Energy storage power station fire safety monitoring system based on big data Internet of Things

Abstract

The invention discloses an energy storage power station fire safety monitoring system based on a big data Internet of Things, and relates to the technical field of fire safety monitoring. A multi-dimensional sensor network is adopted to monitor parameters in real time, a multi-dimensional parameter model under normal working conditions is established based on a multivariable Gaussian distribution model, and the real-time monitoring of the parameters is realized. In the early stage of thermal runaway, abnormity is recognized through the Mahalanobis distance, an early warning signal can be triggered immediately when the abnormity occurs, the defects that in a traditional method, the monitoring means is single, and response lags behind are overcome, fire early warning is more accurate and timely, and the concentration of harmful gas is monitored in real time in combination with a gas detection module; sensor data and gas concentration data are fused through Bayesian inference, the fire risk is evaluated more comprehensively, the comprehensiveness of fire risk evaluation is remarkably improved, a fire area is positioned through a fire source positioning algorithm, and the fire extinguishing range is adjusted in real time according to the fire risk level and the dynamic change of fire source diffusion.

Description

Technical Field

[0001] The present invention relates to the technical field of fire safety monitoring, and in particular to a fire safety monitoring system for an energy storage power station based on big data Internet of Things. Background Art

[0002] In the fire safety monitoring of energy storage power stations, current monitoring solutions mostly rely on a single sensor type, such as temperature or smoke sensors to monitor fire hazards, and then manage the battery status through the battery management system (BMS). However, with the expansion of the scale of energy storage power stations and the diversification of battery types, this solution exposes some problems when dealing with complex fire hazards.

[0003] First, single temperature monitoring is difficult to capture the thermal runaway process inside the energy storage battery, especially systems with thermal runaway risks such as lithium batteries and sodium-sulfur batteries. Temperature sensors can usually only monitor the battery surface or ambient temperature and cannot capture abnormal temperature rise inside the battery in time. In addition, traditional systems lack gas monitoring and cannot detect the unique harmful gases released by the battery. Finally, the BMS and fire protection system operate independently, and the fire protection system cannot obtain key battery status information from the BMS, resulting in a delayed fire warning response.

[0004] The traditional solution to this problem is to install temperature sensors or overheating protection devices. Once the temperature exceeds the threshold, an alarm will be triggered or the system will automatically shut down. However, such solutions are limited to single parameter monitoring and cannot conduct comprehensive analysis of multiple parameters such as internal battery pressure and current fluctuations. At the same time, some systems rely on smoke sensors to detect smoke or open flames generated by fires, but lack gas monitoring. Even if gas sensing equipment is installed, most of them can only detect conventional gases such as CO or CO2, and cannot identify decomposition products unique to energy storage batteries. Although the BMS can activate the protection mechanism when the battery is overloaded, overheated or the voltage is abnormal, these measures are not coordinated with the fire protection system and cannot form a complete early warning network. Therefore, there is an urgent need for a fire safety monitoring system for energy storage power stations based on the big data Internet of Things to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a fire safety monitoring system for energy storage power stations based on big data Internet of Things to solve the problems of current fire safety monitoring, such as single temperature monitoring is difficult to cope with complex thermal runaway processes, lack of multi-parameter monitoring leads to early warning lag, and cannot dynamically adapt to different battery types, gas monitoring equipment lacks pertinence and cannot detect specific harmful gases, missing opportunities for early fire prevention, and slow response speed, BMS and fire protection systems fail to work together, and information cannot be communicated.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] On the one hand, the present invention provides a fire safety monitoring system for energy storage power stations based on big data Internet of Things, which includes:

[0009] The data acquisition module includes a sensor network and a gas detection module. The data acquisition module is responsible for real-time collection of the operating parameters of the energy storage power station.

[0010] The sensor network consists of temperature, pressure, current and gas sensors covering the battery pack and its surrounding environment. The operating parameters monitored by the data acquisition module include battery temperature, pressure, current, charge and discharge rate, and potential harmful gas concentration.

[0011] The anomaly detection module is used to analyze and process the operating parameters obtained from the data acquisition module, establish a multidimensional model under normal working conditions of the battery, and perform real-time anomaly detection.

[0012] The sensor data fusion module integrates and analyzes the voltage, current and temperature operating data collected by the battery management system BMS with the multi-parameter monitoring data of the sensor network.

[0013] In the sensor data fusion module, the internal data of the BMS and the external sensor data are processed based on the data fusion algorithm to generate a comprehensive analysis report to evaluate the fire risk level of the battery.

[0014] The control response module intelligently triggers the automatic fire extinguishing system based on the risk assessment of the sensor data fusion module, and performs zoned fire extinguishing operations according to the location and severity of the fire. It combines the fire source location algorithm and fire level assessment to dynamically adjust the fire extinguishing range.

[0015] Furthermore, in the anomaly detection module, machine learning and statistical analysis algorithms are used to compare the collected multiple parameters, identify abnormal patterns, and detect potential thermal runaway, short circuits, and other failure risks. When the data does not match the model or exceeds the set safety threshold, an early warning is triggered.

[0016] On the other hand, the present invention provides a fire safety monitoring method for an energy storage power station based on big data Internet of Things, comprising:

[0017] Step S1, multi-parameter monitoring,

[0018] The sensor network is used to monitor the battery parameters in the energy storage power station, including temperature, pressure, current, and charge and discharge rate. A multidimensional model of normal working conditions is established based on the battery parameters. Once abnormal data is detected, an early warning signal is triggered immediately.

[0019] Step S2, gas monitoring enhancement,

[0020] When a potential thermal runaway or abnormal state is detected in step S1, the gas detection module is automatically activated to detect whether there is harmful gas, and the gas detection result is combined with the parameter data of step S1.

