradation, the partial discharge is active, and the potential fire risk is high. The lower the value, the more slight the decomposition of the insulating oil and the more stable the grounding potential, indicating good insulation status and safe and reliable operation of the equipment.

[0113] The fire risk index is compared with the preset fire risk threshold:

[0114] When the fire risk index is greater than or equal to the preset fire risk threshold, it indicates that the insulation system of the marked discharge equipment has deteriorated significantly and there is a fire hazard. It is determined that the equipment number of the marked discharge equipment should be stored in the fire early warning and handling table for subsequent inspection and maintenance operations.

[0115] When the fire risk index is less than the preset fire risk threshold, it indicates that the insulation system of the discharge equipment is in relatively good condition, the degree of decomposition of the insulating oil is low and the grounding potential is stable, the fire hazard is low, and it is determined that there is no need to store the equipment number of the marked discharge equipment in the fire early warning and handling table.

[0116] It should be noted that the weighting coefficients are used to balance the contribution ratios of the insulating oil decomposition factor and the insulation degradation factor to the fire risk index. Based on a large amount of historical operating data and fire cases, the sensitivity and early warning capability of the insulating oil decomposition rate and grounding potential fluctuations to equipment fires are analyzed to determine the relative importance of each factor, which is then used as the weighting coefficients. The fire risk threshold is set by collecting the fire risk index of the marked discharge equipment, analyzing its numerical distribution characteristics, and using the quantile method (such as the 95th percentile). The fire early warning and handling table is a structured database table that stores the marked discharge equipment number and related maintenance information, used to support the inspection management and maintenance scheduling of the photovoltaic power generation system.

[0117] Example 2, as Figure 2As shown, the photovoltaic power generation fire early warning and response system based on multimodal perception is used to implement a photovoltaic power generation fire early warning and response method based on multimodal perception. It includes a loss calculation module, an acoustic correction module, an insulation assessment module, and a maintenance judgment module. The functions of each module are as follows:

[0118] The loss calculation module is used to call the circuit loss level of each discharge device after the photovoltaic panel outputs current, detect the output current data of each discharge device and calculate the reverse power loss of each discharge device.

[0119] The acoustic correction module is used to collect the discharge acoustic data of each discharge device and count the discharge noise frequency. It combines the reverse power loss and the discharge noise frequency to generate the fire warning coefficient of each discharge device, and corrects the circuit loss level based on the fire warning coefficient.

[0120] The insulation assessment module is used to screen and mark discharge equipment according to the corrected circuit loss level. It retrieves the equipment number and grounding potential data of the marked discharge equipment from the photovoltaic power generation data log, and assesses the insulation degradation trend of the marked discharge equipment based on the grounding potential data.

[0121] The maintenance judgment module is used to detect the hydrogen concentration data inside the marked discharge equipment, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the equipment number of the marked discharge equipment in the fire early warning and handling table based on the insulation degradation trend.

[0122] Furthermore, the following procedures shall be performed on each marked discharge device stored in the fire early warning response table;

[0123] For example, call the device number of each marked discharge device, find the fire monitoring camera with the corresponding device number, and enable real-time monitoring;

[0124] The location of the fire monitoring cameras was determined by the researchers based on the spatial distribution of the discharge equipment within the photovoltaic array and the fire heat source diffusion path around the discharge equipment. The specific number and model of the cameras are not limited and will not be elaborated here.

[0125] Furthermore, the fire monitoring camera is a video monitoring execution device of the remote response system. It continuously acquires composite images of the marked device in visible light and weak light through its built-in optical imaging module and red and blue photosensitive detection unit. The remote response system includes video monitoring execution device, response medium spraying device, spraying pressure control module, response command receiving and response module, and remote link communication module, etc.

[0126] The real-time monitoring images are transmitted to the visualization port of the local response system, and the fire monitoring camera is used to monitor and collect the fire light signals in the monitoring images in real time.

[0127] The local response system is used to perform fire signal analysis, threshold comparison, and response command triggering.

[0128] Specifically, the fire signal refers to a cluster of high-brightness pixels generated by combustion in the monitoring screen. The ratio of its brightness component to its chromaticity component meets the preset fire spectral characteristics. The preset fire spectral characteristics are that the energy of the red component is significantly higher than that of the green and blue components, the proportion of red reaches a set threshold, and the overall brightness level exceeds the set minimum fire brightness standard. The specific set threshold and minimum fire brightness standard are common knowledge to those skilled in the art and will not be elaborated here.

[0129] The fire signal is captured and the built-in clock is used to accumulate the duration of the fire signal to obtain the duration of the fire.

[0130] The duration of the fire was compared and analyzed with the preset handling threshold.

[0131] If the duration of the fire is greater than or equal to the disposal threshold, the disposal medium spraying device of the remote disposal system will be activated to perform disposal operations on the marked device.

