Fire-fighting equipment and material optimization scheduling method based on big data

Abstract

The application relates to the field of firefighting technology and discloses a fire-fighting equipment and material optimized scheduling method based on big data, which comprises the following steps: step 1, collecting the temperature, voltage and cumulative working hour data of fire-fighting equipment, obtaining real-time wind speed, humidity and temperature data of a meteorological department, and obtaining building density, road connectivity and historical fire point coordinates in geographic information; step 2, predicting the future fire risk probability according to the collected meteorological data, geographic information and historical fire point coordinates, and simultaneously based on the temperature, voltage and cumulative working hour data of the fire-fighting equipment. The application adopts a dynamic risk prediction and cross-region collaborative pre-deployment technical scheme, predicts the fire risk probability by real-time fusion of meteorological, geographic and historical disaster data, and generates a pre-deployment scheme based on the risk level and the equipment health state, so that the technical effects of improving the resource coverage precision of high-risk areas and optimizing the cross-region collaborative efficiency are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fire fighting technology, in particular to a fire fighting equipment and material optimization scheduling method based on big data. BACKGROUND

[0002] Fire fighting equipment and material scheduling is a process of scientifically deploying resources such as fire vehicles, rescue equipment, and fire extinguishing equipment through an intelligent system. With the help of Internet of Things, big data, and artificial intelligence technology, equipment status, traffic conditions, and disaster data are collected in real time, demand is evaluated using prediction models, and the best scheduling scheme is automatically generated through optimization algorithms, with dynamic adjustment supported, significantly improving response speed, resource utilization, and rescue success rate, and realizing the transition from traditional experience-based decision-making to data-driven intelligent decision-making, providing efficient and reliable protection for urban emergency management.

[0003] For example, the existing Chinese patent with publication number CN118569448B discloses an optimization scheduling method for fire fighting equipment and materials, which adopts a multi-objective optimization model, integrates genetic algorithms, and balances factors such as time, equipment utilization rate, and failure rate to achieve efficient scheduling. By integrating multiple real-time data sources such as sensors, traffic, and fire alarm information, deep analysis is performed to ensure that the scheduling scheme can dynamically adapt to emergency situations. AI and machine learning are combined to intelligently analyze historical data, identify scheduling problems, and provide suggestions to reduce human decision-making errors. This method realizes digital management of the entire life cycle of equipment, accurately predicts maintenance and replacement cycles, improves efficiency and safety, and introduces dynamic optimization and self-learning mechanisms to continuously adjust and improve the system to adapt to changes in various scenarios and task requirements, ensuring continuous improvement in the accuracy and efficiency of scheduling decisions.

[0004] For example, the existing Chinese patent with publication number CN109865231B discloses a mobile modular intelligent fire fighting duty support equipment method and related products, which includes obtaining liquid level sensing data in the target fire truck through a liquid level sensor, obtaining the first position of the target fire truck through a position acquisition module, and obtaining the running state of the target fire truck through a fire cloud platform. When the liquid level sensing data meets the preset conditions, the scheduling strategy of the target fire truck is determined based on the running state and the first position. In this way, the fire trucks in the fire station can be uniformly monitored and scheduled by determining the corresponding scheduling strategy based on the position information and running state of the fire truck when the liquid level sensing data meets the preset conditions, making fire rescue and fire protection more intelligent and efficient.

[0005] The above-mentioned patents have the following deficiencies:

[0006] On the one hand, the traditional fire-fighting material scheduling mainly relies on historical data and static rules, lacks the dynamic prediction ability of the fire area about to occur, and at the same time, cannot deploy equipment to the boundaries of cities and provinces adjacent to the fire area, resulting in slow response speed of fire-fighting material scheduling when facing large fires, thereby reducing the fire extinguishing efficiency.

[0007] On the other hand, the current fire-fighting equipment scheduling is mainly based on the allocation of material types and inventory quantity, but ignores the actual state of the equipment, such as aging degree, maintenance cycle, remaining power, usage frequency and other key information. The neglected information can cause equipment with substandard performance and near-end-of-life to be misassigned to the rescue front, delay the best disposal opportunity of the fire, and even cause secondary fires.

[0008] Therefore, the present application provides a fire-fighting equipment material optimization scheduling method based on big data to solve the above problems. SUMMARY

[0009] In view of the deficiencies of the prior art, the present application provides a fire-fighting equipment material optimization scheduling method based on big data to solve the problems raised in the background art.