[0021] Step S3, BMS and sensor network data fusion,

[0022] The battery management system (BMS) of the energy storage power station monitors the core parameters of the battery in real time, including voltage, current and temperature, and integrates and analyzes the sensor network data in steps S1 and S2, and gives a fire risk level based on the evaluation results.

[0023] Step S4, intelligent automatic fire extinguishing response,

[0024] When the data fusion analysis in step S3 determines that there is a fire risk in the battery, the fire extinguishing response is automatically triggered. According to the fire risk level, the fire area is located in combination with the fire source location algorithm to dynamically adjust the fire extinguishing range.

[0025] Step S5, multi-level emergency response,

[0026] After the fire extinguishing system is activated, the control response module adjusts the emergency measures according to the speed of fire spread and temperature changes: automatically triggering local power outages, total power outages or emergency evacuation mechanisms, and the remote control system stops the charging and discharging of related battery units.

[0027] Furthermore, in step S1, the multi-parameter monitoring method is:

[0028] The multidimensional vector x(t) of battery parameters is collected from the sensor network, where t is time. The battery detection parameters include temperature T(t), pressure P(t), current I(t), and charge and discharge rate C(t). The parameter set is represented as a four-dimensional vector: Each parameter changes with time and is recorded in real time by the sensor network. Under normal working conditions, all parameters x(t) are within the safe range. The multivariate Gaussian distribution model is used to describe the joint distribution of parameters. Under normal working conditions, the multidimensional parameter x(t) of the battery obeys a Gaussian distribution with a mean of μ and a covariance matrix of Σ, which can be expressed as:

[0029] Where k is the number of dimensions, here k = 4, μ is the mean vector of each parameter Σ is the covariance matrix, which indicates the correlation between the parameters. |Σ| is the determinant of the covariance matrix. -1 is the inverse matrix of the covariance matrix. The mean and covariance matrix of each parameter are obtained based on the historical data training period. T Indicates the mean temperature μ under normal working conditions P Indicates the mean pressure under normal working conditions, μ IIndicates the average current under normal working conditions, μ X represents the mean value of the charge and discharge rate under normal working conditions, Σ represents the covariance matrix between the parameters,

[0030] Calculate the Mahalanobis distance of the real-time acquired parameter x(t) relative to the normal model:

[0031] Among them, when D M (x) exceeds the set threshold δ, indicating that the current state is abnormal and triggers an early warning signal, D M (x) represents the Mahalanobis distance between the current parameter state and the normal state. An adaptive adjustment algorithm is used to regularly update μ and Σ. Here, the recursive least squares (RLS) method and Kalman filter method are used.

[0032] Furthermore, in step S2, when thermal runaway or abnormal state is detected, the gas detection module is automatically activated, and the gas detection is combined with the data of step S1:

[0033] When the Mahalanobis distance D M (x) When the value exceeds the set threshold δ, the battery is considered to have a potential abnormality and the gas detection module is triggered. The triggering conditions are:

[0034] D M (x)>δ, where D M (x) is the current parameter x(t) = (T(t)P(t)I(t)C(t)) T The Mahalanobis distance from the normal operating condition, δ is the detection threshold of thermal runaway or abnormal state,

[0035] After the trigger conditions are met, the gas detection module automatically activates to detect the concentration of harmful gases, including carbon monoxide and hydrogen. The concentration of harmful gases obtained by the gas detection module is The concentration of harmful gases is modeled as a linear response function:

[0036] g i (t) = α i f(T(t),P(t),I(t))+∈ i (t), where g i (t) represents the concentration of the i-th harmful gas, α i is the response coefficient, f(T(t), P(t), I(t)) is a comprehensive function based on temperature T(t), pressure P(t), and current I(t), which represents the impact of parameter abnormality on gas generation, ∈ i (t) is the detection error, which follows a normal distribution

[0037] Furthermore, in step S2, after activating the gas detection module, the gas concentration data g(t) is fused with the multi-parameter monitoring data x(t) in step S1:

[0038] Assume that p(F|x(t)) is the probability of failure based on battery parameter x(t), and p(F|g(t)) is the probability of failure based on gas concentration g(t). According to Bayes' theorem, the final failure probability is expressed as:

[0039] Among them, p(F|x(t)) is the fault probability based on the abnormality detection in step S1, p(g(t)|F) is the gas concentration distribution given by the gas detection module when the known fault F occurs, and p(f(t)) is the marginal distribution of gas concentration. As a normalization constant,

[0040] According to the fused failure probability p(F|x(t),g(t)), the fire risk level is dynamically adjusted. When p(F|x(t),g(t)) exceeds the risk threshold θ, further emergency response is triggered.

[0041] Furthermore, in step S3, the parameters monitored by the BMS include: voltage V(t), current I(t) and temperature T(t).

[0042] The parameters monitored by the sensor network include: Among them, T ext (t) represents the temperature detected by the external temperature sensor, P(t) represents the pressure outside the battery, and I ext (t) represents the current value of the external current sensor, C(t) represents the charge and discharge rate of the battery,

[0043] In step S3, the BMS parameters are combined with the multidimensional data of the sensor network using weighted linear fusion. The fused parameters are:

[0044] z(t)=w V V(t)+w T T ext (t)+w I I(t)+w P P(t)+w C C(t),

[0045] Among them, w V ,w T ,w I ,w P ,w C is the weight of the corresponding parameter,

[0046] The fused parameter z(t) is compared with the multidimensional model of normal working status to evaluate the fire risk of the current battery. The risk probability model is used to quantify the risk by combining Bayesian inference and Markov decision process. R(t) is defined as the fire risk level. The risk level is dynamically adjusted according to the abnormality of the sensor fusion data.

[0047] The probability of fire risk is estimated as:

[0048] where p(F|z(t)) is the probability of fire occurrence based on the fused data, γ is the sensitivity parameter, and μ z is the mean value of the fused data under normal conditions, representing normal working conditions.