[0132] If the duration of the fire is less than the disposal threshold, there is no need to activate the disposal medium spraying equipment of the remote disposal system.

[0133] It should be noted that the treatment threshold was set by our researchers based on the statistical value of the combustion diffusion rate and the heat resistance limit of the marking equipment in the fire scene, and will not be elaborated here.

[0134] Furthermore, this invention can introduce a multimodal collaborative sensing structure consisting of an acoustic wave detection unit, a radar parameter acquisition unit, a visualization imaging module, and an AI computing module into the decision-making process. The acoustic wave detection unit samples the gas pulsation characteristics of the fire field near the marking device based on changes in sound pressure disturbance waveforms. The radar parameter acquisition unit uses pulse echo delay and its energy attenuation curve to perform non-contact detection of local fire source morphology changes and smoke density distribution. The visualization imaging module provides a continuous sequence of composite images of visible light and low illumination. The AI ​​computing module performs fusion modeling and dynamic weight allocation on the above-mentioned multi-source features. Through a multimodal fusion function based on convolutional feature tensors, acoustic disturbance spectra, and radar echo matrices, a fire risk characterization quantity is generated and used as an auxiliary means to determine whether to enter the response process. This multimodal collaborative sensing structure is only used to enhance the stability of fire signal recognition and the reliability of response decision-making. Its specific hardware model and algorithm implementation are not limited and are common knowledge to those skilled in the art, and will not be elaborated here.

[0135] Furthermore, the treatment medium injection device is a pressurized liquid storage tank injection mechanism containing treatment medium, wherein the treatment medium is a fire extinguishing inhibitor used to quickly cover the combustion source and suppress continued combustion;

[0136] Furthermore, the treatment operation involves directional spraying, covering, and cooling of the treatment medium along the outer surface of the marking equipment and its surrounding potential combustion diffusion area to interrupt the combustion chain and reduce the local temperature. The specific spray intensity and duration are not limited, but are determined by the experimenters based on the brightness change trend and duration of the on-site flame signal, and will not be elaborated here.

[0137] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0138] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0139] In this document, the singular forms “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that terms such as “comprising / including” or “having” specify the presence of the stated features, integrals, steps, operations, components, parts, or combinations thereof, but do not preclude the possibility of the presence or addition of one or more other features, integrals, steps, operations, components, parts, or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0140] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0141] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for early warning and handling of photovoltaic power generation fires based on multimodal sensing, characterized in that: Includes the following steps: Step S1: After the photovoltaic panel outputs current, call the circuit loss level of each discharge device, detect the output current data of each discharge device, and calculate the reverse power loss of each discharge device. Step S2: Collect discharge acoustic data of each discharge device and count the discharge noise frequency. Generate the fire warning coefficient of each discharge device by combining the reverse power loss and the discharge noise frequency. Correct the circuit loss level based on the fire warning coefficient. Step S3: Screen and mark the discharge equipment according to the corrected circuit loss level, retrieve the equipment number and grounding potential data of the marked discharge equipment through the photovoltaic power generation data log, and evaluate the insulation degradation trend of the marked discharge equipment based on the grounding potential data; Step S4: Detect the hydrogen concentration data inside the marking discharge device, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the device number of the marking discharge device in the fire early warning and handling table based on the insulation degradation trend.

2. The photovoltaic power generation fire early warning and response method based on multimodal sensing according to claim 1, characterized in that: In step S1, after the photovoltaic module starts to output current through photoelectric conversion, the circuit loss level of each discharge device is called. The circuit loss level is a set parameter determined by each discharge device during the operation calibration stage based on the device's rated power, conductor resistance characteristics, and long-term heat loss characteristics. Output current data is collected in real time by current sensors and voltage sensors configured at the output terminals of each discharge device. The output current data includes instantaneous current intensity and output potential difference. Instantaneous current intensity refers to the rate of charge flow across the conductor cross-section at the output end of a discharge device; its physical definition is the amount of charge passing through the conductor cross-section per unit time. Output potential difference refers to the potential difference between the output terminal of a discharge device and the reference ground, which reflects the voltage level at the output terminal of the device during energy transmission.

3. The photovoltaic power generation fire early warning and handling method based on multimodal sensing according to claim 2, characterized in that: In step S1, phase synchronization analysis is performed on the instantaneous current intensity and output potential difference to extract the phase difference between voltage and current in each discharge device; Based on the output current data, the reverse power loss of each discharge device is calculated according to the power transmission theory formula. The specific calculation process is as follows: using the instantaneous current intensity, output potential difference, and phase difference within a preset sampling period as inputs, the reverse power loss is calculated, as shown in the following expression: ； in, For reverse power loss, These are the start and end times of the preset sampling period, respectively. For the output potential difference, Instantaneous current intensity This represents the phase difference.

