Perfluorohexanone fire extinguishing agent fire extinguishing method and system applied to power switch cabinet fire

Abstract

The invention discloses a perfluorohexanone fire extinguishing agent fire extinguishing method and system applied to a power switch cabinet fire, and belongs to the technical field of power system energy storage equipment.The method comprises the steps that an arc light signal is detected through an arc light detection device, and a partial discharge signal is detected through an ultrasonic partial discharge detection device; meanwhile, temperature data and smoke concentration data in the cabinet are detected through a temperature and smoke detection device; signal fusion is executed, and a comprehensive fire hazard grade judgment value is generated; comparing the comprehensive fire hazard grade judgment value with a preset fire hazard grade threshold value, and judging the fire hazard grade; and the release mechanism is controlled to release the perfluorohexanone fire extinguishing agent. According to the invention, through multi-signal fusion and a physical power model, the fire stage is dynamically discriminated, the dosage of the perfluorohexanone is adaptively controlled, the fire extinguishing accuracy and efficiency are improved, the false alarm rate is reduced, and data recording and intelligent management are supported.

Description

Technical Field

[0001] The present invention relates to the technical field of power system energy storage equipment, and in particular to a fire extinguishing method and system using a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet. Background Art

[0002] Existing technologies typically use smoke detectors, temperature sensors, or infrared detection equipment to monitor potential fires inside power switchgear cabinets. These traditional solutions rely primarily on the alarm signal from a single sensor, triggering an alarm or activating fire extinguishing devices when an anomaly is detected. Some systems also incorporate ultrasonic partial discharge detection to detect potential electrical faults within high-voltage switchgear. Existing technologies typically employ gaseous extinguishing agents, such as heptafluoropropane, carbon dioxide, or perfluorohexanone, through concentrated release.

[0003] However, the above-mentioned existing technologies have obvious limitations. On the one hand, the detection method of a single sensor has the problem of missed detection or high false alarm rate, which cannot effectively judge the development stage and actual severity of the fire; on the other hand, the fire extinguishing strategy is usually a fixed dose or a simple single release, which lacks the ability to adjust the release amount according to the dynamic level of the fire, which easily leads to insufficient or wasteful use of fire extinguishing agents and cannot adapt to the variability of fire evolution. In addition, the signal processing in the existing methods is relatively rough, and there is a lack of deep fusion and comprehensive analysis of different detection signals, resulting in inaccurate judgment of the fire level, which affects the timeliness and effectiveness of fire extinguishing decisions. Therefore, how to achieve accurate fire detection based on multi-source signal fusion and dynamically adjust the fire extinguishing agent release strategy accordingly has become a technical problem that needs to be solved urgently in this field. Summary of the Invention

[0004] To solve the above technical problems, a method for extinguishing fires in power switch cabinets using a perfluorohexanone fire extinguishing agent is proposed, comprising: arranging an arc light detection device, an ultrasonic partial discharge detection device, and a temperature and smoke detection device in the power switch cabinet; detecting arc light signals by the arc light detection device, detecting partial discharge signals by the ultrasonic partial discharge detection device, and detecting temperature data and smoke concentration data in the cabinet by the temperature and smoke detection device; synchronously transmitting the arc light signal, partial discharge signal, temperature data, and smoke concentration data to a data processing module, performing signal fusion, and generating a comprehensive fire level judgment value; comparing the comprehensive fire level judgment value with a preset fire level threshold to judge the fire level; when the fire level is lower than or equal to a first preset level threshold, controlling a release mechanism to release a first dose of perfluorohexanone fire extinguishing agent; when the fire level is higher than the first preset level threshold, controlling the release mechanism to release a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose; after the fire extinguishing agent is released, recording the detection data and release parameters, and outputting them to an external monitoring terminal through a communication module.

[0005] As a preferred solution of the perfluorohexanone fire extinguishing agent fire extinguishing method described in the present invention for power switch cabinet fires, the execution of signal fusion includes converting the detected arc light signal into arc light radiation power; converting the detected partial discharge signal into partial discharge power, and making corrections based on the changing trend of the partial discharge signal; calculating the heat release power based on the detected temperature data; and calculating the smoke combustion release power based on the detected smoke concentration data.

[0006] As a preferred solution of the method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent according to the present invention, the arc radiation power is calculated as follows:

[0007]

[0008] Where P1(t) represents the arc light radiation power; S1(t) represents the arc light signal intensity; d eff (t) represents the equivalent detection distance; η1(t) represents the detector response correction coefficient; the calculated partial discharge power is expressed as,

[0009]

[0010] Where P2(t) represents the partial discharge power; q(t) represents the discharge charge; U(t) represents the discharge voltage; Δt represents the measurement time interval; λ2 represents the partial discharge change rate correction coefficient; represents the charge change rate; the heat release power is calculated as,

[0011]

[0012] Where P3(t) represents the heat release power; m(t) represents the mass of the burning material; c s represents specific heat capacity; Indicates the temperature change rate; the smoke combustion release power is calculated as,

[0013]

[0014] Wherein, P4(t) represents the power released by smoke combustion; k4 represents the conversion coefficient of smoke growth energy; Indicates the rate of change of smoke concentration.

[0015] As a preferred embodiment of the method for extinguishing a fire with a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet according to the present invention, before generating the comprehensive fire level judgment value, the method further includes calculating the time change rate of each of the arc radiation power P1(t), the partial discharge power P2(t), the heat release power P3(t) and the smoke combustion release power P4(t), respectively, to obtain and Calculate the sum of the arc radiation power change rate and the partial discharge power change rate to obtain:

[0016]

[0017] Calculate the sum of the heat release power change rate and the smoke combustion release power change rate to obtain:

[0018]

[0019] Compare R1(t) with R2(t) to determine the fire stage, which can be expressed as:

[0020]

[0021] Wherein, k is a proportional coefficient, which is used to adjust the comparative sensitivity between the sum of the arc radiation power change rate and the partial discharge power change rate and the sum of the heat release power change rate and the smoke combustion release power change rate, and is determined based on fire history data or training data.