[0049] According to the probability of fire occurrence p(F|z(t)), the fire risk is divided into different levels:

[0050] p(F|z(t))<θ1, indicating low risk,

[0051] θ1≤p(F|z(t))<θ2, indicating medium risk,

[0052] p(F|z(t))≥θ2, indicating high risk,

[0053] Among them, θ1 and θ2 are risk level thresholds set based on historical data and on-site needs.

[0054] When assessing fire risk, the fused fire risk level R(t) is dynamically adjusted according to changes in real-time data. When the system detects an increase in fire risk, that is, p(F|z(t)) exceeds a specific threshold, further response measures are automatically triggered.

[0055] Furthermore, in step S4, the intelligent automatic fire extinguishing response mode is:

[0056] Based on the fire risk probability p(F|z(t)) obtained in step S3, when the fire risk reaches the set high risk threshold θ2, the fire extinguishing response system is automatically triggered.

[0057] Once the fire risk triggers a fire extinguishing response, the specific location of the fire is determined based on the fire source location algorithm. Let the fire source location r f As the coordinates of the fire point, the weighted distance sum is used to combine the positions of each sensor point r i And the corresponding sensor data to infer the fire source location:

[0058] Among them, r f is the estimated location of the fire source, r i is the position of the ith sensor, T i is the temperature data detected by the i-th sensor, wi is the weight of the sensor,

[0059] Uncertainty modeling of fire location using measurement noise:

[0060] in, is the variance of fire source location, r i is the position of the ith sensor, r f is the estimated location of the fire source,

[0061] According to the results of fire source location and uncertainty analysis, the fire extinguishing range is dynamically adjusted. f The fire source position r f The dynamic radius r is the center f , the radius is based on the fire source location uncertainty σ r And the fire risk probability p(F|z(t)) is dynamically adjusted:

[0062] r f =α·σ r +β·p(F|z(t)), where r f is the dynamically adjusted fire extinguishing radius, α is the coefficient for adjusting the radius based on the uncertainty of fire source location, β is the coefficient for adjusting the radius based on fire risk, and σ r is the uncertainty of fire source location, p(F|z(t)) is the risk probability of fire,

[0063] Fire extinguishing range A f Finally, f The circular area with a radius of r is the center f ,

[0064] According to the fire risk level and fire extinguishing range, the corresponding fire extinguishing means are selected, and the fire extinguishing level is dynamically defined according to the fire risk probability p(F|z(t)):

[0065] If p(F|z(t))≤θ1, L f =1,

[0066] If θ1 <p(F|z(t))≤θ2,L f =2,

[0067] If p(F|z(t))>θ2, L f =3,

[0068] Among them, L f is the fire extinguishing level, corresponding to light fire level 1, moderate fire level 2 and severe fire level 3, θ1 and θ2 are the set fire risk thresholds,

[0069] Each fire extinguishing level corresponds to a different fire extinguishing response strategy. Level 1 fires are extinguished by gas, level 2 fires are extinguished by water sprinkler systems, and level 3 fires activate all automatic fire extinguishing systems.

[0070] Furthermore, in step S5, the control response module adjusts the emergency measures according to the fire spread speed and temperature change:

[0071] Calculate the fire source diffusion speed based on the dynamic changes of the fire source position:

[0072] where v f (t) is the fire spreading speed, r f (t1) and r f (t2) are the positions of the fire source at time t1 and t2, t2-t1 is the time difference, |·| represents the Euclidean distance of the vector, and the fire source position r f (t) obtained by the fire source location algorithm,

[0073] Calculate the rate of temperature change:

[0074] in, is the temperature change rate, T(t) is the temperature at the current time t, T(t0) is the temperature at the initial time t0, t-t0 is the time interval,

[0075] Based on the fire spread velocity v f (t) and temperature change rate To determine whether a partial power outage or a full power outage is triggered, two key thresholds are set:

[0076] Fire source spread speed threshold v local and v global , and the temperature change rate threshold and

[0077] Local power outage triggering conditions: v f (t)≥v local or

[0078] Total power outage triggering conditions: v f (t)≥v global or

[0079] where v local is the local power outage threshold for the fire spread rate, v global is the total power outage threshold for the speed of fire spread, is the local blackout threshold for the temperature change rate, is the total blackout threshold for the temperature change rate,

[0080] When the fire spreads and the temperature changes abnormally, the system triggers the emergency evacuation mechanism. The evacuation condition is set as the temperature change rate or the fire spread speed exceeds the high risk threshold. and v evac :

[0081] v f (t)≥v evac or

[0082] Among them, v evac is the evacuation threshold of the fire spread speed, It is the evacuation threshold of the temperature change rate. When the evacuation condition is met, the evacuation alarm is automatically activated and the evacuation procedure is executed.

[0083] Furthermore, in step S5, in order to avoid further aggravation of the accident when the fire spreads, the charging and discharging operations are remotely controlled, and the triggering of the remote control is combined with the power outage triggering condition. When the power outage or evacuation mechanism is activated, the charging and discharging operations are automatically shut down. When the remote control system is triggered by a local power outage, a total power outage or an evacuation, the charging and discharging power is set to 0.

[0084] The beneficial effects of the present invention are:

[0085] The present invention adopts a multidimensional sensor network to monitor parameters in real time, and adopts a multivariate Gaussian distribution model to establish a multidimensional parameter model under normal working conditions. In the early stage of thermal runaway, the Mahalanobis distance is used to identify anomalies, and an early warning signal can be triggered immediately when an anomaly occurs, thereby making up for the shortcomings of the traditional method of single monitoring means and delayed response, and making fire early warning more accurate and timely.

[0086] The present invention combines a gas detection module to monitor the concentration of harmful gases in real time, and fuses sensor data with gas concentration data through Bayesian inference to more comprehensively evaluate fire risks, thereby significantly improving the comprehensiveness of fire risk assessment.

