Mine air compressor fire prevention and extinguishing system

Abstract

The application discloses a mine air compressor fire prevention and extinguishing system, which realizes predictive processing of potential thermal runaway risk of the air compressor cavity by constructing an early hidden danger identification mechanism based on micro-plasma disturbance characteristics. A micro-disturbance monitoring module analyzes gas ionization discharge waveform to form a disturbance characteristic sequence reflecting hot spot trend; a deposition inversion module calculates combustible material deposition-thermal release inversion quantity based on temperature gradient ionization response; a coupling inference module generates a fire extinguishing early warning trend value by combining pressure and temperature rise change; a trigger judgment module calculates a predictive fire extinguishing trigger index according to the trend value and outputs an instruction when the threshold is reached; an oxygen reduction control module adjusts an oxygen release inhibition coefficient according to oxygen concentration and humidity to implement quantitative oxygen reduction; and an explosion-proof interlocking module starts an explosion-proof action when the adjusted oxygen concentration is lower than a safety threshold. The system realizes full-link fire prevention and extinguishing control from hidden danger identification, trend prediction to active combustion suppression and explosion-proof, and significantly improves the safety and reliability of the underground air compressor.

Description

Technical Field

[0001] This invention relates to the field of air compressor technology, and in particular to a fire prevention and extinguishing system for mining air compressors. Background Technology

[0002] In existing technologies, fire prevention and extinguishing methods for mining air compressors mainly rely on conventional sensors for monitoring temperature, smoke, and flame, combined with traditional fixed threshold alarms and mechanical interlocking methods to prevent equipment overheating, combustion, or explosion risks. These systems are typically based on the direct acquisition and classification of environmental parameters, triggering spraying, power outages, or explosion-proof actions through simple logical judgments. Their overall structure is primarily passive, and their applicability is limited to clearly defined abnormal temperature rises or smoke / fire conditions.

[0003] However, traditional systems have limited ability to identify early-stage, latent heat accumulation, changes in combustible material deposition, and complex gas states within cavities, making it difficult to make judgments before hotspots generate obvious external signals. Furthermore, existing fire suppression control strategies often rely on single parameters and cannot predict thermal runaway risks based on multi-dimensional coupling trends, resulting in delayed fire suppression triggering. Oxygen reduction control methods also suffer from crude adjustment mechanisms, making it difficult to achieve stable and controllable fire suppression effects. The lack of quantitative correlation between interlocking actions and oxygen reduction treatments limits the reliability of fire prevention and suppression.

[0004] Based on the above shortcomings, a fire prevention and extinguishing system that can achieve earlier identification, more accurate prediction and more stable control is needed. Summary of the Invention

[0005] This application provides a fire prevention and extinguishing system for mining air compressors to improve the safety and reliability of underground air compressors.

[0006] This application provides a fire prevention and extinguishing system for a mining air compressor, comprising: The system includes the following modules: a micro-plasma monitoring module for collecting micro-plasma disturbance intensity of the air compressor's operating cavity and analyzing ionization discharge waveforms to obtain a disturbance feature sequence characterizing the hotspot formation trend; a deposition inversion module for receiving the disturbance feature sequence and calculating the combustible deposition-heat release inversion quantity based on the response relationship between ionization characteristics and temperature gradient to characterize the combustible accumulation risk; a coupling inference module for receiving the combustible deposition-heat release inversion quantity and generating a fire extinguishing warning trend value by combining operating pressure fluctuations and exhaust temperature rise rate; a trigger judgment module for generating a predictive fire extinguishing trigger index based on the fire extinguishing warning trend value and outputting a fire extinguishing trigger command when a threshold is reached; an oxygen reduction control module for receiving the fire extinguishing trigger command and adjusting the oxygen release inhibition adjustment coefficient based on the cavity oxygen concentration, humidity, and the predictive fire extinguishing trigger index to implement quantitative oxygen reduction and output the adjusted oxygen concentration value; and an explosion-proof interlock module for receiving the adjusted oxygen concentration value and performing an explosion-proof interlock action to maintain the cavity's explosion-proof state when it falls below a safety threshold.

[0007] The beneficial effects of this application mainly include: (1) By analyzing the micro-plasma disturbance and discharge waveform of the cavity gas through the micro-perturbation monitoring module, potential hot spot trends can be identified before the appearance of traditional temperature and smoke signals, enabling the system to have early prediction capabilities and improving the controllability of thermal runaway. (2) The deposition inversion module calculates the combustible deposition-heat release inversion amount based on ionization characteristics and temperature gradient, which can quantitatively evaluate the state of the cavity inner wall and oil mist deposition, thereby identifying hidden heat storage and latent combustion risks that cannot be detected by conventional sensors, and improving the accuracy of fire prevention and extinguishing. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of a fire prevention and extinguishing system for a mining air compressor provided in the first embodiment of this application. Detailed Implementation

[0009] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific implementations disclosed below.

[0010] The first embodiment of this application provides a fire prevention and extinguishing system for a mining air compressor. Please refer to... Figure 1 This figure is a schematic diagram of the first embodiment of this application. The following is in conjunction with... Figure 1 The first embodiment of this application provides a detailed description of a fire prevention and extinguishing system for a mining air compressor.

[0011] The fire prevention and extinguishing system for the mining air compressor includes a micro-disturbance monitoring module 101, a sedimentation inversion module 102, a coupling inference module 103, a trigger judgment module 104, an oxygen reduction control module 105, and an explosion-proof interlocking module 106.

[0012] The micro-perturbation monitoring module 101 is used to collect the micro-plasma perturbation intensity of the cavity gas in the air compressor operating cavity and analyze the ionization discharge waveform to obtain a perturbation characteristic sequence characterizing the hot spot formation trend.

[0013] The micro-plasma monitoring module 101 is used for highly sensitive disturbance monitoring of the gas state within the operating cavity of an air compressor. Structurally, it includes a plasma disturbance acquisition unit, a discharge waveform analysis unit, and a trend characteristic output unit. Micro-plasma disturbance refers to an extremely low-energy unstable ionization phenomenon generated within a high-pressure mechanical cavity due to the combined effects of gas composition, local temperature gradient, combustible vapor concentration, and the triboelectric effect between the metal wall and oil mist. This ionization phenomenon exists in the form of weak discharges, ionization flash points, and spark precursors within a range invisible to the naked eye. Its energy is far lower than that of visible flames or obvious discharges, and it will not damage the equipment, but it can serve as an early indicator of hot spot formation. The micro-plasma monitoring module 101 collects and analyzes this weak ionization phenomenon, enabling the system to identify hot spot trends tens of seconds to several minutes in advance, before combustibles begin to oxidize or even reach smoke or high-temperature thresholds.

[0014] In this embodiment, the perturbation monitoring module 101 has a set of ionization detection electrodes fixed by high-temperature resistant insulating material installed inside the air compressor cavity. These electrodes form a monitoring area over the internal space of the cavity. Under normal operation, the gas inside the cavity is unlikely to undergo significant ionization. However, when the local temperature rises slightly or the concentration of combustible vapor increases, gas molecules will experience very slight excitation and ionization, resulting in minute voltage fluctuations between the detection electrodes. The amplitude of these fluctuations is typically between tens and hundreds of microvolts, requiring a plasma perturbation acquisition unit for collection.

[0015] The acquisition unit is equipped with a high-precision ADC (Analog-to-Digital Converter) of at least 24 bits and continuously samples the voltage across the electrodes at a sampling rate between 10 kHz and 200 kHz. Such a high sampling rate is used because the micro-plasma discharge phenomenon has an extremely short duration, typically 0.01 to 0.2 milliseconds; if the sampling rate is insufficient, the waveform cannot be fully captured. During the sampling process, the module also employs digital phase-locked loop (PLL) filtering technology to remove background noise and power frequency interference from the acquired signal, ensuring that the remaining signal contains only the transient spikes corresponding to weak ionization events.

[0016] The acquired signal then enters the discharge waveform analysis unit, which performs feature decomposition on the sampled waveform. An ionization discharge waveform is a curve showing the change of an electrical signal generated by a gas ionization event over time. Its typical shape includes a steep rising edge, a brief peak plateau, and a gradually declining recovery segment. Different hotspot formation stages lead to differences in waveform characteristics; for example, the rising edge is slow in the early stages of a hotspot, while the rising edge tends to be vertical during the acceleration phase, with increased peak amplitude and multiple secondary peaks. The discharge waveform analysis unit extracts features from the sampled signal using wavelet transform, instantaneous energy analysis, and time-frequency distribution analysis, calculating physical quantities such as peak amplitude, rising edge slope, energy concentration, and frequency band distribution from the waveform.

[0017] Here is an example. Suppose the acquisition module detects a transient signal with a peak value of 80 microvolts during a certain sampling period. The time it takes for the signal to rise from 10 microvolts to 80 microvolts within the sampling period is 0.02 milliseconds. The rise slope can be calculated using the following formula: Slope = (80 − 10) microvolts ÷ 0.02 milliseconds = 3500 microvolts / millisecond. The analysis unit records this value as a hotspot-sensitive parameter for subsequent judgment. If the detected energy concentration deviates from the normal operating state by more than a set threshold (e.g., exceeding five standard deviations of the normal statistical mean), the analysis unit will further mark this waveform segment as belonging to a hotspot development trend.

[0018] After parsing, the trend feature output unit compresses multiple time-continuous waveform features into a "perturbation feature sequence." This perturbation feature sequence is not a simple data list; it contains the dynamic trend of micro-plasma perturbations over a period of time, including dimensions such as periodicity, peak value variation, waveform component variation, and energy distribution variation. This sequence can be seen as a quantitative representation of the hotspot formation rate. If the perturbation feature sequence shows phenomena such as gradually increasing peak values, increasingly steep rising edges, and the emergence of secondary peaks, it indicates that the temperature rise rate at a certain point in the air compressor cavity exceeds the normal operating range, combustible vapor is undergoing accelerated excitation in a localized area, and a hotspot is forming.

[0019] In practical deployments, the perturbation monitoring module 101 can adjust its monitoring sensitivity according to different compressor models and cavity volumes. For example, in larger air compressors, the gas disturbance area formed by hotspots is wider, requiring increased electrode spacing and the use of a higher-gain preamplifier to ensure that weak perturbations are not drowned out by noise. The module uses a real-time update method when outputting the perturbation feature sequence, enabling the subsequent deposition inversion module to capture the most subtle changes in a timely manner. By continuously analyzing these perturbation features, the system can achieve sub-early risk identification that is unattainable by traditional methods, providing a reliable foundation for subsequent combustible material deposition inversion and trend assessment.

[0020] In a preferred embodiment of this application, the perturbation monitoring module is specifically implemented as a continuous data acquisition and analysis link, used to monitor micro-plasma disturbances in the operating cavity of the air compressor with high temporal resolution under normal operating conditions, and gradually generate a perturbation feature sequence that can characterize the trend of hotspot formation.

[0021] During operation, the perturbation monitoring module first generates multiple independent micro-plasma perturbation acquisition windows within the air compressor's operating cavity using a time-domain segmented acquisition method. The time-domain segmented acquisition method refers to the module dividing the continuous monitoring time into several non-overlapping time slices of fixed length; each time slice is referred to as a micro-plasma perturbation acquisition window in this specification. For example, one second of monitoring time can be divided into one hundred acquisition windows of ten milliseconds each, with each acquisition window corresponding to a specific angle range of the air compressor shaft or a repetitive compression cycle stage. In this invention, the length of the acquisition window is pre-set based on the sensor sampling frequency and the duration of the micro-plasma discharge, ensuring that the entire process of a typical discharge event can be completely contained within one or a few adjacent windows, thereby guaranteeing the continuity and distinguishability of subsequent analysis. The module records the raw electrical signals output by the sensor within each acquisition window and arranges all micro-plasma perturbation signals acquired within that time slice in chronological order to form a corresponding initial micro-plasma perturbation signal sequence. The initial micro-plasma disturbance signal sequence here refers to a discrete signal set consisting of several voltage sample values ​​and their time labels ordered by sampling time, which preserves the complete waveform shape of ionization discharge from start to finish.

