A multi-information fusion fire extinguishing training evaluation method

Abstract

The present application relates to the field of fire extinguishing evaluation, in particular to a multi-information fusion fire extinguishing training evaluation method, comprising the following steps: after completing the calibration operation, the weight correlation coefficient is evaluated; the time characteristic value is extracted to calculate the fire extinguishing training time quality score, the burner temperature control quality score, the environmental temperature control quality score, the fire-fighting water use quality score and the gas consumption quality score; based on the obtained quality scores, weighted fusion is carried out to obtain the standardized output of the comprehensive evaluation result and the corresponding horizontal grade division. The present application constructs the quality score of the time dimension through hierarchical weighting, realizes the fine disassembly of the ability of different stages of fire extinguishing operation, and through the weighted fusion of multi-dimensional sub-scores, the standardized output of the comprehensive evaluation result and the horizontal grade division are completed.

Description

Technical Field

[0001] This invention relates to the field of fire extinguishing assessment, and more specifically to a multi-information fusion method for fire extinguishing training assessment. Background Technology

[0002] Various types of live-fire training equipment are widely used in fire training in fields such as emergency firefighting and maritime shipping. Currently, the objective assessment of firefighters' firefighting effectiveness relies on a single source of information, primarily burner temperature and video images. The method of assigning scores based on burner temperature is problematic because the temperature sensor's location is fixed, allowing trainees to employ targeted tactical maneuvers to achieve higher scores, thus negating the training's purpose. As for assessments based on video images, obstructions from objects and personnel in the training area, as well as the effects of water mist, vapor, and high temperatures, make it difficult to judge the effectiveness of the firefighting efforts. Summary of the Invention

[0003] This invention addresses the technical problems existing in the prior art by providing a multi-information fusion method for fire extinguishing training and evaluation.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: a multi-information fusion fire extinguishing training evaluation method, comprising the following steps: Step S1, calibrating the fire extinguishing training time characteristic benchmark parameters based on the fire extinguishing training test, and collecting full data of the burner temperature change over time to complete the calibration of the burner temperature control benchmark parameters, and calibrating the ambient temperature control benchmark parameters according to the ambient temperature change, and calibrating the resource consumption characteristic benchmark parameters based on the fire water usage and gas consumption, and evaluating the weighted correlation coefficient after completing the calibration operation; Step S2: Based on the burner temperature control reference parameters, fire extinguishing training time characteristic reference parameters, and ambient temperature control reference parameters calibrated in Step S1, the fire extinguishing training is divided into time sequences, and time characteristic values ​​are extracted. Based on the extracted time characteristic values, the fire extinguishing training time quality score, burner temperature control quality score, ambient temperature control quality score, fire water usage quality score, and gas consumption quality score are calculated. Step S3: Based on the obtained quality scores, perform weighted fusion to obtain a standardized output of the comprehensive evaluation results, and classify the training level levels accordingly.

[0005] In a preferred embodiment, step S1 obtains the entry time for each test based on fire extinguishing training experiments. With attack time Full sample data, for entry time Perform a normal distribution fitting operation on the full sample data, calculate the arithmetic mean of the full sample data, and use the obtained arithmetic mean as the entry time. The expected value of the corresponding normal distribution Then calculate the full sample data and the expected value. The arithmetic mean of the squares of the deviations is used as the entry time. The corresponding normal distribution variance ; Then regarding the attack time Perform a normal distribution fitting operation to obtain the attack time. The expected value of the corresponding normal distribution variance of normal distribution Meanwhile, based on the safety regulations and operational requirements corresponding to the training subjects, and by setting the maximum allowable duration for fire extinguishing training, the benchmark parameters for the time characteristics of fire extinguishing training are calibrated. This duration is the maximum time threshold from the issuance of the fire extinguishing command to the determination of fire extinguishing failure. Therefore, this application obtains the normal distribution parameters of the time characteristics that conform to the actual operation level of the training subject through large-sample experiments, and provides an evaluation benchmark for the subsequent calculation of entry time quality score and attack time quality score; Step S1 involves collecting the burner temperature throughout the entire attack time. Using all available data, construct an ideal temperature control curve for the burner. The horizontal axis of the curve represents the attack time t3, and the vertical axis represents the standard ideal temperature value of the fire scene at the corresponding moment. Weighting coefficients are set for the data collected by each temperature sensor, and a burner temperature threshold is also established. The calibration of the burner temperature control reference parameters is completed. The weighting coefficient is set according to the correlation between the sensor's position on the burner and the core combustion area. The higher the correlation, the larger the weighting coefficient. The full data of the burner temperature change over time is collected to determine the temperature critical value that can stably determine the effective extinguishing effect of the extinguishing agent on the burner. This critical value is set as the threshold of the burner temperature. Step S2 involves collecting ambient temperature data from sensors at different heights and locations around the fire site throughout the entire attack period. Using all available temperature change data, construct an ideal ambient temperature control curve. The horizontal axis of the curve represents the attack time. The time axis is represented by the vertical axis, which is the standard ideal temperature value of the fire scene environment at the corresponding moment. A weighting coefficient is set for the data collected by each ambient temperature sensor.