[0022] As a preferred embodiment of the perfluorohexanone fire extinguishing agent fire extinguishing method for power switch cabinet fires according to the present invention, the generating of the comprehensive fire level judgment value includes: selecting the current stage s according to the fire stage judgment result, the corresponding parameter is the early stage or the development stage; based on the current stage s, selecting the first set of change rate weighting coefficients a i (s) and the second set of cumulative weight coefficients b i(s), and time window T(s), where i = 1 to 4; according to P1(t) to P4(t) and their change rates, respectively calculate the first judgment value D1(t) and the second judgment value D2(t); jointly solve the first judgment value D1(t) and the second judgment value D2(t) to obtain the comprehensive fire level judgment value E(t), which is expressed as:

[0023] E(t)=D1(t) p D2(t) q

[0024] Among them, p and q are coefficients determined based on historical fire data.

[0025] As a preferred embodiment of the fire extinguishing method using perfluorohexanone fire extinguishing agent for electric switch cabinet fires according to the present invention, wherein:

[0026] The first judgment value is expressed as,

[0027]

[0028] Where i represents the detection parameter number; a i (s) represents the change weight coefficient of the i-th parameter under the current fire stage; s represents the current fire stage, early stage or development stage; P i (t) represents the i-th physical power parameter; represents the first-order derivative of the i-th physical power parameter with respect to time t; ∑ represents the sum of the four physical parameters;

[0029] The second judgment value is expressed as,

[0030]

[0031] Among them, b i (s) represents the cumulative weight coefficient of the i-th parameter at the current fire stage s; P i (τ) represents the value of the i-th physical power parameter at time τ; T(s) represents the length of the integration time window determined according to stage s; τ represents the integration variable, integrated from tT(s) to t; ∫ represents the integration operation.

[0032] As a preferred embodiment of the fire extinguishing method of the present invention using perfluorohexanone fire extinguishing agent for electric switch cabinet fire, wherein: a i (s) and b iThe determination process of (s) includes collecting historical data containing different fire stages and levels; training the target error functions of the first judgment value D1(t) and the second judgment value D2(t) according to the fire stage, so that the mean square error between D1(t) and D2(t) and the historical fire level label is minimized; using ridge regression or Bayesian regression for parameter fitting, and setting coefficient boundaries to prevent overfitting; T(s) is determined by analyzing the signal period of historical data, preferably using Fourier transform or autocorrelation analysis, and selecting a time range covering the main energy change period.

[0033] Another object of the present invention is to provide a system that improves the efficiency of short-term high-frequency energy storage. The present invention solves the problem that existing fire detection methods mostly rely on a single sensor signal, resulting in insufficient detection accuracy, missed reports and false alarms, and the inability to accurately judge the evolution of the fire. Existing fire extinguishing systems generally use a fixed dose to release fire extinguishing agents, and lack the ability to dynamically adjust the release dose according to the actual development of the fire, resulting in low efficiency in the use of fire extinguishing agents or insufficient fire extinguishing effects. The existing technology lacks a method for deep fusion analysis of multi-source detection data, and cannot effectively integrate information from different physical quantities to improve the accuracy of fire level judgment. In the existing fire response process, there is a lack of a systematic recording and intelligent transmission mechanism for fire extinguishing process detection data, which affects the subsequent analysis and the overall system intelligence level.

[0034] As a preferred solution of the system for improving the short-term high-frequency energy storage efficiency described in the present invention, it is characterized by comprising: an acquisition unit, configured to install an arc light detection device, an ultrasonic partial discharge detection device, and a temperature and smoke detection device in the power switch cabinet; detecting arc light signals by the arc light detection device, detecting partial discharge signals by the ultrasonic partial discharge detection device, and detecting temperature data and smoke concentration data in the cabinet by the temperature and smoke detection device; a data transmission module, configured to synchronously transmit the arc light signal, partial discharge signal, temperature data, and smoke concentration data to a data processing module, perform signal fusion, and generate a comprehensive fire level judgment value; the data processing module, configured to compare the comprehensive fire level judgment value with a preset fire level threshold to judge the fire level; when the fire level is lower than or equal to a first preset level threshold, controlling the release mechanism to release a first dose of perfluorohexanone fire extinguishing agent; when the fire level is higher than the first preset level threshold, controlling the release mechanism to release a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose; and after completing the release of the fire extinguishing agent, the communication module records the detection data and release parameters and outputs them to an external monitoring terminal.

[0035] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of a fire extinguishing method using a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet are implemented.

[0036] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a fire extinguishing method using a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet.

[0037] The beneficial effects of the present invention are as follows: First, by integrating the four types of signals, namely arc light, partial discharge, temperature and smoke, not only the accuracy of fire detection is significantly improved, but also the false alarm rate and missed alarm rate caused by single signal detection can be effectively reduced. Secondly, by establishing a physical power calculation model and combining the rate of change and the integral characteristics within the time window, the stage of the fire can be dynamically identified, the early stage and the development stage can be distinguished, and more refined fire process identification can be achieved. In addition, the present invention adaptively adjusts the release dose of perfluorohexanone fire extinguishing agent according to the different levels of fire, thereby improving the utilization efficiency of the fire extinguishing agent, avoiding waste of resources, and ensuring that the fire extinguishing effect is timely and effective. Furthermore, the present invention also facilitates subsequent operation and maintenance and accident tracing through the recording and transmission function of the fire extinguishing process data, thereby enhancing the manageability and intelligence level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 An overall flow chart of a fire extinguishing method using perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0040] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0041] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent, comprising:

[0042] Step 1: Install arc detection devices, ultrasonic partial discharge detection devices, and temperature and smoke detection devices in the power switch cabinet;

[0043] Step 2: Detect arc light signals through the arc light detection device, detect partial discharge signals through the ultrasonic partial discharge detection device, and detect the temperature data and smoke concentration data inside the cabinet through the temperature and smoke detection device;

[0044] Step 3: Synchronously transmit the arc light signal, partial discharge signal, temperature data, and smoke concentration data to the data processing module, perform signal fusion, and generate a comprehensive fire level judgment value;

[0045] Step 4: Compare the comprehensive fire level judgment value with the preset fire level threshold to determine the fire level;

[0046] Step 5: When the fire level is lower than or equal to the first preset level threshold, the release mechanism is controlled to release a first dose of perfluorohexanone fire extinguishing agent; when the fire level is higher than the first preset level threshold, the release mechanism is controlled to release a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose;

[0047] Step 6: After the fire extinguishing agent is released, the detection data and release parameters are recorded and output to the external monitoring terminal through the communication module.

