Search

CN-121095585-B - Fire-fighting facility safety evaluation optimization system and method based on big data

CN121095585BCN 121095585 BCN121095585 BCN 121095585BCN-121095585-B

Abstract

The invention discloses a fire-fighting equipment safety evaluation optimization system and method based on big data, and belongs to the technical field of big data. The method comprises the steps of collecting image features through a contour detection algorithm, comparing feature data of a hydrant interface with feature data of a water hose interface, carrying out adaptation, monitoring appearance states when the hydrant interface is matched with the water hose interface, constructing a water pipe network digital twin model, marking abnormal positions, simulating fire scenes, combining water pipe network distribution and water source positions, triggering a preset hierarchical supply strategy, combining the abnormal positions, abstracting to be the top points and the edges of a graph structure, constructing a weighted directed graph model, carrying out path searching in the graph structure when a certain section of water pipe is judged to be blocked in water flow, screening a standby water supply path, sending control instructions to intelligent valves on the standby water supply path by using a reinforcement learning algorithm, opening corresponding valves according to path sequences, and closing independent branches to form a closed water supply channel from the water source to a target hydrant.

Inventors

  • MA JIE
  • Wu Qipei
  • XIA QING

Assignees

  • 江苏正安消防检测评估服务有限公司

Dates

Publication Date
20260508
Application Date
20250828

Claims (9)