[0006] In a preferred embodiment, step S1 collects the attack time in each trial. Fire water usage Gas consumption The full sample data on fire water usage The full sample data was fitted with a normal distribution to complete the calibration of the burner temperature control baseline parameters. Specifically: Calculate fire water usage The arithmetic mean of the full sample data, and used as the fire water usage. The expected value of the corresponding normal distribution Then calculate the fire water usage. With mathematical expectation The arithmetic mean of the squares of the deviations is used as the fire water usage. The corresponding normal distribution variance ; Then the gas consumption Perform a normal distribution fitting operation to obtain the gas consumption. The expected value of the corresponding normal distribution variance of normal distribution .

[0007] In a preferred embodiment, step S1 is based on the calibrated and obtained fire extinguishing training time characteristic reference parameters, burner temperature control reference parameters, ambient temperature control reference parameters, and includes fire water usage. Gas consumption The resource consumption characteristic benchmark parameters are used to set the corresponding correlation coefficients. , , , and ; And based on the entry time of the fire extinguishing training time characteristic benchmark parameters and attack time Set the corresponding correlation coefficient , All correlation coefficients range from 0 to 1, and the arithmetic sum of the correlation coefficients at the same level is 1.

[0008] In a preferred embodiment, step S2 calculates the real-time weighted average temperature of the burner based on the temperature data collected in real time from all temperature sensors on the burner, and according to the weighting coefficient of the data collected by each temperature sensor calibrated in step S1. The step also includes the following specific settings: When the real-time weighted average temperature of the burner remains within the preset stable temperature range for a preset period of time, a fire extinguishing command is issued. From the moment the fire extinguishing command is issued, the real-time weighted average temperature of the burner is continuously monitored. When a continuous downward trend in the real-time weighted average temperature of the burner is observed, and the decrease reaches a preset effective decrease threshold, the burner temperature is determined to have started to decrease. The time interval between the moment the fire extinguishing command is issued and the moment the burner temperature begins to decrease is recorded, and this time interval is determined as the entry time. Meanwhile, monitoring entry time If, during the process, the timed duration reaches the maximum allowable duration for fire extinguishing training and it is still not determined that the burner temperature has started to drop, then the fire extinguishing training will be directly deemed a failure. The real-time weighted average temperature of the burner is continuously collected at a fixed sampling period, and the real-time weighted average temperature of the burner at each sampling moment is compared with the burner temperature threshold. The comparison revealed that the real-time weighted average temperature of the burner at the current sampling time was lower than the burner temperature threshold. At that time, the sampling period at the current sampling moment is determined as the effective fire extinguishing period. The total duration of the accumulated effective fire extinguishing periods is used to obtain the cumulative effective fire extinguishing time. The cumulative effective fire extinguishing time is compared with the preset effective stable fire extinguishing time. If the cumulative effective fire extinguishing time is greater than the preset effective stable fire extinguishing time, the fire is determined to be successfully extinguished. The time interval from the moment the burner temperature begins to drop to the moment the fire is successfully extinguished is recorded and determined as the attack time. The firefighting training was divided into time sequences, and the attack time was extracted. Entry Time The time characteristic value.

[0009] In a preferred embodiment, step S2 involves dividing the time sequence of fire extinguishing training and extracting time feature values ​​into the time sequence. With attack time Then, based on the entry time Normal distribution mathematical expectation variance of normal distribution , construct entry time probability density function ,include: in, To enter time, For entry time The corresponding normal distribution variance For entry time The corresponding normal distribution expectation exists. The probability density at time is: The final entry time quality score is: And attack time The density function is: in, For the attack time, Time for attack The corresponding normal distribution variance, For attack time The corresponding normal distribution expectation is obtained. The probability density at time is: The final entry time quality score is: ; Based on the entry time quality score and the attack time quality score, and combined with the entry time determined in step S1. and attack time Correlation coefficient Achieve high scores in firefighting training time and quality. : in, To achieve time-quality performance, For the quality of attack time, , The correlation coefficient.

