An intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency

Abstract

The present invention discloses an intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency, comprising an adsorption-type climbing robot, a multi-parameter sensor, a liquid nitrogen injection mechanism and a control system. The multi-parameter sensor establishes a multi-sensor confidence weighted decision-making mechanism to screen and feed back valid data; the adjustment mechanism is configured to adjust the orientation of the hose nozzle in three angular directions (X, Y, and Z); the control system, through the cooperation of a built-in fire situation analysis module and an energy density assessment model, breaks through the limitations of traditional single-dimensional detection and achieves real-time matching of fire extinguishing parameters and fire characteristics; after the first liquid nitrogen injection, the control system automatically evaluates the fire extinguishing effect and immediately implements a secondary injection if there is a risk of re-ignition, effectively preventing the fire from reigniting and improving the reliability of fire extinguishing.

Description

Technical Field

[0001] The present invention relates to the technical field of safety protection of electrochemical energy storage systems, and in particular to an intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency. Background Art

[0002] Currently, liquid nitrogen protection technology for electrochemical energy storage systems has become a research hotspot. Among them, extremely early monitoring and early warning of thermal runaway and dynamic tracking have become the main means to prevent thermal runaway. However, existing technologies mostly use fixed sensor networks to achieve thermal runaway process monitoring. Fixed sensors are limited by their installation location and cannot cover the multi-layer structure and hidden areas of battery racks, resulting in monitoring blind spots and high deployment costs. In addition, some scholars have conducted extensive research on inspection mechanisms based on preset paths. Track-mounted robots rely on fixed rails for inspection operations, and their flexibility cannot adapt to the dynamic inspection needs of complex battery clusters. Traditional inspection equipment mostly uses single or low-parameter detection modes, lacking multi-dimensional early perception and warning capabilities for the early characteristics of thermal runaway in lithium-ion batteries. In summary, existing technologies cannot achieve full-space perception of the energy storage compartment when monitoring and warning of thermal runaway in energy storage batteries. They have defects such as limited detection range, difficulty in adapting the monitoring mechanism to the dynamically changing energy storage environment, and delayed thermal runaway identification. Therefore, an intelligent inspection-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency is urgently needed to solve the above problems. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency, which can effectively solve the problems existing in the above-mentioned prior art.

[0004] To solve the above technical problems, the present invention adopts the following technical solution: an intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency, comprising:

[0005] An adsorption-type climbing robot has multiple climbing joints, each of which is equipped with an electromagnetic suction cup at its end;

[0006] A multi-parameter sensor is installed at the traveling end of the adsorption-type climbing robot, and a multi-sensor confidence weighted decision-making mechanism is established for the multi-parameter sensor to screen and feed back valid data;

[0007] The liquid nitrogen spraying mechanism comprises a liquid nitrogen tank assembled on the body of the adsorption climbing robot and a spraying hose with a hose nozzle connected to the liquid nitrogen tank, and

[0008] The adsorption climbing robot is further provided with an adjustment mechanism, the spray hose is connected to the adjustment mechanism, and the adjustment mechanism is configured to adjust the direction of the hose nozzle in three angular directions of XYZ;

[0009] The control system shall have at least the following functions:

[0010] Based on the initial valid data, the built-in fire situation assessment module predicts the fire risk level. Based on the predicted level, the adsorption climbing robot is controlled to quickly move from the initial inspection position to the fault area.

[0011] Control multi-parameter sensors to implement secondary precise detection and feedback secondary valid data;

[0012] Based on the secondary effective data, the battery energy release rate is calculated in real time through the built-in energy density evaluation model, the injection strategy is formed, and the regulating mechanism is controlled to perform corresponding actions.

[0013] Preferably, the plurality of climbing joints are installed at both ends of the fuselage in two groups, and are divided into a front climbing joint and a rear climbing joint, and the front climbing joint is connected to the fuselage through a mirror pitch joint; and

[0014] Each of the climbing joints includes a servo joint, an upper arm and a lower arm. The servo joint is installed on the fuselage. One end of the upper arm is connected to the servo joint through the upper arm joint, and the other end is connected to the lower arm through the lower arm joint. An electromagnetic suction cup is installed on the end of the lower arm away from the upper arm.