[0022] After obtaining the initial micro-plasma disturbance signal sequence, the disturbance monitoring module inputs it to the waveform granularity analysis unit to perform fine-grained structural analysis on the rising edge of the ionization discharge waveform within the acquisition window. The rising edge of the ionization discharge waveform refers to the time interval during which the discharge event jumps from the background noise level to the peak voltage; this interval reflects the excitation speed and intensity of the ionization process. In this invention, to meticulously characterize the internal variation structure of the rising edge, the waveform granularity analysis unit divides each rising edge into several continuous granular segments at equal time intervals. For example, a rising edge lasting twenty microseconds is divided into twenty granular segments, each lasting one microsecond. Within each granular segment, the analysis unit calculates the actual change value and rate of change of the voltage within that segment and compares these values ​​with the reference rising edge data obtained by the equipment during the calibration phase. The reference rising edge data refers to the average rising shape obtained through long-term statistical analysis of a standard discharge waveform acquired under healthy operating conditions where the air compressor is normal and there are no obvious deposition hotspots in the cavity.

[0023] After completing the analysis of all granular segments, the waveform granularity analysis unit constructs a waveform granularity feature set. The waveform granularity feature set is a dataset containing multiple structured features used to characterize the details of micro-plasma discharge behavior within a specific acquisition window. In this invention, the waveform granularity feature set includes at least the following types of features: first, the number of segments in the rising edge granularity segment whose voltage increase is several times higher than the reference increase; this feature characterizes the proportion of high-excitation regions during ionization; second, the ratio of the voltage increase of the largest single granular segment in the rising edge to the reference increase; this characterizes the peak excitation intensity; third, the average increase of several granular segments before the rising edge; this characterizes whether ionization starts rapidly in the early stages; and fourth, the ratio of the duration of the entire rising edge to the reference duration; this characterizes whether the overall excitation process is accelerated or delayed. Each of the above features is stored in numerical form and associated with the corresponding acquisition window number, constituting the waveform granularity feature set for that acquisition window.

[0024] For example, within a data acquisition window, the rise time of a micro-plasma discharge is 20 microseconds, divided into 20 granular segments. In the reference rise time, the average voltage increase of each granular segment is set to one unit. In this discharge, the analysis results show that eight of the first ten granular segments have an increase of three times the reference value, three of the last ten segments have an increase of twice the reference value, the largest single granular segment has an increase of four times the reference value, and the overall rise time duration is the same as the reference. Therefore, the waveform granularity feature set of this acquisition window must contain at least the following defined values: eight granular segments with an increase greater than three times the reference value, eleven granular segments with an increase greater than twice the reference value, a maximum increase ratio of four, an average increase of approximately three times the reference value for the first ten segments, an average increase of approximately 1.5 times the reference value for the last ten segments, and an overall rise time duration ratio of one. These features are stored as explicit numerical fields, without any vague or qualitative descriptions.

[0025] After obtaining the waveform granularity feature set for each acquisition window, the perturbation monitoring module deduces the disturbance influence region of the micro-plasma discharge event in the cavity gas based on these feature sets. Here, the disturbance influence region refers to several virtual regions spatially divided within the air compressor's operating cavity. Each region corresponds to a specific compression stroke or a section of pipeline space. During system configuration, the module establishes a time-phase correspondence for each disturbance influence region based on the air compressor's mechanical structure and flow field simulation results. That is, within a certain angle range of the rotating shaft or at a certain stage of the compression cycle, the ionization activity in a specific spatial region within the cavity will primarily be reflected within one or several acquisition windows. Through this correspondence, the module can map the waveform granularity feature set of a specific acquisition window to a specific disturbance influence region.

[0026] The local disturbance contribution is a quantitative indicator calculated for each disturbance-affected region, representing its contribution to the overall micro-plasma disturbance activity within a monitoring period. In a monitoring period, the module first summarizes the waveform granularity features of all acquisition windows mapped to the same disturbance-affected region, calculating a disturbance intensity value for each region. This disturbance intensity value can be obtained by accumulating several key feature items of the acquisition windows within that region, such as a weighted sum of the "number of granular segments with an amplification greater than three times the baseline" and the "maximum amplification ratio" across all acquisition windows in that region, thus obtaining a value representing the overall ionization intensity of that region. Next, the module normalizes the disturbance intensity values ​​of all disturbance-affected regions, ensuring that the local disturbance contribution of each region equals the proportion of its disturbance intensity to the total disturbance intensity of all regions. The resulting local disturbance contribution value is between zero and one, and the sum of the contributions of all regions equals one, facilitating direct use in subsequent trend analysis.

[0027] For example, within a monitoring period, the system divides the cavity into three disturbance-affected areas, denoted as Area A, Area B, and Area C. Based on the aggregated calculation of granular features from multiple acquisition windows, the disturbance intensity of Area A is 50, the disturbance intensity of Area B is 30, and the disturbance intensity of Area C is 20, resulting in a total disturbance intensity of 100. In this invention, the module defines the local disturbance contribution of Area A as 50 divided by 100, corresponding to 0.5; the local disturbance contribution of Area B as 30 divided by 100, corresponding to 0.3; and the local disturbance contribution of Area C as 20 divided by 100, corresponding to 0.2. In this way, it can be clearly seen that Area A is the main disturbance concentration area within this period, Area B is a secondary hotspot area, and the disturbance in Area C is relatively weak.

[0028] After obtaining the local disturbance contribution of each affected area, the micro-disturbance monitoring module generates a disturbance feature sequence characterizing the hotspot formation trend based on these contributions. The disturbance feature sequence is a time-updated numerical sequence. It can select the maximum value among the local disturbance contributions, the contribution of a predetermined key area, or a combination of contributions from multiple areas, arranged chronologically according to the monitoring period. This sequence describes whether the hotspot is continuously strengthening and whether it is concentrating in a fixed area. For example, if the local disturbance contribution of area A is 0.4, 0.55, 0.7, and 0.78 in four consecutive monitoring periods, the module will write 0.4, 0.55, 0.7, and 0.78 into the disturbance feature sequence in chronological order, correspondingly deriving the quantitative trend information that "the hotspot is continuously strengthening and concentrating in area A." This disturbance feature sequence serves as input to the sedimentation inversion module, enabling subsequent modules to extrapolate changes in combustible material deposition and heat release, achieving a seamless transition from micro-disturbance monitoring to sedimentation inversion.

[0029] The deposition inversion module 102 is used to receive the perturbation feature sequence and calculate the combustible deposition-heat release inversion amount based on the response relationship between ionization features and temperature gradient, so as to characterize the combustible accumulation risk.

[0030] The deposition inversion module 102 is used to quantify and deduce the potential combustible material deposition status within the cavity after the perturbation monitoring module 101 outputs the perturbation characteristic sequence, in order to obtain a combustible material deposition-heat release inversion quantity that can characterize the risk of combustible material accumulation. This inversion quantity is derived from three types of directly measurable changes: the degree of gas ionization enhancement, the change in ionization excitation rate, and the change in the density of ionization perturbation events.

[0031] The degree of gas ionization enhancement is used to characterize the amount of volatile components released after sediments are continuously heated. When oil mist residues, solid combustibles, or other volatile sediments are present in the cavity, these substances release more molecules that participate in ionization upon heating, causing the peak amplitude of ionization events to be significantly higher than the normal operating baseline. The sediment inversion module 102 obtains the ionization enhancement factor by comparing the average amplitude of the ionization peaks in the perturbation characteristic sequence with the baseline amplitude recorded by the system under long-term stable operation. The larger the ionization enhancement factor, the more substances participate in ionization per unit time, thus indicating a higher sediment quantity or a significant thermally excited state.

[0032] The change in ionization excitation rate is used to characterize the enhancement of the local temperature gradient within the cavity. When heat accumulation occurs in a local area, the rate of change of the rising edge of the ionization signal will significantly increase. The sedimentation inversion module 102 obtains the excitation rate enhancement factor by comparing the time required for the rising edge of the current disturbance event to reach its peak from the starting point with the duration of the baseline rising edge. If the rising edge change time is significantly shortened, it indicates that the local temperature gradient is enhanced, and the sediment is in a more active thermal excitation environment.

[0033] The density change of ionization perturbation events reflects the frequency of unstable excitation behavior in sediments during heating. Heating, pyrolysis, evaporation, or surface chemical changes in sediments significantly increase the number of localized micro-discharge events. The sediment inversion module 102 calculates the number of perturbation events per unit time and compares it to the event frequency of the system baseline to obtain the density change factor. A higher density change factor indicates a more significant energy release process is occurring in the sediments.

[0034] The deposition inversion module 102 takes the three types of changes mentioned above as input, normalizes them, and merges them according to a preset ratio to obtain the final combustible material deposition-heat release inversion quantity. The purpose of normalization is to convert measurement results of different dimensions and magnitudes into continuous values ​​between zero and one, so that the three types of changes can be compared and derived under the same dimension. Then, the module merges the three types of changes according to the set weights to form a single index representing the degree of deposition and the trend of heat release. The closer the index is to one, the more severe the combustible material deposition in the cavity and the stronger its instantaneous thermal activation behavior; the closer the index is to zero, the lower the deposition amount or the less obvious the thermal activation trend.

[0035] The following is a specific example. For instance, within a five-second monitoring window of continuous air compressor operation, the deposition inversion module 102 receives a perturbation characteristic sequence output from the perturbation monitoring module 101. This sequence contains 120 micro-plasma perturbation events recorded during this time period, each event including basic characteristics such as peak amplitude, rise time, and event timestamp. The module first determines the long-term baseline data of this model during normal healthy operation, where the perturbation peak value is typically maintained at around 20 units, the rise time is approximately 30 microseconds, and the event density averages once per second. The deposition inversion module compares the acquired data with the baseline to calculate three changes: ionization enhancement degree, ionization excitation rate change, and perturbation event density change rate.

[0036] During this monitoring window, the perturbation sequence received by the module showed that the peak amplitude exceeded 60 units forty times within five seconds, with most remaining stable between 60 and 70 units. Since the normal peak value is 20 units, the module determined this peak change to be approximately a three-fold increase in ionization. A higher degree of ionization enhancement indicates that the heated deposits within the cavity are releasing a larger amount of volatile substances, making the gas more susceptible to ionization.

[0037] The module then analyzed the rise time. Within the same monitoring window, the average rise time was approximately ten microseconds, significantly lower than the normal thirty microseconds. Based on this, the module determined that the ionization excitation rate had increased approximately threefold, indicating that the deposits or oil mist within the cavity had entered a higher excitation state after continuous heating, and the local temperature gradient was sufficient to cause the ionization event to occur at a faster rate than usual.

[0038] The module continued to analyze the density of disturbance events. A total of 120 events were recorded within a five-second window, averaging 24 events per second, compared to only one per second under baseline conditions. The module determined the density change to be a 24-fold increase, reflecting significant unstable excitation behavior in the sediment under heated conditions, a typical state of heat storage and impending release.

[0039] After obtaining the three changes, the sediment inversion module aggregates them into a single combustible sediment-thermal release inversion value. The module first maps the three-fold increase in ionization, the three-fold acceleration of ionization activation, and the twenty-four-fold change in event density to three normalized values ​​of 0.75, 0.75, and 1, respectively. The closer the normalized value is to one, the closer the physical change is to the danger boundary. In this example, the module averages the three changes to form the final inversion value, combining 0.75, 0.75, and 1 to obtain a composite value of approximately 0.83. This value indicates that the sediments within the cavity are not only abundant but also exhibit significantly enhanced volatile release and activation behavior upon heating, classifying it as belonging to a clearly medium-to-high accumulation risk level.