[0010] To achieve the above purpose, the present application is implemented by the following technical scheme: a fire-fighting equipment material optimization scheduling method based on big data, comprising:

[0011] Step one, collect the temperature, voltage and cumulative working hour data of the fire-fighting equipment, and obtain the real-time wind speed, humidity and temperature data of the meteorological department, and the building density, road connectivity and historical fire point coordinates in geographic information;

[0012] Step two, according to the collected meteorological data, geographic information and historical fire point coordinates, predict the future fire risk probability, and based on the temperature, voltage and cumulative working hour data of the fire-fighting equipment, calculate the equipment health score and classify it as available, warning and disabled;

[0013] In the step two, the prediction of fire risk probability and the calculation of equipment health score further comprise:

[0014] Sub-step , multi-source disaster risk prediction model construction:

[0015] Input the collected meteorological data

wind speed , humidity , ambient temperature

building density , road connectivity , historical fire point coordinates

[0016] ,

[0017] Wherein, is the fusion feature, is the long short-term memory network to extract meteorological time series features, is the graph convolution network to model geographical space correlation, is the feature splicing operation;

[0018] Output area risk probability:

[0019] ,

[0020] Wherein, is function, , is the trainable weight and bias, is the area fire risk probability at time

[0021] Sub-step , fire equipment health score calculation:

[0022] Input the equipment temperature , voltage , cumulative working hours collected in steps , calculate the health score:

[0023] ,

[0024] Wherein, is the health score, , is the life factor and state deviation factor weight, is the equipment design life, , is the temperature and voltage reference value when the equipment is working normally;

[0025] Health score classification:

[0026] Available: ≥ ;

[0027] Warning: ≤ < ;

[0028] Disable: < ;

[0029] ​Sub-step , risk-health data fusion and classification output:

[0030] Risk probability and health score are associated by area, generating a risk-health matrix:

[0031] ,

[0032] Wherein, is the risk probability and equipment health score association matrix of the area , is the probability of fire occurring in the area at time t, is the highest health score of available equipment in the area ;

[0033] Output classification results:

[0034] High-risk area: ≥ and ≥ ;

[0035] Medium-risk area: ≤ < or < ;

[0036] Low-risk area: < ;

[0037] Step three, according to the predicted fire risk probability and equipment health score, generate cross-area fire-fighting equipment pre-deployment scheme;

[0038] In the step three, generating cross-area fire-fighting equipment pre-deployment scheme further comprises:

[0039] Sub-step , area risk priority division:

[0040] Input fire risk probability , divide the area priority by threshold:

[0041] Priority= ,

[0042] Output target area set ;

[0043] Sub-step , health equipment screening and demand matching:

[0044] Input equipment health score , screening available equipment ≥ 80], matching regional demand by type: ,

[0045] wherein, is the region adjustable available equipment set, is the first firefighting equipment individual, is the function category of the firefighting equipment , is the required equipment type of the region ;

[0046] sub-step , multi-objective resource optimization model solving:

[0047] Construct a mixed integer programming model, the objective function is to minimize the allocation cost and health penalty term:

[0048] ,

[0049] wherein, is the number of equipment allocated from the warehouse to the region , is the allocation cost, is the health penalty coefficient, is the health score conversion factor, is the inventory status of the firefighting equipment , is the set of all firefighting equipment warehouses, is the set of target regions requiring pre-deployment of firefighting equipment;

[0050] The constraint conditions include:

[0051] Coverage requirement: ≥ , is the estimated demand of the region ;

[0052] Warehouse capacity: ≤ , is the maximum capacity of the warehouse ;

[0053] Step four, according to the generated pre-deployment scheme, combined with real-time traffic conditions and dangerous area coordinates, generate dynamic rescue path and equipment allocation instructions;

[0054] Step 5: Based on the actual equipment usage data and path execution results, update the fire risk prediction model and equipment health scoring model, and generate equipment maintenance and decommissioning instructions.

[0055] Preferably, in step one, collecting fire-fighting equipment status data and disaster risk data further includes:

[0056] Sub-step Firefighting equipment sensor data acquisition and preprocessing:

[0057] Temperature sensors, voltage sensors, and time counters are deployed in fire-fighting equipment to collect temperature data. ,Voltage Cumulative working hours Calculation formula:

[0058] ,

[0059] ,

[0060] ,

[0061] in, For the first Second temperature sampling value, The number of samples per unit time. The number of samples per unit time. For the first Secondary voltage sampling value For equipment in time The work status;

[0062] Sub-step Disaster risk data integration and spatiotemporal alignment:

[0063] Real-time wind speed is obtained from the meteorological department's interface. ,humidity Ambient temperature Building density is extracted from geographic information databases. Road connectivity Coordinates of historical fire sites Calculation formula:

[0064] ,

[0065] ,

[0066] ,

[0067] Data time alignment: Linear interpolation is used to compensate for time deviations in asynchronous data.