[0048] In step 1, arc detection devices should be installed in areas where electrical components may short-circuit, arc, or break down, including but not limited to the main busbar area of ​​the switchgear, circuit breaker interfaces, and cable inlets and outlets. This allows for rapid capture of high-intensity arc light signals. The detectors should be installed toward key electrical components and avoid obstructions to ensure effective reception of optical signals.

[0049] Ultrasonic partial discharge detection devices are preferably installed near busbar connection points, circuit breaker contacts, and other insulators where electrical breakdown may occur. These areas are prone to partial discharge during early electrical faults. Ultrasonic sensors should be placed in a space that covers the main insulation components and takes into account the minimum attenuation of the sound wave propagation path.

[0050] Temperature detectors are preferably installed near cable connectors, busbar connectors, and load switches, as these locations are prone to abnormal temperature rises due to overload or poor contact. The number of temperature detectors depends on the switchgear capacity and component layout, and should cover all critical connections and heat-prone areas.

[0051] Smoke detection devices should be installed on top of the switchgear or above the air convection path. Since smoke tends to diffuse upward, a top-mounted installation can capture smoke particles released in the early stages of a fire earlier. At the same time, forced ventilation ducts or obstructed areas that could affect detection should be avoided.

[0052] In step 2, the arc light signal is detected using an arc light detection device. This device is preferably a photodetector with a broad spectral response, covering the typical arc light wavelength range (200 nm to 1100 nm). The device is equipped with a fast-response circuit, preferably with a response time of less than 1 millisecond, to capture short-duration, high-intensity arc events. The detection signal is output as a voltage or current, filtered, amplified, and then fed into a data processing module.

[0053] Partial discharge signals are detected using an ultrasonic partial discharge detection device with a frequency response range of 30kHz to 300kHz, covering common partial discharge signal frequencies. The detector installation position is optimized to reduce mechanical vibration and external noise interference. The collected ultrasonic signal undergoes preamplification and bandpass filtering to extract the partial discharge signature and convert it into analog or digital input for the data processing module.

[0054] The temperature inside the switchgear is detected by a temperature detection device, preferably a thermocouple or thermistor element, with a measurement range of -20°C to +150°C, meeting the need for monitoring temperature changes within the switchgear. The temperature signal undergoes analog-to-digital conversion to form a continuous digital data stream, which is transmitted in real time to the data processing module.

[0055] Smoke concentration data is detected by a smoke detection device. The smoke detection device can use a photoelectric smoke sensor or an ion smoke sensor with a response time of no more than 5 seconds and a sensitivity that meets the preset standard. The collected smoke signal is digitally encoded and transmitted to the data processing module.

[0056] In step 3, performing signal fusion includes converting the detected arc light signal into arc light radiation power;

[0057] Converting the detected partial discharge signal into partial discharge power and making corrections based on the change trend of the partial discharge signal;

[0058] Calculate the heat release power based on the detected temperature data;

[0059] The smoke combustion release power is calculated based on the detected smoke concentration data.

[0060] The arc radiation power is calculated as:

[0061]

[0062] Where P1(t) represents the arc light radiation power; S1(t) represents the arc light signal intensity; d eff (t) represents the equivalent detection distance; η1(t) represents the detector response correction coefficient;

[0063] The calculated partial discharge power is expressed as,

[0064]

[0065] Where P2(t) represents the partial discharge power; q(t) represents the discharge charge; U(t) represents the discharge voltage; Δt represents the measurement time interval; λ2 represents the partial discharge change rate correction coefficient; represents the rate of change of charge;

[0066] The calculated heat release power is expressed as,

[0067]

[0068] Where P3(t) represents the heat release power; m(t) represents the mass of the burning material; c s represents specific heat capacity; Indicates the rate of temperature change;

[0069] The calculation of smoke combustion release power is expressed as,

[0070]

[0071] Wherein, P4(t) represents the power released by smoke combustion; k4 represents the conversion coefficient of smoke growth energy; Indicates the rate of change of smoke concentration.

[0072] It should be noted that during implementation, the collected arc light signals, partial discharge signals, temperature data, and smoke concentration data are preferably converted into arc light radiation power, partial discharge power, heat release power, and smoke combustion release power, respectively, to unify physical dimensions for subsequent data fusion processing. This choice is based on the dynamic characteristics of different physical phenomena during fire development and can accurately reflect the multi-dimensional information of fire initiation and evolution. In the prior art, common methods include single-sensor threshold determination, multi-signal simple weighted averaging, and neural network-based discrimination models. Single-threshold determination relies solely on triggering an alarm when a single sensor exceeds a fixed threshold, resulting in a high false alarm rate for short-term fluctuations or interference signals and a slow response to complex fire situations. Multi-signal simple weighted averaging, while incorporating multiple sources of signals, has fixed weights for each signal and cannot be dynamically adjusted according to the fire stage or development trend, resulting in insufficient sensitivity and poor adaptability to stage characteristics. While machine learning models such as neural networks have achieved complex discrimination functions in certain applications, they suffer from insufficient interpretability and fail to meet the requirements of power equipment safety management for algorithm transparency and engineering verifiability.

[0073] Furthermore, the method of the present application overcomes the shortcomings of the above-mentioned traditional methods by converting signal physical quantities and unifying power dimensions, and establishes a unified energy expression method between different physical signals, making subsequent trend analysis and continuity analysis possible, and ensuring the engineering interpretability and feasibility of implementation of the method, which is a substantial technological advancement.

[0074] In specific implementations, the arc radiation power is calculated using the signal strength collected by the arc detector as input data. The calculation is based on the equivalent distance between the detector and the monitored electrical component and the detector response correction factor. The equivalent distance is determined based on the installation location and on-site structural parameter measurements. The response correction factor is determined based on sensitivity data provided by the sensor manufacturer and in combination with on-site calibration results. The partial discharge power is calculated based on the discharge charge and discharge voltage collected by the sensor, combined with the measurement time interval. The charge change rate and change rate correction factor are also included. The change rate correction factor is obtained by statistically fitting historical partial discharge event data or through experimental measurement. The thermal release power is calculated using the material mass and specific heat capacity of the switchgear. This is set based on the known mass or standard configuration of the flammable materials within the switchgear, and is selected based on standard values ​​for engineering materials. The temperature change rate is calculated using continuous temperature measurement data. The smoke combustion release power is calculated based on the smoke concentration change rate and the smoke growth energy conversion factor, determined through experimental or simulation testing, reflecting the conversion ratio between the concentration of a specific type of smoke particle and the energy released.