[0040] After obtaining an inversion value of 0.83, the deposition inversion module outputs this value as a quantitative characterization result of the cavity accumulation risk to the coupled inference module 103. Since the inversion value in this example is significantly higher than the system's set medium risk range, this result can be directly used to trigger the subsequent trend prediction program, indicating that the air compressor is entering a critical stage of thermal runaway formation.

[0041] In summary, the sedimentation inversion module 102 performs in-depth analysis on the data provided by the perturbation monitoring module, converting complex ionization perturbation behavior into quantifiable combustible deposition amounts and heat release trends, enabling the system to accurately identify the material basis before hotspot formation.

[0042] In a preferred embodiment, the sediment inversion module is configured as a processing link that performs step-by-step analysis and quantification derivation around the perturbation feature sequence, which is used to convert the perturbation feature sequence output by the micro-perturbation monitoring module into a combustible sediment-heat release inversion quantity that can directly reflect the degree of sediment accumulation and the intensity of instantaneous heat release.

[0043] The disturbance feature sequence can be understood as a set of numerical values ​​or vectors with a time order. Each element corresponds to the comprehensive intensity of micro-plasma disturbances within a monitoring period, specifically calculated by the micro-disturbance monitoring module based on factors such as the contribution of local disturbances and waveform granularity characteristics. A continuous monitoring period refers to a continuous time interval divided by the system according to a fixed duration (e.g., one second or half a second). Each time interval completes a full data acquisition and processing cycle. Therefore, the disturbance feature sequence is actually a series of ionization intensity scalars arranged in time. In this embodiment, ionization intensity can be understood as the overall strength of ionization discharge activity within the monitoring period. Its dimension can be obtained by normalizing the number of rising edge granularity segment amplifications, peak amplifications, etc., resulting in a dimensionless value. A larger value indicates a more pronounced thermal excitation of combustible materials within the cavity.

[0044] The sedimentation inversion module first performs windowed aggregation of ionization intensity changes over consecutive monitoring periods based on the perturbation feature sequence. Windowed aggregation refers to not directly using the ionization intensity of a single monitoring period, but rather treating multiple adjacent monitoring periods as an aggregation unit, and synthesizing the ionization intensity changes within these periods to form an aggregated index reflecting the rate of change of the thermal state of the combustible material. For example, three consecutive monitoring periods can be selected as an aggregation window, with the ionization intensities of the first three periods recorded as a lower, middle, and higher value, respectively. By comparing the differences between the last two periods and the previous period, the overall increase and rate of increase of the ionization intensity within that time period are obtained. The resulting comprehensive result is defined as the aggregated ionization change value, which quantifies the rate of change of the thermal state of the combustible material over a longer time scale using a specific numerical value. For a specific example, if the ionization intensities of the perturbation feature sequence are 0.3, 0.5, and 0.9 in three consecutive monitoring periods, the aggregated value of ionization change can be obtained as follows: First, calculate the increase in the second period relative to the first period, which is 0.5 minus 0.3, or 0.2. Then, calculate the increase in the third period relative to the second period, which is 0.9 minus 0.5, or 0.4. Finally, average these two increases to obtain an average increase that represents the overall trend of ionization intensity change across the three periods. In this case, the average increase is 0.3. In this embodiment, this average increase can be directly used as the aggregated value of ionization change, or it can be simply amplified to combine with the subsequent temperature gradient sensitivity coefficient.

[0045] After obtaining the ionization change aggregation value, the deposition inversion module inputs this aggregation value to the temperature gradient response analysis unit. The temperature gradient response analysis unit analyzes the change in the cavity operating temperature gradient over time and matches this change with the ionization change aggregation value to extract the sensitivity of the deposits to temperature gradient changes. Here, the temperature gradient refers to the magnitude of temperature change along the gas flow path or along the metal wall within the air compressor operating cavity. For example, it could be the temperature difference between the inlet and outlet divided by the corresponding distance, or the temperature difference at the same location between adjacent monitoring cycles divided by the time interval. To make the processing more explicit, this embodiment can select the temperature difference between the outlet and inlet as the main temperature gradient indicator and record the increase or decrease of this temperature gradient in adjacent monitoring cycles. When the temperature gradient increases significantly over several cycles, it indicates the presence of a continuous heat source inside the cavity, most likely combustible deposits continuously releasing heat; when the temperature gradient change is not significant or decreases slightly, it indicates a limited heat source activity.

[0046] The temperature gradient response analysis unit judges the change in the current temperature gradient based on the changes in a pre-calibrated reference temperature gradient. For example, under normal equipment conditions with no significant sediment accumulation, the temperature gradient change generally fluctuates within a narrow range. When the current temperature gradient change is detected to be higher than this range for several consecutive cycles, it indicates an abnormal heating trend in the current operating condition. In this case, the unit determines a temperature gradient sensitivity coefficient based on factors such as the increase rate and duration of the temperature gradient. This coefficient represents the degree of response of the sediment to changes in the temperature gradient; the larger the value, the more sensitive the sediment is to the current temperature rise, and the more intense the heat release. For example, during the equipment calibration phase, if the increase in temperature gradient for three consecutive cycles is within 1.5 times the reference value, it is classified as a "low-sensitivity state," and the corresponding temperature gradient sensitivity coefficient can be preset to one. If the increase in temperature gradient for three consecutive cycles reaches approximately twice the reference value, it is classified as a "medium-sensitivity state," and the corresponding coefficient can be set to one and a half. If the increase in temperature gradient for three consecutive cycles exceeds two and a half times or even approaches three times the reference value, it is classified as a "high-sensitivity state," and the corresponding coefficient can be set to two. In practical implementation, those skilled in the art only need to compare the measured temperature gradient change with the reference range of the calibration phase and select the corresponding value from three or more coefficient levels based on the comparison result.

[0047] After obtaining the temperature gradient sensitivity coefficient, the sediment inversion module performs thermosensitive amplification processing on the ionization change aggregation value based on this coefficient, generating the ionization excitation enhancement corresponding to the local sediment thermal release process. This thermosensitive amplification processing does not involve complex mathematical derivations, but rather applies an amplification level to the ionization change aggregation value based on the temperature gradient sensitivity coefficient, thus amplifying the impact of ionization changes when there is a significant thermosensitive response. For example, when the temperature gradient sensitivity coefficient is one, it indicates only a slight anomaly in the temperature gradient; in this case, the ionization change aggregation value is not amplified and is directly considered as the current ionization excitation enhancement. When the temperature gradient sensitivity coefficient is one and a half, it indicates a significant but not severe temperature gradient anomaly; in this case, the ionization change aggregation value can be increased by half, amplifying it to 1.5 times its original value. When the temperature gradient sensitivity coefficient is two, it indicates a severe temperature gradient anomaly and a strong thermosensitive response in the sediment; in this case, the ionization change aggregation value can be directly doubled, making it twice its original value. Continuing with the previous example, suppose the ionization change aggregation value within a certain aggregation window is 0.3. The temperature gradient response analysis unit determines that the current state is medium sensitivity, and the corresponding temperature gradient sensitivity coefficient is 1 and 1 / 2. Then the ionization excitation enhancement can be obtained by "increasing half of the original 0.3". The added part is 0.15, and the final ionization excitation enhancement is 0.45.

[0048] After completing the thermal amplification process, the sediment inversion module couples the ionization activation enhancement with the local perturbation contribution sequence in the perturbation characteristic sequence to obtain the combustible sedimentation-heat release inversion quantity that truly characterizes the degree of sediment accumulation and instantaneous heat release. The local perturbation contribution sequence here refers to the time-ordered data sequence calculated in the aforementioned micro-perturbation monitoring module, used to characterize the contribution of different spatial regions to the overall ionization activity. Each value is between zero and one, and the sum of the contributions from all regions equals one. During the sedimentation inversion process, to more accurately reflect "where there is more deposition and where there is more intense release," the module weights the ionization activation enhancement according to the spatial distribution of the local perturbation contribution. For a specific example, within a given aggregation window, the ionization excitation enhancement is 0.45. The local perturbation contribution sequence shows that regions A, B, and C contribute 0.5, 0.3, and 0.2%, respectively. The depositional inversion module would then assume that region A accounts for more than half of the ionization excitation enhancement within that time period, while regions B and C account for the remaining 30% and 20%, respectively. Therefore, when calculating the combustible sedimentation-heat release inversion, the 0.45 enhancement is allocated to each region in proportions of 0.5, 0.3, and 0.2. Furthermore, the module can accumulate the allocation results across multiple aggregation windows over time to obtain the cumulative depositional-heat release contribution of each region within the observation period, which serves as a component of the combustible sedimentation-heat release inversion.

[0049] Finally, the sedimentation inversion module summarizes the results obtained by accumulating over time and weighting by space to form a combustible sedimentation-heat release inversion quantity that simultaneously reflects the degree of sedimentation and the intensity of instantaneous heat release. This inversion quantity can be represented by one or more specific values, such as providing a comprehensive sedimentation-heat release index for each key area, or providing a comprehensive index representing the global sedimentation-heat release degree for the entire system.

[0050] The coupling inference module 103 is used to receive the combustible material deposition-heat release inversion quantity and generate a fire extinguishing early warning trend value by combining the operating pressure fluctuation and the exhaust temperature rise rate.

[0051] The coupling inference module 103 is used to further analyze the thermal runaway trend within the air compressor cavity after the combustion deposit-heat release inversion quantity is output by the deposition inversion module 102, in order to generate a fire extinguishing early warning trend value that can characterize the possibility of future short-term fire extinguishing needs. To enable those skilled in the art to clearly understand the implementation logic of this module, it is necessary to first define the three types of input information received and processed by this module: air compressor operating pressure fluctuation, exhaust temperature rise rate, and combustion deposit-heat release inversion quantity. Operating pressure fluctuation refers to the magnitude of pressure changes generated by the air compressor during the compression cycle, including periodic and non-periodic fluctuations. Under normal conditions, this fluctuation range is stable and follows a certain pattern. However, when local heat accumulation occurs within the cavity or the deposits exhibit active heat release behavior, the pressure fluctuation will show slight but measurable anomalies. The exhaust temperature rise rate reflects the rate at which the air compressor exhaust temperature rises over time. This parameter is related to the presence of a continuous heat source within the cavity. When the deposits in the cavity have entered a state of significant thermal activation, the rate of increase in exhaust temperature usually increases significantly.

[0052] The core function of the coupled inference module 103 is to combine the combustible material deposition-heat release inversion quantity with the two operating parameters mentioned above to determine whether the air compressor is developing towards thermal runaway. The combustible material deposition-heat release inversion quantity provides the degree of accumulation and thermal activation occurring at the material level inside the cavity, while the operating pressure fluctuation and exhaust temperature rise rate provide the changes occurring at the mechanical operation level and the gas dynamic level. By simultaneously observing whether these changes show a trend of increasing in the same direction within the same time window, the coupled inference module 103 generates a fire extinguishing early warning trend value that reflects the speed and intensity of thermal runaway development.

[0053] To further illustrate its operation, the module quantifies the three inputs and combines them into a single trend value. First, the module receives a combustible material deposition-heat release inversion value, a continuous value between zero and one, representing the degree of deposition and thermal activation. Next, the module reads the real-time amplitude of the operating pressure fluctuation and, by comparing the current fluctuation amplitude with the baseline fluctuation amplitude under long-term stable operation, obtains a pressure fluctuation increase factor. A higher pressure fluctuation increase factor indicates a stronger disturbance in the cavity caused by thermal unevenness or gas expansion. The module then reads the exhaust temperature rise rate and compares it with the baseline temperature rise rate to obtain the temperature rise rate increase factor. During the development stage of air compressor thermal runaway, the temperature rise rate often increases much more than normal; therefore, this factor is crucial for trend analysis.