[0068] ,

[0069] wherein, is a real-time wind speed value at time , is a current timestamp, is a wind speed value observed at time point , is a time point of the th data collection, is a wind speed value observed at time point , is a time point of the th data collection;

[0070] sub-step , multi-source data fusion and feature storage:

[0071] After aligning the equipment data and disaster data by timestamp, a spatio-temporal fusion data block is generated and stored in a spatio-temporal database, with the time and coordinates as index fields, and the calculation formula being:

[0072] ,

[0073] wherein, is the multi-source fusion data block after spatio-temporal alignment.

[0074] Preferably, in the step four, the generation of the dynamic rescue path and equipment allocation instruction further comprises:

[0075] sub-step , real-time traffic and dangerous area data integration:

[0076] The target area coordinates in the pre-deployment scheme are input, and the traffic speed and congestion coefficient of the traffic condition are obtained in real time, as well as the dangerous area coordinates , the geofence radius is calculated, and the dangerous area geofence is generated: ,

[0077] wherein, is the geofence radius, is the dangerous substance diffusion speed, is a preset safety response time threshold value;

[0078] sub-step , multi-target dynamic path optimization model construction:

[0079] Based on the reinforcement learning model, the state space is defined as , action space is path node selection, reward function is:

[0080] ,

[0081] wherein, reward function, path estimation time consumption, maximum allowable time consumption, , , time, health, safety weight, indication function, index set, used to limit the object of summation operation;

[0082] Substep , health equipment allocation instruction generation:

[0083] Input equipment health score , screening available equipment[ ≥80], combined with optimized path, generate allocation instruction, instruction attached path navigation coordinate sequence :

[0084] ≥ ,

[0085] wherein, equipment list to the area , available equipment set that can be allocated in the area , the first fire-fighting equipment individual , function category of fire-fighting equipment , required equipment type of the area .

[0086] Preferably, in step five, updating the model and generating maintenance retirement instructions further comprise:

[0087] Substep , actual scheduling data acquisition and feature extraction:

[0088] Acquire the use data of actual allocation equipment, including cumulative working hour increment , failure times , and path execution time difference ;

[0089] Extract features: ,

[0090] wherein, feature set, is the cumulative man-hour increment, is the number of failures, is the actual time consumption minus the planned time consumption;

[0091] sub-step , risk prediction and health assessment model increment update:

[0092] updating the fire risk prediction model:

[0093] ,

[0094] wherein, is the mean squared error loss function of the fire risk prediction model, is the probability of fire occurrence predicted by the fire risk prediction model, is the actual state of whether a fire has occurred in the area updating the equipment health score model:

[0095]

[0096] ,

[0097] wherein, is the predicted health score of the fire-fighting equipment , is the mean squared error loss function of the equipment health score model, is the actual health score of the fire-fighting equipment ;

[0098] joint optimization: ,

[0099] wherein, is the learning rate, is the new parameter updated by gradient descent, is the old parameter before model update, is the gradient of the loss function with respect to the model parameters ;

[0100] sub-step , equipment retirement priority calculation and instruction generation:

[0101] calculating the retirement priority: wherein, is the retirement priority score of the fire-fighting equipment , is the cumulative man-hours of the fire-fighting equipment , is the design life of the fire-fighting equipment , is the fire-fighting equipment​​ a predicted health score of the equipment;

[0102] generating retirement instructions and maintenance instructions:

[0103] retirement instructions: ≥ ;

[0104] maintenance instructions: < ≤ .

[0105] Preferably, in the step two, the calculation of the equipment health score comprises a life factor and a state deviation factor, the life factor is a proportion of the remaining life to the total life, and the state deviation factor is a normalized Euclidean distance between the sensor data and the reference value.

[0106] In the step three, the objective function of the mixed integer programming model is to minimize the deployment cost and a health penalty term, and the constraint conditions comprise covering the demand of the high-risk area and calling the equipment with a health score higher than a preset value.

[0107] Preferably, in the step four, the reward function of the multi-objective reinforcement learning model comprises an exponential decay term of path time consumption, an accumulated term of equipment health score and a penalty term of entering a dangerous area.

[0108] In the step five, the calculation of the equipment retirement priority comprises a linear combination of the proportion of accumulated working hours and the health score decay value, and the automatic retirement is triggered when the priority exceeds a threshold.

[0109] A terminal device comprises a processor and a memory, the memory stores a computer program, and the computer program is executed by the processor to realize the fire-fighting equipment and material optimization scheduling method based on big data.

[0110] A storage medium stores a computer program, and the computer program is executed by a processor to realize the fire-fighting equipment and material optimization scheduling method based on big data.