[0010] In a preferred embodiment, step S2 will determine the attack time. At each sampling moment, the measured temperature values ​​of all temperature sensors on the burner are collected. Based on the weighting coefficients set for each temperature sensor's data, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average measured temperature of the burner at each sampling moment. This is then used to calculate the attack time. The measured weighted average temperature of the burner at all sampling times is arranged in chronological order to construct a sequence of measured weighted average temperatures of the burner. Compare each sampling time in the measured weighted average temperature sequence of the burner with the ideal temperature control curve of the burner. The burner temperature control deviation value at each sampling time is obtained by subtracting the corresponding ideal temperature value. ; attack time Burner temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total temperature control deviation values, then divide the total temperature control deviation values ​​by the attack time. The duration of the time is used to obtain the burner temperature control quality score. .

[0011] In a preferred embodiment, step S2 relates to the attack time. At each sampling moment within the fire scene, the measured temperature values ​​of all environmental temperature sensors deployed within the fire scene are collected. Based on the weighting coefficients of all environmental temperature sensors, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average environmental temperature of the fire scene at the corresponding sampling moment. This is then used to calculate the attack time. The measured weighted average ambient temperature of the fire site at all sampling times is arranged in chronological order to construct a sequence of measured weighted average ambient temperatures of the fire site. The measured weighted average ambient temperature at each sampling time in the fire scene measurement weighted average ambient temperature sequence is compared with the ideal ambient temperature control curve. The difference between the corresponding ideal ambient temperature values ​​is used to obtain the ambient temperature control deviation values ​​for all sampling times. Time of attack Ambient temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total environmental temperature control deviation values, and divide by the attack time. The duration of the environmental temperature control results in quality assessment. .

[0012] In a preferred embodiment, step S2 is based on gas consumption. Normal distribution mathematical expectation and normal distribution variance Build gas consumption probability density function Then calculate the gas consumption. equal to mathematical expectation baseline probability density And based on the baseline probability density and probability density function Obtain gas consumption quality scores And determined by fire water usage Normal distribution mathematical expectation and mathematical expectation Build fire water usage probability density function and baseline probability density Ultimately, the quality score for gas consumption was obtained. .

[0013] In a preferred embodiment, step S3 obtains a comprehensive evaluation result based on a weighted fusion formula, including: ; in, To assess the quality and duration of firefighting training, For the quality performance of burner temperature control, For the quality performance of ambient temperature control, For the quality performance of fire water use, For the quality performance of gas consumption, , , , and The correlation coefficient.

[0014] The beneficial effects of this invention are: it integrates the characteristic parameters of five core dimensions—firefighting training time, burner temperature control, ambient temperature control, fire water usage, and gas consumption—to construct a multi-dimensional evaluation system. By constructing quality scores in the time dimension through hierarchical weighting, it achieves a refined breakdown of capabilities at different stages of firefighting operations. Through the weighted fusion of scores from multiple dimensions, it completes the standardized output and level classification of comprehensive evaluation results. This allows for both horizontal comparison and grading of trainees' capabilities through comprehensive scores and precise identification of individual capability shortcomings through scores from each sub-item, providing a quantitative basis for training review and targeted improvement of training programs. Attached Figure Description

[0015] Figure 1 This is a logical schematic diagram of the present invention; Figure 2 This is a schematic diagram of the configuration of the live fire training chamber in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] As attached Figure 1-2 As shown, this embodiment provides a multi-information fusion method for firefighting training and evaluation, including the following steps: Step S1: Based on the fire extinguishing training test, calibrate the benchmark parameters of the fire extinguishing training time, collect the full data of the burner temperature change over time, complete the calibration of the burner temperature control benchmark parameters, calibrate the ambient temperature control benchmark parameters according to the ambient temperature change, calibrate the resource consumption characteristic benchmark parameters based on the fire water usage and gas consumption, and evaluate the weighted correlation coefficient after completing the calibration operation. Step S2: Based on the burner temperature control reference parameters, fire extinguishing training time characteristic reference parameters, and ambient temperature control reference parameters calibrated in Step S1, the fire extinguishing training is divided into time sequences, and time characteristic values ​​are extracted. Based on the extracted time characteristic values, the fire extinguishing training time quality score, burner temperature control quality score, ambient temperature control quality score, fire water usage quality score, and gas consumption quality score are calculated. Step S3: Based on the obtained quality scores, perform weighted fusion to obtain a standardized output of the comprehensive evaluation results, and classify the corresponding level.

