An emergency breaking operation state intelligent early warning method

Abstract

The application provides an emergency breaking operation state intelligent early warning method, relates to the technical field of industrial information and data processing, and acquires scene basic data of a target emergency breaking task, classifies building collapse rescue breaking, confined space rescue breaking, high-risk facility rescue breaking and traffic obstacle removal breaking scenes, divides core breaking zones, structure response zones, personnel exposure zones, equipment support zones and evacuation buffer zones, forms quantile statistical benchmarks based on similar historical samples, calculates similar sample quantile deviation indexes, equipment object coupling response coefficients, multi-window instability trend degrees and perception credible correction degrees, and fuses to generate statistical early warning values of each partition operation state, and outputs partition early warning strategies according to early warning levels and strategy dominant items. The method improves the accuracy of multi-source industrial data integration processing, statistical early warning and differentiated disposal of the emergency breaking site.

Description

Technical Field

[0001] This invention relates to the field of industrial information and data processing technology, specifically to an intelligent early warning method for emergency demolition operations. Background Technology

[0002] Emergency demolition operations are widely used in industrial site rescue scenarios such as building collapse rescue, confined space rescue, high-risk facility emergency repair, and traffic obstruction removal. These operations involve multiple methods, including hydraulic shearing, impact demolition, and cutting demolition, and are accompanied by complex conditions such as component crack propagation, local displacement, structural vibration, dust and smoke obstruction, flammable gas risks, personnel approaching the core work area, and dynamic changes in evacuation routes. With the development of industrial information and data processing technologies, demolition equipment controllers, pressure sensors, crack displacement gauges, laser rangefinders, vibration sensors, gas sensors, personnel positioning tags, and edge computing terminals are increasingly being used for on-site data acquisition and status monitoring. Industrial data integration services are also used to aggregate equipment operation information, object response information, environmental disturbance information, and personnel exposure information, providing a data foundation for status assessment, information processing, and storage support services during emergency demolition processes.

[0003] Existing emergency demolition early warning methods largely rely on single-sensor alarms, manual experience judgment, or fixed threshold comparisons, making it difficult to adapt to data differences under various demolition scenarios, risk levels in different zones, and operating conditions of different equipment. While multi-source industrial data can be collected, the sampling frequency, timestamps, units, and validity of different sensors vary. There is a lack of unified time synchronization, anomaly screening, missing data completion, and normalization mechanisms for emergency demolition operations, making it difficult to form reusable statistical analysis datasets. Existing methods directly compare current pressure, vibration, dust, and personnel distance data with fixed thresholds, ignoring the statistical distribution differences of historical samples under the same scenario, zone, object, and equipment conditions, easily leading to false alarms and missed alarms. Existing early warning logic struggles to simultaneously evaluate the coupling relationship between equipment output and object response, the continuous instability trend of object response, and the impact of dust, smoke, noise, and visibility on the reliability of data collection, resulting in a lack of interpretable mapping between early warning levels and actual response actions. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent early warning method for emergency demolition operations to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent early warning of emergency demolition operation status, comprising the following steps: S1. Obtain the basic scenario data for the target emergency breaching mission. This data includes the type of object to be breached, the type of equipment used for breaching, the location of the breaching point, the on-site spatial conditions, the type of hazard source, the location of personnel, and the location of evacuation routes. Based on this data, classify the scenario and divide the target emergency breaching area into a core breaching zone, a structural response zone, a personnel exposure zone, an equipment protection zone, and an evacuation buffer zone. Generate a scenario partition index table. S2. According to the scenario partition index table, collect the operation data of the demolition equipment, the response data of the object being demolished, the on-site environmental disturbance data, the personnel exposure status data, and historical demolition sample data. Perform time synchronization, anomaly filtering, missing data completion, and normalization processing to generate an emergency demolition operation status dataset. S3. Based on the scene classification results, partitioning results, the type of the object being demolished, and the type of demolition equipment, perform similar sample filtering on the historical demolition sample data to generate a similar historical sample dataset. Based on the similar historical sample dataset, form a quantile statistical benchmark set. S4. Based on the emergency demolition operation status dataset and quantile statistical benchmark set, calculate the quantile deviation index of similar samples, the equipment-object coupling response coefficient, the multi-window instability trend degree, and the perceived reliability correction degree. The quantile deviation index of similar samples is used to characterize the degree of deviation of the current operation status from the statistical distribution of similar historical samples. The equipment-object coupling response coefficient is used to characterize the degree of coupling anomaly between the output of the demolition equipment and the response of the demolished object. The multi-window instability trend degree is used to characterize the instability trend of the response of the demolished object over time. The perceived reliability correction degree is used to characterize the reliability of the data collected on site under the influence of dust, smoke, noise, and visibility. S5. Based on the quantile deviation index of similar samples, the coupling response coefficient of equipment objects, the instability trend degree of multi-window and the perception credibility correction degree, calculate the statistical early warning value of the operation status corresponding to each partition, generate the regular monitoring level, the key attention level or the linkage disposal level according to the statistical early warning value of the operation status, and generate the scenario partition early warning strategy based on the scenario classification results, partition results, partition early warning level and strategy dominant item.

[0006] Compared with the prior art, the beneficial effects of the present invention are: This method integrates equipment operation data, response data of the objects being breached, environmental disturbance data, personnel exposure status data, and historical breach sample data from the emergency demolition site into a unified scenario partitioning index table for processing. Furthermore, it generates an emergency demolition operation status dataset through time synchronization, anomaly filtering, missing data completion, and normalization. This transforms dispersed on-site sensor data into statistically comparable, storable, and reusable industrial information processing data, improving data consistency and computational reliability in emergency demolition scenarios for industrial data integration services.

[0007] This method filters similar historical demolition samples based on scene classification results, zoning results, the type of object being demolished, and the type of demolition equipment, forming a quantile statistical benchmark set. The quantile deviation index of similar samples is used to evaluate the degree of deviation of the current operational status from the statistical distribution of similar historical samples. This avoids the limitations of relying solely on fixed thresholds for alarms, making early warning judgments more statistically adaptable to different scenarios, zones, objects, and equipment conditions, and reducing the risk of false alarms caused by fluctuations in normal operations.