  1. 1. The fire-fighting facility safety evaluation optimization method based on big data is characterized by comprising the following steps of: Comparing characteristic data of the hydrant interface and the water hose interface, and acquiring an adaptation solution when adaptation deviation exists; when the fire hydrant interface is matched with the water hose interface, monitoring the appearance state of the exposed part of the water pipe positioned at the key node of the fire-fighting water pipe network of the building, summarizing the appearance states, constructing a digital twin model of the water pipe network, evaluating the pressure, flow and smoothness of fire-fighting water in each area, marking abnormal positions, simulating fire scenes based on the digital twin model of the water pipe network, and triggering a preset grading replenishment strategy based on the simulated fire scenes and combining the distribution of the water pipe network and the water source positions; in the water source replenishing process, combining the abnormal positions, abstracting key nodes in the fire-fighting water pipe network into vertexes in the graph structure, abstracting water pipe sections among the nodes into edges, and constructing a weighted directed graph model; Extracting basic attributes of corresponding water pipe sections from each side as initial parameters, wherein the physical attributes comprise the length, the diameter and the material of the water pipe sections, and the state attributes are combined with an abnormal position mark, so that when the water pipe sections are in a normal state, the current pressure value and the current flow value are given, and when the water pipe sections are in an abnormal state, the abnormal degree parameters and the corresponding correction coefficients are recorded; Based on the initial attribute parameters, a weight quantization formula is established, the weight value comprehensively reflects the water supply efficiency and the reliability of the water pipe section, and the calculation formula is as follows: Wherein R is a basic resistance coefficient, reflects the water flow resistance of the water pipe section due to the length and the diameter, and the larger the value is, the larger the resistance is, and the calculation formula is Wherein k is a resistance constant, L is the length of the water pipe section, and D is the diameter of the water pipe section; S is a state correction coefficient, and the weight is adjusted according to the abnormality degree of the water pipe section, and the larger the value is, the worse the state is, P is the weight is adjusted according to the importance of the area where the water pipe section is located, and the smaller the value is, the higher the priority is, the weights of the vertexes, the oriented sides and the sides are recorded into a graph structure database, and a weighted oriented graph model is constructed; When a certain section of water pipe is judged to be blocked, the available fire hydrant around the fire starting point is taken as a target node, a main water source and a standby water source are taken as starting nodes, path searching is carried out in the graph structure, the path with the highest efficiency is screened out to be taken as a standby water supply path, a reinforcement learning algorithm is used for sending a control instruction to intelligent valves on the standby water supply path, the corresponding valves are opened according to the path sequence, and an independent branch is closed, so that a closed water supply channel from the water source to the target fire hydrant is formed.
  2. 2. The method for optimizing fire protection equipment safety assessment based on big data according to claim 1, wherein the step of acquiring the image features of the hydrant interface through the contour detection algorithm comprises the following steps: the method comprises the steps of shooting a fire hydrant interface at multiple angles, collecting an image data set containing the shape, caliber size, thread type, buckle number and distribution position of the interface, optimizing the image quality through an image preprocessing algorithm based on the image data set, and eliminating interference factors; The method comprises the steps of carrying out full contour scanning on an image by using a contour detection algorithm, extracting all closed contour areas, setting screening conditions according to physical characteristics of a fire hydrant interface, removing interference contours which do not accord with the characteristics, judging the degree of regularity of the contours according to the ratio of the circumferences to the areas of the contours, reserving the continuous-edge interface main body contours, simplifying edge curves on the interface main body contours by using a Fabry-Perot algorithm, reserving characteristic points with curvature change meeting the conditions, calculating core parameters of the interface, including obtaining the diameter of the contours by using a minimum circumcircle algorithm, positioning the edge contours of an inner ring of the interface, calculating the diameter of the inner ring, identifying periodic contour fluctuation formed by threads or buckles, calculating the intervals between adjacent characteristic points, counting the number of buckle characteristic points, determining interface buckle distribution characteristics, converting each calculated parameter into a unified unit, calibrating with a preset image scale, and removing abnormal values caused by image blurring or shielding to serve as the image characteristics of the fire hydrant interface.
  3. 3. The fire protection facility safety evaluation optimization method based on big data according to claim 1, wherein the comparing characteristic data of the hydrant interface and the water hose interface, when there is an adaptation deviation, obtaining an adaptation solution, comprises: Setting an adaptive threshold range for each characteristic parameter, comparing the characteristic parameters of the hydrant interface and the water hose interface item by item according to a set comparison rule, judging complete adaptation when all the parameters are in the threshold range, recording specific values of deviation parameters and marking deviation types when the parameters are beyond the threshold range, retrieving a preset solution database according to the recognized deviation parameters and deviation types, carrying out adaptation verification on the matched solutions, and outputting an adaptation solution.
  4. 4. The method for optimizing fire protection equipment safety assessment based on big data according to claim 1, wherein the monitoring of the appearance state of the exposed part of the water pipe at the key node of the fire protection water pipe network of the building when the fire hydrant interface is matched with the water hose interface comprises the following steps: the key nodes of the building fire-fighting water pipe network comprise branch pipe joints, valve interfaces and floor riser exposed sections, camera shooting parameters are set, a timing shooting period is set, the collected water pipe appearance images are preprocessed, a Canny edge detection algorithm is used for carrying out feature extraction on the preprocessed images, the color change of the water pipe surface is identified through an HSV color space model, a gray level co-occurrence matrix is used for calculating texture consistency of the water pipe surface, abnormal textures are identified, protruding, recessed or deformed areas of the water pipe surface are located through the edge detection algorithm, actual sizes of the identified abnormal areas are converted through pixel scales, and the water pipe appearance state of each key node is evaluated according to the extracted appearance feature parameters and in combination with preset judging standards to obtain the appearance state.
  5. 5. The method for optimizing fire protection facility safety evaluation based on big data according to claim 1, wherein the step of summarizing appearance states, constructing a digital twin model of a water pipe network, evaluating pressure, flow and smoothness of fire protection water in each area, and marking abnormal positions comprises the steps of: Based on the network layout of the fire fighting water pipe in the building CAD drawing, mapping the physical structure of the water pipe, and converting the pipeline section, the valve and the fire hydrant into basic components of a digital twin model; defining topological association among components according to actual connection relation, binding a unique identifier for each component, and constructing a digital twin model of the water pipe network corresponding to the ID of the physical node; The method comprises the steps of dividing a digital twin model into evaluation units according to building functions, evaluating each evaluation unit, calculating average pressure values of all pipeline sections in the unit, comparing the average pressure values with a designed pressure threshold of the area, marking the average pressure values as pressure anomalies when the average pressure values are lower than the designed pressure threshold, analyzing balance relation between total water inflow and water outflow in the unit based on flow sensor data of each node, marking the balance relation as flow anomalies when the average pressure values exceed a set threshold, calculating actual path of the pipeline through fluid dynamics simulation by combining appearance states, pressure change trends and flow change trends, marking the average pressure values as smoothness blockage when the path reduction rate is larger than the set threshold, and tracing specific anomaly nodes and marking anomaly positions through topological relation of the digital twin model for the evaluated anomaly units.