[0054] After obtaining the three continuous variables mentioned above, the coupling inference module 103 normalizes them, unifying them to the same numerical range so that they can be directly compared. The three normalized data are then weighted according to the coupling weights set by the module. The weight of the combustible material deposition-heat release inversion quantity is usually slightly higher than the other two, because it directly reflects the internal state of combustible materials. The pressure fluctuation increase factor and the temperature rise rate increase factor reflect the external manifestations of mechanical and thermal behavior, respectively. The weighted sum of the three values ​​forms a fire extinguishing warning trend value. This trend value changes continuously between zero and one. The closer it is to one, the stronger the thermal excitation behavior inside the air compressor, and the closer it is to a thermal runaway state.

[0055] The following is a practical example. For instance, in a five-second real-time monitoring window, the combustible deposit-heat release inversion value output by the deposition inversion module 102 is 0.83, indicating significant internal deposits and a state of significant thermal excitation. Within the same window, the instantaneous amplitude of the operating pressure fluctuation is twice that of the baseline, which the module normalizes to 0.6. The exhaust temperature rise rate increases from one degree per second to three degrees per second, a threefold increase, which the module normalizes to 0.75. The module then weights and combines 0.83, 0.6, and 0.75 according to a set ratio. This ratio ensures that all three types of data are represented, but the impact of the deposition-heat release inversion value is more significant. After weighted combining, the module obtains a fire extinguishing warning trend value of approximately 0.73. This trend value indicates that the air compressor is showing a relatively clear trend of thermal runaway development, falling within the range where the system needs to enter a warning state.

[0056] After the above process is completed, the coupling inference module 103 outputs the fire extinguishing early warning trend value to the trigger judgment module 104, which will further determine whether the fire extinguishing process needs to be triggered in advance based on this trend value. By accurately coupling the internal state of combustibles with external operating parameters, the coupling inference module 103 provides an accurate, continuous and quantifiable basis for the selection of fire extinguishing timing, enabling the system to take timely suppression measures at the critical stage before the actual fire source forms.

[0057] In a preferred embodiment, the coupling inference module receives the combustible deposition-heat release inversion quantity output by the deposition inversion module, and combines it with the data on the change of air compressor operating pressure over time and the exhaust temperature rise rate to perform step-by-step quantification of the correlation between the three, and finally generates a fire extinguishing early warning trend value that can characterize the direction and speed of thermal runaway risk evolution in the near future.

[0058] First, the coupled inference module constructs a deposition-heat release time series divided by monitoring period based on the combustible deposit-heat release inversion quantity. In this invention, the monitoring period is the time unit for the system to collect and process the air compressor's operating status once at a fixed time interval, such as one second or five seconds. Each monitoring period corresponds to a specific time tag and a set of measurement data. The deposition-heat release time series is formed by arranging the combustible deposit-heat release inversion quantities obtained in each monitoring period in chronological order, creating a one-dimensional numerical sequence with a temporal order. To reflect the trend of deposition-heat release changing over time, the module further calculates the deposition-heat release inversion increment between adjacent monitoring periods. The so-called deposition-heat release inversion increment refers to the combustible deposit-heat release inversion quantity of the later monitoring period minus the inversion quantity of the previous monitoring period. If the result is positive, it indicates that the deposition-heat release has increased between these two monitoring periods; if the result is negative, it indicates that the deposition-heat release has decreased. In this invention, each such inversion increment can be directly regarded as the deposition-heat release growth intensity value within that time period, or multiple increments can be smoothed to obtain a more stable deposition-heat release growth intensity value when needed. For a simple numerical example, in three consecutive monitoring periods, the combustible material deposition-heat release inversion amounts are 0.2, 0.5, and 0.9, respectively. Then, the deposition-heat release inversion increment of the second period relative to the first period is 0.3, and the deposition-heat release inversion increment of the third period relative to the second period is 0.4. The module can use these two increments as the deposition-heat release growth intensity value for the corresponding time period, or it can take their average to obtain 0.35, which can be used as the overall deposition-heat release growth intensity value for the three periods. Which method is used can be preset and kept constant during the system calibration phase.

[0059] After obtaining the deposition-heat release growth intensity value, the coupling inference module compares this growth intensity value with the time-varying data of the air compressor operating pressure to extract the pressure fluctuation component that changes in the same direction as the deposition-heat release change within the same monitoring period. Here, the time-varying data of the air compressor operating pressure refers to the chamber pressure or exhaust pressure value recorded in each monitoring period. Generally, the pressure at the beginning and end of each monitoring period can be selected, or the average pressure within that period can be selected. For ease of analysis, the module first calculates the pressure change of each monitoring period relative to the previous period. For example, subtracting the average pressure of the first period from the average pressure of the second period yields the pressure fluctuation for that time period. When the deposition-heat release growth intensity value is positive and the pressure change is also positive, it indicates that the enhancement of deposition-heat release and the increase in pressure occur simultaneously, which is a co-directional change. When the deposition-heat release growth intensity value is negative and the pressure change is also negative, it indicates that the weakening of deposition and the decrease in pressure occur simultaneously, also a co-directional change. If one is positive and the other is negative, it is an inverse change. In this invention, the coupling inference module only considers the pressure change under the aforementioned unidirectional change condition as a pressure fluctuation component directly related to deposition-heat release. Cases of inverse changes or where one component changes while the other remains almost unchanged are considered to have a weaker coupling relationship and can be set to zero or excluded from subsequent calculations. Within each monitoring cycle, the module records the pressure change that meets the unidirectional change condition as the coupled pressure fluctuation index for that cycle, used to characterize the degree of pressure response to changes in deposition-heat release. For example, in three consecutive cycles, the deposition-heat release growth intensity values ​​are 0.3, 0.4, and 0.1, respectively, while the corresponding pressure changes are two units, one unit, and -0.5 units, respectively. In the first cycle, the growth intensity is positive and the pressure change is positive, indicating a unidirectional change, and the corresponding coupled pressure fluctuation index is set to two units. In the second cycle, the growth intensity is positive and the pressure change is positive, indicating a unidirectional change, and the corresponding index is set to one unit. In the third cycle, the growth intensity is positive but the pressure change is negative, indicating an inverse change, and the corresponding coupled pressure fluctuation index is set to zero. In this way, the module extracts the pressure fluctuation component that is highly correlated with deposition-heat release behavior from the raw data of operating pressure changes over time.

[0060] After obtaining the coupled pressure fluctuation index, the coupling inference module also performs a one-to-one pairing operation between this index and the exhaust temperature rise rate obtained within the corresponding monitoring period to obtain the pressure-temperature rise coupling response strength that comprehensively reflects the combined effect of the operating pressure fluctuation response and the exhaust temperature rise rate. Here, the exhaust temperature rise rate refers to the rate at which the exhaust temperature rises within a monitoring period. For example, it can be represented by dividing the difference between the exhaust temperature in the current period and the exhaust temperature in the previous period by the monitoring period duration. A larger temperature rise rate indicates a faster temperature rise at the exhaust end. In each monitoring period, the module pairs the coupled pressure fluctuation index of that period with the exhaust temperature rise rate of that period. A simple product or weighted combination method can be used to reflect the combined effect strength of the two. For ease of explanation in this embodiment, a product method can be used to define the pressure-temperature rise coupling response strength. That is, when the coupled pressure fluctuation index and the exhaust temperature rise rate are both large in a certain period, the pressure-temperature rise coupling response strength will obtain a higher value, reflecting that the enhanced deposition-heat release, pressure response, and temperature rise all point to a high-risk state within that period. Continuing with the previous example, assuming the coupled pressure fluctuation index is two units in the first cycle and the exhaust temperature rise rate is three degrees Celsius per minute, the pressure-temperature rise coupling response intensity for that cycle can be understood as two times three, corresponding to six units. In the second cycle, the coupled pressure fluctuation index is one unit, and the exhaust temperature rise rate is two degrees Celsius per minute, so the pressure-temperature rise coupling response intensity for that cycle is two units. In the third cycle, the coupled pressure fluctuation index is zero. Even if the exhaust temperature rise rate has a certain value, since it does not form a response in the same direction as the deposition-heat release change, its pressure-temperature rise coupling response intensity is still recorded as zero. Through this cycle-by-cycle paired calculation, the module generates a quantitative index for each monitoring cycle that can simultaneously reflect the degree of "resonance" between pressure and temperature changes.

[0061] After obtaining the pressure-temperature rise coupled response intensity, the coupling inference module also needs to combine the continuous change trend of the deposition-heat release growth intensity value within a preset time window to generate a fire extinguishing early warning trend value that ultimately characterizes the direction and speed of thermal runaway risk evolution in the near future. The preset time window refers to a fixed-length time interval selected by the system, such as the most recent three or five monitoring periods. The module only considers data within this window when calculating the fire extinguishing early warning trend value. The module first collects the pressure-temperature rise coupled response intensity and deposition-heat release growth intensity value for each monitoring period within this time window. Then, it superimposes these two values ​​for each period to obtain the comprehensive risk intensity value for that period. Alternatively, a higher weight can be assigned to one of the factors as needed; for example, in coal mine conditions where exhaust temperature rise is more sensitive, the contribution of the pressure-temperature rise coupled response intensity can be appropriately amplified. Next, the module averages or calculates a weighted average of all comprehensive risk intensity values ​​within the time window to obtain a value representing the overall risk level of that time window. The larger this value, the more prominent the deposition, pressure rise, and rapid temperature increase are during this period. To further reflect the "evolution direction" and "evolution speed," the module can also compare the average comprehensive risk intensity of the first and second halves of the time window. If the average value of the second half is significantly higher than that of the first half, it indicates that the risk is accelerating, and the fire extinguishing warning trend value can be increased by a certain percentage based on the original average value. If the average value of the second half is similar to that of the first half, it indicates that the risk remains stable, and the fire extinguishing warning trend value can be kept at the basic level. If the average value of the second half is lower than that of the first half, it indicates that the risk has eased, and the fire extinguishing warning trend value can be appropriately reduced. Taking three monitoring cycles as an example, assuming that within a certain time window, the deposition-heat release growth intensity values ​​of the three cycles are 0.3, 0.4, and 0.5, respectively, and the corresponding pressure-temperature rise coupling response intensities are 2, 3, and 4, respectively, then the comprehensive risk intensity values ​​of the three cycles can be obtained by adding the growth intensity to the coupling response intensity, which are 2.3, 3.4, and 4.5, respectively. The module can first average these three values ​​to obtain a basic fire extinguishing warning trend value of approximately 3.4, and then observe its changes over time. Since the value shows a clear upward trend, it indicates that the risk is continuously increasing. The module can then adjust the basic trend value upward by a certain percentage under the rules set by the system, for example, by 20%, and finally output a fire extinguishing warning trend value of approximately 4.08 to the trigger judgment module.

[0062] The trigger judgment module 104 is used to generate a predictive fire extinguishing trigger index based on the fire extinguishing early warning trend value, and output a fire extinguishing trigger command when the threshold is reached.

[0063] The trigger judgment module 104 performs further logical processing on the fire extinguishing warning trend value output by the coupling inference module 103 after the trend value is output, in order to generate a predictive fire extinguishing trigger index that represents the urgency of fire extinguishing. When the index reaches a preset activation condition, a fire extinguishing trigger command is output. The predictive fire extinguishing trigger index is a quantitative value used to indicate whether the thermal runaway process within the air compressor cavity is approaching an irreversible point. Its value is typically set between zero and one; the closer it is to one, the more urgent the fire extinguishing intervention should be.

[0064] The trigger judgment module 104 first receives the fire extinguishing warning trend value. This trend value is derived from the comprehensive analysis of three factors—deposition inversion amount, pressure fluctuation, and exhaust temperature rise rate—by the coupled inference module 103, and is a risk value with continuously changing characteristics. The trigger judgment module uses this trend value as the basic input to determine whether the air compressor is approaching a sensitive stage before thermal runaway. However, the warning trend value only reflects the state change within the current time window and cannot directly determine whether to immediately trigger fire extinguishing. Therefore, the trigger judgment module needs to further analyze the rate and direction of change of this trend value over continuous time to form a predictable trigger index.