[0008] This method simultaneously calculates the coupling response coefficient of the equipment and the object, the multi-window instability trend, and the perceived reliability correction. It incorporates factors such as mismatch between equipment output and object response, continuous increase in object response, and low data reliability caused by dust, smoke, noise, and visibility into a unified statistical early warning value. Based on quantile anomalies, coupling anomalies, trend anomalies, and low-reliability anomalies, it determines the dominant strategy term. This allows for the transformation of early warning levels into differentiated handling strategies such as load reduction, contact point verification, adjustment of demolition angle, segmented demolition, ventilation and smoke extraction, redundant data acquisition, remote control, and expansion of personnel boundaries, improving the interpretability, executability, and efficiency of on-site verification of early warning results. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the overall technical blueprint of the present invention. Figure 2 This is a flowchart illustrating the scene classification and region division process of this invention. Figure 3 This is a flowchart of the multi-source data acquisition and preprocessing process of this invention. Figure 4 This is a flowchart illustrating the process of forming historical samples and quantile benchmarks similar to those used in this invention. Figure 5 This is a flowchart illustrating the calculation process for the four types of statistical indicators of this invention. Figure 6 This is a flowchart of the partitioned early warning level and strategy output of the present invention. Detailed Implementation

[0010] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0011] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below. Example

[0012] Figure 1It shows the spatial relationships of the emergency demolition operation site, demolition equipment, the object to be demolished, personnel, sensor acquisition units, edge computing terminals, as well as the core demolition area, structural response area, personnel exposure area, equipment protection area, and evacuation buffer zone. Figures 2 to 6 The process is shown in detail, including scene classification and regional division, multi-source data collection and preprocessing, formation of similar historical samples and quantile benchmarks, calculation of four types of statistical indicators, and output of regional early warning levels and strategies.

[0013] Please see Figures 1 to 6 The present invention provides a technical solution: This embodiment illustrates the overall execution flow of an intelligent early warning method for emergency demolition operations, applicable to emergency demolition scenarios such as building collapse rescue demolition, confined space rescue demolition, high-risk facility emergency demolition, and traffic obstacle removal demolition. The on-site configuration includes hydraulic pressure sensors, equipment controllers, triaxial vibration sensors, crack displacement gauges, laser rangefinders, tilt sensors, dust sensors, smoke sensors, combustible gas sensors, noise sensors, visibility detection devices, personnel positioning tags, and edge computing terminals. Specifically, it includes... S1. Obtain the basic scenario data for the target emergency breaching mission. This data includes the type of object to be breached, the type of breaching equipment, the location of the breaching point, the on-site spatial conditions, the type of hazard source, the location of personnel, and the location of evacuation routes. Based on this basic scenario data, classify the target emergency breaching mission into scenarios and divide the target emergency breaching area into a core breaching zone, a structural response zone, a personnel exposure zone, an equipment protection zone, and an evacuation buffer zone, generating a scenario partition index table.

[0014] S2. Collect demolition equipment operation data, demolition target response data, on-site environmental disturbance data, personnel exposure status data, and historical demolition sample data according to the scenario partition index table. Perform time synchronization, anomaly filtering, missing data completion, and normalization on the collected data to generate an emergency demolition operation status dataset.

[0015] S3. Based on the scene classification results, partitioning results, the type of the object being demolished, and the type of the demolition equipment, the historical demolition sample data is filtered to generate a similar historical sample dataset, and a quantile statistical benchmark set is formed based on the similar historical sample dataset.

[0016] S4. Based on the emergency demolition operation status dataset and quantile statistical benchmark set, calculate the quantile deviation index of similar samples, the coupling response coefficient of equipment and object, the multi-window instability trend degree and the perceived credibility correction degree, respectively. These are used to characterize the degree of deviation of the current operation status from similar historical samples, the degree of coupling anomaly between the output of the demolition equipment and the response of the demolished object, the instability trend of the response of the demolished object over time, and the credibility of the data collected on site under the influence of dust, smoke, noise and visibility.

[0017] S5. Based on the quantile deviation index of similar samples, the coupling response coefficient of equipment objects, the instability trend degree of multi-window and the perception credibility correction degree, calculate the statistical early warning value of the operation status corresponding to each partition, generate the regular monitoring level, the key attention level or the linkage disposal level according to the statistical early warning value of the operation status, and generate the scenario partition early warning strategy based on the scenario classification results, partition results, partition early warning level and strategy dominant item.

[0018] This embodiment sequentially completes scene classification and partitioning, data collection and processing, screening of similar samples, calculation of statistical indicators, and output of partition early warning strategies. The output results of the previous stage are used as the basis for processing in the next stage.

[0019] Example 2: This embodiment is a detailed description of step S1 in Embodiment 1.

[0020] Let the set of scenario elements for the current emergency demolition task be denoted as... Scene element collection The data was extracted from on-site investigation records, 3D scanning results, demolition operation plans, personnel positioning data, and hazard identification results. Scene elements included building collapse, underground space, utility tunnel space, storage tanks and pipelines, pressurized containers, road obstacles, vehicle crushing structures, restricted ventilation, flammable gases, and single evacuation route.

[0021] The first The set of preset scene elements is denoted as The set of pre-defined scenario elements is formed by summarizing samples of historical building collapse rescue and demolition, confined space rescue and demolition, high-risk facility emergency demolition, and traffic obstacle removal and demolition. This will be combined with the current target emergency demolition task... The scene matching degree between preset scenes is denoted as The Jaccard similarity is calculated using the following formula:

[0022] in, This indicates the current target emergency demolition task and the first The number of scene elements commonly included in a class of preset scenes. This indicates the current target emergency demolition task and the first The number of scene elements after merging preset scenes.