  6. 6. The fire protection facility safety evaluation optimization method based on big data according to claim 1, wherein the simulating a fire scene based on the digital twin model of the water pipe network comprises: the method comprises the steps of collecting typical fire case data of different types of buildings, extracting fire intensity, fire position characteristics and floor distribution rules, establishing a fire parameter library, setting triggering conditions for each fire intensity type, dividing the fire intensity into particularly important fire intensity, large fire intensity and general fire intensity, superposing building space layout information in a virtual space of a digital twin model of a water pipe network, mapping scene variables in the fire intensity parameter library to corresponding positions of the digital twin model of the water pipe network, defining interaction logic of fire intensity scenes and the water pipe network, setting influences of fire smoke diffusion on a sensor, starting a fire simulation engine, and simulating changes of the fire intensity along with time based on set initial parameters.
  7. 7. The method for optimizing fire protection equipment safety assessment based on big data according to claim 1, wherein when a certain section of water pipe is determined to be blocked in water flow, the method uses available fire hydrants around a fire point as target nodes, uses a main water source and a standby water source as starting nodes, performs path searching in a graph structure, screens out a path with highest efficiency as a standby water supply path, and comprises the following steps: When a sensor detects that the water flow of a certain section of water pipe is blocked, a graph structure edge corresponding to the blocked water pipe section is positioned by combining a water pipe network digital twin model, and the graph structure edge is marked as a failure edge; according to the position and topological relation of the failure edge, expanding the associated edge which is possibly affected by the association to determine the area which cannot participate in path searching in the graph structure, taking the corresponding vertexes of the main water source and the standby water source in the graph structure into a starting node set, and when the path from the main water source to the blocked area is partially failed, preferentially retaining the standby water source nodes with normal states; The Di Jie St algorithm is called, a starting node set is taken as a starting point, a target node set is taken as an ending point, all feasible paths are searched in the graph structure, no invalid edge is included in the feasible paths, the total weight and the estimated water delivery efficiency of each path are calculated, paths with the total weight exceeding a threshold value are removed, candidate paths are reserved, the candidate paths are ordered according to the estimated water delivery efficiency from high to low, and the candidate path with the highest estimated water delivery efficiency is selected as a standby water supply path.
  8. 8. The method for optimizing security assessment of fire-fighting equipment based on big data according to claim 1, wherein the step of using reinforcement learning algorithm to send control command to intelligent valves on the standby water supply path, opening corresponding valves in path sequence, closing independent branches, and forming a closed water supply channel from water source to target fire hydrant, comprises: The method comprises the steps of modeling a valve control process of a standby water supply path into a Markov decision process, defining environment elements, including a state space, wherein the state space comprises the current state of all intelligent valves on the standby water supply path, the real-time pressure and the water flow speed of each pipeline section and the water demand of a target hydrant, the action space comprises opening, closing and maintaining operation sets of each intelligent valve, an objective function comprises the steps of taking a fastest closed water supply channel as a core, quantifying indexes comprising total time of valve operation, standard reaching time of path water pressure and irrelevant branch closing rate, constructing a neural network fitting Q by adopting a deep reinforcement learning algorithm, inputting the neural network fitting Q into a state space feature, outputting the state space feature into the Q value of each action, generating a training sample by utilizing historical fire simulation data and valve operation logs, training the deep reinforcement learning model, acquiring the real-time state of the standby water supply path after fire occurrence as the initial state, inputting the trained deep reinforcement learning model, outputting the Q value of each possible action, selecting the action with the largest Q value as the current optimal operation, generating an instruction sequence according to the path sequence, preferably opening the water source end valves, sequentially closing the intermediate valves, finally closing all the branch valves irrelevant to the paths, guaranteeing the preset water supply channel, sending the instruction sequence to the preset water supply channel, and sending the instruction to the target hydrant when the water supply channel is judged to reach the target water supply condition after the target water supply channel is completed.
  9. 9. A fire protection equipment safety evaluation optimization system based on big data, using the fire protection equipment safety evaluation optimization method based on big data as set forth in any one of claims 1 to 8, comprising: The adaptation solution generating module comprises an image characteristic acquisition unit and an adaptation solution generating unit, wherein the image characteristic acquisition unit acquires the image characteristic of the fire hydrant interface through a contour detection algorithm; The abnormal position marking and supplying strategy executing module comprises an appearance state monitoring unit, an abnormal position marking unit, a fire scene simulating unit and a supplying strategy executing unit, wherein the appearance state monitoring unit monitors the appearance state of the exposed part of a water pipe positioned at a key node of a fire-fighting water pipe network of a building when a fire hydrant interface is matched with a water hose interface, the abnormal position marking unit gathers the appearance states, builds a water pipe network digital twin model, evaluates the pressure, flow and smoothness of fire-fighting water in each area and marks the abnormal position, the fire scene simulating unit simulates a fire scene based on the water pipe network digital twin model, and the supplying strategy executing unit triggers a preset grading supplying strategy based on the simulated fire scene and combining the water pipe network distribution and the water source position; The directed graph construction module comprises a directed graph construction unit, wherein the directed graph construction unit is used for abstracting key nodes in a fire-fighting water pipe network into vertexes in a graph structure by combining abnormal positions in the water source supply process, abstracting water pipe segments among the nodes into edges, and constructing a weighted directed graph model; The closed water supply channel generation module comprises a standby water supply path screening unit and a closed water supply channel generation unit, wherein when a certain section of water pipe is judged to be blocked in water flow, the standby water supply path screening unit takes available fire hydrants around a fire starting point as target nodes, takes a main water source and a standby water source as starting nodes, performs path searching in a graph structure, screens out the path with highest efficiency as a standby water supply path, and sends a control instruction to intelligent valves on the standby water supply path by using a reinforcement learning algorithm, opens corresponding valves according to the path sequence, and closes independent branches to form the closed water supply channel from the water source to the target fire hydrants.