[0087] The present invention locates the fire area through a fire source location algorithm, and adjusts the fire extinguishing range in real time according to the dynamic changes of the fire risk level and the fire source spread, effectively controls the fire, avoids unnecessary excessive fire extinguishing, and improves the efficiency of fire extinguishing and the flexibility of response.

[0088] The present invention adaptively adjusts the model of the normal working state of the battery through the recursive least squares method RLS and the Kalman filter method. As the battery ages or the working conditions change, the mean and covariance matrix of the monitoring model are dynamically updated, which greatly enhances the long-term stability of the system and the accuracy of detection.

[0089] The present invention calculates the fire source diffusion speed and temperature change rate, determines the severity of the fire spread, and automatically triggers a local power outage, a full power outage or an emergency evacuation mechanism according to the fire spread speed and temperature change, while automatically and remotely stopping the charging and discharging operations of related battery units to avoid further expansion of the accident. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0091] Figure 1 This is a schematic diagram of the structure of the fire safety monitoring system of the energy storage power station based on the big data Internet of Things of the present invention;

[0092] Figure 2 It is a schematic flow chart of the fire safety monitoring method of energy storage power station based on big data Internet of Things of the present invention. DETAILED DESCRIPTION

[0093] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0094] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0095] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0096] Example 1, reference Figure 1 This embodiment provides a fire safety monitoring system for an energy storage power station based on big data Internet of Things, including:

[0097] The data acquisition module includes a sensor network and a gas detection module. The data acquisition module is responsible for real-time collection of the operating parameters of the energy storage power station.

[0098] The sensor network consists of temperature, pressure, current and gas sensors covering the battery pack and its surrounding environment. The operating parameters monitored by the data acquisition module include battery temperature, pressure, current, charge and discharge rate, and potential harmful gas concentration.

[0099] The anomaly detection module is used to analyze and process the operating parameters obtained from the data acquisition module, establish a multidimensional model under normal working conditions of the battery, and perform real-time anomaly detection.

[0100] In the anomaly detection module, machine learning and statistical analysis algorithms are used to compare the collected multiple parameters, identify abnormal patterns, and detect potential thermal runaway, short circuits, and other failure risks. When the data does not match the model or exceeds the set safety threshold, an early warning is triggered.

[0101] The sensor data fusion module integrates and analyzes the voltage, current and temperature operating data collected by the battery management system BMS with the multi-parameter monitoring data of the sensor network.

[0102] In the sensor data fusion module, the internal data of the BMS and the external sensor data are processed based on the data fusion algorithm to generate a comprehensive analysis report to evaluate the fire risk level of the battery.

[0103] The control response module intelligently triggers the automatic fire extinguishing system based on the risk assessment of the sensor data fusion module, and performs zoned fire extinguishing operations according to the location and severity of the fire. It combines the fire source location algorithm and fire level assessment to dynamically adjust the fire extinguishing range.

[0104] Example 2, reference Figure 2 This embodiment provides a fire safety monitoring method for an energy storage power station based on big data Internet of Things, comprising the following steps:

[0105] Step S1, multi-parameter monitoring,

[0106] The sensor network is used to monitor the battery parameters in the energy storage power station, including temperature, pressure, current, and charge and discharge rate. A multidimensional model of normal working conditions is established based on the battery parameters. Once abnormal data is detected, an early warning signal is triggered immediately.

[0107] In step S1, the multi-parameter monitoring method is:

[0108] The multidimensional vector x(t) of battery parameters is collected from the sensor network, where t is time. The battery detection parameters include temperature T(t), pressure P(t), current I(t), and charge and discharge rate C(t). The parameter set is represented as a four-dimensional vector: Each parameter changes with time and is recorded in real time by the sensor network. Under normal working conditions, all parameters x(t) are within the safe range. The multivariate Gaussian distribution model is used to describe the joint distribution of parameters. Under normal working conditions, the multidimensional parameter x(t) of the battery obeys a Gaussian distribution with a mean of μ and a covariance matrix of Σ, which can be expressed as:

[0109] Where k is the number of dimensions, here k = 4, μ is the mean vector of each parameter Σ is the covariance matrix, which indicates the correlation between the parameters. |Σ| is the determinant of the covariance matrix. -1 is the inverse matrix of the covariance matrix. The mean and covariance matrix of each parameter are obtained based on the historical data training period. T Indicates the mean temperature μ under normal working conditions P Indicates the mean pressure under normal working conditions, μ I Indicates the average current under normal working conditions, μ c represents the mean value of the charge and discharge rate under normal working conditions, Σ represents the covariance matrix between the parameters,

[0110] Calculate the Mahalanobis distance of the real-time acquired parameter x(t) relative to the normal model:

[0111] Among them, when D M (x) exceeds the set threshold δ, indicating that the current state is abnormal and triggers an early warning signal, D M (x) represents the Mahalanobis distance between the current parameter state and the normal state. An adaptive adjustment algorithm is used to regularly update μ and Σ. Here, the recursive least squares method (RLS) and Kalman filter method are used.

[0112] Specifically, based on real-time monitoring of a multidimensional sensor network, the battery's temperature, pressure, current, and charge and discharge rate are collected simultaneously. A multidimensional model under normal working conditions is established through a multivariate Gaussian distribution model to dynamically capture system operation anomalies. The Mahalanobis distance is used to promptly detect the potential thermal runaway risk of the battery and immediately trigger a warning signal.

[0113] Step S2, gas monitoring enhancement,

[0114] When a potential thermal runaway or abnormal state is detected in step S1, the gas detection module is automatically activated to detect whether there is harmful gas, and the gas detection result is combined with the parameter data of step S1.