[0015] Preferably, the suction force of the electromagnetic chuck can be dynamically adjusted according to the surface material and contact pressure of the cabin, and the adjustment formula is:

[0016]

[0017] Where B is the magnetic induction intensity of the working air gap; A is the effective adsorption area; μ0=4π×10 -7 is the vacuum permeability; μ r is the relative magnetic permeability of the adsorbed material, that is, the correction coefficient of the surface material; k is the magnetic leakage coefficient.

[0018] Preferably, the multi-sensor confidence weighted decision-making mechanism includes:

[0019] First level verification: check whether the data is within the preset range;

[0020] Second level verification: Check whether the logical relationship between temperature and gas production data is established, and determine the logical consistency of multiple parameters;

[0021] The third level of verification: compare the difference between the current data and the mean of historical data, and eliminate data with a difference greater than the preset range;

[0022] Perform triple verification in sequence, and feed back the data that passes the triple verification as valid data, and update the historical data with the valid data.

[0023] Preferably, a mist arrester is arranged between the spray hose and the liquid nitrogen tank, and the mist arrester is configured to reduce the impact impulse of high-speed liquid nitrogen.

[0024] Preferably, the adjustment mechanism includes a horizontal servo installed on the fuselage, a first single-axis servo, a second single-axis servo and a third single-axis servo installed on a segmented bracket along the vertical direction, and a mounting frame, the spray hose is assembled on the mounting frame, the horizontal servo is configured to control the mounting frame to adjust in the XY axis direction, the first single-axis servo and the second single-axis servo control the segmented structure robotic arm to achieve Z-axis direction adjustment of the spray hose, and the third single-axis servo controls the direction of the hose nozzle, and the linkage and cooperation realize all-round liquid nitrogen spraying.

[0025] Preferably, the fire situation analysis module performs the following functions:

[0026] Load the pre-trained anomaly detection model and fire extinguishing database, and train the anomaly detection model;

[0027] Load the extinguishing agent-distance-angle three-dimensional database to store the extinguishing parameters corresponding to different fire levels;

[0028] Simulate and train anomaly detection models, generate random training data, and fit the model;

[0029] Construct multi-dimensional feature vectors for machine learning models, extract valid data and calculate cross-features;

[0030] Based on a hybrid analysis of supervised and unsupervised learning, the fire severity is determined using anomaly detection models and a rule engine. If the anomaly detection results conflict with the rule engine results, a review mechanism is triggered.

[0031] The fire extinguishing parameters are obtained from the database according to the fire level, and the posture compensation is calculated based on the current posture of the suction climbing robot to control the suction climbing robot to move.

[0032] Preferably, the energy density assessment execution function includes:

[0033] Initialize the energy density evaluation model; set the battery parameters;

[0034] Based on the improved Bernardi heat production model, ohmic heat, polarization heat, and entropy heat are calculated. The steady-state equations of the original model are replaced by non-steady-state partial differential equations. The finite element method is combined to realize the dynamic distribution of heat production in the three-dimensional space of the human body and the total heat production power is obtained by summing them.

[0035] The battery internal resistance is calculated based on the temperature through the battery internal resistance temperature compensation model;

[0036] Based on the liquid nitrogen injection parameter decision table, the direction, distance, dosage, speed, duration and pattern of the liquid nitrogen injection are determined according to the heat generation power to form an injection strategy.

[0037] Preferably, the energy density evaluation further comprises: comparing the change values ​​of temperature and gas concentration before and after injection, and when the change value is less than a preset value, performing a secondary liquid nitrogen injection action.

[0038] Beneficial effects: The present invention cooperates with the adsorption climbing robot, multi-parameter sensor and liquid nitrogen injection mechanism to achieve wide-range, multi-dimensional and extremely early perception and warning of the thermal runaway process, and through the multi-sensor confidence weighted decision-making mechanism, screens and feeds back effective data to ensure the reliability and anti-interference ability of the multi-parameter sensor data feedback. The built-in fire analysis module and energy density assessment model are used to break through the limitations of traditional single-dimensional detection and realize real-time matching of fire extinguishing parameters and fire characteristics. After the first liquid nitrogen injection, the control system automatically evaluates the fire extinguishing effect. If there is a risk of re-ignition, a secondary injection is immediately implemented to effectively prevent the fire from re-igniting and improve the reliability of fire extinguishing.