[0023] Because emergency demolition scenarios have complex characteristics, the scenario matching threshold is determined by a trade-off between the historical scenario sample misclassification rate, the manual review rate of complex scenarios, and the number of similar samples retained, with a preferred value of 0.55. When the value is not lower than 0.55, the current task is assigned to the corresponding scenario. When the match score is below 0.55, it is marked as a composite demolition scenario and enters manual review. When only one scenario matches the scene matching threshold, the current task is assigned to that scenario. When multiple scenarios match the scene matching threshold, the scenario with the highest match score is selected as the primary scenario, and the remaining scenarios that match the threshold are marked as related scenarios. When the difference between the highest and second-highest match scores does not exceed 0.05, the current task is marked as a composite demolition scenario and enters manual review. Building collapse rescue demolition scenarios mainly correspond to wall cracks, damaged beams and columns, tilted floors, and trapped personnel. Confined space rescue demolition scenarios mainly correspond to underground spaces, pipe gallery spaces, limited ventilation, and single passageways. High-risk facility emergency demolition scenarios mainly correspond to storage tanks and pipelines, flammable gases, pressurized containers, and high-temperature components. Traffic obstacle removal demolition scenarios mainly correspond to road blockages, vehicle crushing, bridge and tunnel obstacles, and traffic needs.

[0024] After completing the scenario classification, the target emergency demolition area is delineated. The effective radius of the demolition equipment is denoted as... The unit is meters (m), obtained from the nameplate parameters of the demolition equipment and the equipment control terminal. The safe correction distance for debris scattering is denoted as... The unit is meters (m), determined by the material type of the object being demolished, the demolition method, historical debris scattering records, and on-site shielding conditions. If historical records are insufficient, the safety correction distance range corresponding to the material type and demolition method is used. The on-site safety redundancy distance is denoted as... The position is m, determined based on the minimum station interval of rescue personnel, equipment rotation margin, and temporary evacuation margin. The radius of the core demolition zone is denoted as... The unit is m, and it is composed of

[0025] When similar historical debris scattering records exist, adjust the debris scattering safety distance. Determine using the following formula:

[0026] in, This indicates the distance of debris scattering in similar historical demolition samples, in units of... The 90th quantile, in meters. This represents the on-site occlusion correction factor, which is dimensionless. It is taken as 1.0 when there is no occlusion, 0.7 to 0.9 when reliable occlusion exists, and 1.1 to 1.3 when there is a rebound surface or narrow space. These represent the lower and upper limits of the values ​​for the corresponding material type and demolition method, respectively, both in meters (m). When historical records are insufficient, the safety correction distance is determined according to the safety correction distance value table formed by the material type and demolition method, as shown in Table 1 below.

[0027] Table 1: Value table

[0028] The structural response zone is the area outside the core breaching zone where displacement, tilting, crack propagation, vibration transmission, and localized fall responses may occur. The personnel exposure zone is the area where rescue personnel, equipment operators, and support staff enter and remain. The equipment protection zone is the area containing breaching equipment, power sources, control terminals, and backup equipment. The evacuation buffer zone is the passageway connecting the personnel exposure zone and safety exits. These zoning zones, together with the scenario classification results, form the spatial index basis for subsequent data collection, sample selection, and strategy output.

[0029] Example 3: This embodiment is a detailed description of step S2 in Embodiment 1.

[0030] Operational data of the demolition equipment was collected by hydraulic pressure sensors, equipment controllers, current transformers, temperature sensors, and triaxial vibration sensors. Response data of the object being demolished was collected by crack displacement gauges, laser rangefinders, tilt sensors, and triaxial vibration sensors. Environmental disturbance data was collected by dust sensors, smoke sensors, combustible gas sensors, noise sensors, and visibility detection devices. Personnel exposure data was collected by UWB personnel positioning tags and positioning base stations. Based on the scene partitioning index table generated in Example 2, this example categorizes data collection into four types: equipment operation, object response, environmental disturbance, and personnel exposure. The correspondence between the various types of collected data, the collecting equipment, and the units of measurement is shown in Table 2 below.

[0031] Table 2: Multi-source Acquisition Data Table

[0032]

[0033]

[0034]

[0035]

[0036] The time alignment values ​​and linear interpolation padding values ​​mentioned above maintain the same dimensions as the corresponding original collected data, and are robust anomaly screening values. All values ​​are dimensionless. Data with time alignment intervals or missing times exceeding two sampling periods, as well as fields with no effective span between reference upper and lower limits, are marked as data or fields to be reviewed and are not included in the calculation of current statistical indicators. This embodiment forms an emergency demolition operation status dataset based on this.

[0037] Example 4: This embodiment is a detailed description of step S3 in Embodiment 1.

[0038] The current emergency demolition task is the first Each sample feature is denoted as , will the The first historical demolition sample Each sample feature is denoted as The data is derived from the scene classification and partitioning results obtained in Example 2, and the equipment operation data, personnel exposure status data, and basic data of the objects being demolished obtained in Example 3. These data are derived from corresponding fields in historical demolition sample data. Sample features include scene type, area type, type of object being demolished, type of demolition equipment, structural dimensions, rated power of equipment, personnel density, hazard level, access restriction level, and structural damage level. Personnel density is calculated from the number of personnel in the current partition obtained from UWB personnel positioning tags and positioning base stations, as well as the current partition area in the scene partition index table. The number of people in each partition is recorded as follows: The unit is a person, and the first The area of ​​each partition is denoted as Units are , will the The population density of each zone is denoted as The unit is people The specific calculation formula is as follows:

[0039] Calculate the temporary difference values ​​for discrete features, continuous features, and ordered hierarchical features according to the sample feature type, and then map the corresponding temporary difference values ​​to the difference value of the q-th historical demolition sample on the r-th sample feature. .

[0040] For discrete features such as scene type, area type, type of object being demolished, and type of demolition equipment, the temporary difference value of the discrete features is denoted as... The consistency difference rule is used for calculation, and the specific calculation formula is as follows:

[0041] When they match, it means that the historical demolition sample is a perfect match for the current task in terms of that feature, and the difference value is 0. If there is an inconsistency, it indicates that the historical demolition sample does not match the current task in terms of this feature, and the difference value is taken as 1. This calculation result is a dimensionless value.