[0075] In an optional embodiment of the present application, the detector response correction coefficient η1(t) is obtained by calibration with a standard light source. The specific steps are as follows: Under laboratory conditions, the arc light detector is installed 1 meter away from a standard adjustable intensity light source, and different known radiation intensities (e.g. 5W / m 2 ,10W / m 2 ,20W / m 2 ,50W / m 2 ) and record the detector output signal S1(t). With the known radiation intensity as the independent variable and the detector response signal as the dependent variable, the least square method is used for linear or nonlinear fitting to obtain the response sensitivity curve. In actual use, according to the installation position and the actual measured equivalent distance d e ff(t), according to the inverse square law and the sensitivity curve, calculate the correction factor η1(t) under specific installation conditions. For example, a certain type of detector at 50W / m 2 Output 2V signal under illumination, calibration fitting results in η1(t)=0.04W / (m 2 ·V).

[0076] In an optional embodiment of the present application, the partial discharge change rate correction coefficient λ2 is obtained by fitting historical discharge data. During the test phase, the discharge charge q(t) and charge change rate dq / dt data of a typical switchgear under different load conditions are collected, and the amount of data is not less than 50 groups. The following empirical fitting formula is used: P2(t) = (q(t)·U(t) / Δt)·(1+λ2·dq / dt). With the known discharge power as the target, the fitting error is minimized and λ2 is calculated by least squares regression. For example, in the sample data, the best fit λ2 = 0.75. The collected data can be recorded by combining the output of the ultrasonic detector with the charge detector, and Δt is set to 1 second.

[0077] In an optional embodiment of the present application, the smoke growth energy conversion coefficient K4 is obtained through a standard combustion experiment. In a controlled laboratory environment, the same insulating material as that inside the switch cabinet (such as polyvinyl chloride, cross-linked polyethylene cable sheath) is selected for a combustion experiment. The relationship between the smoke concentration change rate dc / dt and the released heat during combustion is recorded, and the experiment is repeated no less than 10 times. With the smoke concentration change rate as the independent variable and the released heat as the dependent variable, a linear relationship is established to calculate the conversion coefficient k4. For example, when dc / dt is 0.1% / s in the experiment, the heat released is 200W, then k4 = 2000W / %. The final coefficient k4 is determined based on the average value of the experimental results.

[0078] Based on the arc radiation power P1(t), partial discharge power P2(t), heat release power P3(t) and smoke combustion release power P4(t), the respective time change rates are calculated to obtain and

[0079] Calculate the sum of the arc radiation power change rate and the partial discharge power change rate to obtain:

[0080]

[0081] Calculate the sum of the heat release power change rate and the smoke combustion release power change rate to obtain:

[0082]

[0083] Compare R1(t) with R2(t) to determine the fire stage, which can be expressed as:

[0084]

[0085] Wherein, k is a proportional coefficient, which is used to adjust the comparative sensitivity between the sum of the arc radiation power change rate and the partial discharge power change rate and the sum of the heat release power change rate and the smoke combustion release power change rate, and is determined based on fire history data or training data.

[0086] It should be noted that, based on the physical characteristics of a fire's onset and evolution, the dominance of fire signals at different stages varies significantly. In the early stages of a fire, electrical anomalies (such as arcing and partial discharge) typically appear first, manifesting as instantaneous energy release and a decrease in electrical insulation performance. Their rate of change is significantly higher than that of other physical quantities. Therefore, the present invention preferably sums the arcing radiation power change rate dP1 / dt and the partial discharge power change rate dPP2 / dt to form a first characteristic quantity, R1(t), to reflect the dangerous trend in the early stages. Correspondingly, during the fire's development, heat release and smoke generation gradually become the primary energy release pathways, manifested by a continuously increasing heat release power change rate dP3 / dt and smoke combustion release power change rate dP4 / dt. Therefore, the sum of these two factors forms a second characteristic quantity, R2(t), to reflect the fire's development trend and cumulative effects. The combination of R1(t) and R2(t) not only clearly distinguishes fire stages, but also dynamically adjusts the judgment criteria based on the rate of change of each physical signal, ensuring a reasonable response basis for each fire stage.

[0087] Furthermore, by dividing the above-mentioned signal change rates and constructing a combination of R1(t) and R2(t), dynamic fusion of multi-source information can be achieved, overcoming many limitations of traditional methods. In the prior art, the fire stage is usually triggered by a single signal threshold, or relies on a fixed weighted average judgment, lacking adaptive adjustment of the evolution of the dominance of different signals over time, which can easily lead to false alarms in the early stages or delayed responses in the development stage. The present invention divides the dominant parameters based on physical mechanisms through real-time calculation of the signal change rate, which can quickly identify electrical anomalies in the early stages of a fire and give priority to responding to energy accumulation during the development of the fire, significantly improving the accuracy and timeliness of fire stage identification. In addition, the use of change rates rather than single values ​​for comparison enhances sensitivity to trends and improves the system's response speed to rapidly evolving fire conditions, thereby effectively reducing the risk of misjudgment and missed judgments and improving the intelligence and reliability of the overall system.

[0088] In an optional embodiment of the present invention, the proportionality coefficient k can be obtained through controlled experiments and historical data fitting. First, based on the typical operating environment and expected load conditions of the switchgear, representative insulation materials and load components are selected, and a fire simulation experiment is designed. In the experiment, data on the rate of change of arc radiation power, partial discharge power, heat release power, and smoke combustion release power are recorded to form a time series data set. The known fire stage labels are matched with the real-time calculated values ​​of R1(t) and R2(t), and a classification algorithm such as logistic regression or support vector machine (SVM) is used to fit the optimal demarcation coefficient k of R1 and R2 in different stages. In addition, the rationality of the k value is verified through statistical analysis combined with historical fire case data, and the adjustable range is set according to different equipment types or operating conditions. For example, for switchgear with a rated capacity of 1000A to 2500A, the k value obtained by fitting is generally between 1.2 and 2.0. The specific value can be further optimized and set according to the protection objectives and safety level requirements.