[0039] In addition, by designing a multi-degree-of-freedom adjustment mechanism, all-round injection and precise coverage of the source of thermal runaway are ensured, and under the control of the control system, extremely early monitoring and early warning of the thermal runaway process of the energy storage battery and dynamic process tracking are achieved. By building an integrated collaborative system of perception and early warning, efficient blocking and prevention of re-ignition, the stable operation of the electrochemical energy storage system is effectively protected. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0041] In the attached figure:

[0042] Figure 1 It is a structural schematic diagram of the intelligent patrol-type liquid nitrogen fire extinguishing device of the present invention;

[0043] Figure 2 It is a structural schematic diagram of the adsorption-type climbing robot of the present invention;

[0044] Figure 3 It is a schematic structural diagram of the spray hose of the present invention;

[0045] Figure 4 It is a structural schematic diagram of the liquid nitrogen tank of the present invention;

[0046] Figure 5 It is a structural schematic diagram of the regulating mechanism of the present invention;

[0047] Figure 6 This is a working flow chart of the intelligent patrol-type liquid nitrogen fire extinguishing device of the present invention.

[0048] Numbers in the figure: 1. Forearm; 2. Fuselage; 3. Electromagnetic suction cup; 4. Nose; 5. Control system; 6. Fixing device; 7. Liquid nitrogen tank; 8. Mist arrester; 9. First single-axis servo; 10. Horizontal servo; 11. Multi-parameter sensor; 12. Second single-axis servo; 13. Third single-axis servo; 14. Fixing bracket; 15. Spray hose; 16. Hose nozzle; 17. Upper bearing bracket; 18. Support spring; 19. Bearing; 20. Lower bearing bracket; 21. Short copper column; 22. Bearing disc; 23. Long copper column; 24. Fixing screw; 25. Rubber pad; 26. Rubber pad screw; 27. Servo joint; 28. Upper arm joint; 29. ​​Upper arm; 30. Forearm joint; 31. Mirror pitch joint; 32. Heat sink; 33. Mist controller; 34. Connector; 35. Flow control valve; 36. Connecting device. DETAILED DESCRIPTION

[0049] The following describes the embodiments of the present invention in conjunction with the accompanying drawings. The terms used in the embodiments of the present invention are only used to explain the specific embodiments of the present invention and are not intended to limit the present invention. The following describes the embodiments of the present application in conjunction with the accompanying drawings.

[0050] Example: Figure 1 As shown, an intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency includes a mobile adsorption climbing robot as a base, a liquid nitrogen injection mechanism for performing liquid nitrogen fire extinguishing, an adjustment mechanism for controlling the liquid nitrogen injection direction, and a control system 5 for connecting and controlling various functional components;

[0051] For the adsorption climbing robot, it has multiple climbing joints, and each climbing joint end is equipped with an electromagnetic suction cup 3. Figure 2 As shown, it specifically includes four climbing joints, which are symmetrically distributed at both ends of the suction climbing robot body, with the moving end of the suction climbing robot as the front end and the other end as the rear end, divided into front climbing joints and rear climbing joints, and the front climbing joint is connected to the fuselage through a mirror pitch joint; the lifting height of the front climbing joint can be controlled within a certain range in the vertical direction, each climbing joint includes a servo joint 27, an upper arm 29 and a small arm 1, the servo joint 27 is installed on the fuselage 2, one end of the upper arm 29 is connected to the servo joint 27 through the upper arm joint 28, and the other end is connected to the small arm 1 through the small arm joint 30, and the small arm 1 is equipped with an electromagnetic suction cup 3 at the end away from the upper arm 29, and the servo joint 27, the upper arm 29 and the small arm 1 can refer to Figure 2 The settings shown in the figure can flexibly adjust the position and direction of each arm and electromagnetic suction cup;

[0052] The suction force of the electromagnetic chuck 3 can be dynamically adjusted according to the surface material and contact pressure of the cabin. The adjustment formula is:

[0053]

[0054] Where B is the magnetic induction intensity of the working air gap; A is the effective adsorption area; μ0=4π×10 -7 is the vacuum permeability; μ r is the relative magnetic permeability of the adsorbed material, that is, the correction coefficient of the surface material; k is the magnetic leakage coefficient;

[0055] Ensure the robot's stable movement at different tilt angles and on complex surfaces, while having sufficient carrying capacity to meet the needs of different firefighting tasks and realize the full-space movement of the electrochemical energy storage cabin.