[0018] Step S1: Obtain the entry time for each test based on the fire extinguishing training experiment. With attack time Full sample data, for entry time Perform a normal distribution fitting operation on the full sample data, calculate the arithmetic mean of the full sample data, and use the obtained arithmetic mean as the entry time. The expected value of the corresponding normal distribution Then calculate the full sample data and the expected value. The arithmetic mean of the squares of the deviations is used as the entry time. The corresponding normal distribution variance ; Then regarding the attack time Perform a normal distribution fitting operation to obtain the attack time. The expected value of the corresponding normal distribution variance of normal distribution Meanwhile, based on the safety regulations and operational requirements corresponding to the training subjects, and by setting the maximum allowable duration for fire extinguishing training, the benchmark parameters for the time characteristics of fire extinguishing training are calibrated. This duration is the maximum time threshold from the issuance of the fire extinguishing command to the determination of fire extinguishing failure. Therefore, this application obtains the normal distribution parameters of the time characteristics that conform to the actual operation level of the training subject through large-sample experiments, and provides an evaluation benchmark for the subsequent calculation of entry time quality score and attack time quality score; Step S1: Collect burner temperature throughout the attack time. Using all available data, construct an ideal temperature control curve for the burner. The horizontal axis of the curve represents the attack time t3, and the vertical axis represents the standard ideal temperature value of the fire scene at the corresponding moment. Weighting coefficients are set for the data collected by each temperature sensor, and a burner temperature threshold is also established. The calibration of the burner temperature control reference parameters is completed. The weighting coefficient is set according to the correlation between the sensor's position on the burner and the core combustion area. The higher the correlation, the larger the weighting coefficient. The full data of the burner temperature change over time is collected to determine the temperature critical value that can stably determine the effective extinguishing effect of the extinguishing agent on the burner. This critical value is set as the threshold of the burner temperature. Step S2 collects ambient temperature data from sensors at different heights and locations around the fire site throughout the entire attack period. Using all available temperature change data, construct an ideal ambient temperature control curve. The horizontal axis of the curve represents the attack time. The time axis is represented by the vertical axis, which is the standard ideal temperature value of the fire scene environment at the corresponding moment. A weighting coefficient is set for the data collected by each ambient temperature sensor.

[0019] Step S1: Collect the attack time in each trial. Fire water usage Gas consumption The full sample data on fire water usage The full sample data was fitted with a normal distribution to complete the calibration of the burner temperature control baseline parameters. Specifically: Calculate fire water usage The arithmetic mean of the full sample data, and used as the fire water usage. The expected value of the corresponding normal distribution Then calculate the fire water usage. With mathematical expectation The arithmetic mean of the squares of the deviations is used as the fire water usage. The corresponding normal distribution variance ; Then the gas consumption Perform a normal distribution fitting operation to obtain the gas consumption. The expected value of the corresponding normal distribution variance of normal distribution .

[0020] Step S1 is based on the calibrated and obtained fire extinguishing training time characteristic reference parameters, burner temperature control reference parameters, ambient temperature control reference parameters, and fire water usage data. Gas consumption The resource consumption characteristic benchmark parameters are used to set the corresponding correlation coefficients. , , , and ; And based on the entry time of the fire extinguishing training time characteristic benchmark parameters and attack time Set the corresponding correlation coefficient , All correlation coefficients range from 0 to 1, and the arithmetic sum of the correlation coefficients at the same level is 1.

[0021] Step S2 calculates the real-time weighted average temperature of the burner based on the temperature data collected in real time from all temperature sensors on the burner and according to the weighting coefficients of the data collected by each temperature sensor as calibrated in step S1. It also includes the following specific settings: When the real-time weighted average temperature of the burner remains within the preset stable temperature range for a preset period of time, a fire extinguishing command is issued. From the moment the fire extinguishing command is issued, the real-time weighted average temperature of the burner is continuously monitored. When a continuous downward trend in the real-time weighted average temperature of the burner is observed, and the decrease reaches a preset effective decrease threshold, the burner temperature is determined to have started to decrease. The time interval between the moment the fire extinguishing command is issued and the moment the burner temperature begins to decrease is recorded, and this time interval is determined as the entry time. Meanwhile, monitoring entry time If, during the process, the timed duration reaches the maximum allowable duration for fire extinguishing training and it is still not determined that the burner temperature has started to drop, then the fire extinguishing training will be directly deemed a failure. The real-time weighted average temperature of the burner is continuously collected at a fixed sampling period, and the real-time weighted average temperature of the burner at each sampling moment is compared with the burner temperature threshold. The comparison revealed that the real-time weighted average temperature of the burner at the current sampling time was lower than the burner temperature threshold. At that time, the sampling period at the current sampling moment is determined as the effective fire extinguishing period. The total duration of the accumulated effective fire extinguishing periods is used to obtain the cumulative effective fire extinguishing time. The cumulative effective fire extinguishing time is compared with the preset effective stable fire extinguishing time. If the cumulative effective fire extinguishing time is greater than the preset effective stable fire extinguishing time, the fire is determined to be successfully extinguished. The time interval from the moment the burner temperature begins to drop to the moment the fire is successfully extinguished is recorded and determined as the attack time. The firefighting training was divided into time sequences, and the attack time was extracted. Entry Time The time characteristic value.