[0065] To achieve this function, the trigger judgment module monitors the changes in the fire extinguishing warning trend value across multiple adjacent monitoring windows. The duration of these monitoring windows can be set according to the operating structure of the air compressor; in this embodiment, each window can be understood as corresponding to five seconds of continuous data. The trigger judgment module compares the trend values ​​in several consecutive windows, observing whether the trend value shows a continuous increase, whether the increase is increasing, and whether the rate of increase is accelerating. If the trend value is not only high in a certain window but also shows a significant and continuous increase in subsequent windows, the module will classify this change as an acceleration of the thermal runaway development rate.

[0066] After obtaining information on trend changes, the trigger judgment module combines the trend value with the rate of trend growth according to preset logic to derive a predictive fire extinguishing trigger index. To ensure the index's calculability, the trigger judgment module combines the absolute risk level represented by the trend value itself with the time sensitivity represented by the rate of trend increase in a certain proportion. Specifically, when the trend value is high but the rate of trend change is slow, the trigger index will usually remain in a high but not yet triggered range; when the rate of trend change is very fast, even if the trend value has not yet reached a high level, a high trigger index may be generated to respond in advance to potential rapid temperature escalation risks. This design ensures that the system can respond differently to fire risks with different rates of development.

[0067] The following is a specific example. Within a continuous 15-second monitoring cycle, the fire extinguishing warning trend value output by the coupled inference module 103 is 0.58 in the first window, 0.68 in the second window, and 0.73 in the third window. The trigger judgment module first confirms that the overall trend value is continuously rising, indicating that the thermal excitation behavior inside the cavity is intensifying. The module then observes that the rate of increase in the trend value changes from 0.1 in the first window to 0.05 in the second, indicating that the rate of increase has slightly slowed but remains high. Based on this, the module uses the current trend value of 0.73 and its continuous growth rate as input, and generates a predictive fire extinguishing trigger index through a proportional merging method. If the trend value contribution is approximately 0.73, and the growth rate contribution is approximately 0.6, the module will combine the two in its internal logic to obtain a trigger index of approximately 0.69. This value indicates a high probability of the system entering thermal runaway in the near future, requiring close monitoring.

[0068] The trigger judgment module then compares the trigger index with the trigger threshold set internally by the system. The trigger threshold is set considering the air compressor structure, the fire extinguishing system response time, and the safety boundary. In this embodiment, when the trigger index reaches 0.7, the system determines that the risk has entered a state requiring early intervention. Since the trigger index in this example is approximately 0.69, which is very close to the trigger limit, the module will continue to observe the trend changes in the next monitoring window and decide whether to cross the threshold based on the actual changes. When the trigger index reaches or exceeds the threshold, the trigger judgment module will output a fire extinguishing trigger command to the next module immediately, enabling the oxygen reduction control module 105 to intervene in advance to change the oxygen concentration in the cavity and suppress the thermal ignition process that may develop into an open flame. The fire extinguishing trigger command carries a predictive fire extinguishing trigger index.

[0069] In this way, the trigger judgment module 104 not only makes a simple judgment on the trend value of the coupled inference module, but also derives a predictive trigger index based on the absolute height and continuous change characteristics of the trend value, so that the system can take suppression measures in advance before the danger is fully formed, thereby improving the overall fire extinguishing response speed and stability.

[0070] In this embodiment, the fire extinguishing warning trend value is a set of time-ordered risk quantification values ​​output by the coupled inference module according to the monitoring cycle. Each trend value corresponds to the combined effect of deposition-heat release behavior, operating pressure fluctuations, and exhaust temperature rise within a monitoring cycle. After receiving the fire extinguishing warning trend values ​​from several consecutive monitoring cycles, the trigger judgment module first performs time smoothing processing on the continuous changes of these trend values ​​within a preset time window. The preset time window can be set according to the system's response sensitivity requirements. For example, the most recent five monitoring cycles can be selected as the time window, so that the smoothing process can reflect the changes in risk in a short period of time without being overly sensitive to single occasional fluctuations. The purpose of time smoothing processing is to eliminate the abrupt impact of extreme values ​​within a single cycle, obtaining a more stable and smoothed warning risk value that can represent the current comprehensive risk level.

[0071] Time smoothing can be implemented using a simple and clear averaging method: summing the fire extinguishing warning trend values ​​corresponding to all monitoring periods within a preset time window, and then dividing by the number of monitoring periods to obtain a window average. In this invention, this window average is defined as the smoothed warning risk value, used to characterize the baseline intensity of the overall risk level over a current period. For example, at a certain moment, the system selects the most recent five monitoring periods as the time window, with fire extinguishing warning trend values ​​of 0.5, 0.6, 0.7, 0.8, and 0.9 respectively. The trigger judgment module first adds these five values ​​together, obtaining a sum of 3.5, and then divides 3.5 by 5 to obtain 0.7. At this point, 0.7 is the smoothed warning risk value within the window, indicating that in the most recent five monitoring periods, considering deposition-heat release, pressure fluctuations, and temperature rise, the overall risk level of the current system is relatively high but has not yet reached an extreme state.

[0072] After obtaining the smoothed early warning risk value, the trigger judgment module also needs to consider the maximum change range of the fire extinguishing early warning trend value within the same time window to reflect the speed and severity of the risk increase or decrease. In this invention, the maximum change range is defined as the maximum absolute value of the difference between the fire extinguishing early warning trend values ​​between any two adjacent monitoring periods within the time window. Specifically, the module will calculate the difference between the second period minus the first period, the third period minus the second period, the fourth period minus the third period, and the fifth period minus the fourth period in sequence, then take the absolute value of these differences, and find the largest one. Continuing with the above numerical example, if the five trend values ​​are 0.5, 0.6, 0.7, 0.8, and 0.9, then the differences between adjacent periods are 0.1, 0.1, 0.1, and 0.1, respectively, and the absolute value is still 0.1. Therefore, the maximum change range is 0.1. In another scenario, if the trend values ​​are 0.5, 0.6, 0.9, 0.8, and 0.9, then the differences are 0.1, 0.3, -0.1, and 0.1 respectively, with absolute values ​​of 0.1, 0.3, 0.1, and 0.1. At this point, the maximum change is 0.3, indicating that there was a significant surge in risk between the second and third cycles.

[0073] The trigger judgment module jointly evaluates the smoothed warning risk value and the maximum change range to obtain a risk assessment quantity that reflects the combined effect of the absolute strength and the rate of increase of the risk. The risk assessment quantity is a comprehensive indicator that considers both whether the overall risk level is high or low over a given period and whether the risk is changing rapidly. In this implementation, the risk assessment quantity can be obtained by weighted summation of the smoothed warning risk value and the maximum change range. Specifically, the weighting ratio can be pre-set during the system calibration stage. For example, the smoothed warning risk value can have a slightly higher weight to highlight the absolute strength of the risk, while the maximum change range can have a slightly lower weight to supplement the reflection of the rate of increase of the risk. For instance, the weight of the smoothed warning risk value can be set to two, and the weight of the maximum change range to one. That is, when calculating the risk assessment quantity, the smoothed warning risk value is multiplied by two, and then the maximum change range is added. Finally, it is normalized or kept as the original value as needed. Using the first example above, if the smoothed warning risk value is 0.7 and the maximum change range is 0.1, the risk assessment quantity can be obtained by multiplying 0.7 by two to get 1.4, and then adding 0.1 to get 1.5. For the second example, assuming the smoothed early warning risk value is 0.74 and the maximum change is 0.3, the risk assessment amount is 0.74 multiplied by 2, which is 1.48, plus 0.3, resulting in 1.78. Clearly, 1.78 is greater than 1.5, indicating that in the second case, even if the smoothed risk levels are similar, the overall risk assessment amount is correspondingly higher due to the larger single-cycle surge.

[0074] After obtaining the risk assessment value corresponding to the current monitoring period, the trigger judgment module also needs to refer to the highest risk assessment value recorded in historical monitoring periods to calculate the predictive fire extinguishing trigger index. The highest historical risk assessment value here is the maximum risk assessment value that has occurred in all monitoring periods since the system was put into operation, or within a certain reset period, and is used as an upper limit reference value for risk comparison. When calculating the predictive fire extinguishing trigger index, the module considers two aspects: first, the degree of closeness of the current risk assessment value to the highest historical risk assessment value; and second, the growth rate of the current risk assessment value relative to the risk assessment value of the previous monitoring period. To obtain a numerically intuitive and clearly defined index, the module can proceed as follows: First, compare the current risk assessment with the highest historical risk assessment. For example, divide the current risk assessment by the highest historical risk assessment to obtain a ratio between zero and one. The closer this ratio is to one, the closer the current risk is to the most dangerous state the system has ever experienced. Second, calculate the difference between the current risk assessment and the risk assessment of the previous monitoring period, and compare this difference with the risk assessment of the previous period. For example, divide the difference by the risk assessment of the previous period to obtain a relative proportion representing the growth rate. The larger this proportion, the faster the current risk is rising. Subsequently, the trigger judgment module can average the two ratios to obtain a comprehensive index that takes into account both the "absolute level of current risk" and the "current rate of risk growth." This comprehensive index serves as the base value for the predictive fire suppression trigger index. To illustrate with a specific numerical example, suppose the historical highest risk assessment value is 2.0, the risk assessment value of the previous period is 1.2, and the risk assessment value of the current period is 1.6. Then, the ratio of the current risk assessment value to the historical highest risk assessment value is 1.6 divided by 2.0, which equals 0.8. The increase in the current period compared to the previous period is 1.6 minus 1.2, which equals 0.4. Dividing 0.4 by 1.2 from the previous period yields 0.33. The module can take the average of 0.8 and 0.33, i.e., 0.565, as the base value for the unadjusted predictive fire extinguishing trigger index. If the system needs to give higher sensitivity to rapid increases under certain types of operating conditions, the weight of the growth rate ratio can be slightly increased to make the predictive fire extinguishing trigger index slightly more sensitive to rapid surges.

[0075] Finally, the trigger judgment module compares the predictive fire extinguishing trigger index with the dynamic trigger threshold, which is adaptively adjusted based on the air compressor's operating conditions. It outputs a fire extinguishing trigger command when the index reaches or exceeds the dynamic trigger threshold. The dynamic trigger threshold here is not a fixed constant, but a parameter that adjusts in real time according to the air compressor's operating conditions. During system calibration, different threshold ranges can be preset according to different operating conditions. For example, when the air compressor is under low load and the temperature and pressure are relatively stable, the dynamic trigger threshold can be set slightly higher to avoid excessively frequent fire extinguishing actions when the risk is low. When the air compressor is under high load or frequent start-stop operations, the dynamic trigger threshold can be appropriately lowered to make the system more sensitive to potential thermal runaway risks. Adaptive adjustment can be achieved by analyzing the average load, average pressure, and average exhaust temperature rise rate over a recent period. For example, when the average load and average exhaust temperature rise rate of the last ten monitoring cycles exceed the preset standard values, the system lowers the current dynamic trigger threshold from 0.7 to 0.65; otherwise, it maintains or slightly increases it.

[0076] In summary, once the predictive fire extinguishing trigger index reaches or exceeds the current dynamic trigger threshold, the trigger judgment module immediately generates a fire extinguishing trigger command and outputs it to the oxygen reduction control module. For example, under a certain operating condition, if the dynamic trigger threshold is set to 0.65 using an adaptive algorithm, and the previously calculated predictive fire extinguishing trigger index is 0.565, the system will not issue a fire extinguishing trigger command because the index is still below the threshold; it will only continue monitoring. In the following monitoring cycles, the risk assessment value further increases, causing the predictive fire extinguishing trigger index to be updated to 0.75. At this point, since 0.75 is higher than 0.65, the trigger judgment module will immediately output a fire extinguishing trigger command, driving the oxygen reduction control module into the quantitative oxygen reduction and environmental combustion suppression control stage.