Description

Fire-fighting facility safety evaluation optimization system and method based on big data Technical Field The invention relates to the technical field of big data, in particular to a fire-fighting facility safety evaluation optimization system and method based on big data. Background The fire safety is a core link of building safety management, rapid adaptation of a fire hydrant and a water hose and stable water supply of a fire water pipe network are key to fire suppression, the existing fire control system relies on manual inspection maintenance, and when a complex building structure and a fire burst occur, fire extinguishing time is delayed due to problems of interface adaptation deviation, unknown water pipe network state, blocked water supply paths and the like, and emergency response efficiency is required to be improved through an intelligent technology. The fire hydrant and the water hose interface have various specifications, and the prior art relies on manual identification and matching of fire fighters, so that the adaptation deviation is easily caused by insufficient experience or misoperation in emergency. The existing system mostly adopts a mode of combining regular manual inspection with local sensors to monitor the fire-fighting water pipe, so that full network coverage cannot be realized, the appearance state of key nodes depends on naked eyes for observation, and early hidden danger is missed easily due to insufficient inspection frequency. When a fire disaster occurs, if a certain section of water pipe is blocked due to high-temperature deformation or breakage, the prior art generally depends on firefighters to select a standby path according to field experience, and global analysis on the network topology structure of the whole water pipe is lacking. Disclosure of Invention The invention aims to provide a fire-fighting facility safety evaluation optimization system and method based on big data, so as to solve the problems in the prior art. In order to achieve the above purpose, the present invention provides the following technical solutions: in a first aspect, the application provides a fire protection facility safety evaluation optimization method based on big data, which comprises the following steps: Comparing characteristic data of the hydrant interface and the water hose interface, and acquiring an adaptation solution when adaptation deviation exists; when the fire hydrant interface is matched with the water hose interface, monitoring the appearance state of the exposed part of the water pipe positioned at the key node of the fire-fighting water pipe network of the building, summarizing the appearance states, constructing a digital twin model of the water pipe network, evaluating the pressure, flow and smoothness of fire-fighting water in each area, marking abnormal positions, simulating fire scenes based on the digital twin model of the water pipe network, and triggering a preset grading replenishment strategy based on the simulated fire scenes and combining the distribution of the water pipe network and the water source positions; in the water source replenishing process, combining the abnormal positions, abstracting key nodes in the fire-fighting water pipe network into vertexes in the graph structure, abstracting water pipe sections among the nodes into edges, and constructing a weighted directed graph model; When a certain section of water pipe is judged to be blocked, the available fire hydrant around the fire starting point is taken as a target node, a main water source and a standby water source are taken as starting nodes, path searching is carried out in the graph structure, the path with the highest efficiency is screened out to be taken as a standby water supply path, a reinforcement learning algorithm is used for sending a control instruction to intelligent valves on the standby water supply path, the corresponding valves are opened according to the path sequence, and an independent branch is closed, so that a closed water supply channel from the water source to the target fire hydrant is formed. With reference to the first aspect, in a first implementation manner of the first aspect of the present application, the capturing, by a contour detection algorithm, image features of a hydrant interface includes: the method comprises the steps of shooting a fire hydrant interface at multiple angles, collecting an image data set containing the shape, caliber size, thread type, buckle number and distribution position of the interface, optimizing the image quality through an image preprocessing algorithm based on the image data set, and eliminating interference factors; The method comprises the steps of carrying out full contour scanning on an image by using a contour detection algorithm, extracting all closed contour areas, setting screening conditions according to physical characteristics of a fire hydrant interface, removing interference contours which do not accord with the