[0022] In some other specific implementations, during the monitoring of attack time During the process, if the total time elapsed from the moment the fire extinguishing command is issued reaches the maximum allowed time for fire extinguishing training, but the accumulated effective fire extinguishing time still does not reach the preset effective and stable fire extinguishing time, then the fire extinguishing training will be directly judged as a failure and the evaluation process will be terminated. Step S2 involves dividing the fire extinguishing training into time sequences and extracting time feature values ​​for input. With attack time Then, based on the entry time Normal distribution mathematical expectation variance of normal distribution , construct entry time probability density function ,include: in, To enter time, For entry time The corresponding normal distribution variance For entry time The corresponding normal distribution expectation exists. The probability density at time is: The final entry time quality score is: And attack time The density function is: in, For the attack time, Time for attack The corresponding normal distribution variance, For attack time The corresponding normal distribution expectation is obtained. The probability density at time is: The final entry time quality score is: ; Based on the quality scores of entry time and attack time, and combined with the entry time determined in step S1 and attack time Correlation coefficient Achieve high scores in firefighting training time and quality. : in, To achieve time-quality performance, For the quality of attack time, , The correlation coefficient.

[0023] Step S2 will determine the attack time. At each sampling moment, the measured temperature values ​​of all temperature sensors on the burner are collected. Based on the weighting coefficients set for each temperature sensor's data, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average measured temperature of the burner at each sampling moment. This is then used to calculate the attack time. The measured weighted average temperature of the burner at all sampling times is arranged in chronological order to construct a sequence of measured weighted average temperatures of the burner. Compare each sampling time in the measured weighted average temperature sequence of the burner with the ideal temperature control curve of the burner. The burner temperature control deviation value at each sampling time is obtained by subtracting the corresponding ideal temperature value. ; attack time Burner temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total temperature control deviation values, then divide the total temperature control deviation values ​​by the attack time. The duration of the time is used to obtain the burner temperature control quality score. .

[0024] Step S2: Attack Time At each sampling moment within the fire scene, the measured temperature values ​​of all environmental temperature sensors deployed within the fire scene are collected. Based on the weighting coefficients of all environmental temperature sensors, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average environmental temperature of the fire scene at the corresponding sampling moment. This is then used to calculate the attack time. The measured weighted average ambient temperature of the fire site at all sampling times is arranged in chronological order to construct a sequence of measured weighted average ambient temperatures of the fire site. The measured weighted average ambient temperature at each sampling time in the fire scene measurement weighted average ambient temperature sequence is compared with the ideal ambient temperature control curve. The difference between the corresponding ideal ambient temperature values ​​is used to obtain the ambient temperature control deviation values ​​for all sampling times. Time of attack Ambient temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total environmental temperature control deviation values, and divide by the attack time. The duration of the environmental temperature control results in quality assessment. .

[0025] Step S2 is based on gas consumption. Normal distribution mathematical expectation and normal distribution variance Build gas consumption probability density function Then calculate the gas consumption. equal to mathematical expectation baseline probability density And based on the baseline probability density and probability density function Obtain gas consumption quality scores And determined by fire water usage Normal distribution mathematical expectation and mathematical expectation Build fire water usage probability density function and baseline probability density Ultimately, the quality score for gas consumption was obtained. .

[0026] Step S3 obtains the comprehensive evaluation result based on the weighted fusion formula, including: ; in, To assess the quality and duration of firefighting training, For the quality performance of burner temperature control, For the quality performance of ambient temperature control, For the quality performance of fire water use, For the quality performance of gas consumption, , , , and The correlation coefficient.

[0027] For some other specific implementations, please refer to the appendix. Figure 2 This application constructs a real fire training facility. A real fire burner 4 is installed in compartment 1, and three thermocouples 5 are arranged in the real fire burner 4. A thermocouple 6 for measuring ambient temperature is installed on the bulkhead of compartment 1 at a distance of 1.5 meters from the deck. Fire hoses 7 are used to carry out fire extinguishing training. The amount of fire water used is measured by flow meter 2, and the amount of gas consumed is measured by flow meter 3.

[0028] The flowchart for evaluating the performance of live fire training is attached. Figure 1 As shown. After the live-fire fire drill begins, the burner is ignited. When the average temperature of the three thermocouples inside the burner reaches the preset value, indicating that the fire has stabilized, the command to begin extinguishing the fire is issued to the trainee firefighters. When the firefighters spray fire-fighting water onto the live-fire burner, the temperature of the thermocouples inside the burner will drop significantly. The time from issuing the extinguishing command to the temperature drop of the burner is called the entry time. This demonstrates the firefighters' reaction speed.