[0077] The oxygen reduction control module 105 is used to receive the fire extinguishing trigger command and adjust the oxygen release inhibition adjustment coefficient according to the oxygen concentration and humidity of the cavity and the predictive fire extinguishing trigger index, thereby implementing quantitative oxygen reduction and outputting the adjusted oxygen concentration value.

[0078] The main function of the oxygen reduction control module 105 is to determine the appropriate oxygen reduction control intensity based on the fire extinguishing trigger command output by the trigger judgment module 104, combined with the current oxygen concentration in the air compressor cavity, the cavity humidity, and the urgency level reflected by the trigger index, and to generate a clear control command to guide the oxygen reduction execution unit to reduce the cavity oxygen concentration.

[0079] During operation, this module first receives a fire extinguishing trigger command from the trigger judgment module 104, and simultaneously collects real-time oxygen concentration and humidity data from the cavity monitoring unit. For ease of explanation, oxygen concentration in this system is divided into three levels: low oxygen, medium oxygen, and high oxygen. Low oxygen corresponds to a concentration range below 16%, medium oxygen corresponds to 16 to 20%, and high oxygen corresponds to above 20%. This classification is based on common combustion conditions inside the air compressor cavity. When the oxygen concentration is higher than 20%, oil mist, deposits, and metal surface temperature rise rapidly increase the ignition sensitivity, making hot spots more likely to develop. Therefore, the high oxygen level corresponds to the high-risk state determined by the system.

[0080] Humidity is also categorized into three levels: humid, moderate, and dry, reflecting the water vapor content within the cavity. A dry environment indicates a lack of moisture that can absorb heat, making it more difficult to naturally suppress hotspots once they form; a humid environment, on the other hand, can reduce the rate of temperature rise and diffusion to some extent. Therefore, humidity level is one of the important factors the system uses to assess actual thermal risk. Similar to oxygen concentration, the three humidity levels are mapped to risk labels: humid corresponds to "weak," moderate to "medium," and dry to "strong."

[0081] The predictive fire suppression trigger index is also divided into three levels: slight increase, significant increase, and rapid increase. This index level is determined by the trigger judgment module 104 based on the fire suppression warning trend value output by the coupled inference module 103, reflecting the rate of change in the hotspot development trend. A rapidly increasing index level typically indicates that the heat release of the sediment has changed significantly, and the hotspot has transformed from initial point-like growth to area or sheet-like expansion; therefore, it is assigned the highest risk level to ensure timely intervention of oxygen reduction control.

[0082] When the oxygen reduction control module 105 simultaneously obtains the oxygen concentration level, humidity level, and trigger index level, it maps each level to a preset risk label. For example, high oxygen corresponds to "strong," medium oxygen to "medium," and low oxygen to "weak"; dry to "strong," moderate to "medium," and humid to "weak"; a sharp increase to "strong," a significant increase to "medium," and a slight increase to "weak." These three labels together form a single adjustment input combination, and each combination corresponds to a control level that has been fixed at the system's factory. This method ensures that the adjustment result does not rely on calculation but on a clearly defined, fixed mapping table.

[0083] The system categorizes all possible combinations into three comprehensive risk levels, corresponding to mild, moderate, and severe oxygen reduction levels. The highest level of comprehensive risk is determined when two or three of the three input labels are "severe"; the lowest level is when all three labels are "weak"; and all other cases are classified as moderate. Based on the comprehensive risk level, the system directly looks up the corresponding oxygen release inhibition adjustment coefficient, for example, 0.25 for mild, 0.5 for moderate, and 0.8 for severe. These coefficients were solidified during the design phase based on extensive experimental data and are used to control the reduction in oxygen supply or the introduction of inert dilution gas by the oxygen reduction execution unit.

[0084] The following is a complete operational example. During a certain detection cycle, the trigger judgment module 104 outputs a fire extinguishing trigger command, indicating that the hotspot development trend has reached a point where immediate adjustment of the cavity environment is required. The oxygen reduction control module 105 reads that the cavity oxygen concentration at this time is 22%, which belongs to the high oxygen level, therefore its risk label is "strong". Simultaneously, it reads that the cavity humidity is approximately 20%, which corresponds to the dryness level, therefore the humidity risk label is also "strong". Then, it reads the trigger index as 0.78, which exceeds the threshold for a sharp increase, therefore its risk label is also "strong".

[0085] Since all three risk labels are "strong," the module directly determines the overall risk intensity to be the highest level. In the factory-preset strategy, the system sets the oxygen release inhibition adjustment coefficient corresponding to the highest level to 0.8. This value indicates that the oxygen reduction execution unit needs to reduce the oxygen supply by approximately 80% in this control operation, while simultaneously supplementing a certain proportion of nitrogen or other inert diluent gases, so that the oxygen concentration in the chamber drops to a level unfavorable to combustion within a very short time. For example, if the chamber volume and the original oxygen concentration are determined to decrease by less than five percentage points, the oxygen reduction execution unit will, according to the corresponding execution strategy, directly inject a fixed amount of inert gas and suppress oxygen supply based on this adjustment coefficient.

[0086] In a preferred embodiment of the present invention, after receiving the fire extinguishing trigger command output by the trigger judgment module, the oxygen reduction control module is designed as a closed-loop quantitative oxygen reduction control unit. It is used to dynamically determine a reasonable target oxygen concentration setpoint based on the actual working conditions of the current air compressor operating cavity, calculate the oxygen release inhibition adjustment coefficient that matches the target, and drive the oxygen reduction execution unit to adjust the oxygen-containing gas supply or inert gas injection in stages until the actual oxygen concentration in the cavity is stably close to the target oxygen concentration setpoint. Finally, the adjusted oxygen concentration value is given as the result feedback of this round of oxygen reduction control process.

[0087] When the system enters the oxygen reduction control phase, the oxygen reduction control module first responds to the fire extinguishing trigger command and locks the corresponding monitoring period according to the preset control cycle. The preset control cycle refers to the time unit used by the module when performing a complete oxygen reduction control calculation and action; for example, it can be set to ten, twenty, or thirty seconds. The length of the control cycle is pre-calibrated during system design based on the air compressor cavity volume, gas circulation refresh time, and measurement sensor response time, and remains fixed on the same equipment once determined. After locking a control cycle, the module calls the oxygen concentration sensor and humidity sensor within that cycle to collect real-time cavity oxygen concentration and humidity data within the air compressor's operating cavity. To ensure data representativeness, the module can collect raw data multiple times within a control cycle and obtain the current cavity oxygen concentration and humidity corresponding to that control cycle at the end of the cycle by averaging or removing obvious outliers. The module combines these two data points with the corresponding control cycle time stamp, current air compressor load status, and whether it is in a high-temperature warning zone, among other operating parameters, to form a current operating condition dataset containing multi-dimensional information. In this stage, the main task of the oxygen reduction control module is to accurately know basic information such as "how much oxygen is in the cavity, how humid is the air, and whether the air compressor is under light or heavy load" to provide a real working condition basis for subsequent calculations.

[0088] After the current operating condition dataset is constructed, the oxygen reduction control module correlates this dataset with the predictive fire suppression trigger index given by the trigger judgment module to obtain the target oxygen concentration setpoint characterizing the required oxygen reduction range for this control. The predictive fire suppression trigger index, defined in the aforementioned module, is a dimensionless quantity that comprehensively reflects the current level of thermal runaway risk and its growth rate; a larger value indicates that the risk is closer to or exceeds historical danger levels. During the correlation calculation, the oxygen reduction control module comprehensively considers the current cavity oxygen concentration, the predictive fire suppression trigger index, and the air compressor's load conditions. For example, under low-load conditions, to avoid excessive impact on production, even with a high trigger index, the target oxygen concentration setpoint can be set slightly higher than the minimum oxygen concentration required for strict combustion suppression; while under high-load conditions with a clearly deteriorating trend, the target oxygen concentration setpoint can be set significantly lower than the normal operating level to suppress combustion conditions as quickly as possible. For a specific example, in a certain control cycle, the current cavity oxygen concentration is measured at 20.5% by volume, and the predictive fire suppression trigger index is 0.75, a value close to the high-risk level pre-recorded by the system. The oxygen reduction control module can set the target oxygen concentration for the control cycle to 12 percent according to a pre-set rule. This means that the system aims to reduce the oxygen concentration in the chamber from a level close to that of ambient air to a level that is clearly insufficient to support continuous combustion after several control cycles of oxygen reduction action.

[0089] After obtaining the target oxygen concentration setpoint, the oxygen reduction control module enters the calculation stage of the oxygen release inhibition adjustment coefficient. This adjustment coefficient is used to quantify the intensity of work required by the oxygen reduction execution unit within the current control cycle; simply put, it's the control scale indicating "how much the valve should be closed and how much inert gas should be added." To calculate this coefficient, the module first compares the difference between the target oxygen concentration setpoint and the current cavity oxygen concentration. This difference directly reflects the adjustment range in the oxygen concentration dimension of this control cycle. For example, in the above example, if the current oxygen concentration is 20.5 percent and the target setpoint is 12 percent, the difference is 8.5 percent, indicating that the oxygen concentration needs to be reduced by 8.5 percentage points. Considering the difference alone only reflects "how much is still needed," but it cannot reflect the impact of humidity on combustion and oxygen reduction effects. Therefore, this invention introduces a humidity compensation curve to correct the oxygen release inhibition adjustment coefficient.

[0090] A humidity compensation curve is a pre-established curve based on extensive experimental data during equipment calibration. It reflects the differences in oxygen reduction required to maintain the same flame suppression effect under varying humidity conditions. Generally, higher humidity in the chamber increases water vapor content, which inhibits flame propagation to some extent, thus slightly reducing the required oxygen reduction under the same combustible conditions. Conversely, lower humidity results in drier air, making it easier for the combustible surface to heat and release volatile components, requiring a greater oxygen reduction to achieve the same flame suppression effect. The humidity compensation curve can be constructed segmented or graded, for example, dividing the humidity range into three levels: below 30%, 30% to 70%, and above 70%. Different correction coefficients are assigned to each level to amplify or reduce the oxygen release suppression adjustment coefficient. In practice, the oxygen reduction control module uses the humidity value from the current operating condition dataset to find the corresponding correction coefficient in the humidity compensation curve. This coefficient is then combined with the base adjustment intensity based on the oxygen concentration difference to obtain the final oxygen release suppression adjustment coefficient used to control the execution unit.

[0091] For example, continuing the previous example, assuming the current operating condition data shows a concentrated humidity of 80%, according to the pre-set humidity compensation curve, the correction coefficient for high humidity conditions is 0.8, indicating that under high humidity conditions, the oxygen reduction effort required to achieve the same flame suppression effect can be appropriately reduced. Simultaneously, according to internal rules, based on the oxygen concentration difference of 8.5%, the system divides the basic adjustment intensity into a "medium-high intensity" range. For example, the basic oxygen release inhibition adjustment coefficient for this range is 0.9, meaning that if humidity is not considered, the oxygen reduction execution unit needs to operate at 90% intensity. At this point, the oxygen reduction control module combines the basic adjustment coefficient of 0.9 with the humidity correction coefficient of 0.8. This can be achieved simply by multiplication or by combining according to preset weights, such as multiplying to obtain 0.72. This means that under high humidity conditions, in order to achieve the predetermined target oxygen concentration, the execution unit's operating intensity can be adjusted from the original 0.9 to 0.72, thereby saving inert gas consumption and avoiding excessive impact on the equipment. In this process, the oxygen release inhibition adjustment coefficient is a value between zero and one.