[0111] The present application provides a fire-fighting equipment and material optimization scheduling method based on big data. It has the following advantages:

[0112] 1. The present application adopts a dynamic risk prediction and cross-regional collaborative pre-deployment technical solution, predicts the fire risk probability by real-time fusion of meteorological, geographical and historical disaster data, and generates a pre-deployment scheme based on the risk level and the equipment health status, thereby improving the resource coverage accuracy of high-risk areas and optimizing the cross-regional collaborative efficiency. Compared with the technical solution in the prior art which relies on static preplan or manual experience scheduling, the present application solves the problems of mismatch between resource allocation and dynamic disaster risk and lag of cross-regional collaborative response.

[0113] 2. The application adopts a technical scheme of dynamic evaluation of equipment health state and whole life cycle closed-loop management, calculates a health score in real time through multi-source sensor data, classifies and controls, combines a retirement priority algorithm to trigger a maintenance or retirement instruction, achieves the technical effect of guaranteeing the availability of allocated equipment and reducing the occupation of inefficient resources, and solves the problems of lagging equipment state sensing and increased rescue risk caused by misallocation of faulty equipment compared with the technical scheme of periodic manual inspection or fixed life determination in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0114] Figure 1 The flowchart of the application. DETAILED DESCRIPTION

[0115] To enable personnel skilled in the art to understand the application scheme, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the application.

[0116] The application will be described in detail below with reference to the drawings:

[0117] Embodiment:

[0118] Please refer to the accompanying drawings Figure 1 , the embodiment of the application provides a fire-fighting equipment and material optimization scheduling method based on big data, characterized in that it comprises:

[0119] Step 1, collect the temperature, voltage, cumulative working hour data of the fire-fighting equipment, and obtain the real-time wind speed, humidity, temperature data of the meteorological department, and the building density, road connectivity and historical fire point coordinates in geographic information;

[0120] Substep , fire-fighting equipment sensor data acquisition and preprocessing:

[0121] Deploy temperature sensors, voltage sensors and working hour counters in the fire-fighting equipment to collect temperature , voltage , cumulative working hours , the calculation formula is:

[0122] ,

[0123] ,

[0124] ,

[0125] wherein, the first temperature sampling value, the first temperature sampling value, the number of sampling times per unit time, the number of sampling times per unit time, the first voltage sampling value, the first voltage sampling value, the working state of the equipment at time ;

[0126] sub-step Disaster risk data integration and spatio-temporal alignment:

[0127] Real-time wind speed , humidity , and ambient temperature are obtained from the meteorological department interface, and building density , road connectivity , and historical fire point coordinates are extracted from the geographic information database, and the calculation formula is:

[0128] ,

[0129] ,

[0130] ,

[0131] Data time alignment: linear interpolation method is used to compensate for time deviation for non-synchronous data:

[0132] ,

[0133] wherein, is the real-time wind speed value at time , is the current timestamp, is the wind speed value observed at time point , is the time point of the first data collection, is the wind speed value observed at time point , is the time point of the first data collection, is the wind speed value observed at time point ;

[0134] sub-step Multi-source data fusion and feature storage:

[0135] After aligning the equipment data and disaster data by timestamp, a spatio-temporal fusion data block is generated and stored in the spatio-temporal database, with the index fields being time and coordinates , and the calculation formula is:

[0136] ,

[0137] wherein, is the spatio-temporal aligned multi-source fusion data block;

[0138] Step two, according to the collected meteorological data, geographic information and historical fire point coordinates, the future fire risk probability is predicted, and based on the temperature, voltage and cumulative working hour data of the fire fighting equipment, the equipment health score is calculated and classified as available, warning and disabled;

[0139] In step two, the calculation of the equipment health score includes a life factor and a state deviation factor, the life factor is the proportion of the remaining life to the total life, and the state deviation factor is the normalized Euclidean distance between the sensor data and the reference value;

[0140] Sub-step , multi-source disaster risk prediction model construction:

[0141] Input the collected meteorological data

wind speed , humidity , ambient temperature

building density , road connectivity , historical fire point coordinates

[0142] ,

[0143] wherein, is the fusion feature, is the long short-term memory network to extract meteorological time sequence features, is the graph convolution network to model the geographical space correlation, is the feature splicing operation;

[0144] Output regional risk probability:

[0145] ,

[0146] wherein, is the function, , are trainable weights and biases, is the regional fire risk probability in time ;

[0147] Sub-step Calculation of health score for fire-fighting equipment:

[0148] Input step to collect equipment temperature ,Voltage Cumulative working hours Calculate the health score:

[0149] ,

[0150] in, Rate your health. , The weights are the lifespan factor and the state deviation factor. For the design life of the equipment, , The temperature and voltage reference values ​​are for normal operation of the equipment;