[0042] For continuous features such as structural dimensions, equipment rated power, and personnel density, structural dimensions were extracted from 3D scanning results, on-site distance measurement results, and construction drawings; equipment rated power was read from the demolition equipment nameplate parameters and the equipment control terminal. The first [data point] in similar historical sample data... The 95th quantile of a feature of a continuous sample is denoted as Both were calculated from historical demolition sample data. The first... The reference span of a continuous sample feature is denoted as The specific calculation formula is as follows:

[0043] Using the difference between the 95th quantile and the 5th quantile as a reference span can reduce the impact of extreme historical samples on the calculation of the difference value. The stability constant of a continuous sample feature is denoted as Unit and number Consistent characteristics of consecutive samples. Stability constant. It is not used as a sample matching feature in the comparison, but only in the first step. Reference span of features of a continuous sample Keep the denominator stable when it is too small or zero. The minimum distinguishable measure of a feature of a continuous sample is denoted as Determined by the corresponding sensor resolution, database storage accuracy, or minimum scale of engineering records. Stability constant. The preferred value is determined by the following formula:

[0044] The first The first historical demolition sample in the The temporary differences in continuous features across consecutive samples are denoted as follows: The specific calculation formula is as follows:

[0045] in, absolute value and To ensure consistent units, negative values ​​are avoided when historical sample feature values ​​are less than the current task feature value, by using the absolute value term. The constraint allows temporary differences in continuous features to be limited to between 0 and 1.

[0046] Based on the ordered hierarchical characteristics such as hazard level, access restriction level, and structural damage level, the current target emergency demolition task will be classified as follows: Each level number is denoted as , will the The first historical demolition sample Each level number is denoted as , will the The total number of levels of an ordered hierarchical feature is denoted as The temporary difference value of the ordered rank feature is denoted as The specific calculation formula is as follows:

[0047] The hazard level is determined based on the proportion of combustible gas concentration to the alarm concentration, the condition of pressurized containers, and the condition of high-temperature components. The access restriction level is determined based on the number of evacuation routes, the minimum passage width, and the proportion of obstructions. The structural damage level is determined based on the crack propagation rate, component displacement, tilt angle, and structural vibration amplitude.

[0048] After obtaining the corresponding temporary difference value, the uniform difference value of the q-th historical demolition sample on the r-th sample feature is denoted as... The corresponding temporary difference value is selected according to the sample feature type. The specific calculation formula is as follows:

[0049] These are temporary difference values ​​for discrete features, continuous features, and ordered hierarchical features, respectively; all three are dimensionless. Therefore, the unified difference value... It is also a dimensionless value, and its range is from 0 to 1. Through the above method, this embodiment avoids applying the same... Instead of repeatedly assigning values, it calculates temporary difference values ​​of different types, and obtains them uniformly according to the sample feature type. .

[0050] After calculating the uniform difference values ​​for all sample features, the sample matching degree of the q-th historical demolition sample is denoted as . The number of sample features participating in the matching is denoted as The sample matching degree is calculated using the average difference in reverse similarity. The specific calculation formula is as follows:

[0051] in, The closer it is to 1, the more similar the q-th historical demolition sample is to the current target emergency demolition task.

[0052] Because similar historical samples need to have high consistency in terms of scene, partition, object, and device, and the number of samples cannot be too small, the preferred value for the sample matching threshold is set to 0.72, determined by a trade-off between sample retention rate and baseline stability. When, the q-th historical demolition sample is included in the same historical sample dataset, when When the threshold is reached, the q-th historical breach sample is excluded. This threshold avoids using historical samples with excessively large differences to form a statistical benchmark.

[0053] After creating a dataset of similar historical samples, peak impact, crack propagation rate, dust concentration, and duration of personnel stay are extracted from these historical samples to form quantile statistical benchmarks. The data within the current statistical window is then used to determine the quantile baselines. The statistical characteristic value of the project is denoted as Target statistical characteristic value The valid collected data or valid normalized data obtained from Example 3 are processed through window statistical processing within the current statistical window. Window statistical processing includes at least one of peak value extraction, mean calculation, duration accumulation, rate of change calculation, or area / frequency statistics. For example, the impact peak value is obtained by extracting the peak value from the equipment output pressure or equipment vibration data within the current statistical window; the crack propagation rate is obtained by calculating the rate of change from the crack displacement gauge data within the current statistical window; the dust concentration is obtained by calculating the mean or peak value from the dust sensor data within the current statistical window; and the personnel dwell time is obtained by accumulating the duration from the personnel positioning data within the current statistical window. The same historical sample dataset is then used to calculate the peak value. The set of sample values ​​for the statistical characteristics of an objective is denoted as .

[0054] when When using original units, The same primitive units are also used. When When using normalized values, The same normalization method is also used. The same historical sample dataset is used in the [number]th [period]. The total number of samples under the statistical characteristics of the project is denoted as ,Will less than or equal to The number of samples is denoted as , will the The empirical quantile of the statistical characteristics of the project target is denoted as The empirical distribution function is used for calculation, and the specific calculation formula is as follows:

[0055] in, Indicates the first in a dataset of similar historical samples The total number of samples for the statistical characteristics of the project objectives. less than or equal to The number of samples. Both the numerator and denominator are the number of samples.

[0056] The first The number of target statistical features involved in the calculation for each partition is denoted as . The deviation of the quantile from the starting point is denoted as... Dimensionless. The first... The quantile deviation index of similar samples in each partition is denoted as . The specific calculation formula is as follows:

[0057] in, Indicates the target statistical feature number that participates in the quantile deviation calculation. Indicates the first Empirical quantile values ​​of statistical characteristics of project objectives This indicates that the quantile deviates from the starting point. This indicates the number of target statistical features involved in the calculation. Since data below the 80th quantile typically fall within the normal fluctuation range of similar historical samples, and only those exceeding the 80th quantile exhibit significant abnormal deviations, therefore... The preferred value is set to 0.80, which can reduce false alarms caused by normal shock fluctuations and short-term dust fluctuations.