[0092] Once the oxygen release inhibition adjustment coefficient is determined, the oxygen reduction control module will drive the oxygen reduction actuator to adjust the amount of oxygen-containing gas supplied or inert gas injected into the air compressor operating chamber based on this coefficient. The oxygen reduction actuator can consist of one or more controllable flow valves, mixing valves, inert gas cylinder inlet valves, and related control actuators. In this invention, its specific structural form is not limited, but its operating intensity is uniformly constrained by the oxygen release inhibition adjustment coefficient. Specifically, when the oxygen release inhibition adjustment coefficient is 0.72, the opening degree of the relevant valves in the oxygen reduction actuator will be adjusted to approximately 72% of the maximum oxygen reduction operating state. For example, the inert gas nitrogen or carbon dioxide injection valve will be opened to 72% of its maximum flow rate, while the fresh air supply valve will be moderately closed, reducing the flow rate of oxygen-containing gas entering the chamber, thereby continuously reducing the oxygen concentration in the chamber throughout the entire control cycle. To avoid excessive instantaneous shocks during the oxygen reduction process, the system can adopt a phased approach to complete the oxygen reduction. For example, in the first control cycle, the oxygen reduction action is performed according to the current adjustment coefficient. After the cycle ends, the oxygen concentration in the cavity is remeasured. If it is not close to the target value, the adjustment coefficient is recalculated based on the new oxygen concentration and humidity, and the next control cycle begins. This process continues until the measured oxygen concentration in the cavity is close to the target oxygen concentration setting value.

[0093] During the oxygen reduction process, the oxygen reduction control module continuously monitors changes in the cavity oxygen concentration. When it detects that the cavity oxygen concentration has gradually decreased from the initial 20.5% to approximately 13%, and stabilizes within one or two control cycles, the module can determine that the target oxygen concentration setpoint has been essentially achieved, and thus ends the current phased quantitative oxygen reduction process. At the end of oxygen reduction, the module calls the oxygen concentration sensor to perform one or more final value measurements, averaging the measured final cavity oxygen concentrations to obtain the adjusted oxygen concentration value for this round of oxygen reduction control. For example, if the target value is 12%, and multiple measurements at the end of oxygen reduction yield 12.3%, 12.1%, and 11.9%, the module can average these three values ​​to obtain approximately 12.1% as the adjusted oxygen concentration value. This value will be used as feedback for this round of oxygen reduction control to subsequent modules, particularly providing a basis for judgment for the explosion-proof interlocking module. When the adjusted oxygen concentration value is lower than the preset safety threshold, the explosion-proof interlocking module can enter a stable explosion-proof maintenance mode.

[0094] Through the above continuous measurement, calculation and execution process, the oxygen reduction control module of the present invention realizes a complete control closed loop from the fire extinguishing trigger command to the determination of the target oxygen concentration setpoint, to the calculation of the oxygen release inhibition adjustment coefficient, and finally to the quantitative oxygen reduction and the output of the adjusted oxygen concentration result.

[0095] The explosion-proof interlocking module 106 is used to receive the adjusted oxygen concentration value and perform an explosion-proof interlocking action to maintain the explosion-proof state of the cavity when the oxygen concentration value is lower than the safety threshold.

[0096] The explosion-proof interlock module 106 is used to further determine whether the air compressor cavity has entered a state requiring forced explosion-proof protection after the oxygen concentration in the cavity has been adjusted to a new stable value by the oxygen reduction control module 105. When the adjusted oxygen concentration value is lower than the system's preset safety threshold, the module will immediately execute the explosion-proof interlock action to put the air compressor in explosion-proof condition, so as to prevent hot spots, residual combustibles, mechanical friction sparks, or oil mist spontaneous combustion from triggering the risk of deflagration in a low-oxygen but not completely inertized environment.

[0097] In actual operation, this module continuously receives the adjusted oxygen concentration value output by the oxygen reduction control module 105. During the design phase, the safety threshold for explosion-proof determination in this system is fixed at a range that has been repeatedly verified through experiments; for example, 500% is typically used as the critical threshold. When the oxygen concentration in the cavity falls below this threshold, it indicates that the cavity is in an extremely low-oxygen environment. Although combustion conditions are severely restricted, small-scale pressure pulses may still be generated due to localized high temperatures or residual particle activation. Therefore, it is necessary to immediately enter the explosion-proof interlock state to ensure that the cavity's sealing, pressure release path, and explosion-proof actuator all enter the highest level of protection.

[0098] When determining triggering conditions, the explosion-proof interlocking module does not only rely on the absolute value of the adjusted oxygen concentration, but also considers the stability window of the adjusted oxygen concentration over time. For example, if the adjusted oxygen concentration fluctuates significantly within a monitoring cycle, especially by about five percentage points, the module will consider this an indication that the gas stratification in the cavity has not yet been uniformly mixed. In this case, even if the value is slightly higher than the threshold for a short period, the module will still execute the explosion-proof action in advance based on the fluctuation to ensure that the cavity does not cause secondary risks due to oxygen concentration rebound or the formation of local oxygen-rich areas.

[0099] To illustrate the specific working process of this module, a directly applicable example is given below. During a certain oxygen reduction adjustment process, the oxygen reduction control module 105 reduces the oxygen concentration in the cavity from 22% to approximately 4% through an enhanced oxygen reduction strategy. This value is received by the explosion-proof interlock module 106, which immediately compares it with the system's fixed 5% safety threshold. Since 4% is below the threshold, the module immediately enters explosion-proof preparation mode. Before executing the explosion-proof interlock action, the module performs three consecutive verifications of the 4% concentration, with a verification cycle of one second, to confirm that the concentration value is not caused by short-term flow disturbances. If the three verification results all fluctuate within 0.5% of 4%, the module determines that the gas mixture in the cavity has basically stabilized and then sends an explosion-proof interlock signal to the explosion-proof execution unit.

[0100] After the explosion-proof interlock signal is triggered, the explosion-proof actuator will sequentially complete operations such as locking the explosion-proof door, closing the pressure relief valve, stopping the external gas supply channel of the cavity, and forcibly deploying the internal explosion-proof lining according to the system's preset action sequence. Throughout the process, the explosion-proof interlock module continuously monitors the micro-pressure changes inside the cavity. If a short-term pressure rise is detected due to heat release or structural rebound, it will immediately coordinate the pressure relief path to release to a safe value in a very short time, ensuring that the cavity remains in a completely explosion-proof state.

[0101] To further illustrate the details of the explosion-proof judgment, a more complex scenario is presented below. When the adjusted oxygen concentration initially drops to 4.8%, but then fluctuates between 4.6 and 5.2% multiple times within the next two seconds, the explosion-proof interlock module interprets this as an indication that the inert gas inside the cavity has not yet fully diffused. Although the instantaneous lowest point is below the threshold, the fluctuation amplitude suggests that there may still be a latent oxygen-rich area of ​​approximately 5.5%. Based on this, the explosion-proof interlock module will not immediately execute the explosion-proof action but will continue to monitor the gas mixing state. In a continuous verification three seconds later, when the oxygen concentration stabilizes at approximately 4.7%, the module will determine that the oxygen reduction mixing conditions meet the requirements and then execute the explosion-proof operation. This process ensures that the explosion-proof action occurs after the gas composition has truly stabilized, improving the reliability of the explosion-proof effect.

[0102] By constructing the above decision-making mechanism, the explosion-proof interlocking module 106 can not only trigger the explosion-proof action based on a clear oxygen concentration threshold, but also determine whether it is necessary to enter the explosion-proof state in advance by monitoring the trend of oxygen concentration change after adjustment, thereby avoiding the potential deflagration risk that may be caused by hot spots not being completely extinguished, local residual oil mist not being diffused, or cavity pressure fluctuations.

[0103] In a preferred embodiment of the present invention, the explosion-proof interlock module is set as the end safety barrier of the entire mine air compressor fire prevention and extinguishing system. Its working start point is to receive the adjusted oxygen concentration value output by the oxygen reduction control module, and on this basis, combined with the operating chamber pressure and machine base vibration acceleration data, to comprehensively assess the flame suppression margin and mechanical impact risk under the current working conditions. Finally, it drives the explosion-proof valve, the intake and exhaust isolation gate, and the motor shutdown circuit to perform explosion-proof interlock actions in a predetermined sequence according to a clear action level, and after the action is completed, it gives an explosion-proof status confirmation signal that can be called by the host computer or safety control system.

[0104] After the oxygen reduction control module completes a round of phased quantitative oxygen reduction, it outputs an adjusted oxygen concentration value. This value is obtained by averaging the cavity oxygen concentration measured multiple times at the end of this round of oxygen reduction, reflecting the actual oxygen content in the air compressor's operating cavity. When the explosion-proof interlock module receives the adjusted oxygen concentration value, it first compares this value with a preset safety threshold. The preset safety threshold is a fixed or graded upper limit of oxygen concentration determined during the system design and test calibration phase based on factors such as the structural characteristics of the mining air compressor, typical combustible materials, and the explosion limits of coal dust or oil mist. When the oxygen concentration in the operating cavity is lower than this threshold, it is considered that the chemical reaction conditions are no longer conducive to the continued development of combustion or explosion. For example, in a certain type of underground coal mine air compressor system, the preset safety threshold can be set to 14% by volume. When the oxygen concentration is lower than 14%, it can be considered that it has entered a relatively safe flame suppression zone.

[0105] To further quantify the safety margin at the current adjusted oxygen concentration, the explosion-proof interlocking module also calculates a remaining safety margin characterizing the current flame suppression capacity of the cavity. The remaining safety margin can be understood as the size of the safety space between "just reaching the safety threshold" and "the current actual oxygen reduction level," generally represented by the difference between the two. For example, if the preset safety threshold is 14%, and the adjusted oxygen concentration measured after this round of oxygen reduction is 12.1%, then the difference is 1.9%, which can be defined as the remaining safety margin under the current operating condition. The larger the difference, the more sufficient the oxygen reduction, and the greater the room for flame suppression in the cavity; a difference close to zero or zero indicates that although the safety threshold requirement is just met, a sudden increase in pressure or vibration could easily cause it to revert to an unstable state. The explosion-proof interlocking module uses this remaining safety margin as one of the important inputs for subsequent interlocking risk calculations.

[0106] After obtaining the remaining safety margin, the explosion-proof interlocking module does not solely rely on oxygen concentration to determine whether to execute the interlocking action. Instead, it comprehensively considers the operating cavity pressure and base vibration acceleration data collected during the corresponding monitoring period. The operating cavity pressure reflects the internal gas pressure; when the pressure abnormally increases, even if the oxygen concentration has decreased, residual combustibles may still create dangerous pressure and shock waves locally. The base vibration acceleration reflects the overall mechanical system's vibration condition; an abnormally large increase in vibration acceleration may indicate internal mechanical faults such as impact, friction, or jamming. Combined with residual combustibles and high temperatures, this significantly increases the risk of localized ignition. Therefore, this invention correlates the remaining safety margin with the operating cavity pressure and base vibration acceleration to obtain a comprehensive explosion-proof interlocking risk measure that reflects the degree of pressure and vibration risk under the current flame suppression margin.

[0107] In practical implementation, the explosion-proof interlock module can first compare the operating chamber pressure and the base vibration acceleration with their respective normal upper limits set during the calibration phase. For example, in a certain model of equipment, the normal upper limit for the operating chamber pressure can be set to one MPa; when the monitored pressure reaches one MPa or higher, it is considered an abnormal pressure range. The normal upper limit for the base vibration acceleration can be set to five meters per second squared; when this value exceeds this limit, it is considered an abnormal vibration range. Then, the module will construct a comprehensive risk level based on the proximity of the current pressure value to the normal upper limit, the proximity of the current vibration acceleration value to the normal upper limit, and the size of the remaining safety margin. For example, when the remaining safety margin is large, even if the pressure and vibration slightly exceed the limits, the increase in the explosion-proof interlock risk level can be appropriately controlled; when the remaining safety margin is small or even close to zero, as long as there is a moderate abnormality in pressure or vibration, the explosion-proof interlock risk level will increase significantly. For a concrete example, suppose the remaining safety margin after this round of oxygen reduction is 1.9%, the operating chamber pressure is 0.95 MPa, slightly below the upper limit of 1 MPa, and the base vibration acceleration is 4.5 m / s², slightly below the upper limit of 5 m / s². This operating condition can be judged as "high oxygen margin, mechanical condition is tight but not serious", and the explosion-proof interlock risk level can be assigned a moderately low value. Now suppose another operating condition, where the remaining safety margin is only 0.2%, the operating chamber pressure has reached more than 1 MPa, and the vibration acceleration is also close to or slightly exceeds the upper limit of 5 m / s². In this case, the explosion-proof interlock module can assign an explosion-proof interlock risk level close to the highest level to trigger a more intense explosion-proof interlock action.