[0151] Health score classification:

[0152] Available: ≥ ;

[0153] Warning: ≤ < ;

[0154] Disabled: < ;

[0155] Sub-step Risk-health data fusion and classification output:

[0156] Risk probability With health score Generate a risk-health matrix by region:

[0157] ,

[0158] in, For the region The correlation matrix between risk probability and equipment health score. For the region The probability of a fire occurring at time t. For the region The highest health rating of available equipment;

[0159] Output classification results:

[0160] High-risk areas: ≥ and ≥ ;

[0161] High-risk area: ≤ < or < ;

[0162] Low-risk area: < ;

[0163] Step three, generate the pre-deployment scheme of fire-fighting equipment across areas according to the predicted fire risk probability and equipment health score;

[0164] In step three, the objective function of the mixed integer programming model is to minimize the deployment cost and health penalty term, and the constraint conditions include covering the demand of high-risk areas and calling equipment with a health score higher than a preset value;

[0165] Sub-step , area risk priority division:

[0166] Input fire risk probability , divide the area priority according to the threshold value:

[0167] Priority= ,

[0168] Output target area set ;

[0169] Sub-step , health equipment screening and demand matching:

[0170] Input equipment health score , screen available equipment[ ≥80] and match the area demand by type: ,

[0171] Wherein, is the set of available equipment that can be deployed in the area , the th fire-fighting equipment individual, is the functional category of the fire-fighting equipment , and is the required equipment type of the area ; Sub-step

[0172] , multi-objective resource optimization model solution: Construct a mixed integer programming model, and the objective function is to minimize the deployment cost and health penalty term:

[0173]

[0174] ,

[0175] ​in, For warehouse Transferred to the region The number of equipment To allocate costs, As a health penalty coefficient, As a health score conversion factor, For firefighting equipment Inventory status, This is a collection of all fire-fighting equipment warehouses. A collection of target areas that require the pre-deployment of fire-fighting equipment;

[0176] The constraints include:

[0177] Coverage requirements: ≥ , For the region Forecast demand;

[0178] Warehouse capacity: ≤ , For warehouse Maximum capacity;

[0179] Step 4: Based on the generated pre-deployment plan, and combined with real-time traffic conditions and the coordinates of dangerous areas, generate dynamic rescue routes and equipment allocation instructions.

[0180] In step four, the reward function of the multi-objective reinforcement learning model includes an exponential decay term for path time, an accumulation term for equipment health score, and a penalty term for entering dangerous areas.

[0181] Sub-step Real-time traffic and hazardous area data integration:

[0182] Input the coordinates of the target area in the pre-deployment plan to obtain real-time traffic conditions and speed. With congestion coefficient and the coordinates of the danger zone Calculate the geofence radius and generate a geofence for the danger zone: ,

[0183] in, The radius of the geofence. For the rate of diffusion of hazardous substances, To preset a safety response time threshold;

[0184] Sub-step Construction of a multi-objective dynamic path optimization model:

[0185] Based on the reinforcement learning model, the state space is defined as follows: The action space is path node selection, and the reward function is:

[0186] ,

[0187] in, For the reward function, To estimate the time required for the path, To the maximum allowable time, , , Weighted by time, health, and safety, For indicator functions;

[0188] Sub-step Health equipment allocation order generated:

[0189] Enter equipment health rating Filter available equipment [ ≥80], combined with the optimized path, generate a transfer instruction, with the instruction appended with a path navigation coordinate sequence. :

[0190] ≥ ,

[0191] in, To be transferred to the region Equipment list, For the region A collection of available and deployable equipment. For the first Individual firefighting equipment For firefighting equipment Functional categories For the region Required equipment type;

[0192] Step 5: Based on the actual equipment usage data and path execution results, update the fire risk prediction model and equipment health scoring model, and generate equipment maintenance and decommissioning instructions.

[0193] In step five, the calculation of equipment retirement priority includes a linear combination of the cumulative working hours percentage and the health score decay value. When the priority exceeds the threshold, automatic retirement is triggered.

[0194] Sub-step Actual scheduling data collection and feature extraction:

[0195] Collect actual usage data of allocated equipment, including cumulative working hours increments. Number of failures and path execution time difference ;

[0196] Feature extraction: ,

[0197] in, For feature set, To accumulate working hours increment, Number of failures This represents the difference between the actual time taken and the planned time taken.