[0029] After the burner temperature begins to drop, the average temperature of three thermocouples is taken and compared with a set threshold (obtained experimentally). If the temperature is lower than the threshold, the extinguishing agent spraying is considered effective. The sampling period of the information acquisition system is... Then the cumulative effective fire extinguishing time is n. .

[0030] A fire extinguishing command is issued, and a timer begins. If this time exceeds the set maximum training time, the fire extinguishing fails. When n... The fire is considered successfully extinguished if the preset effective extinguishing time is reached but the maximum training time is not exceeded. The attack time is the time from when the burner temperature begins to drop until the fire is successfully extinguished. .

[0031] The system then recorded the entry time. Attack time The sampling period is The burner temperature T and the cabin ambient temperature sequence, fire water usage Gas consumption .

[0032] Therefore, this application further discloses the relevant evaluation process as follows: (1) Calculate the quality score of training time Training time quality score is determined by entry time and attack time The eigenvalues ​​are obtained through calculation, i.e. in, To achieve time-quality performance, For the quality of attack time, , The correlation coefficient.

[0033] 1) Entry time quality score and entry time Regarding the entry time It follows a normal distribution, and its expected value (mean) is... and variance Obtained through experiments.

[0034] but The density function is: get The probability density at time is: The quality score for entry time can be obtained as follows: 2) Offensive time quality score and offensive time Regarding the entry time It follows a normal distribution, and its expected value (mean) is... and variance Obtained through experiments.

[0035] but The density function is: get The probability density at time is: The quality score for entry time can be obtained as follows: (2) Calculate the burner temperature control quality score The burner temperature is taken as the average temperature of three thermocouples, and the burner temperature threshold is... Ideal temperature control curve Obtained experimentally. During fire drills, the burner temperature was recorded as being below [a certain value]. time n This refers to the effective fire extinguishing time. When the effective time accumulates to a set value, the fire is considered successfully extinguished. The actual fire extinguishing time is the attack time. .

[0036] Attack time Inside, the burner temperature is compared with the ideal temperature control curve at the sampling time. The difference is , This represents the current burner temperature. The burner temperature control quality score is given by the following formula: , =1, 2, 3... in, This represents the number of sampling points; (3) Calculate the environmental temperature control quality score In this example, the ambient temperature is 1, and the ideal ambient temperature control curve is shown. Obtained through experiments. Attack time. Inside, at the sampling time, the ambient temperature is compared with the ideal ambient temperature control curve. The difference is , The ambient temperature is the current value. The quality score for ambient temperature control is given by the following formula: , =1, 2, 3... (4) Calculate the quality score of fire water use Fire water usage is obtained from flow meters installed on the fire water pipeline, and the attack time is also considered. Internal fire water usage It follows a normal distribution, and its expected value (mean) is... and variance Obtained through experiments.

[0037] but The density function is: get The probability density at time is: The quality score for fire water usage is as follows: (5) Calculate the quality score of gas consumption Gas consumption is obtained from a flow meter installed on the gas pipeline, attack time Internal fuel consumption It follows a normal distribution, and its expected value (mean) is... and variance Obtained through experiments.

[0038] but The density function is: get The probability density at time is: The quality score for gas consumption can be obtained as follows: (6) Calculate the fire extinguishing training assessment score The formula for calculating fire extinguishing training results is: Among them, the quality score of fire extinguishing training time Burner temperature control quality performance Ambient temperature control quality performance Fire water usage quality performance Gas consumption quality performance The correlation coefficient is obtained from calculation steps (1) to (5). , , , , Acquired through experience.

[0039] Furthermore, regarding the correlation coefficient proposed in this application... , , , , Even with prior experience, other techniques can still be used to provide guidance, such as selecting N sets of complete and effective real-world datasets for firefighting training, including training time, quality, and performance scores. Burner temperature control quality performance Ambient temperature control quality performance Fire water usage quality performance Gas consumption quality performance A five-dimensional performance matrix is ​​established for N sets of samples. For the i-th dimension, the maximum and minimum values ​​of all N sets of samples in that dimension are extracted. The original values ​​of the i-th dimension are standardized using the extreme value standardization formula. The arithmetic mean and standard deviation of the standardized data in the i-th dimension are calculated to obtain the standard deviation. The coefficient of variation of the i-th dimension is then calculated. The information independence quantification value of each dimension is calculated using the Pearson correlation coefficient matrix. The contribution quantification value of the fire extinguishing results of each dimension is then calculated, and the correlation coefficient is obtained after normalization.