[0056] A multi-parameter sensor 11 is installed at the head 4 of the adsorption climbing robot. The multi-parameter sensor 11 is used to detect key parameters such as the temperature, gas production and flue gas of the energy storage battery within a specified range, and establish a multi-sensor confidence weighted decision-making mechanism (setting a triple verification logic) to feedback effective information;

[0057] The multi-sensor confidence weighted decision-making mechanism includes:

[0058] First level verification: check whether the data is within the preset range;

[0059] Second level verification: Check whether the logical relationship between temperature and gas production data is established, and determine the logical consistency of multiple parameters;

[0060] The third level of verification: compare the difference between the current data and the mean of historical data, and eliminate data with a difference greater than the preset range;

[0061] Perform triple verification in sequence, and feed back the data that passes the triple verification as valid data, and update the historical data with the valid data.

[0062] In a specific case, the main code of the triple check logic module that performs the above functions is as follows:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] Based on the above, a reasonable fluctuation range is dynamically calculated based on historical data, and the logical consistency of the data is verified using the temperature-gas production empirical formula to avoid misjudgments caused by single sensor failure. Time series analysis is used to identify sudden anomalies (such as transient sensor failures) to ensure the reliability and anti-interference ability of multi-parameter sensor data feedback;

[0070] The liquid nitrogen spraying mechanism includes a liquid nitrogen tank 7 assembled on the body of the adsorption climbing robot and a spray hose 15 with a hose nozzle 16 connected to the liquid nitrogen tank.

[0071] refer to Figure 1 and Figure 3-Figure 4 As shown, a fixing device 6 is installed on the fuselage 2 to support the liquid nitrogen tank 7. The fixing device 6 can adopt multiple fixing structures according to needs, including mechanical locking, elastic buffering and anti-slip design, to ensure that the liquid nitrogen tank 7 is stably fixed in a fast-moving or vibrating environment, avoid safety accidents caused by the liquid nitrogen tank 7 falling off, and improve the overall reliability of the device;

[0072] The output end of the liquid nitrogen tank 7 is connected to the spray hose 15 via a connecting device 36, and a mist arrester 8 is installed between the connecting device 36 and the spray hose 15 to reduce the impact impulse of the high-speed liquid nitrogen; a flow control valve 35 is installed between the mist arrester and the mist arrester 8, and a mist controller 33 is installed between the spray hose 15 and the hose nozzle 16;

[0073] For the adjustment mechanism, refer to Figure 5 As shown, it includes a first single-axis servo 9, a second single-axis servo 12, and a third single-axis servo 13 mounted on the fuselage and vertically mounted on a segmented bracket, as well as a fixed bracket 14. The spray hose 15 is assembled on the fixed bracket 14. The horizontal servo 10 is connected to the metal main steering wheel, which is connected to a large bearing, which is connected to the gimbal. The fixed bracket 14 is fixed to the gimbal. The rotation of the servo output shaft can control the rotation of the fixed bracket, achieving adjustment of the fixed bracket 14 in the X and Y axes. The first and second single-axis servos 9 and 12 are connected to small metal servos on both sides, which are connected to the segmented bracket via bolts. The spray hose can be adjusted in the Z axis by controlling the segmented bracket through the rotation of the servo output shaft. The third single-axis servo 13 is connected to the notch on the side of the hose nozzle via its output shaft. The rotation of the output shaft controls the direction of the hose nozzle, and the spray hose is coordinated and coordinated to achieve omnidirectional liquid nitrogen spraying.

[0074] An upper bearing bracket 17, a lower bearing bracket 20, and a bearing 19 are mounted on the bottom of the first single-axis servo 9. A support spring 18 is mounted between the upper and lower bearing brackets 17 and 20. The bottom of the lower bearing bracket 19 is connected to a bearing disc 22 via a plurality of short copper columns 21. The bottom of the bearing disc 22 is assembled to the fuselage 2 via a long copper column 23, fixing screws 24, rubber washers 25, and rubber washers screws 26. The horizontal servo 10 is mounted below the bearing disc 22.

[0075] The control system 5 has at least the following functions:

[0076] Based on the initial valid data, the built-in fire situation assessment module predicts the fire risk level. Based on the predicted level, the adsorption climbing robot is controlled to quickly move from the initial inspection position to the fault area.