[0108] After obtaining the explosion-proof interlock risk level, the explosion-proof interlock module needs to map this continuous value to a preset classification range to generate a discrete explosion-proof interlock action level indicator. The preset classification range is generally determined during the system design and testing phases and can be divided into, for example, low-risk, medium-risk, high-risk, and extremely high-risk ranges, each corresponding to a specific interlock action level. For example, the explosion-proof interlock risk level can be internally normalized to between zero and one. When the risk level is between zero and 0.3, it falls into the low-risk range, requiring only pre-closing or reducing the opening of some valves; when the risk level is between 0.3 and 0.6, it falls into the medium-risk range, requiring the closure of some pipelines leading to non-hazardous areas and preparation for shutdown; when the risk level is between 0.6 and 0.8, it falls into the high-risk range, requiring rapid closure of most explosion-proof valves and intake / exhaust isolation gates; and when the risk level is above 0.8, it is considered an extremely high-risk range, requiring simultaneous disconnection of the motor shutdown circuit and bringing the entire system into a fully explosion-proof state. The explosion-proof interlocking action level indicator is a specific level identifier. For example, it can be represented by levels one, two, three, and four to indicate the four different interlocking levels mentioned above. It will directly determine the type, number, and sequence of subsequent actions.

[0109] After the explosion-proof interlock action level indicator is generated, the explosion-proof interlock module will sequentially drive the explosion-proof valves connected to the air compressor operating cavity, the intake and exhaust isolation gates, and the motor shutdown circuit to the closed or disconnected state according to the level sequence determined by the indicator. Taking a typical three-level interlock action as an example, when the explosion-proof interlock risk level is classified as medium risk and the corresponding interlock action level indicator is level two, the explosion-proof interlock module will first control the closure of some explosion-proof valves leading to non-critical branches to block the connection between the air compressor operating cavity and the external pipeline network or other equipment. Then, it will drive the intake and exhaust isolation gates to gradually close, separating the air compressor operating cavity from the external atmospheric environment from the intake and exhaust sides. Finally, if necessary, it will issue a load reduction or shutdown warning signal, but will not immediately disconnect the motor shutdown circuit to retain limited space for manual intervention. When the explosion-proof interlock risk level is at its highest level, corresponding to level four of the interlock action level indicator, the explosion-proof interlock module will close all explosion-proof valves and intake / exhaust isolation gates in parallel or rapidly in sequence within a very short time. Simultaneously, it will directly cut off the motor stop circuit, stopping the motor and thus minimizing the application of mechanical or compressive energy to the cavity under high-risk conditions. The specific sequence and delay of these actions can be determined through parameter configuration during system design. For example, in level two actions, a certain slow-closing time can be set for gate closure, while in level four actions, gate closure and the stop command can be triggered almost simultaneously to achieve the fastest explosion-proof response.

[0110] During the execution of interlocking actions, the explosion-proof interlocking module also monitors the status feedback signals of each actuator in real time, such as the on / off position signals of explosion-proof valves, the position signals of inlet and outlet isolation gates, and the disconnection confirmation signals of motor stop circuits. When it is detected that all actuators corresponding to the current action level have reached the predetermined state, the explosion-proof interlocking module generates an explosion-proof status confirmation signal and outputs it to the upper-level safety control system or local alarm unit. The explosion-proof status confirmation signal can be a specific logical quantity, or it can carry information such as the current explosion-proof level, a list of shut-down equipment, and a timestamp for subsequent accident analysis or recovery operations. For example, after the fourth-level interlocking action is completed, the explosion-proof status confirmation signal can include a marker that "all main and branch explosion-proof valves are closed, inlet and outlet isolation gates are closed, motors have stopped, and the cavity is in a fully explosion-proof state," so that operators and the upper-level system will not mistakenly believe that the system is still running.

[0111] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

Claims

1. A fire prevention and extinguishing system for a mining air compressor, characterized in that, include: The perturbation monitoring module is used to collect the intensity of micro-plasma perturbation of the gas in the air compressor operating cavity and analyze the ionization discharge waveform to obtain a perturbation characteristic sequence that characterizes the hot spot formation trend. The deposition inversion module is used to receive the perturbation feature sequence and calculate the combustible deposition-heat release inversion amount based on the response relationship between ionization features and temperature gradient, so as to characterize the combustible accumulation risk. The coupling inference module is used to receive the combustible material deposition-heat release inversion quantity and combine it with the operating pressure fluctuation and exhaust temperature rise rate to generate a fire extinguishing early warning trend value; The trigger judgment module is used to generate a predictive fire extinguishing trigger index based on the fire extinguishing early warning trend value, and output a fire extinguishing trigger command when the threshold is reached; The oxygen reduction control module is used to receive the fire extinguishing trigger command and adjust the oxygen release inhibition adjustment coefficient according to the oxygen concentration and humidity of the cavity and the predictive fire extinguishing trigger index, thereby implementing quantitative oxygen reduction and outputting the adjusted oxygen concentration value. The explosion-proof interlocking module is used to receive the adjusted oxygen concentration value and perform an explosion-proof interlocking action to maintain the explosion-proof state of the cavity when the oxygen concentration value is lower than the safety threshold.

2. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The perturbation monitoring module is specifically used for: Multiple independent micro-plasma disturbance acquisition windows are generated within the air compressor operating cavity using a time-domain segmented acquisition method, and an initial micro-plasma disturbance signal sequence is obtained within each acquisition window. The initial micro-plasma disturbance signal sequence is input into the waveform granularity analysis unit, and fine-grained structure analysis is performed on the rising edge of the ionization discharge waveform within the acquisition window to generate a waveform granularity feature set characterizing the local ionization activity. The disturbance influence region of micro-plasma discharge events in cavity gas is derived based on waveform granularity feature set, and the local disturbance contribution of hotspot germination location is determined by the response difference between different acquisition windows. Based on the contribution of the local disturbance, a disturbance feature sequence representing the trend of hotspot formation is generated.

3. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The sedimentation inversion module is specifically used for: Based on the perturbation feature sequence, the ionization intensity change of the continuous monitoring period is windowed and aggregated to form the aggregated value of ionization change characterizing the rate of change of the thermal state of combustibles. The ionization change aggregation value is input into the temperature gradient response analysis unit. By segmenting and matching the increase or decrease of the cavity operating temperature gradient at different time periods, the temperature gradient sensitivity coefficient reflecting the thermal sensitivity tendency of the sediment is obtained. The ionization change polymerization value was thermally amplified based on the temperature gradient sensitivity coefficient to generate the ionization excitation enhancement corresponding to the local sediment heating and release process. By coupling the ionization-induced enhancement amount with the local perturbation contribution sequence in the perturbation characteristic sequence, the combustible sedimentation-heat release inversion amount characterizing the degree of sediment accumulation and instantaneous heat release is obtained.

4. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The coupling inference module is specifically used for: Based on the inversion of combustible material deposition-heat release, a deposition-heat release time series divided by monitoring period is constructed, and the deposition-heat release inversion increment between adjacent monitoring periods is calculated to obtain the deposition-heat release growth intensity value. The deposition-heat release growth intensity value is compared with the change data of air compressor operating pressure over time. The pressure fluctuation component that changes in the same direction as the deposition-heat release growth intensity value in the same monitoring period is extracted to generate a coupled pressure fluctuation index that characterizes the degree of pressure response to changes in deposition-heat release. The coupled pressure fluctuation index is paired with the exhaust temperature rise rate obtained in the corresponding monitoring period to obtain the pressure-temperature rise coupled response intensity that comprehensively reflects the combined effect of the operating pressure fluctuation response and the exhaust temperature rise rate. Based on the continuous changing trends of the pressure-temperature rise coupling response intensity and the deposition-heat release growth intensity within a preset time window, a fire extinguishing early warning trend value is generated to characterize the direction and speed of thermal runaway risk evolution in the near future.

5. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The triggering judgment module is specifically used for: Based on the continuous changes of the fire extinguishing warning trend value within a preset time window, time smoothing is performed to generate a smoothed warning risk value that represents the current comprehensive risk level. By jointly assessing the maximum change in the smoothed early warning risk value and the fire extinguishing early warning trend value within the same time window, a risk assessment quantity reflecting the combined effect of the absolute intensity and rate of increase of the risk is obtained. The predictive fire extinguishing trigger index is calculated based on the risk assessment volume and the highest risk assessment volume recorded in the historical monitoring period, so that the predictive fire extinguishing trigger index simultaneously reflects the degree of proximity of the current risk to the highest historical risk and the current risk growth rate. The predictive fire extinguishing trigger index is compared with the dynamic trigger threshold obtained by adaptive adjustment based on the air compressor operating conditions. When the predictive fire extinguishing trigger index reaches or exceeds the dynamic trigger threshold, a fire extinguishing trigger command is output.

6. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The oxygen reduction control module is specifically used for: Upon receiving the fire extinguishing trigger command, the corresponding monitoring period is locked according to the preset control cycle, and real-time oxygen concentration and humidity data of the air compressor operating cavity are collected to form a current operating status dataset. The current operating condition dataset is correlated with the predictive fire extinguishing trigger index to obtain the target oxygen concentration set value that represents the required oxygen reduction range; Based on the difference between the target oxygen concentration setpoint and the current cavity oxygen concentration, and combined with the corresponding correction coefficient of the humidity in the preset humidity compensation curve, calculate the oxygen release inhibition adjustment coefficient used to control the working intensity of the oxygen reduction execution unit. Based on the oxygen release inhibition adjustment coefficient, the oxygen reduction actuator adjusts the amount of oxygen-containing gas supplied or inert gas injected into the air compressor operating chamber to implement phased quantitative oxygen reduction until the oxygen concentration in the chamber approaches the target oxygen concentration set value, and outputs the adjusted oxygen concentration value characterizing the oxygen reduction result at the end of the oxygen reduction process.

7. The fire prevention and extinguishing system for mining air compressors according to claim 1, characterized in that, The explosion-proof interlocking module is specifically used for: Upon receiving the adjusted oxygen concentration value, the adjusted oxygen concentration value is compared with the preset safety threshold to calculate the remaining safety margin that characterizes the current flame suppression margin of the cavity. The remaining safety margin is correlated with the operating cavity pressure and base vibration acceleration data collected during the corresponding monitoring period to obtain the explosion-proof interlock risk quantity, which reflects the comprehensive risk level of cavity pressure and vibration under the current flame suppression margin. Based on the level range to which the explosion-proof interlock risk level belongs within the preset grading range, an explosion-proof interlock action level indication quantity is generated, so that the explosion-proof interlock action level indication quantity clearly indicates the level sequence of the explosion-proof interlock actions that need to be performed. According to the explosion-proof interlock action level indication, the explosion-proof valves, intake and exhaust isolation gates and motor stop circuits connected to the air compressor operating cavity are driven in a predetermined sequence to enter the closed or cut-off state in sequence until the cavity is in an explosion-proof state isolated from the external combustible gas environment. After the action is completed, an explosion-proof status confirmation signal is output to characterize the completion of the explosion-proof interlock action.

Images