[0198] Sub-step Incremental updates to risk prediction and health assessment models:

[0199] Update the fire risk prediction model:

[0200] ,

[0201] in, Let be the mean squared error loss function of the fire risk prediction model. Areas predicted by fire risk prediction model The probability of a fire occurring. For the region The actual state of whether a fire actually occurs;

[0202] Update equipment health rating model:

[0203] ,

[0204] in, For firefighting equipment Predicted health score To equip the health scoring model with the mean squared error loss function, For firefighting equipment Actual health score;

[0205] Joint optimization: ,

[0206] in, For learning rate, The new parameters are updated using gradient descent. These are the old parameters before the model update. For the loss function with respect to model parameters The gradient;

[0207] Sub-step Equipment retirement priority calculation and instruction generation:

[0208] Calculate retirement priority: ,in, For firefighting equipment Retirement priority score For fire fighting equipment Cumulative man-hours, For fire fighting equipment Design life, For fire fighting equipment Predicted health score;

[0209] Generating retirement instructions and maintenance instructions:

[0210] Retirement instructions: ≥ ;

[0211] Maintenance instructions: < ≤ .

[0212] Step Benefits, by comprehensive collection and preprocessing of multi-source data, the application establishes a high-quality spatio-temporal fusion data basis in the first time. Synchronization alignment of sensor data and meteorological, geographic information eliminates time deviation caused by asynchronous collection, and reduces data noise and redundancy. After data storage in the spatio-temporal database, the retrieval efficiency is greatly improved. Overall, step Provides complete and reliable data support for subsequent risk prediction and health assessment, ensuring the accuracy and real-time of the scheduling method.

[0213] Step Benefits, after deep fusion of disaster risk and equipment health in two dimensions, step Can realize the dual classification judgment of regional and equipment state. The risk prediction model can quickly identify high-risk areas, and the health score mechanism ensures that the equipment allocated is in the best available state. The risk-health matrix formed by the superposition of the two provides decision basis for fine scheduling. Avoid waste caused by concentrated resources, and prevent faulty equipment from entering the front line of rescue.

[0214] Step Benefits, based on mixed integer programming cross-regional pre-deployment, considering the dual goals of minimum cost and minimum health penalty. The model fully considers warehouse capacity, regional demand and equipment health level, realizes the optimal matching of resource allocation. In addition, priority division and type matching ensure that the equipment needed by high-risk areas is in place in time, reducing allocation conflicts. Overall, step While ensuring the efficiency of rescue, effectively controlling the cost of transportation and the loss of equipment life.

[0215] Step Benefits, by integrating real-time traffic congestion and dangerous area information, step The dynamic generation of safe and efficient rescue routes is enabled. The multi-target reinforcement learning model autonomously learns and optimizes after weighing the path time consumption, equipment health, and entry risk, thereby improving the intelligent level of path decision-making. The generated allocation instructions are accompanied by navigation coordinates, which shortens the response time and enhances the on-site operability of rescue personnel. The flexibility and on-site safety of rescue are significantly improved.

[0216] Step The closed-loop feedback mechanism enables the risk prediction and health assessment model to continuously self-correct. Based on incremental updates of actual allocation and execution data, the model accuracy is continuously improved. The equipment retirement priority algorithm enables automatic triggering of maintenance and retirement, thereby precisely controlling the entire life cycle of the equipment. This step optimizes maintenance resource allocation, avoids inefficient or faulty equipment from occupying storage and transportation channels, and lays a solid foundation for the next round of scheduling.

[0217] A terminal device includes a processor and a memory. The memory stores a computer program that, when executed by the processor, implements a big data-based fire-fighting equipment and material optimization scheduling method.

[0218] A storage medium stores a computer program that, when executed by a processor, implements a big data-based fire-fighting equipment and material optimization scheduling method.

[0219] The terminal device integrates a high-performance processor and a large-capacity memory, and is pre-installed with the big data scheduling algorithm of the present application. It can quickly complete data acquisition, preprocessing, risk prediction, optimization scheduling, and path planning and other full-process calculations locally. Compared with solutions that rely on cloud computing, localized processing significantly shortens response time, ensuring that scheduling decisions can be autonomously executed in network-limited or emergency situations. At the same time, by caching the model and historical data in the device, the system can implement edge intelligence, reducing communication bandwidth pressure and improving system robustness and security. The terminal device supports modular expansion and remote upgrading, and can flexibly adapt to the needs of various fire stations, greatly enhancing the convenience and maintainability of equipment deployment.

[0220] The storage medium completely encapsulates the fire-fighting equipment and material optimization scheduling software of the present application, facilitating rapid deployment and installation on various general-purpose computing platforms. With the plug-and-play feature, users can directly run the software on servers, workstations, or portable terminal devices without complex environment configuration. Moreover, the storage medium supports partitioning and encryption management, ensuring the integrity and confidentiality of the algorithm and data, and enabling data isolation according to different permission requirements to improve system information security protection capabilities. Through centralized storage and unified update mechanism, maintenance personnel can easily push new versions and continuously iterate algorithms, ensuring that the scheduling method is always in the optimal state.