[0089] The generating of the comprehensive fire grade judgment value includes:

[0090] According to the above fire stage judgment results, the current stage s is selected, and the corresponding parameters are early stage or development stage;

[0091] Based on the current stage s, select the first set of change rate weighting coefficients a i (s) and the second set of cumulative weight coefficients b i (s), and time window T(s), where i = 1 to 4;

[0092] Calculate the first judgment value D1(t) and the second judgment value D2(t) according to P1(t) to P4(t) and their change rates respectively;

[0093] The first judgment value D1(t) and the second judgment value D2(t) are jointly solved to obtain the comprehensive fire level judgment value E(t), which is expressed as:

[0094] E(t)=D1(t) p D2(t) q

[0095] Among them, p and q are coefficients determined based on historical fire data.

[0096] The first judgment value is expressed as,

[0097]

[0098] Where i represents the detection parameter number; a i (s) represents the change weight coefficient of the i-th parameter under the current fire stage; s represents the current fire stage, early stage or development stage; Pi (t) represents the i-th physical power parameter; represents the first-order derivative of the i-th physical power parameter with respect to time t; ∑ represents the sum of the four physical parameters;

[0099] The second judgment value is expressed as,

[0100]

[0101] Among them, b i (s) represents the cumulative weight coefficient of the i-th parameter at the current fire stage s; P i (τ) represents the value of the i-th physical power parameter at time τ; T(s) represents the length of the integration time window determined according to stage s; τ represents the integration variable, integrated from tT(s) to t; ∫ represents the integration operation.

[0102] a i (s) and b i The determination process of (s) includes collecting historical data containing different fire stages and levels; training the target error functions of the first judgment value D1(t) and the second judgment value D2(t) according to the fire stage, so that the mean square error between D1(t) and D2(t) and the historical fire level labels is minimized; using ridge regression or Bayesian regression for parameter fitting, and setting coefficient boundaries to prevent overfitting;

[0103] T(s) is determined by analyzing the signal period of historical data, preferably using Fourier transform or autocorrelation analysis, and selecting a time range covering the main energy change period.

[0104] It should be noted that, based on the varying patterns of physical signal characteristics during the fire's development stages, the present invention preferably employs both rate of change and integral calculation methods to construct the first judgment value D1(t) and the second judgment value D2(t), respectively. In the early stages of a fire, the rate of change of electrical anomaly signals, such as arc radiation power and partial discharge power, accelerates significantly, exhibiting a strong trend. The rate of change can reflect the sudden increase in the degree of danger. Therefore, the first judgment value D1(t) is calculated based on the time-varying rate of each power signal to characterize the degree of danger trend. During the fire's development phase, the absolute value and rate of change of heat release power and smoke combustion release power increase significantly, indicating a continuous accumulation of anomaly signals. The second judgment value D2(t) is calculated using integral calculation, comprehensively measuring the persistence and cumulative effect of the anomaly signals. Compared to the single-signal threshold judgment or fixed weighted average commonly used in traditional methods, the combined judgment method of the present invention considers both the transient change trend of the signal and the accumulated energy information, significantly improving the sensitivity and accuracy of fire level judgment. This can effectively reduce false alarm and missed alarm rates, particularly in complex electrical environments, and enhance the reliability of the system response and its engineering applicability.

[0105] Furthermore, in order to ensure the objectivity and adaptability of the judgment value calculation, the present invention sets the first set of weighted coefficients a according to the fire stage. 11 ,a 12 ,a 13 ,a 14 and the second set of weighted coefficients a 21 ,a 22 ,a 23 ,a 24 , and the corresponding time windows T1, T2. The weighting coefficients are determined by statistical regression of historical fire data or simulation experiment data. Specifically, in the test phase, signal data containing different fire stages are collected, the power change rate and cumulative value of each signal are calculated, and the least squares method or other regression analysis methods are used to determine the contribution ratio of each signal to the judgment value at different stages, and then determine the weighting coefficients. For example, in the early stage, the weights of arc light and partial discharge signals are higher, while the weights of heat release and smoke signals are relatively low; in the development stage, the weights of heat release and smoke signals are increased to better reflect the energy accumulation process. The time windows T1 and T2 are set according to the typical time scale of fire evolution. T1 is preferably 5 to 10 seconds to quickly respond to early signal changes, and T2 is preferably 15 to 60 seconds to fully cover the energy accumulation process in the development period.

[0106] In an optional embodiment of the present invention, the generation of the comprehensive fire level judgment value E(t) adopts the joint solution of D1(t) and D2(t). This product relationship can effectively express the combined effect of the dangerous trend degree and the abnormal persistence degree, and enhance the ability to identify the severity of the fire evolution. Unlike the traditional weighted average or simple linear combination, the product function has the effect of amplifying the dangerous trend and abnormal accumulation that exist simultaneously, thereby improving the judgment sensitivity under severe fire conditions. The exponential coefficients p and q can be obtained by fitting the training data. In the specific process, a historical data set containing a variety of fire types and development speeds is selected, and the error minimization method or machine learning algorithm (such as genetic algorithm or particle swarm optimization) is applied to optimize p and q to achieve the best fit between the judgment value E(t) and the actual fire level. The specific value is adjusted according to the operating conditions of the switchgear and the fire response strategy.

[0107] It should be noted that in the method of the present invention, in order to achieve the rationality and accuracy of the comprehensive fire level judgment value E(t), it is necessary to optimize and determine the power coefficients p and q of the first judgment value D1(t) and the second judgment value D2(t) based on fire history data. Traditional methods often use empirical coefficient settings or global search methods such as particle swarm optimization (PSO) and genetic algorithms (GA). However, such methods have slow convergence speeds when the parameter dimension is low, and are prone to falling into local optimality or relying on complex hyperparameter adjustments. In addition, it is difficult to achieve data-driven adaptive optimization based on fire characteristics, which limits the accuracy of parameter fitting and the engineering interpretability of the model.