[0077] Control multi-parameter sensors to implement secondary precise detection and feedback secondary valid data;

[0078] Based on the secondary effective data, the battery energy release rate is calculated in real time through the built-in energy density evaluation model to form an injection strategy, and the adjustment mechanism is controlled to perform corresponding actions, that is, the horizontal servo 10, the first uniaxial servo 9, the second uniaxial servo 12 and the third uniaxial servo 13 are controlled to act according to the corresponding parameters of the injection strategy.

[0079] The fire situation analysis module performs the following functions:

[0080] Load the pre-trained anomaly detection model and fire extinguishing database, and train the anomaly detection model;

[0081] Load the extinguishing agent-distance-angle three-dimensional database to store the extinguishing parameters corresponding to different fire levels;

[0082] Simulate and train anomaly detection models, generate random training data, and fit the model;

[0083] Construct multi-dimensional feature vectors for machine learning models, extract valid data and calculate cross-features;

[0084] Based on a hybrid analysis of supervised and unsupervised learning, the fire severity is determined using anomaly detection models and a rule engine. If the anomaly detection results conflict with the rule engine results, a review mechanism is triggered.

[0085] The fire extinguishing parameters are obtained from the database according to the fire level, and the posture compensation is calculated based on the current posture of the suction climbing robot to control the suction climbing robot to move.

[0086] In a specific case, the main code that performs the above functions is as follows:

[0087] "High":{"agent":"LN2","distance":0.5,"angle":30,"dose":300},"Medium":{"agent":"LN2","distance":0.7,"angle":20,"dose":

[0088] 200},

[0089] "Low":{"agent":"LN2","distance":1.0,"angle":10,"dose":100}

[0090] }

[0091] def_train_anomaly_detector(self):

[0092] """

[0093] Simulate training of anomaly detection models.

[0094] Generate random training data and fit the model.

[0095] """

[0096] training_data=np.random.rand(100,5)

[0097] self.anomaly_detector.fit(training_data)

[0098] def_extract_features(self,data):

[0099] """

[0100] Construct multidimensional feature vectors for use in machine learning models.

[0101] Extract data such as temperature, gas, smoke, etc. and calculate cross-features.

[0102] :param data: data dictionary collected by the sensor

[0103] :return: Characteristic vector array

[0104] """

[0105] return np.array([

[0106] data.get('temperature',0),

[0107] data.get('gas',0),

[0108] data.get('smoke',0),

[0109] data.get('temperature',0)*data.get('gas',0),#cross features

[0110] data.get('smoke',0) / (data.get('temperature',0)+1e-5)#Prevent division by zero

[0111] ]).reshape(1,-1)

[0112] def predict_fire_level(self,data):

[0113] """

[0114] Hybrid analysis based on supervised learning + unsupervised learning.

[0115] Use anomaly detection models and rule engines to determine the severity of fire.

[0116] :param data: data dictionary collected by the sensor

[0117] :return: Fire level string

[0118] """

[0119] #Unsupervised anomaly detection

[0120] features=self._extract_features(data)

[0121] is_anomaly=self.anomaly_detector.predict(features)

[0122] #Rule engine judgment

[0123] if data.get('temperature',0)>90and data.get('gas',0)>600:

[0124] level="Critical"

[0125] elif data.get('temperature',0)>80and data.get('gas',0)>500:

[0126] level="High"

[0127] else:

[0128] level="Low"

[0129] #If the anomaly detection and rule engine results conflict, the review mechanism is triggered

[0130]

[0131]

[0132] Combining unsupervised anomaly detection (Isolation Forest) with a rule engine; adjusting the injection angle according to the robot's real-time posture to solve the positioning error problem of traditional fixed injection, and introducing nonlinear features such as temperature-gas production product and flue gas-temperature ratio to improve model sensitivity.

[0133] Energy density assessment functions include:

[0134] Initialize the energy density evaluation model; set the battery parameters;

[0135] Based on the improved Bernardi heat production model, ohmic heat, polarization heat, and entropy heat are calculated. The steady-state equations of the original model are replaced by non-steady-state partial differential equations. The finite element method is combined to realize the dynamic distribution of heat production in the three-dimensional space of the human body and the total heat production power is obtained by summing them.