[0115] In step S2, when thermal runaway or abnormal conditions are detected, the gas detection module is automatically activated and the gas detection is combined with the data from step S1:

[0116] When the Mahalanobis distance D M (x) When the value exceeds the set threshold δ, the battery is considered to have a potential abnormality and the gas detection module is triggered. The triggering conditions are:

[0117] D M (x)>δ, where D M (x) is the current parameter x(t) = (T(t)P(t)I(t)C(t)) T The Mahalanobis distance from the normal operating condition, δ is the detection threshold of thermal runaway or abnormal state,

[0118] After the trigger conditions are met, the gas detection module automatically activates to detect the concentration of harmful gases, including carbon monoxide and hydrogen. The concentration of harmful gases obtained by the gas detection module is The concentration of harmful gases is modeled as a linear response function:

[0119] g i (t) = α i f(T(t),P(t),I(t))+∈ i (t), where g i (t) represents the concentration of the i-th harmful gas, α i is the response coefficient, f(T(t), P(t), I(t)) is a comprehensive function based on temperature T(t), pressure P(t), and current I(t), which represents the impact of parameter abnormality on gas generation, ∈ i (t) is the detection error, which follows a normal distribution

[0120] In step S2, after activating the gas detection module, the gas concentration data g(t) is fused with the multi-parameter monitoring data x(t) in step S1:

[0121] Assume that p(F|x(t)) is the probability of failure based on battery parameter x(t), and p(F|g(t)) is the probability of failure based on gas concentration g(t). According to Bayes' theorem, the final failure probability is expressed as:

[0122] Among them, p(F|x(t)) is the fault probability based on the abnormality detection in step S1, p(g(t)|F) is the gas concentration distribution given by the gas detection module when the known fault F occurs, and p(g(t)) is the marginal distribution of gas concentration. As a normalization constant,

[0123] According to the fused failure probability p(F|x(t),g(t)), the fire risk level is dynamically adjusted. When p(F|x(t),g(t)) exceeds the risk threshold θ, further emergency response is triggered.

[0124] Specifically, when a potential abnormality is detected, the gas detection module is automatically activated to further detect the concentration of harmful gases. By combining the multi-dimensional data of the sensor and the gas concentration data, the Bayesian inference algorithm is used for data fusion to confirm the fire risk and dynamically adjust the fire risk level based on real-time information, providing a multi-dimensional information source for fire risk assessment.

[0125] Step S3, BMS and sensor network data fusion,

[0126] The battery management system (BMS) of the energy storage power station monitors the core parameters of the battery in real time, including voltage, current and temperature, and integrates and analyzes the sensor network data in steps S1 and S2, and gives a fire risk level based on the evaluation results.

[0127] In step S3, the parameters monitored by the BMS include: voltage V(t), current I(t) and temperature T(t).

[0128] The parameters monitored by the sensor network include: Among them, T ext (t) represents the temperature detected by the external temperature sensor, P(t) represents the pressure outside the battery, and I ext (t) represents the current value of the external current sensor, C(t) represents the charge and discharge rate of the battery,

[0129] In step S3, the BMS parameters are combined with the multidimensional data of the sensor network using weighted linear fusion. The fused parameters are:

[0130] z(t)=w V V(t)+w T T ext (t)+w I I(t)+w P P(t)+w C C(t),

[0131] Among them, w V ,w T ,w I ,w P ,w C is the weight of the corresponding parameter,

[0132] The fused parameter z(t) is compared with the multidimensional model of normal working status to evaluate the fire risk of the current battery. The risk probability model is used to quantify the risk by combining Bayesian inference and Markov decision process. R(t) is defined as the fire risk level. The risk level is dynamically adjusted according to the abnormality of the sensor fusion data.

[0133] The probability of fire risk is estimated as:

[0134] where p(F|z(t)) is the probability of fire occurrence based on the fused data, γ is the sensitivity parameter, and μ z is the mean value of the fused data under normal conditions, representing normal working conditions.

[0135] According to the probability of fire occurrence p(F|z(t)), the fire risk is divided into different levels:

[0136] p(F|z(t))<θ1, indicating low risk,

[0137] θ1≤p(F|z(t))<θ2, indicating medium risk,

[0138] p(F|z(t))≥θ2, indicating high risk,

[0139] Among them, θ1 and θ2 are risk level thresholds set based on historical data and on-site needs.

[0140] When assessing fire risk, the fused fire risk level R(t) is dynamically adjusted according to the changes in real-time data. When the system detects an increase in fire risk, that is, p(F|z(t)) exceeds a certain threshold, further response measures are automatically triggered.

[0141] Specifically, the voltage, current, and temperature data of the battery management system BMS are weighted linearly fused with the sensor network data to quantify the fire risk and dynamically adjust the risk level. The Markov decision process and Bayesian inference model are combined to evaluate different fire risk levels and take corresponding measures based on the risk level.

[0142] Step S4, intelligent automatic fire extinguishing response,

[0143] When the data fusion analysis in step S3 determines that there is a fire risk in the battery, the fire extinguishing response is automatically triggered. According to the fire risk level, the fire area is located in combination with the fire source location algorithm to dynamically adjust the fire extinguishing range.

[0144] In step S4, the intelligent automatic fire extinguishing response mode is:

[0145] Based on the fire risk probability p(F|z(t)) obtained in step S3, when the fire risk reaches the set high risk threshold θ2, the fire extinguishing response system is automatically triggered.