[0221] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A big data-based fire-fighting equipment and material optimization scheduling method, characterized in that, The method comprises the following steps: Step 1: Collecting temperature, voltage, and cumulative working hour data of fire-fighting equipment, and obtaining real-time wind speed, humidity, and temperature data from the meteorological department, as well as building density, road connectivity, and historical fire point coordinates from geographic information; Step 2: Predicting the future fire risk probability based on the collected meteorological data, geographic information, and historical fire point coordinates, and calculating the equipment health score based on the temperature, voltage, and cumulative working hour data of the fire-fighting equipment, and classifying it as available, warning, and disabled; In the step 2, the prediction of the fire risk probability and the calculation of the equipment health score further comprise: Sub-step Multi-source disaster risk prediction model construction: Input the collected meteorological data 【wind speed , humidity , ambient temperature 】 and geographical information 【building density , road connectivity , historical fire point coordinates 】, construct spatiotemporal fusion features, wherein, , is the longitude and latitude coordinates of the first historical fire point: ， wherein, is a fusion feature, is a long short-term memory network extracting meteorological time series features, is a graph convolution network modeling geographical spatial correlation, is a feature splicing operation; Outputting the regional risk probability: , wherein, is a function, , are trainable weights and biases, is a region at time a fire risk probability; Sub-step Firefighting equipment health score calculation: Input step collected equipment temperature , voltage , cumulative hours , calculate health score: ， wherein, is a health score, , is a life factor and a state deviation factor weight, is an equipment design life, , is a temperature and voltage reference value when the equipment is operating normally; Health score classification: Available: ≥ ; Warning: ≤ < ; Disable: < ; Sub-step Risk-health data fusion and classification output: Risk probabilities are calculated with health scores associated by region, generating a risk-health matrix: ， wherein, is a matrix of risk probabilities for the zones is a matrix of risk probabilities for the zones is a matrix of risk probabilities for the zones is a probability of fire occurrence at time t, is a matrix of risk probabilities for the zones is a maximum health score of available equipment in the zone; Outputting the classification result: High risk area: ≥ and ≥ ; High risk area: ≤ < or < ; Low risk areas: ;​ Step 3: Generating a cross-regional fire-fighting equipment pre-deployment plan based on the predicted fire risk probability and the equipment health score; In the step 3, generating the cross-regional fire-fighting equipment pre-deployment plan further comprises: Sub-step , regional risk prioritization: Input fire risk probability Prioritize zones by threshold Priority , Output target region set ; Sub-step Health equipment screening and demand matching: Input equipment health score , screen available equipment ≥ 80], match regional demand by type: , wherein, is a region a set of available equipment that can be dialled, is a first firefighting equipment individual, is a type of equipment a functional category of firefighting equipment, is a region a required equipment type; Sub-step , multi-objective resource optimization model solving: Building a mixed integer programming model, with the objective function being to minimize the allocation cost and the health penalty term: ， in, For warehouse Transferred to the region The number of equipment To allocate costs, As a health penalty coefficient, As a health score conversion factor, For firefighting equipment Inventory status, This is a collection of all fire-fighting equipment warehouses. For target areas that require pre-deployment of fire-fighting equipment, This is an indexed set used to limit the objects to be summed. The constraint conditions include: Coverage needs: ≥ , for areas estimated needs; Warehouse capacity: ≤ , for the warehouse maximum capacity; Step 4: Generating a dynamic rescue path and equipment allocation instruction based on the generated pre-deployment plan, combined with real-time traffic conditions and dangerous area coordinates; Step 5: Updating the fire risk prediction model and the equipment health score model based on the actual allocation of equipment usage data and path execution effect, and generating equipment maintenance and retirement instructions. 2.The fire-fighting equipment and material optimized dispatching method based on big data according to claim 1, characterized in that, In the step 1, collecting fire-fighting equipment status data and disaster risk data further comprises: Sub-step Firefighting equipment sensor data collection and pre-processing: Temperature sensors, voltage sensors and time counters are deployed in fire fighting equipment to collect temperature , voltage , cumulative time , and calculation formula ， ， ， wherein, is the first temperature sampling value, is the first temperature sampling value, is the number of sampling times per unit time, is the number of sampling times per unit time, is the first voltage sampling value, is the first voltage sampling value, is the working state of the equipment at time is the working state of the equipment at time Sub-step Disaster risk data integration and spatio-temporal alignment: Real-time wind speed is obtained from the meteorological department's interface. ,humidity Ambient temperature Building density is extracted from geographic information databases. Road connectivity Coordinates of historical fire sites Calculation formula: ， ， ， Data time alignment: using linear interpolation method to compensate for time deviation for non-synchronous data: ， wherein is a real-time wind speed value at time is a current time stamp, is a wind speed value observed at time point is a time point of the th data collection, is a wind speed value observed at time point is a time point of the th data collection;​​​ Sub-step , multi-source data fusion and feature storage: After the equipment data and disaster data are timestamp-aligned, a spatio-temporal fusion data block is generated and stored in a spatio-temporal database, with a time index field With the coordinates The calculation formula is: ， wherein, is the spatio-temporally aligned multi-source fused data block.