[0136] The battery internal resistance is calculated based on the temperature through the battery internal resistance temperature compensation model;

[0137] Based on the liquid nitrogen injection parameter decision table; determine the direction, distance, dosage, speed, duration and pattern of liquid nitrogen injection according to the heat generation power to form an injection strategy;

[0138] In a specific case, the main code that performs the above functions is as follows:

[0139]

[0140] Set the battery parameters such as nominal voltage, rated capacity and temperature rise coefficient.

[0141] """

[0142] self.battery_params={

[0143] "voltage_nom":3.2,#Nominal voltage (V)

[0144] "capacity":100,#rated capacity (Ah)

[0145] "thermal_coeff":0.05#Temperature rise coefficient (W / ℃)

[0146] }

[0147] def calculate_energy_release(self,realtime_data):

[0148] """

[0149] Calculate real-time heating power based on the improved Bernardi heating model.

[0150] Calculate the ohmic heat, polarization heat, and entropy heat and sum them to get the total heat generation power.

[0151] :param realtime_data: real-time data dictionary containing voltage, current and temperature

[0152] :return:Total heat generation power (W)

[0153] """

[0154] #Input: voltage (V), current (A), temperature (℃)

[0155] I=realtime_data.get('current',0)

[0156] V=realtime_data.get('voltage',0)

[0157] T=realtime_data.get('temperature',0)

[0158] #Ohmic heat + polarization heat + entropy heat (simplified model)

[0159] Q_ohm=I**2*self._get_internal_resistance(T)

[0160] Q_polar=0.1*V*I#empirical coefficient

[0161] Q_entropy=I*T*self.battery_params['thermal_coeff']

[0162] total_power=Q_ohm+Q_polar+Q_entropy#Total heat generation power (W)

[0163] return total_power

[0164] def_get_internal_resistance(self,temp):

[0165] """

[0166] Battery internal resistance temperature compensation model.

[0167] Calculate the battery internal resistance based on temperature.

[0168] :param temp: battery temperature (℃)

[0169] :return:Battery internal resistance

[0170] """

[0171] return 0.001*(1+0.005*(25-temp))#25℃ reference internal resistance

[0172] def determine_nitrogen_params(self,power):

[0173] """

[0174] Liquid nitrogen injection parameter decision table.

[0175] The direction, distance, dosage, speed, duration and pattern of liquid nitrogen spraying are determined according to the heat generation power.

[0176] :param power:Heating power (W)

[0177] :return: Liquid nitrogen injection parameter dictionary

[0178] """

[0179] if power>1000:

[0180] return{"dose":800,"duration":5,"mode":"pulse_high"}

[0181] elif power>500:

[0182] return{"dose":500,"duration":3,"mode":"pulse_medium"}

[0183] else:

[0184] return{"dose":300,"duration":2,"mode":"continuous"}

[0185] def evaluate_effect(self,pre_data,post_data):

[0186] """

[0187] Fire extinguishing effect evaluation (comparison of parameter change rates before and after injection).

[0188] Compare the changes in temperature and gas concentration before and after spraying to determine the fire extinguishing effect.

[0189] :param pre_data: data dictionary before injection

[0190] :param post_data: data dictionary after injection

[0191] :return: Fire extinguishing effect evaluation result string

[0192] """

[0193] if'temperature'not in pre_data or'temperature'not in post_data or\

[0194] 'gas'not in pre_data or'gas'not in post_data:

[0195] print("Warning:Incomplete data for evaluation.")

[0196] return "Incomplete_data"

[0197] delta_T=pre_data['temperature']-post_data['temperature']

[0198] delta_gas=pre_data['gas']-post_data['gas']

[0199] if delta_T>10and delta_gas>200:

[0200] return "Effective"

[0201] else:

[0202] return "Need_secondary"

[0203] By integrating the Bernardi heat generation equation, the thermal runaway dynamics process is quantified; modeling based on the battery's electrochemical characteristics avoids the uncertainty of black-box models; breaking through the limitations of traditional single-dimensional detection, and establishing a dynamic optimization model for fire extinguishing effectiveness to achieve real-time matching of fire extinguishing parameters and fire characteristics; and after the first liquid nitrogen injection, the control system automatically evaluates the fire extinguishing effect. If there is a risk of re-ignition, a second injection is immediately implemented to effectively prevent the fire from reigniting and improve fire extinguishing reliability.

[0204] Based on the functions of the above modules, in a specific case, the following program is used to simulate hardware interface data acquisition:

[0205] def get_sensor_data():

[0206] """

[0207] Simulates acquiring data from the Modbus sensor protocol.