[0146] Once the fire risk triggers a fire extinguishing response, the specific location of the fire is determined based on the fire source location algorithm. Let the fire source location r f As the coordinates of the fire point, the weighted distance sum is used to combine the positions of each sensor point r i And the corresponding sensor data to infer the fire source location:

[0147] Among them, rf is the estimated location of the fire source, r i is the position of the ith sensor, T i is the temperature data detected by the i-th sensor, w i is the weight of the sensor,

[0148] Uncertainty modeling of fire location using measurement noise:

[0149] in, is the variance of fire source location, r i is the position of the ith sensor, r f is the estimated location of the fire source,

[0150] According to the results of fire source location and uncertainty analysis, the fire extinguishing range is dynamically adjusted. f The fire source position r f The dynamic radius r is the center f , the radius is based on the fire source location uncertainty σ r And the fire risk probability p(F|z(t)) is dynamically adjusted:

[0151] r f =α·σ r +β·p(F|z(t)), where r f is the dynamically adjusted fire extinguishing radius, α is the coefficient for adjusting the radius based on the uncertainty of fire source location, β is the coefficient for adjusting the radius based on fire risk, and σ r is the uncertainty of fire source location, p(F|z(t)) is the risk probability of fire,

[0152] Fire extinguishing range A f Finally, f The circular area with a radius of r is the center f ,

[0153] According to the fire risk level and fire extinguishing range, the corresponding fire extinguishing means are selected, and the fire extinguishing level is dynamically defined according to the fire risk probability p(F|z(t)):

[0154] If p(F|z(t))≤θ1, L f =1,

[0155] If θ1 <p(F|z(t))≤θ2,L f =2,

[0156] If p(F|z(t))>θ2, L f =3,

[0157] Among them, L fis the fire extinguishing level, corresponding to light fire level 1, moderate fire level 2 and severe fire level 3, θ1 and θ2 are the set fire risk thresholds,

[0158] Each fire extinguishing level corresponds to a different fire extinguishing response strategy. Level 1 fires are extinguished by gas, level 2 fires are extinguished by water sprinkler systems, and level 3 fires activate all automatic fire extinguishing systems.

[0159] Specifically, after the fire risk assessment reaches the set threshold, the system automatically triggers an intelligent fire extinguishing response, locates the fire source through a fire source location algorithm, and dynamically adjusts the fire extinguishing scope and strategy in combination with uncertainty analysis to improve the accuracy of the fire extinguishing system and avoid excessive fire extinguishing or waste of resources.

[0160] Step S5, multi-level emergency response,

[0161] After the fire extinguishing system is activated, the control response module adjusts the emergency measures according to the speed of fire spread and temperature changes: automatically triggering a partial power outage, a full power outage or an emergency evacuation mechanism, and at the same time the remote control system stops the charging and discharging of the relevant battery units.

[0162] In step S5, the control response module adjusts the emergency measures according to the fire spread speed and temperature changes:

[0163] Calculate the fire source diffusion speed based on the dynamic changes of the fire source position:

[0164] where v f (t) is the fire spreading speed, r f (t1) and r f (t2) are the positions of the fire source at time t1 and t2, t2-t1 is the time difference, |·| represents the Euclidean distance of the vector, and the fire source position r f (t) obtained by the fire source location algorithm,

[0165] Calculate the rate of temperature change:

[0166] in, is the temperature change rate, T(t) is the temperature at the current time t, T(t0) is the temperature at the initial time t0, t-t0 is the time interval,

[0167] Based on the fire spread velocity v f (t) and temperature change rate To determine whether a partial power outage or a full power outage is triggered, two key thresholds are set:

[0168] Fire source spread speed threshold v local and v global , and the temperature change rate threshold and

[0169] Local power outage triggering conditions: v f (t)≥v local or

[0170] Total power outage triggering conditions: v f (t)≥v global or

[0171] where v local is the local power outage threshold for the fire spread rate, v global is the total power outage threshold for the speed of fire spread, is the local blackout threshold for the temperature change rate, is the total blackout threshold for the temperature change rate,

[0172] When the fire spreads and the temperature changes abnormally, the system triggers the emergency evacuation mechanism. The evacuation condition is set as the temperature change rate or the fire spread speed exceeds the high risk threshold. and v evac :

[0173] v f (t)≥v evac or

[0174] Among them, v evac is the evacuation threshold of the fire spread speed, is the evacuation threshold of the temperature change rate. When the evacuation condition is met, the evacuation alarm is automatically activated and the evacuation procedure is executed.

[0175] Specifically, the control response module dynamically adjusts emergency response measures based on the fire spread rate and temperature changes. By calculating the fire spread rate and temperature change rate, it intelligently determines whether it is necessary to trigger a local power outage, a full power outage or an emergency evacuation mechanism. When the fire spreads or the temperature changes abnormally, the power outage and evacuation are quickly initiated to reduce the further spread of the fire and the threat to personnel.

[0176] In step S5, in order to avoid further aggravation of the accident when the fire spreads, the charging and discharging operations are remotely controlled. The triggering of the remote control is combined with the power outage triggering condition. When the power outage or evacuation mechanism is activated, the charging and discharging operations are automatically shut down. When a local power outage, a total power outage or an evacuation is triggered, the remote control system sets the charging and discharging power to 0.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fire safety monitoring system for energy storage power stations based on big data Internet of Things, characterized by: include, The data acquisition module includes a sensor network and a gas detection module. The data acquisition module is responsible for real-time collection of the operating parameters of the energy storage power station. The sensor network consists of temperature, pressure, current and gas sensors covering the battery pack and its surrounding environment. The operating parameters monitored by the data acquisition module include battery temperature, pressure, current, charge and discharge rate, and potential harmful gas concentration. The anomaly detection module is used to analyze and process the operating parameters obtained from the data acquisition module, establish a multidimensional model under normal working conditions of the battery, and perform real-time anomaly detection. The sensor data fusion module integrates and analyzes the voltage, current and temperature operating data collected by the battery management system BMS with the multi-parameter monitoring data of the sensor network. In the sensor data fusion module, the internal data of the BMS and the external sensor data are processed based on the data fusion algorithm to generate a comprehensive analysis report to evaluate the fire risk level of the battery. The control response module intelligently triggers the automatic fire extinguishing system based on the risk assessment of the sensor data fusion module, and performs zoned fire extinguishing operations according to the location and severity of the fire. It combines the fire source location algorithm and fire level assessment to dynamically adjust the fire extinguishing range.