3. The fire-fighting equipment and material optimized dispatching method based on big data according to claim 1, characterized in that, In the step 4, generating a dynamic rescue path and equipment allocation instruction further comprises: Sub-step Real-time traffic and hazardous area data integration: Input target area coordinates in pre-deployment scheme, real-time access to traffic speed With congestion coefficient , and dangerous area coordinates , calculate the radius of the geographic fence, and generate the dangerous area geographic fence: , wherein, is a geo-fence radius, is a dangerous substance diffusion speed, is a preset safety response time threshold; Sub-step , multi-objective dynamic path optimization model construction: Based on the reinforcement learning model, the state space is defined as , the action space is path node selection, and the reward function is: ， wherein, is a reward function, is a path estimation time consumption, is a maximum allowed time consumption, , , is a time, health, safety weight, is an indicator function, is an index set, used to limit the objects of the summation operation; Sub-step , health equipment requisition instruction generation: Input equipment health score , screen available equipment ≥ 80], combined with optimized path, generate allocation instructions, instructions attached path navigation coordinate sequence : ≥ ， wherein, a list of equipment to be dispatched to the area, a list of equipment to be dispatched to the area, a set of available equipment that can be assigned to the area, a set of available equipment that can be assigned to the area, a first fire-fighting equipment individual, a first fire-fighting equipment individual, a function category of the fire-fighting equipment, a function category of the fire-fighting equipment, a required equipment type for the area, a required equipment type for the area.

4. The fire-fighting equipment and material optimized dispatching method based on big data according to claim 1, characterized in that, In the step 5, updating the model and generating maintenance and retirement instructions further comprise: Sub-step Actual scheduling data collection and feature extraction: Collecting actual allocation equipment usage data, including cumulative work hour increment , failure times , and path execution time difference ; Extracted features: , wherein, is a feature set, is a cumulative man-hour increment, is a number of failures, is a difference between actual time consumption and planned time consumption; Sub-step Risk prediction and health assessment model incremental update: Updating the fire risk prediction model: ， wherein, is a mean squared error loss function for the fire risk prediction model, is a region predicted by the fire risk prediction model a probability of a fire occurring, is a region a true state of whether a fire actually occurred in the region; Updating the equipment health score model: ， wherein, is a predicted health score for the fire-fighting equipment , is a mean squared error loss function for the equipment health score model, is an actual health score for the fire-fighting equipment . Joint optimization: , wherein, is the learning rate, is the new parameter updated by gradient descent, is the old parameter before model update, is the gradient of the loss function with respect to the model parameter ; Sub-step , equipment retirement priority calculation and instruction generation: Calculate retirement priority: ,in, For firefighting equipment Retirement priority score For firefighting equipment Total working hours For firefighting equipment Design life, For firefighting equipment Predicted health score; Generating retirement instructions and maintenance instructions: Retire command: ≥ ; Maintenance instructions: < < .

5. The fire-fighting equipment and material optimized dispatching method based on big data according to claim 1, characterized in that, In the step 2, the calculation of the equipment health score includes a life factor and a state deviation factor, the life factor being the proportion of remaining life to total life, and the state deviation factor being the normalized Euclidean distance between sensor data and reference value; In the step 3, the objective function of the mixed integer programming model is to minimize the allocation cost and the health penalty term, and the constraint conditions include covering the demand of high-risk areas and calling equipment with a health score higher than a preset value.

6. The fire-fighting equipment and material optimized dispatching method based on big data according to claim 1, characterized in that, In the step 4, the reward function of the multi-objective reinforcement learning model includes a path time consumption exponential decay term, an equipment health score accumulation term, and a dangerous area entry penalty term; In the step 5, the calculation of the equipment retirement priority includes a linear combination of cumulative working hour proportion and health score decay value, and when the priority exceeds a threshold, automatic retirement is triggered.

7. A terminal device, characterized by comprising: The system comprises a processor and a memory, the memory stores a computer program, and the computer program is executed by the processor to implement the method of claim 1-6.

8. A storage medium, characterized by The storage medium stores a computer program, and the computer program is executed by the processor to implement the method of claim 1-6.

Images