Claims

1. A multi-information fusion method for firefighting training assessment, characterized in that, Includes the following steps: Step S1: Based on the fire extinguishing training test, calibrate the benchmark parameters of the fire extinguishing training time, collect the full data of the burner temperature change over time, complete the calibration of the burner temperature control benchmark parameters, calibrate the ambient temperature control benchmark parameters according to the ambient temperature change, calibrate the resource consumption characteristic benchmark parameters based on the fire water usage and gas consumption, and evaluate the weighted correlation coefficient after completing the calibration operation. Step S2: Based on the burner temperature control reference parameters, fire extinguishing training time characteristic reference parameters, and ambient temperature control reference parameters calibrated in Step S1, the fire extinguishing training is divided into time sequences, and time characteristic values ​​are extracted. Based on the extracted time characteristic values, the fire extinguishing training time quality score, burner temperature control quality score, ambient temperature control quality score, fire water usage quality score, and gas consumption quality score are calculated. Step S3: Based on the obtained quality scores, perform weighted fusion to obtain a standardized output of the comprehensive evaluation results, and classify the corresponding level.

2. The multi-information fusion method for firefighting training and evaluation according to claim 1, characterized in that, Step S1 obtains the entry time for each test based on the fire extinguishing training test. With attack time Full sample data, for entry time Perform a normal distribution fitting operation on the full sample data, calculate the arithmetic mean of the full sample data, and use the obtained arithmetic mean as the entry time. The corresponding normal distribution mathematical expectation Then calculate the full sample data and the expected value. The arithmetic mean of the squares of the deviations is used as the entry time. The corresponding normal distribution variance ; Then regarding the attack time Perform a normal distribution fitting operation to obtain the attack time. The corresponding normal distribution mathematical expectation variance of normal distribution And set the maximum allowable duration for fire extinguishing training, and complete the calibration of the benchmark parameters for the time characteristics of fire extinguishing training; Step S1 involves collecting the burner temperature throughout the entire attack time. Using all available data, construct an ideal temperature control curve for the burner. Set the weighting coefficient for the data collected by each temperature sensor and set the burner temperature threshold. Complete the calibration of the burner temperature control reference parameters; Step S2 involves collecting ambient temperature data from sensors at different heights and locations around the fire site throughout the entire attack period. Using complete temperature change data, construct an ideal ambient temperature control curve. Set the weighting coefficient for the data collected by each ambient temperature sensor.

3. The multi-information fusion method for firefighting training and evaluation according to claim 2, characterized in that, Step S1 involves collecting the attack time in each trial. Fire water usage Gas consumption The full sample data on fire water usage The full sample data was fitted with a normal distribution to complete the calibration of the burner temperature control baseline parameters. Specifically: Calculate fire water usage The arithmetic mean of the full sample data, and used as the fire water usage. The corresponding normal distribution mathematical expectation Then calculate the fire water usage. With mathematical expectation The arithmetic mean of the squares of the deviations is used as the fire water usage. The corresponding normal distribution variance ; Then the gas consumption Perform a normal distribution fitting operation to obtain the gas consumption. The corresponding normal distribution mathematical expectation variance of normal distribution .

4. The fire extinguishing training evaluation method based on multi-information fusion according to claim 1, characterized in that, Step S1 is based on the calibrated and obtained fire extinguishing training time characteristic reference parameters, burner temperature control reference parameters, ambient temperature control reference parameters, and fire water usage. Gas consumption The resource consumption characteristic benchmark parameters are used to set the corresponding correlation coefficients. , , , and ; And based on the entry time of the fire extinguishing training time characteristic benchmark parameters and attack time Set the corresponding correlation coefficient , .

5. The fire extinguishing training evaluation method based on multi-information fusion according to claim 2, characterized in that, Step S2, based on the real-time temperature data collected from all temperature sensors on the burner and according to the weighting coefficients of the data collected by each temperature sensor calibrated in step S1, calculates the real-time weighted average temperature of the burner, and also includes the following specific settings: When the real-time weighted average temperature of the burner remains within the preset stable temperature range for a preset period of time, a fire extinguishing command is issued. From the moment the fire extinguishing command is issued, the real-time weighted average temperature of the burner is continuously monitored. When a continuous downward trend in the real-time weighted average temperature of the burner is observed, and the decrease reaches a preset effective decrease threshold, the burner temperature is determined to have started to decrease. The time interval between the moment the fire extinguishing command is issued and the moment the burner temperature begins to decrease is recorded, and this time interval is determined as the entry time. ; The real-time weighted average temperature of the burner is continuously collected at a fixed sampling period, and the real-time weighted average temperature of the burner at each sampling moment is compared with the burner temperature threshold. The comparison revealed that the real-time weighted average temperature of the burner at the current sampling time was lower than the burner temperature threshold. At that time, the sampling period at the current sampling moment is determined as the effective fire extinguishing period. The total duration of the accumulated effective fire extinguishing periods is used to obtain the cumulative effective fire extinguishing time. The cumulative effective fire extinguishing time is compared with the preset effective stable fire extinguishing time. If the cumulative effective fire extinguishing time is greater than the preset effective stable fire extinguishing time, the fire is determined to be successfully extinguished. The time interval from the moment the burner temperature begins to drop to the moment the fire is successfully extinguished is recorded and determined as the attack time. The firefighting training was divided into time sequences, and the attack time was extracted. Entry Time The time characteristic value.