[0108] In a preferred embodiment of the present invention, based on the characteristics of the power switchgear, a method for solving the coefficients p and q determined based on historical fire data is preferred. Furthermore, the present invention preferably uses a Bayesian optimization method based on minimum deviation regression to fit the power coefficients p and q, thereby fully utilizing historical data and achieving a parameter solution with minimized error. The specific steps include:

[0109] The first step is data preparation. A historical dataset containing various fire types, stages, and severity levels is selected. This dataset includes P1(t), P2(t), P3(t), and P4(t) data from different time periods, as well as actual fire severity labels. Based on this data, the first judgment value D1(t) and the second judgment value D2(t) are calculated using the aforementioned method. The data is then normalized to eliminate scale differences between different physical quantities.

[0110] The second step is to construct the error function. The error function J(p,q) is defined to evaluate the degree of deviation between the judgment value E(t) and the actual fire level label L(t) under the selected combination of p and q.

[0111] The error function J(p,q) is specifically expressed as:

[0112]

[0113] Where L(t) represents the actual fire level label obtained at time t based on historical records or expert evaluation; N represents the total number of historical data samples; and t represents the sample time index, from 1 to N.

[0114] The goal of the error function is to minimize the mean square error (MSE) between E(t) and L(t) to determine the optimal combination of p and q.

[0115] The third step is the Bayesian optimization process. The search range for the power coefficients p and q is set, preferably p∈[1.0,3.0] and q∈[0.5,3.0]. Gaussian Process Regression is used as a surrogate model. The expected error for different combinations of p and q is predicted based on the Bayesian posterior probability of the error function J(p,q). The expected improvement (EI) acquisition function is used at each iteration to select the next set of test values ​​for p and q.

[0116] The fourth step is iterative optimization. Repeat the error prediction and parameter update process, continuously narrowing the search space of p and q, until the error (p, q) converges to a preset threshold or reaches the maximum number of iterations. Finally, the optimal values ​​of p and q that minimize the error function are determined.

[0117] Step 5: Verification and Application. The determined optimal p and q are applied to an independent validation dataset to evaluate their generalization across different fire types and stages. After verification, the determined p and q are used as fixed coefficients during the operational phase to calculate the comprehensive fire severity judgment value E(t) in real time.

[0118] Through the above method, it is possible to avoid the problem that traditional optimization algorithms are prone to falling into local extreme values ​​or being highly dependent on hyperparameters while keeping the error to a minimum, achieve the best match between fire signal characteristics and judgment parameters, improve the accuracy of the judgment value E(t), engineering interpretability and adaptability, and ensure that the fire level assessment method of the present invention has good implementation effects in various operating scenarios.

[0119] It should be noted that the Bayesian optimization process sets the search range of the power coefficients p and q, preferably p∈[1.0,3.0] and q∈[0.5,3.0]. Gaussian Process Regression (GPR) is used as a proxy model to model the Bayesian posterior probability of the error function J(p,q). The specific steps include:

[0120] First, a Gaussian process regression model is constructed. The radial basis kernel function (RBF) is selected as the kernel function, and the preferred form is:

[0121]

[0122] Among them, x and x′ represent different p / q combinations, σ f is the signal standard deviation hyperparameter, and l is the length scale hyperparameter, both of which are automatically determined by maximizing the marginal likelihood. The initial training samples are obtained by randomly selecting 5 to 10 parameter combinations in the p / q search space and calculating the corresponding error function values ​​J(p,q).

[0123] Secondly, the surrogate model is trained. Using the above initial samples, the GPR surrogate model is trained to obtain the error function prediction mean μ(p,q) and prediction standard deviation σ(p,q) for the currently explored p / q combination.

[0124] Next, the acquisition function is calculated. Expected Improvement (EI) is used as the acquisition function, and its expression is:

[0125] EI(p,q)=(μ min -μ(p,q))·Φ(Z)+σ(p,q)·φ(Z)

[0126] Among them, μ_min is the minimum error among all the currently measured samples, Φ(Z) and are the cumulative distribution function and probability density function of the standard normal distribution, ε is a small constant to prevent division by zero errors.

[0127] Then, parameter selection is performed. Within the search space, the EI values ​​are calculated for the unexplored p / q combinations, and the p / q combination with the largest EI is selected as the next set of test parameters.

[0128] The agent model training, acquisition function calculation and parameter selection steps are iterated repeatedly until the error function J(p,q) converges to a preset threshold or reaches the maximum number of iterations (preferably no more than 100 times), and finally the optimal p and q are determined.

[0129] During the training of the Gaussian process regression model, the kernel function hyperparameter σ f and l are optimized by the Maximum Marginal Likelihood function. Specifically, the marginal likelihood function is:

[0130]

[0131] Among them, y is the error function value J(p,q) of the current sample, X is the corresponding p / q combination, and θ is the hyperparameter set (σ f and l), K is the covariance matrix calculated by the kernel function, and n is the number of samples. Optimize θ using gradient descent or Newton's method to maximize the value of the above function and thus determine the optimal hyperparameters.

[0132] Furthermore, during the Z-normalization process of the Expected Improvement (EI) acquisition function, the value of ε is preferably determined based on the normalization scale of the input data. For example, when the normalized values ​​of E(t) and L(t) are in the interval [0,1], ε = 0.001 is used to avoid zero division errors when σ(p,q) is extremely small, while not affecting the validity of EI.

[0133] In a preferred embodiment of the present invention, the change rate weighting coefficient a i (s) and cumulative weight coefficient b i (s) is determined by fitting based on historical fire data. The specific process is as follows:

[0134] First, data preparation. A historical fire dataset covering different fire stages (early stage and development stage) was selected. The dataset includes:

[0135] The measured values ​​of the physical power parameters P1(t), P2(t), P3(t) and P4(t) at each time point;

[0136] Time change rates of physical power parameters at each time point dP1 / dt, dP2 / dt, dP3 / dt and dP4 / dt;

[0137] The fire stage label (early stage or development stage) and fire level label L(t) corresponding to this time point.