[0058] Based on the quantile deviation starting point Taking 0.80 as an example, the empirical quantile values ​​corresponding to the peak impact, crack propagation rate, dust concentration, and personnel dwell time in the core demolition zone are 0.92, 0.88, 0.76, and 0.84, respectively, with deviation contributions of 0.60, 0.40, 0, and 0.20, respectively. The arithmetic mean of the four deviation contributions is taken to obtain the quantile deviation index for similar samples. It is 0.30.

[0059] This embodiment improves the adaptability of statistical early warning by screening similar historical samples and evaluating empirical quantiles, so that the current demolition status is no longer simply compared with a fixed threshold, but with the historical statistical distribution under the same scene, the same area, the same object and the same equipment conditions.

[0060] Example 5: The seam response takes into account sensor sampling delay and structural propagation delay, therefore... The preferred value is set as The response lag value is denoted as... The specific calculation formula is as follows:

[0061] in, Dividing the two yields a dimensionless proportionality term, therefore It is a dimensionless value, and is limited to the range of 0 to 1 by the outer minimum value function.

[0062] The first The normalized sequence of device outputs for each partition within the current statistical window is denoted as follows: , will the The normalized sequence of object responses for each partition within the current statistics window is denoted as follows: .in, It consists of the normalized values ​​of the device outputs from each sampling point within the current statistical window. Both are composed of the normalized object response values ​​of each sampling point within the current statistical window, and all elements of both are dimensionless values. The number of sampling points within the current statistical window is denoted as... Dimensionless. The intensity value of the sequence output by the device is denoted as... The intensity value of the object's response sequence is denoted as... Both are calculated from the L2 norm of the corresponding normalized sequence, and the specific calculation formulas are as follows:

[0063] in, Represents the normalized output sequence of the device. The 2-norm, Represents the normalized sequence of object responses The 2-norm.

[0064] The strength mismatch value is denoted as The specific calculation formula is as follows:

[0065] Here, 0.001 is a dimensionless stability constant to prevent the denominator from being zero. Since both the numerator and denominator are composed of dimensionless intensity values, ... This is a dimensionless value. The absolute value term is used to prevent negative values ​​when the object's response sequence strength value exceeds the device's output sequence strength value.

[0066] The first The device object coupling response coefficient of each partition is denoted as . In this embodiment, The coupling anomaly is represented by a direction; a larger value indicates a greater mismatch between the device output and the object's response. The specific calculation formula is as follows:

[0067] in, Used to retain the positively correlated portion, when When the value is negative, treat it as 0. and All are dimensionless values, therefore It is a dimensionless value. It simultaneously exhibits three types of coupling anomalies: decreased correlation, increased response lag, and enhanced strength mismatch. Because There was only a slight mismatch. This indicates that at least one coupling factor has shown a significant anomaly. This indicates a severely abnormal coupling relationship; therefore, the preferred value for the first coupling threshold is set to... The preferred value of the second coupling threshold is set to The threshold is determined by the combined distribution of three normalized indicators: reduced correlation, lag response, and intensity mismatch.

[0068] The trend slopes of the object response sequence within the short-term, medium-term, and long-term statistical windows are respectively denoted as... The short-term statistical window length is denoted as... The unit is seconds, and the length of the time statistics window is recorded as... The unit is seconds, and the length of the long-term statistical window is denoted as... The unit is seconds, and the effective sampling frequency of the object response sequence is denoted as . The unit is Hertz. The data was obtained from the timestamps of the crack displacement gauge, laser rangefinder, and triaxial vibration sensor. The [number]th [unit / item] was [data / data / information]. The number of sampling points within the statistical window is denoted as The specific calculation formula is as follows:

[0069] in, Indicates the first Statistical window length, in seconds. The effective sampling frequency of the object's response sequence, measured in Hertz. This indicates rounding down to the nearest integer.

[0070] Short-term, medium-term, and long-term statistical window lengths The preferred time windows are 10s, 60s, and 300s, respectively. The short time window captures instantaneous changes caused by a single contact, impact, and rebound; the medium time window determines whether the object's response continues to increase under multiple breaching actions; and the long time window determines whether a cumulative deviation has formed in a local breach segment. The effective sampling frequency of the object response sequence is... When the frequency is below 1Hz, the window should be extended proportionally according to the requirement that the short-time, medium-time and long-time windows should have no less than 10, 60 and 300 sampling points respectively.

[0071] The first The trend slope of the response sequence of objects within the class statistics window is denoted as... The slope of the linear regression is used for calculation, and the specific calculation formula is as follows:

[0072]

[0073] In the formula, The units for the three types of reference values ​​and their corresponding slopes are all... Therefore, the normalized results are all dimensionless values. Short-term windows are used to identify abrupt changes and allow for higher slopes, while long-term windows are used to identify cumulative deviations and allow for a smaller slope to characterize persistent risk. Therefore, the reference values ​​are set in descending order of window duration.

[0074] The first The multi-window instability trend of each partition is denoted as . The specific calculation formula is as follows:

[0075] because All are dimensionless values, therefore It is a dimensionless value, because The overall response of the object was stable. The response time of the object is continuously increasing. The object response exhibits a clear tendency to become unstable; therefore, the first threshold for the trend is preferably 0.40, and the second threshold for the trend is preferably 0.70.

[0076] The first The dust concentration, smoke concentration, noise intensity, and visibility of each zone are recorded as follows: The data were acquired by a dust sensor, a smoke sensor, a noise sensor, and a visibility detection device, respectively. The corresponding reference values ​​are denoted as follows: And the noise attenuation reference difference is denoted as .in, The above reference values ​​are used to evaluate the impact of the on-site environment on the reliability of the perceived data, and are not used as occupational exposure limits or on-site safety alarm limits. Preferably, The above values ​​are determined based on the effective perception conditions required for visual recognition, laser ranging, crack observation, acoustic-assisted perception, and evacuation path confirmation in near-field demolition operations.