2. According to claim 1, a fire safety monitoring system for energy storage power station based on big data Internet of Things is characterized in that: In the anomaly detection module, machine learning and statistical analysis algorithms are used to compare the collected multiple parameters, identify abnormal patterns, and detect potential thermal runaway, short circuits, and other failure risks. When the data does not match the model or exceeds the set safety threshold, an early warning is triggered.

3. A fire safety monitoring method for energy storage power station based on big data Internet of Things, characterized in that: A fire safety monitoring system for an energy storage power station based on big data Internet of Things according to any one of claims 1 to 2 comprises the following steps: Step S1, multi-parameter monitoring, uses a sensor network to monitor the battery parameters in the energy storage power station, including temperature, pressure, current, and charge and discharge rate. A multi-dimensional model of normal working conditions is established based on the battery parameters. Once abnormal data is detected, an early warning signal is triggered immediately. Step S2, gas monitoring enhancement, when potential thermal runaway or abnormal state is detected in step S1, the gas detection module is automatically activated to detect whether there is harmful gas, and the gas detection result is combined with the parameter data of step S1. Step S3: BMS and sensor network data fusion. The battery management system BMS of the energy storage power station monitors the core parameters of the battery in real time, including voltage, current and temperature, and fuses and analyzes the sensor network data of steps S1 and S2, and gives a fire risk level based on the evaluation results. Step S4, intelligent automatic fire extinguishing response. When the data fusion analysis in step S3 determines that there is a fire risk in the battery, the fire extinguishing response is automatically triggered. According to the fire risk level, the fire area is located in combination with the fire source location algorithm to dynamically adjust the fire extinguishing range. Step S5, multi-level emergency response, after the fire extinguishing system is started, the control response module adjusts the emergency measures according to the fire spread speed and temperature changes: automatically triggering local power outage, full power outage or emergency evacuation mechanism, and the remote control system stops the charging and discharging of related battery units.

4. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 3, characterized in that: In step S1, the multi-parameter monitoring method is: The multidimensional vector x(t) of battery parameters is collected from the sensor network, where t is time. The battery detection parameters include temperature T(t), pressure P(t), current I(t), and charge and discharge rate C(t). The parameter set is represented as a four-dimensional vector: Each parameter changes with time and is recorded in real time by the sensor network. Under normal working conditions, all parameters x(t) are within the safe range. The multivariate Gaussian distribution model is used to describe the joint distribution of parameters. Under normal working conditions, the multidimensional parameter x(t) of the battery obeys a Gaussian distribution with a mean of μ and a covariance matrix of Σ, which can be expressed as: Where k is the number of dimensions, here k = 4, μ is the mean vector of each parameter ∑ is the covariance matrix, which indicates the correlation between the parameters. |∑| is the determinant of the covariance matrix. -1 is the inverse matrix of the covariance matrix. The mean and covariance matrix of each parameter are obtained based on the historical data training period. T Indicates the mean temperature μ under normal working conditions P Indicates the mean pressure under normal working conditions, μ I Indicates the average current under normal working conditions, μ C represents the mean value of the charge and discharge rate under normal working conditions, ∑ represents the covariance matrix between the parameters, Calculate the Mahalanobis distance of the real-time acquired parameter x(t) relative to the normal model: Among them, when D M (x) exceeds the set threshold δ, indicating that the current state is abnormal and triggers an early warning signal, D M (x) represents the Mahalanobis distance between the current parameter state and the normal state, and an adaptive adjustment algorithm is used to regularly update μ and ∑.

5. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 4, characterized in that: In step S2, when thermal runaway or abnormal conditions are detected, the gas detection module is automatically activated and the gas detection is combined with the data from step S1: When the Mahalanobis distance D M (x) When the value exceeds the set threshold δ, the battery is considered to have a potential abnormality and the gas detection module is triggered. The triggering conditions are: D M (x)>δ, where D M (x) is the current parameter x(t) = (T(t) P(t) I(t) C(t)) T The Mahalanobis distance from the normal operating condition, δ is the detection threshold of thermal runaway or abnormal state, After the trigger conditions are met, the gas detection module automatically activates to detect the concentration of harmful gases, including carbon monoxide and hydrogen. The concentration of harmful gases obtained by the gas detection module is The concentration of harmful gases is modeled as a linear response function: g i (t) = α i f(T(t), P(t), I(t))+∈ i (t), where g i (t) represents the concentration of the i-th harmful gas, α i is the response coefficient, f(T(t), P(t), I(t)) is a comprehensive function based on temperature T(t), pressure P(t), and current I(t), which represents the impact of parameter abnormality on gas generation, ∈ i (t) is the detection error, which follows a normal distribution 6. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 5, characterized in that: In step S2, after activating the gas detection module, the gas concentration data g(t) is fused with the multi-parameter monitoring data x(t) in step S1: Assume that p(F|x(t)) is the probability of failure based on battery parameter x(t), and p(F|g(t)) is the probability of failure based on gas concentration g(t). According to Bayes' theorem, the final failure probability is expressed as: Among them, p(F|x(t)) is the fault probability based on the abnormality detection in step S1, p(g(t)|F) is the gas concentration distribution given by the gas detection module when the known fault F occurs, and p(g(t)) is the marginal distribution of gas concentration. As a normalization constant, According to the fused failure probability p(F|x(t), g(t)), the fire risk level is dynamically adjusted. When p(F|x(t), g(t)) exceeds the risk threshold θ, further emergency response is triggered.

7. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 6, characterized in that: In step S3, the parameters monitored by the BMS include: voltage V(t), current I(t) and temperature T(t). The parameters monitored by the sensor network include: Among them, T ext (t) represents the temperature detected by the external temperature sensor, P(t) represents the pressure outside the battery, and I ext (t) represents the current value of the external current sensor, C(t) represents the charge and discharge rate of the battery, In step S3, the BMS parameters are combined with the multidimensional data of the sensor network using weighted linear fusion. The fused parameters are: z(t)=w V V(t)+w T T ext (t)+w I I(t)+w P P(t)+w C C(t), Among them, w V , w T , w I , w P , w C is the weight of the corresponding parameter, The fused parameter z(t) is compared with the multidimensional model of normal working status to evaluate the fire risk of the current battery. The risk probability model is used to quantify the risk by combining Bayesian inference and Markov decision process. R(t) is defined as the fire risk level. The risk level is dynamically adjusted according to the abnormality of the sensor fusion data. The probability of fire risk is estimated as: where p(F|z(t)) is the probability of fire occurrence based on the fused data, γ is the sensitivity parameter, and μ z is the mean value of the fused data under normal conditions, representing normal working conditions. According to the probability of fire occurrence p(F|z(t)), the fire risk is divided into different levels: p(F|z(t))<θ1, indicating low risk, θ1≤p(F|z(t))<θ2, indicating medium risk, p(F|z(t))≥θ2, indicating high risk, Among them, θ1 and θ2 are risk level thresholds set based on historical data and on-site needs. When assessing fire risk, the fused fire risk level R(t) is dynamically adjusted according to changes in real-time data. When the system detects an increase in fire risk, that is, p(F|z(t)) exceeds a specific threshold, further response measures are automatically triggered.

8. The method for fire safety monitoring of energy storage power station based on big data Internet of Things according to claim 7 is characterized in that: In step S4, the intelligent automatic fire extinguishing response mode is: Based on the fire risk probability p(F|z(t)) obtained in step S3, when the fire risk reaches the set high risk threshold θ2, the fire extinguishing response system is automatically triggered. Once the fire risk triggers a fire extinguishing response, the specific location of the fire is determined based on the fire source location algorithm. Let the fire source location r f As the coordinates of the fire point, the weighted distance sum is used to combine the positions of each sensor point r i And the corresponding sensor data to infer the fire source location: Among them, r f is the estimated location of the fire source, r i is the position of the ith sensor, T i is the temperature data detected by the i-th sensor, w i is the weight of the sensor, Uncertainty modeling of fire location using measurement noise: in, is the variance of fire source location, r i is the position of the ith sensor, r f is the estimated location of the fire source, According to the results of fire source location and uncertainty analysis, the fire extinguishing range is dynamically adjusted. f The fire source position r f The dynamic radius r is the center f , the radius is based on the fire source location uncertainty σ r And the fire risk probability p(F|z(t)) is dynamically adjusted: r f =α·σ r +β·p(F|z(t)), where r f is the dynamically adjusted fire extinguishing radius, α is the coefficient for adjusting the radius based on the uncertainty of fire source location, β is the coefficient for adjusting the radius based on fire risk, and σ r is the uncertainty of fire source location, p(F|z(t)) is the risk probability of fire, Fire extinguishing range A f Finally, f The circular area with a radius of r is the center f , According to the fire risk level and fire extinguishing range, the corresponding fire extinguishing means are selected, and the fire extinguishing level is dynamically defined according to the fire risk probability p(F|z(t)): If p(F|z(t))≤θ1, L f =1, If θ1<p(F|z(t))≤θ2, L f =2, If p(F|z(t))>θ2, L f =3, Among them, L f is the fire extinguishing level, corresponding to light fire level 1, moderate fire level 2 and severe fire level 3, θ1 and θ2 are the set fire risk thresholds, Each fire extinguishing level corresponds to a different fire extinguishing response strategy. Level 1 fires are extinguished by gas, level 2 fires are extinguished by water sprinkler systems, and level 3 fires activate all automatic fire extinguishing systems.

9. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 8, characterized in that: In step S5, the control response module adjusts the emergency measures according to the fire spread speed and temperature changes: Calculate the fire source diffusion speed based on the dynamic changes of the fire source position: where v f (t) is the fire spreading speed, r f (t1) and r f (t2) are the positions of the fire source at time t1 and t2, t2-t1 is the time difference, |·| represents the Euclidean distance of the vector, and the fire source position r f (t) obtained by the fire source location algorithm, Calculate the rate of temperature change: in, is the temperature change rate, T(t) is the temperature at the current time t, T(t0) is the temperature at the initial time t0, t-t0 is the time interval, Based on the fire spread velocity v f (t) and temperature change rate To determine whether a partial power outage or a full power outage is triggered, two key thresholds are set: Fire source spread speed threshold v local and v global , and the temperature change rate threshold and Local power outage triggering conditions: v f (t)≥v local or Total power outage triggering conditions: v f (t)≥v global or where v local is the local power outage threshold for the fire spread rate, v global is the total power outage threshold for the speed of fire spread, is the local blackout threshold for the temperature change rate, is the total blackout threshold for the temperature change rate, When the fire spreads and the temperature changes abnormally, the system triggers the emergency evacuation mechanism. The evacuation condition is set as the temperature change rate or the fire spread speed exceeds the high risk threshold. and v evac : v f (t)≥v evac or Among them, v evac is the evacuation threshold of the fire spread speed, It is the evacuation threshold of the temperature change rate. When the evacuation condition is met, the evacuation alarm is automatically activated and the evacuation procedure is executed.

10. A method for monitoring fire safety of energy storage power station based on big data Internet of Things according to claim 9, characterized in that: In step S5, in order to avoid further aggravation of the accident when the fire spreads, the charging and discharging operations are remotely controlled. The triggering of the remote control is combined with the power outage triggering condition. When the power outage or evacuation mechanism is activated, When the remote control system is triggered by a partial power outage, a total power outage or an evacuation, Set the charge and discharge power to 0.

Images