[0138] Secondly, the objective function is constructed. For the weighted coefficient a of the rate of change i (s), construct the objective function J a , so that the mean square error between the first judgment value D1(t) and the fire level label L(t) is minimized. The objective function is expressed as:

[0139]

[0140] For the cumulative weight coefficient b i (s), construct the objective function J_b to minimize the mean square error between the second judgment value D2(t) and the fire level label L(t). The objective function is expressed as:

[0141]

[0142] Where N is the number of samples and t is the sample time index.

[0143] Next, the fitting method is selected. Preferably, Ridge Regression or Bayesian Regression is used for parameter fitting.

[0144] The advantage of ridge regression is that it controls the size of coefficients and avoids overfitting by introducing the L2 regularization term. It is particularly suitable for situations with small samples or highly correlated multiple variables.

[0145] Bayesian regression can introduce prior information to further improve the robustness of the fitting.

[0146] During the fitting process, set a i (s) and bi The search range of (s) is preferably 0 to 5 to prevent extreme values ​​of the coefficient and ensure the physical rationality of the parameters.

[0147] Then, stage differentiation and independent training are performed. The data are divided into the early stage (s=1) and the development stage (s=2) according to the fire stage label.

[0148] The two stages are trained independently to obtain a1(s=1)~a4(s=1) and b1(s=1)~b4(s=1), as well as a1(s=2)~a4(s=2) and b1(s=2)~b4(s=2).

[0149] Finally, verify. i (s) and b i (s) is applied to an independent validation dataset to evaluate the error between D1(t) and D2(t) to ensure the generalization performance of the coefficients. If the error exceeds a preset threshold, adjust the regularization parameters or data preprocessing method and refit.

[0150] In a preferred embodiment of the present invention, the steps of determining the time window T(s) are as follows:

[0151] First, data preparation is performed. The power parameter time series P1(t) to P4(t) from different periods in the historical fire dataset are used and classified according to the fire stage (s = 1 for the early stage and s = 2 for the development stage).

[0152] Secondly, the periodicity analysis method is selected. Fourier transform (FFT) or autocorrelation analysis method is used to extract periodic features of data at different stages.

[0153] Fourier transform can determine the main frequency components of the signal, and autocorrelation analysis can identify the periodicity and duration of the signal.

[0154] Next, identify the main energy cycles. Based on the Fourier spectrum, select the cycle corresponding to the frequency with the highest energy percentage.

[0155] Or determine the main period of the signal based on the first significant peak of the autocorrelation function.

[0156] Then, the time window is set. The time window T(s) is set to cover the main energy variation period, preferably covering more than 80% of the energy distribution.

[0157] For the early stage (s=1), the time window T(1) is selected to range from 5 to 10 seconds to ensure a fast response;

[0158] For the development stage (s=2), the time window T(2) is selected to range from 15 to 60 seconds to reflect the continuous accumulation characteristics of fire energy.

[0159] Finally, perform verification and adjustment. Test the performance of the selected T(s) on an independent validation set. If fluctuations in the judgment results or large errors are found, adjust the time window range or use a sliding window mechanism to optimize the response sensitivity.

[0160] Example 2 is the second embodiment of the present invention, which is different from the previous embodiment in that:

[0161] 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 and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0162] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0163] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be 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 processing in another suitable manner as necessary, and then stored in a computer memory.

[0164] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using a combination of any of the following technologies known in the art: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0165] Embodiment 3, the third embodiment of the present invention, provides a system for improving short-term high-frequency energy storage efficiency, comprising:

[0166] A collection unit is used to set arc detection devices, ultrasonic partial discharge detection devices, and temperature and smoke detection devices in the power switch cabinet;

[0167] The arc light detection device detects arc light signals, the ultrasonic partial discharge detection device detects partial discharge signals, and the temperature and smoke detection devices detect the temperature data and smoke concentration data inside the cabinet;

[0168] The data transmission module is used to synchronously transmit the arc light signal, partial discharge signal, temperature data and smoke concentration data to the data processing module, perform signal fusion, and generate a comprehensive fire level judgment value;

[0169] A data processing module is used to compare the comprehensive fire level judgment value with the preset fire level threshold to determine the fire level;

[0170] When the fire level is lower than or equal to a first preset level threshold, the control release mechanism releases a first dose of perfluorohexanone fire extinguishing agent;

[0171] When the fire level is higher than a first preset level threshold, the control release mechanism releases a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose;

[0172] After completing the release of the fire extinguishing agent, the communication module records the detection data and release parameters and outputs them to the external monitoring terminal.

[0173] It should be noted that the fire extinguishing agent dosage control and data recording process is as follows: First, the fire level is compared with a preset level threshold based on the magnitude of the comprehensive fire level judgment value E(t). When E(t) is lower than or equal to the first preset level threshold E1, the control release mechanism releases a first dose D1 of perfluorohexanone fire extinguishing agent; when E(t) is higher than E1, the control release mechanism releases a second dose D2 of perfluorohexanone fire extinguishing agent, where D2 is greater than D1. The first and second doses D1, D2 are determined based on historical fire suppression test data and optimized on-site operating conditions. The minimum effective dose required for both early-stage and developing-stage fire suppression is preferably determined experimentally and adjusted based on the cabinet volume and gas distribution characteristics.

[0174] Secondly, during the release of the fire extinguishing agent, the following detection data are recorded in real time: the comprehensive fire level judgment value E(t) when the release is triggered; the released dose value (D1 or D2); the fire stage at the time of release (early stage or development stage); the timestamp of the triggering release; and the working status of the release mechanism.

[0175] At the same time, the real-time values ​​of P1(t), P2(t), P3(t) and P4(t) and their change rates are recorded for subsequent data analysis and model updating.

[0176] Finally, the release and detection data are uploaded to an external monitoring terminal via a communication module. Standard industrial communication protocols (such as Modbus, IEC61850, or Ethernet) are preferred. Output data includes E(t) values, dose values, fire stage, timestamp, and release status, ensuring that the external monitoring system can receive and record these data in real time or conduct further analysis.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to 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 method for extinguishing fires in power switch cabinets using a perfluorohexanone fire extinguishing agent, characterized in that: include, Install arc detection devices, ultrasonic partial discharge detection devices, and temperature and smoke detection devices in the power switch cabinet; The arc light detection device detects arc light signals, the ultrasonic partial discharge detection device detects partial discharge signals, and the temperature and smoke detection devices detect the temperature data and smoke concentration data inside the cabinet; The arc light signal, partial discharge signal, temperature data and smoke concentration data are synchronously transmitted to the data processing module, and the signal fusion is performed to generate a comprehensive fire level judgment value; Compare the comprehensive fire level judgment value with the preset fire level threshold to determine the fire level; When the fire level is lower than or equal to a first preset level threshold, the control release mechanism releases a first dose of perfluorohexanone fire extinguishing agent; When the fire level is higher than a first preset level threshold, the control release mechanism releases a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose; After the release of the fire extinguishing agent is completed, the detection data and release parameters are recorded and output to the external monitoring terminal through the communication module.

2. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 1, characterized in that: The performing signal fusion includes: Converting the detected arc light signal into arc light radiation power; Converting the detected partial discharge signal into partial discharge power and making corrections based on the change trend of the partial discharge signal; Calculate the heat release power based on the detected temperature data; The smoke combustion release power is calculated based on the detected smoke concentration data.

3. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 2, characterized in that: The arc radiation power is calculated as: Where P1(t) represents the arc light radiation power; S1(t) represents the arc light signal intensity; d eff (t) represents the equivalent detection distance; η1(t) represents the detector response correction coefficient; The calculated partial discharge power is expressed as, Where P2(t) represents the partial discharge power; q(t) represents the discharge charge; U(t) represents the discharge voltage; Δt represents the measurement time interval; λ2 represents the partial discharge change rate correction coefficient; represents the rate of change of charge; The calculated heat release power is expressed as, Where P3(t) represents the heat release power; m(t) represents the mass of the burning material; c s represents specific heat capacity; Indicates the rate of temperature change; The calculation of smoke combustion release power is expressed as, Wherein, P4(t) represents the power released by smoke combustion; k4 represents the conversion coefficient of smoke growth energy; Indicates the rate of change of smoke concentration.

4. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 3, characterized in that: Before generating the comprehensive fire level judgment value, the method further includes: Based on the arc radiation power P1(t), partial discharge power P2(t), heat release power P3(t) and smoke combustion release power P4(t), the respective time change rates are calculated to obtain and Calculate the sum of the arc radiation power change rate and the partial discharge power change rate to obtain: Calculate the sum of the heat release power change rate and the smoke combustion release power change rate to obtain: Compare R1(t) with R2(t) to determine the fire stage, which can be expressed as: Wherein, k is a proportional coefficient, which is used to adjust the comparative sensitivity between the sum of the arc radiation power change rate and the partial discharge power change rate and the sum of the heat release power change rate and the smoke combustion release power change rate, and is determined based on fire history data or training data.

5. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 4, characterized in that: The generating of the comprehensive fire grade judgment value includes: According to the above fire stage judgment results, the current stage s is selected, and the corresponding parameters are early stage or development stage; Based on the current stage s, select the first set of change rate weighting coefficients a i (s) and the second set of cumulative weight coefficients b i (s), and time window T(s), where i = 1 to 4; Calculate the first judgment value D1(t) and the second judgment value D2(t) according to P1(t) to P4(t) and their change rates respectively; The first judgment value D1(t) and the second judgment value D2(t) are jointly solved to obtain the comprehensive fire level judgment value E(t), which is expressed as: E(t)=D1(t) p ·D2(t) q Among them, p and q are coefficients determined based on historical fire data.

6. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 5, characterized in that: The first judgment value is expressed as, Where i represents the detection parameter number; a i (s) represents the change weight coefficient of the i-th parameter under the current fire stage; s represents the current fire stage, early stage or development stage; P i (t) represents the i-th physical power parameter; represents the first-order derivative of the i-th physical power parameter with respect to time t; ∑ represents the sum of the four physical parameters; The second judgment value is expressed as, Among them, b i (s) represents the cumulative weight coefficient of the i-th parameter at the current fire stage s; P i (τ) represents the value of the i-th physical power parameter at time τ; T(s) represents the length of the integration time window determined according to stage s; τ represents the integration variable, integrated from tT(s) to t; ∫ represents the integration operation.

7. A method for extinguishing a fire in a power switch cabinet using a perfluorohexanone fire extinguishing agent as claimed in claim 6, characterized in that: a i (s) and b i The process of determining (s) includes, Collect historical data covering different fire stages and levels; train the target error functions of the first judgment value D1(t) and the second judgment value D2(t) based on the fire stage, so as to minimize the mean square error between D1(t) and D2(t) and the historical fire level labels; use ridge regression or Bayesian regression for parameter fitting, and set coefficient boundaries to prevent overfitting; T(s) is determined by analyzing the signal period of historical data, preferably using Fourier transform or autocorrelation analysis, and selecting a time range covering the main energy change period.

8. A system for improving short-term high-frequency energy storage efficiency, using a perfluorohexanone fire extinguishing agent fire extinguishing method for power switch cabinet fires as described in any one of claims 1 to 7, characterized in that: include: A collection unit is used to set arc detection devices, ultrasonic partial discharge detection devices, and temperature and smoke detection devices in the power switch cabinet; The arc light detection device detects arc light signals, the ultrasonic partial discharge detection device detects partial discharge signals, and the temperature and smoke detection devices detect the temperature data and smoke concentration data inside the cabinet; The data transmission module is used to synchronously transmit the arc light signal, partial discharge signal, temperature data and smoke concentration data to the data processing module, perform signal fusion, and generate a comprehensive fire level judgment value; A data processing module is used to compare the comprehensive fire level judgment value with the preset fire level threshold to determine the fire level; When the fire level is lower than or equal to a first preset level threshold, the control release mechanism releases a first dose of perfluorohexanone fire extinguishing agent; When the fire level is higher than a first preset level threshold, the control release mechanism releases a second dose of perfluorohexanone fire extinguishing agent, the second dose being greater than the first dose; After completing the release of the fire extinguishing agent, the communication module records the detection data and release parameters and outputs them to the external monitoring terminal.

9. 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 method for extinguishing a fire using a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet according to any one of claims 1 to 7 are implemented.

10. 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 method for extinguishing a fire using a perfluorohexanone fire extinguishing agent applied to a fire in a power switch cabinet according to any one of claims 1 to 7 are implemented.

Images