[0208] In practice, it is necessary to implement specific data reading logic based on the hardware interface.

[0209] :return: Dictionary containing sensor data and timestamp

[0210] """

[0211] current_time = time.time()

[0212] data={

[0213] 'temperature':np.random.randint(0,100),

[0214] 'gas':np.random.randint(0,1000),

[0215] 'smoke':np.random.randint(0,500),

[0216] 'current':np.random.randint(0,10),

[0217] 'voltage':np.random.randint(2,4),

[0218] 'timestamp':current_time

[0219] }

[0220] return data

[0221] Based on the functions of the above modules, in a specific case, the following program is used to simulate ROS robot control:

[0222] def control_robot(action):

[0223] """

[0224] Simulate robot motion control based on ROS robot control interface.

[0225] In practice, it is necessary to implement specific control logic according to the ROS robot control interface.

[0226] :param action: the action string to be performed by the robot

[0227] """

[0228] print(f"Robot is performing action:{action}")

[0229] Based on the functions of the above modules, in a specific case, the following program is used to simulate the main program:

[0230] def main():

[0231] """

[0232] The main program simulates the operation process of the entire system.

[0233] It includes data acquisition, verification, fire situation analysis, energy density assessment and robot control, and secondary operations based on the fire extinguishing effect.

[0234] """

[0235] triple_validator=TripleValidation()

[0236] fire_judger = FireJudgment()

[0237] energy_model=EnergyDensityModel()

[0238] while True:

[0239] sensor_data=get_sensor_data()

[0240] if triple_validator.validate(sensor_data):

[0241] fire_level=fire_judger.predict_fire_level(sensor_data)

[0242] fire_params=fire_judger.match_fire_database(fire_level)

[0243] if fire_params:

[0244] energy_release=energy_model.calculate_energy_release(sensor_dat)

[0245]

[0246] The above code framework needs to be embedded adapted according to specific hardware interfaces (such as ROS robot control and Modbus sensor protocol), and virtual debugging can be performed through digital twin technology to reduce deployment risks.

[0247] refer to Figure 6 As shown, the workflow based on the above device is:

[0248] Under the control system 5's command, the suction-type climbing robot, using its electromagnetic suction cups 3, crawls along the battery cabinet wall at a set path and speed, reaching the designated monitoring area. The multi-parameter sensor 11 measures the temperature, gas production, and smoke parameters of the energy storage batteries within the detection range in real time. After triple-check logic processing, these parameters are fed back to the control system 5, matching the real-time thermal runaway fire situation to the extinguishing agent-distance-angle three-dimensional matching database. Based on the fire analysis results, the suction-type climbing robot quickly moves to the vicinity of the faulty battery within the detection range. The control system 5 issues action commands using an energy density assessment model, determining the direction, distance, speed, and dosage of liquid nitrogen injection based on the real-time fire situation, effectively and accurately preventing thermal runaway in the designated area. The multi-parameter sensor 11 provides feedback on the extinguishing effect of the initial liquid nitrogen injection, and the control system 5 uses the energy density assessment model to independently determine the secondary injection.

[0249] The above describes the embodiments of the present invention in detail with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. After knowing the contents described in the present invention, ordinary technicians in this technical field can make several equivalent changes and substitutions without departing from the principles of the present invention. These equivalent changes and substitutions should also be regarded as falling within the scope of protection of the present invention.

Claims

1. An intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency, characterized in that: include: An adsorption-type climbing robot has multiple climbing joints, each of which is equipped with an electromagnetic suction cup at its end; A multi-parameter sensor is installed at the traveling end of the adsorption-type climbing robot, and a multi-sensor confidence weighted decision-making mechanism is established for the multi-parameter sensor to screen and feed back valid data; The liquid nitrogen spraying mechanism comprises a liquid nitrogen tank assembled on the body of the adsorption climbing robot and a spraying hose with a hose nozzle connected to the liquid nitrogen tank, and The adsorption climbing robot is further provided with an adjustment mechanism, the spray hose is connected to the adjustment mechanism, and the adjustment mechanism is configured to adjust the direction of the hose nozzle in three angular directions of XYZ; The control system shall have at least the following functions: Based on the initial valid data, the built-in fire situation assessment module predicts the fire risk level. Based on the predicted level, the adsorption climbing robot is controlled to quickly move from the initial inspection position to the fault area. Control multi-parameter sensors to implement secondary precise detection and feedback secondary valid data; Based on the secondary effective data, the battery energy release rate is calculated in real time through the built-in energy density evaluation model, the injection strategy is formed, and the regulating mechanism is controlled to perform corresponding actions.

2. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 1 is characterized in that: Multiple climbing joints are installed at both ends of the fuselage in two groups, and are divided into front climbing joints and rear climbing joints, and the front climbing joints are connected to the fuselage through mirror pitch joints; as well as Each of the climbing joints includes a servo joint, an upper arm and a lower arm. The servo joint is installed on the fuselage. One end of the upper arm is connected to the servo joint through the upper arm joint, and the other end is connected to the lower arm through the lower arm joint. An electromagnetic suction cup is installed on the end of the lower arm away from the upper arm.

3. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 2, characterized in that: The suction force of the electromagnetic chuck can be dynamically adjusted according to the surface material and contact pressure of the cabin. The adjustment formula is: Where B is the magnetic induction intensity of the working air gap; A is the effective adsorption area; μ0=4π×10 -7 is the vacuum permeability; μ r is the relative magnetic permeability of the adsorbed material, that is, the correction coefficient of the surface material; k is the magnetic leakage coefficient.

4. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 1 is characterized in that: The multi-sensor confidence weighted decision-making mechanism includes: First level verification: check whether the data is within the preset range; Second level verification: Check whether the logical relationship between temperature and gas production data is established, and determine the logical consistency of multiple parameters; The third level of verification: compare the difference between the current data and the mean of historical data, and eliminate data with a difference greater than the preset range; Perform triple verification in sequence, and feed back the data that passes the triple verification as valid data, and update the historical data with the valid data.

5. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 1 is characterized in that: A mist arrester is arranged between the spray hose and the liquid nitrogen tank, and the mist arrester is configured to reduce the impact impulse of high-speed liquid nitrogen.

6. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 1, characterized in that: The adjustment mechanism includes a horizontal servo installed on the fuselage, a first single-axis servo, a second single-axis servo and a third single-axis servo installed on a segmented bracket along the vertical direction, and a mounting frame, the jet hose is assembled on the mounting frame, the horizontal servo is configured to control the adjustment of the mounting frame in the XY axis direction, and the first single-axis servo, the second single-axis servo and the third single-axis servo jointly cooperate to control the adjustment of the mounting frame in the Z axis direction.

7. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 1, characterized in that: The fire situation analysis module performs the following functions: Load the pre-trained anomaly detection model and fire extinguishing database, and train the anomaly detection model; Load the extinguishing agent-distance-angle three-dimensional database to store the extinguishing parameters corresponding to different fire levels; Simulate and train anomaly detection models, generate random training data, and fit the model; Construct multi-dimensional feature vectors for machine learning models, extract valid data and calculate cross-features; Based on a hybrid analysis of supervised and unsupervised learning, the fire severity is determined using anomaly detection models and a rule engine. If the anomaly detection results conflict with the rule engine results, a review mechanism is triggered. The fire extinguishing parameters are obtained from the database according to the fire level, and the posture compensation is calculated based on the current posture of the suction climbing robot to control the suction climbing robot to move.

8. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 7, characterized in that: The energy density assessment model performs the following functions: Initialize the energy density evaluation model; set the battery parameters; Based on the improved Bernardi heat generation model, ohmic heat, polarization heat and entropy heat are calculated. The steady-state equations of the original model are replaced by non-steady-state partial differential equations. The finite element method is combined to realize the dynamic distribution of heat generation in the three-dimensional space of the cabin and the total heat generation power is obtained by summing them. The battery internal resistance is calculated based on the temperature through the battery internal resistance temperature compensation model; Based on the liquid nitrogen injection parameter decision table, the direction, distance, dosage, speed, duration and pattern of the liquid nitrogen injection are determined according to the heat generation power to form an injection strategy.

9. The intelligent patrol-type liquid nitrogen fire extinguishing device based on dynamic optimization of fire extinguishing efficiency according to claim 8, characterized in that: The energy density evaluation further includes: comparing the change values ​​of temperature and gas concentration before and after the injection, and when the change value is less than a preset value, performing a secondary liquid nitrogen injection action.

Images