6. The multi-information fusion method for firefighting training assessment according to claim 1, characterized in that, Step S2 involves dividing the fire extinguishing training into time sequences and extracting time feature values ​​for input. With attack time Then, based on the entry time Normal distribution mathematical expectation variance of normal distribution , construct entry time probability density function ,include: in, To enter time, For entry time The corresponding normal distribution variance For entry time The corresponding normal distribution expectation exists. The probability density at time is: The final entry time quality score is: And attack time The density function is: in, For the attack time, Time for attack The corresponding normal distribution variance, For attack time The corresponding normal distribution expectation is obtained. The probability density at time is: The final entry time quality score is: ； Based on the entry time quality score and the attack time quality score, and combined with the entry time determined in step S1. and attack time Correlation coefficient Achieve high scores in firefighting training time and quality. : in, To achieve time-quality performance, For the quality of attack time, , The correlation coefficient.

7. The fire extinguishing training evaluation method based on multi-information fusion according to claim 1, characterized in that, Step S2 will occur during the attack time. At each sampling moment, the measured temperature values ​​of all temperature sensors on the burner are collected. Based on the weighting coefficients set for each temperature sensor's data, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average measured temperature of the burner at each sampling moment. This is then used to calculate the attack time. The measured weighted average temperature of the burner at all sampling times is arranged in chronological order to construct a sequence of measured weighted average temperatures of the burner. Compare each sampling time in the measured weighted average temperature sequence of the burner with the ideal temperature control curve of the burner. The burner temperature control deviation value at each sampling time is obtained by subtracting the corresponding ideal temperature value. ; attack time Burner temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total temperature control deviation values, then divide the total temperature control deviation values ​​by the attack time. The duration of the time is used to obtain the burner temperature control quality score. .

8. The fire extinguishing training evaluation method based on multi-information fusion according to claim 2, characterized in that, Step S2 refers to the attack time. At each sampling moment within the fire scene, the measured temperature values ​​of all environmental temperature sensors deployed within the fire scene are collected. Based on the weighting coefficients of all environmental temperature sensors, the measured temperature values ​​of all sensors at the same sampling moment are weighted and summed to obtain the weighted average environmental temperature of the fire scene at the corresponding sampling moment. This is then used to calculate the attack time. The measured weighted average ambient temperature of the fire site at all sampling times is arranged in chronological order to construct a sequence of measured weighted average ambient temperatures of the fire site. The measured weighted average ambient temperature at each sampling time in the fire scene measurement weighted average ambient temperature sequence is compared with the ideal ambient temperature control curve. The difference between the corresponding ideal ambient temperature values ​​is used to obtain the ambient temperature control deviation values ​​for all sampling times. Time of attack Ambient temperature control deviation values ​​corresponding to all sampling times Perform an arithmetic summation to obtain the total environmental temperature control deviation values, and divide by the attack time. The duration of the environmental temperature control results in quality assessment. .

9. The multi-information fusion method for firefighting training and evaluation according to claim 1, characterized in that, Step S2 is based on gas consumption. Normal distribution mathematical expectation and normal distribution variance Build gas consumption probability density function Then calculate the gas consumption. equal to mathematical expectation baseline probability density And based on the baseline probability density and probability density function Obtain gas consumption quality scores And determined by fire water usage Normal distribution mathematical expectation and mathematical expectation Build fire water usage probability density function and baseline probability density Ultimately, the quality score for gas consumption was obtained. .

10. The multi-information fusion method for firefighting training and evaluation according to claim 1, characterized in that, Step S3, which obtains a comprehensive evaluation result based on a weighted fusion formula, includes: ； in, To assess the quality and duration of firefighting training, For the quality performance of burner temperature control, For the quality performance of ambient temperature control, For the quality performance of fire water use, For the quality performance of gas consumption, , , , and The correlation coefficient.

Images