4. The photovoltaic power generation fire early warning and handling method based on multimodal sensing according to claim 1, characterized in that: In step S2, the acoustic wave signal generated by the discharge device in the discharge state is collected by an acoustic sensor array configured at a fixed position inside the discharge device housing, and the discharge acoustic wave data of the discharge device is obtained. The frequency domain energy spectrum is obtained by extracting the energy distribution characteristics of the acoustic signal in the discharge acoustic wave data in different frequency ranges using fast Fourier transform. The frequency domain energy spectrum is screened to identify high-energy discharge pulse signals generated during the discharge process: If the amplitude of the sound wave of each frequency component in the frequency domain energy spectrum exceeds the preset noise threshold, it is determined to be a valid discharge pulse event. The total number of high-energy discharge pulse events is counted and defined as the discharge noise frequency.

5. The photovoltaic power generation fire early warning and response method based on multimodal sensing according to claim 4, characterized in that: In step S2, the reverse power loss and discharge noise frequency are standardized to obtain the power loss factor and noise frequency factor. The product of reverse power loss and discharge noise frequency is used as the fire warning coefficient. The circuit loss level of the discharge equipment is corrected based on the fire early warning coefficient: ； in, This is the corrected circuit loss level. This refers to the circuit loss level of the discharge equipment during the initial operation phase. Fire early warning coefficient, This is the preset correction factor.

6. The photovoltaic power generation fire early warning and handling method based on multimodal sensing according to claim 1, characterized in that: In step S3, the corrected circuit loss level is compared with a preset screening threshold: When the corrected circuit loss level is greater than or equal to the preset screening threshold, the discharge device is screened and marked. If the corrected circuit loss level is less than the preset screening threshold, it will not be screened or marked.

7. The photovoltaic power generation fire early warning and handling method based on multimodal perception according to claim 6, characterized in that: In step S3, the equipment number and grounding potential data of the marked discharge equipment are retrieved through the photovoltaic power generation data log; The equipment number serves as a unique identifier for each discharge device; The grounding potential data is the potential value of the output terminal of the discharge device relative to the reference ground; Calculate the relative standard deviation of the grounding potential data time series, which is the ratio of the standard deviation of the grounding potential data to the mean, and use the relative standard deviation as the insulation degradation trend.

8. The photovoltaic power generation fire early warning and handling method based on multimodal sensing according to claim 1, characterized in that: In step S4, the gas sensor uses the infrared spectroscopy detection principle to measure the volume fraction of hydrogen in the internal operating medium of the marking discharge device, and obtains the hydrogen concentration data inside the marking discharge device. The increment of hydrogen concentration is obtained by subtracting the minimum value from the maximum value of the hydrogen concentration data; The decomposition rate of insulating oil in the marked discharge device was calculated based on the increase in hydrogen concentration and the volume of insulating oil. ; in, This represents the increase in hydrogen concentration. This is the preset solubility coefficient of hydrogen in insulating oil. This represents the volume of the insulating oil.

9. The photovoltaic power generation fire early warning and handling method based on multimodal sensing according to claim 8, characterized in that: In step S4, the decomposition rate of insulating oil and the trend of insulation degradation are standardized to obtain the decomposition factor of insulating oil and the degradation factor of insulation. The fire risk index of the marked discharge equipment is obtained by weighted summation of the insulating oil decomposition factor and the insulation degradation factor. When the fire risk index is greater than or equal to the preset fire risk threshold, the device number of the marked discharge device is stored in the fire early warning and handling table. When the fire risk index is less than the preset fire risk threshold, it is determined that there is no need to store the equipment number of the marked discharge device in the fire early warning and handling table.

10. A photovoltaic power generation fire early warning and response system based on multimodal perception, used to implement the photovoltaic power generation fire early warning and response method based on multimodal perception as described in any one of claims 1-9, characterized in that: It includes a loss calculation module, an acoustic correction module, an insulation assessment module, and a maintenance judgment module. The functions of each module are as follows: The loss calculation module is used to call the circuit loss level of each discharge device after the photovoltaic panel outputs current, detect the output current data of each discharge device and calculate the reverse power loss of each discharge device. The acoustic correction module is used to collect the discharge acoustic data of each discharge device and count the discharge noise frequency. It combines the reverse power loss and the discharge noise frequency to generate the fire warning coefficient of each discharge device, and corrects the circuit loss level based on the fire warning coefficient. The insulation assessment module is used to screen and mark discharge equipment according to the corrected circuit loss level. It retrieves the equipment number and grounding potential data of the marked discharge equipment from the photovoltaic power generation data log, and assesses the insulation degradation trend of the marked discharge equipment based on the grounding potential data. The maintenance judgment module is used to detect the hydrogen concentration data inside the marked discharge equipment, calculate the insulating oil decomposition rate based on the hydrogen concentration data, and determine whether to store the equipment number of the marked discharge equipment in the fire early warning and handling table based on the insulation degradation trend.

Images