[0077] The first The perceived reliability correction degree of each partition is denoted as . The specific calculation formula is as follows:

[0078] Device-object coupling response coefficient and multi-window instability trend The first threshold is preferably 0.40, and the second threshold is preferably 0.70. A value below the first threshold indicates that the overall fluctuation is within the normal range; a value between the two thresholds indicates a significant anomaly; and a value reaching the second threshold indicates a high degree of anomaly. (Perceived credibility correction) The preferred confidence threshold is 0.60. If the value is below this threshold, the data collected in the corresponding partition will be marked as a low confidence verification state.

[0079] Example 6: This embodiment is a detailed description of step S5 in Embodiment 1.

[0080] The first The work status statistical early warning value for each partition is recorded as follows: The normalized Euclidean distance is used to calculate the first... Operation status statistics and early warning values ​​for each partition The specific calculation formula is as follows:

[0081] Indicates the first The device object coupling response coefficient for each partition is derived from Example 5. In this example, the coefficient is taken according to the direction of coupling anomaly; the higher the value, the more mismatched the device output and the response of the damaged object are. Interval.

[0082] because This generally indicates that no significant composite anomaly has been observed among the four statistical indicators. This indicates that at least one indicator is significantly abnormal or that multiple indicators are slightly abnormal. This indicates that multiple indicators deviate simultaneously and require coordinated action; therefore, the preferred value of the preset statistical first threshold is set to... The preferred value of the preset statistical second threshold is set to When the corresponding zone enters the regular monitoring level, At that time, the corresponding partition enters the key attention level. At that time, the corresponding zone will enter the joint response level.

[0083] Based on the quantile deviation index of similar samples obtained in Example 4 And the device object coupling response coefficient obtained in Example 5 Multi-window instability trend and perceived credibility correction In this embodiment, the statistical warning value of the operation status corresponding to each partition is calculated, and the partition warning level is generated accordingly. The corresponding calculation is shown in Table 4 below.

[0084] Table 4: Example Table of Zonal Early Warning Calculation

[0085] The three zones mentioned above show the calculation results for three levels: coordinated response, key focus, and routine monitoring, respectively, which can cover the judgment scenarios for all warning levels.

[0086] quantile deviation index of similar samples Device-object coupling response coefficient Multi-window instability trend The inverse value of perceived credibility correction is used as a quantile anomaly, coupling anomaly, trend anomaly, and low-credibility anomaly. The anomaly with the largest value corresponds to the dominant strategy. When quantile anomaly dominates, the workload is reduced and the benchmark of similar samples is reviewed. When coupling anomaly dominates, the contact point is reviewed and the demolition angle is adjusted. When trend anomaly dominates, the demolition sequence is changed and the personnel boundary is expanded. When low-credibility anomaly dominates, redundant acquisition is enabled and data retesting is performed. When the difference between the largest and second largest anomaly does not exceed 0.05, it is determined to be a composite anomaly dominant state, and the corresponding two types of strategies are executed simultaneously.

[0087] The focus is on mildly modified strategies corresponding to the primary level, and mandatory response strategies corresponding to the secondary level. Based on the scenario type, zone type, and dominant strategy, at least one response action should be selected from the following: equipment downloading, adjustment of demolition methods, enhanced sampling, personnel exposure control, expansion of area boundaries, ventilation and smoke extraction, redundant data collection, and remote control.

[0088] Under the key monitoring level, equipment output pressure is reduced by 10% to 20%, impact frequency by 10% to 15%, equipment feed speed by 15% to 25%, demolition equipment output power by 15% to 25%, cutting or shearing speed by 15% to 20%, personnel single-time dwell time is shortened by 20%, personnel rotation interval is shortened by 30%, sampling frequency is increased to 2 to 3 times the original sampling frequency, personnel boundary is expanded by 1 to 2 meters, and a 30-60 second pause can be made for object response verification. The above actions correspond to the slightly modified range of the equipment's rated adjustable gear, on-site operating procedures, and historical treatment samples.

[0089] Under the coordinated response level, equipment output pressure is reduced by 25% to 40%, impact frequency is reduced by 30% to 40%, demolition equipment output power is reduced by 40% to 60%, single-segment demolition or cutting length is shortened by 30% to 50%, personnel stay duration is shortened by 40% to 60%, on-site personnel number is reduced to 50% to 70% of the original planned number, personnel exposure zone boundary is expanded by 3m to 5m, equipment protection zone safety distance is expanded by 5m to 10m, and remote demolition control, ventilation and smoke extraction, and redundant data collection and verification are activated according to the source of risk. The above actions correspond to a clearly defined risk control zone, and can be implemented at three to four levels of load reduction or operation mode switching.

[0090] When quantile anomalies dominate, reduce workload and review benchmark samples. When coupled anomalies dominate, review the contact point of the demolition head, adjust the demolition angle and feed speed. When trend anomalies dominate, change the demolition sequence, shorten the single-segment demolition length, and expand personnel boundaries. When low-confidence anomalies dominate, first implement ventilation and smoke extraction, dust settling, redundant data acquisition, and data retesting. When compound anomalies dominate, simultaneously implement the corresponding actions for the largest and second-largest anomalies.

[0091] Taking building collapse rescue and demolition as an example, when the core demolition area is at the key attention level and coupled anomalies dominate, the contact point of the demolition head is repositioned, the demolition angle is adjusted by 5° to 15°, the equipment feed speed is reduced by 15% to 25%, and the collection of object response data is paused for 30s to 60s. When the core demolition area or structural response area is at the linkage disposal level, the equipment output pressure is reduced by 25% to 40%, the impact frequency is reduced by 30% to 40%, continuous demolition is adjusted to segmented demolition, the length of a single segment is shortened by 30% to 50%, and the boundary of the personnel exposure area is expanded by 3m to 5m.

[0092] Taking emergency demolition of confined spaces or high-risk facilities as an example, the focus is on reducing the duration of personnel stay, increasing the frequency of gas and environmental data sampling, and reducing equipment output power or cutting speed at the level of coordinated response. At the level of coordinated response, the number of personnel on site should be reduced, ventilation and smoke extraction should be carried out, and continuous gas retesting should be performed. The safety distance of the equipment protection zone should be expanded, and remote demolition control should be activated when necessary. Traffic obstacle removal and demolition should be carried out in the same manner and according to the same level and dominant rules, including segmented obstacle removal, verification of force points, personnel repositioning, and maintenance of evacuation buffer zones.

[0093] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale. Dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization. The algorithm of this invention is implemented as a Python script. Before executing the core logic, the program first executes a data loading module (e.g., using the widely used pandas library in Python) configured to read the aforementioned spreadsheet file and load its contents into the program's working memory (e.g., a DataFrame data structure). Subsequent algorithm steps will directly query and retrieve the required configuration parameters from this in-memory data structure.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent early warning of emergency demolition operation status, characterized in that, The specific steps include, S1. Obtain the basic scenario data for the target emergency breaching mission. This data includes the type of object to be breached, the type of equipment used for breaching, the location of the breaching point, the on-site spatial conditions, the type of hazard source, the location of personnel, and the location of evacuation routes. Based on this data, classify the scenario and divide the target emergency breaching area into a core breaching zone, a structural response zone, a personnel exposure zone, an equipment protection zone, and an evacuation buffer zone. Generate a scenario partition index table. S2. According to the scenario partition index table, collect the operation data of the demolition equipment, the response data of the object being demolished, the on-site environmental disturbance data, the personnel exposure status data, and historical demolition sample data. Perform time synchronization, anomaly filtering, missing data completion, and normalization processing to generate an emergency demolition operation status dataset. S3. Based on the scene classification results, partitioning results, the type of the object being demolished, and the type of demolition equipment, perform similar sample filtering on the historical demolition sample data to generate a similar historical sample dataset. Based on the similar historical sample dataset, form a quantile statistical benchmark set. S4. Based on the emergency demolition operation status dataset and quantile statistical benchmark set, calculate the quantile deviation index of similar samples, the equipment-object coupling response coefficient, the multi-window instability trend degree, and the perceived reliability correction degree. The quantile deviation index of similar samples is used to characterize the degree of deviation of the current operation status from the statistical distribution of similar historical samples. The equipment-object coupling response coefficient is used to characterize the degree of coupling anomaly between the output of the demolition equipment and the response of the demolished object. The multi-window instability trend degree is used to characterize the instability trend of the response of the demolished object over time. The perceived reliability correction degree is used to characterize the reliability of the data collected on site under the influence of dust, smoke, noise, and visibility. S5. Based on the quantile deviation index of similar samples, the coupling response coefficient of equipment objects, the instability trend degree of multi-window and the perception credibility correction degree, calculate the statistical early warning value of the operation status corresponding to each partition, generate the regular monitoring level, the key attention level or the linkage disposal level according to the statistical early warning value of the operation status, and generate the scenario partition early warning strategy based on the scenario classification results, partition results, partition early warning level and strategy dominant item.

2. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that, In step S1, scene classification is performed, specifically including: The set of scene elements for the current target emergency demolition task is denoted as the current scene element set. The current scene element set is extracted from on-site survey records, 3D scanning results, demolition operation plan, personnel positioning data, and hazard source identification results. The preset scene element set is formed by summarizing samples of historical building collapse rescue demolition, confined space rescue demolition, high-risk facility emergency demolition, and traffic obstacle removal demolition. Jaccard similarity is used to calculate the scene matching degree between the current target emergency demolition task and various preset scenes. When the scene matching degree reaches the scene matching threshold, the current target emergency demolition task is assigned to the corresponding scene. When the scene matching degree does not reach the scene matching threshold, the current target emergency demolition task is marked as a composite demolition scene and enters manual review.

3. The intelligent early warning method for emergency demolition operation status according to claim 2, characterized in that, Step S1 involves defining the target emergency demolition area, specifically including: The effective radius of the demolition equipment, the safe correction distance for debris scattering, and the on-site safety redundancy distance are obtained. The effective radius of the demolition equipment is obtained from the nameplate parameters of the demolition equipment and the equipment control terminal. The safe correction distance for debris scattering is determined by the material type of the object being demolished, the demolition method, the historical debris scattering record, and the on-site shielding conditions. The on-site safety redundancy distance is determined based on the minimum standing interval of rescue personnel, the equipment rotation margin, and the temporary evacuation margin. The effective radius of the demolition equipment, the safe correction distance for debris scattering, and the on-site safety redundancy distance are added together to obtain the radius of the core demolition zone. The structural response zone, personnel exposure zone, equipment protection zone, and evacuation buffer zone are determined outside the core demolition zone.

4. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that, Step S2 involves collecting data and generating an emergency demolition operation status dataset, specifically including: Operational data of the demolition equipment is collected through hydraulic pressure sensors, equipment controllers, current transformers, temperature sensors, and triaxial vibration sensors. Response data of the demolished object is collected through crack displacement gauges, laser rangefinders, tilt sensors, and triaxial vibration sensors. Environmental disturbance data is collected through dust sensors, smoke sensors, combustible gas sensors, noise sensors, and visibility detection devices. Personnel exposure status data is collected through UWB personnel positioning tags and positioning base stations. Using the unified clock of the edge computing terminal as the time reference, data collected by different sensors are aligned to a unified sampling time. Data that cannot be aligned within the set sampling period is marked as data to be reviewed.

5. The intelligent early warning method for emergency demolition operation status according to claim 4, characterized in that, Step S2 involves anomaly filtering, missing data completion, and normalization of the collected data, specifically including: Using the median and median absolute deviation of similar collected data within the current sliding window as robust statistical benchmarks, robust anomaly screening values ​​are calculated for the corresponding collected data. When the robust anomaly screening value exceeds the anomaly screening threshold, the corresponding collected value is marked as data to be reviewed. When the robust anomaly screening value does not exceed the anomaly screening threshold, the corresponding collected value is considered valid collected data. For data segments with missing time not exceeding the set number of sampling periods, linear interpolation is used to fill in the missing data using adjacent valid sampling times and their corresponding collected values. For data segments with missing time exceeding the set number of sampling periods, they are marked as data to be reviewed. Normalization reference upper and lower limits are determined according to the sensor's effective range, equipment rated range, on-site safety regulation limits, and historical demolition sample quantile range. Valid collected data are then converted into dimensionless normalized values.

6. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that, Step S3 generates a historical sample dataset of the same type and a quantile statistical benchmark set, specifically including: Extract sample features from the current emergency demolition task. These features include scene type, area type, type of object to be demolished, type of demolition equipment, structural dimensions, rated power of equipment, personnel density, hazard level, access restriction level, and structural damage level. Calculate the temporary difference values ​​for discrete, continuous, and ordered level features according to the sample feature type, and map them uniformly to the uniform difference values ​​of historical demolition samples on the corresponding sample features. Calculate the sample matching degree of historical demolition samples based on the uniform difference values ​​of each sample feature. When the sample matching degree reaches the sample matching threshold, the corresponding historical demolition sample is included in the same historical sample dataset. Extract the peak impact, crack propagation rate, dust concentration, and personnel dwell time from the same historical sample dataset to form a quantile statistical benchmark set.

7. The intelligent early warning method for emergency demolition operation status according to claim 6, characterized in that, Step S4 calculates the quantile deviation index for similar samples, specifically including: Extract the target statistical feature value from the current statistical window. The target statistical feature value comes from valid collected data or valid normalized data. Determine the set of historical sample values ​​corresponding to the target statistical feature in the same historical sample dataset. Calculate the empirical quantile value of the target statistical feature based on the number of samples in the historical sample value set that are not greater than the current target statistical feature value and the total number of samples in the same historical sample dataset. Take the portion of the empirical quantile value that exceeds the quantile deviation benchmark as the deviation contribution. Average the deviation contributions of multiple target statistical features involved in the calculation to obtain the quantile deviation index of the same sample in the corresponding partition.

8. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that, Step S4 involves calculating the coupling response coefficient of the device object, specifically including: The equipment output sequence is formed according to the demolition method. The equipment output sequence corresponding to hydraulic shearing demolition is formed by normalizing the output pressure, the equipment output sequence corresponding to impact demolition is formed by normalizing the impact frequency and equipment vibration amplitude, and the equipment output sequence corresponding to cutting demolition is formed by normalizing the output current. The object response sequence is formed according to the normalized values ​​of crack propagation, displacement, and structural vibration. The response weight is determined according to the main dangerous response type. The correlation coefficient between the equipment output sequence and the object response sequence, the response lag value between the peak value of the equipment output and the peak value of the object response, and the intensity mismatch value between the intensity value of the equipment output sequence and the intensity value of the object response sequence are calculated. The equipment-object coupling response coefficient is obtained by fusing the correlation coefficient, response lag value, and intensity mismatch value. The equipment-object coupling response coefficient is taken according to the direction of coupling anomaly. The larger the value, the greater the mismatch between the equipment output and the object response.

9. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that, Step S4 calculates the multi-window instability trend degree and perceived reliability correction degree, specifically including: Short-term, medium-term, and long-term statistical windows are set up separately. Based on the sampling time and response value of the object response sequence within each statistical window, the short-term trend slope, medium-term trend slope, and long-term trend slope are calculated using linear regression slope. The short-term trend slope, medium-term trend slope, and long-term trend slope are normalized according to their corresponding slope reference values ​​and fused to obtain the multi-window instability trend degree. Dust concentration, smoke concentration, noise intensity, and on-site visibility are obtained. Based on the dust concentration reference value, smoke concentration reference value, noise reference value, noise attenuation reference difference, and visibility reference value, the perception confidence correction degree is calculated. When the perception confidence correction degree is lower than the confidence threshold, the data collected in the corresponding partition is marked as a low confidence verification state.

10. The intelligent early warning method for emergency demolition operation status according to claim 1, characterized in that: Step S5 generates a scene-partitioned early warning strategy, specifically including: Based on the quantile deviation index of similar samples, the coupling response coefficient of equipment objects, the instability trend degree of multi-window, and the perception reliability correction degree, the normalized Euclidean distance is used to calculate the statistical early warning value of the operation status of each zone. According to the statistical early warning value of the operation status, the corresponding zone is determined as a regular monitoring level, a key attention level, or a linkage response level. The anomaly type corresponding to the largest item among quantile anomalies, coupling anomalies, trend anomalies, and low-confidence anomalies is determined as the policy-dominant item. When the difference between the largest item and the second largest item does not exceed the preset tolerance, the corresponding partition is determined as the composite anomaly-dominant state. When the corresponding zone is at a high-priority level, a mild correction strategy is output based on the scenario type and the dominant strategy. This mild correction strategy includes at least one of the following: reducing equipment output pressure by 10% to 20%, reducing impact frequency by 10% to 15%, reducing equipment feed speed by 15% to 25%, reducing demolition equipment output power by 15% to 25%, reducing cutting or shearing speed by 15% to 20%, shortening personnel dwell time by 20%, shortening personnel rotation interval by 30%, increasing sampling frequency to 2 to 3 times the original sampling frequency, expanding personnel boundaries by 1 to 2 meters, and pausing for 30 to 60 seconds for object response verification. When the corresponding zone is at the linkage response level, a mandatory response strategy is output according to the scenario type and the dominant strategy. The mandatory response strategy includes reducing the equipment output pressure by 25% to 40%, reducing the impact frequency by 30% to 40%, reducing the output power of the demolition equipment by 40% to 60%, shortening the single-segment demolition length or single-segment cutting length by 30% to 50%, shortening the single-time stay time of personnel by 40% to 60%, reducing the number of on-site personnel to 50% to 70% of the original planned number, expanding the boundary of the personnel exposure area by 3m to 5m, expanding the safety distance of the equipment protection area by 5m to 10m, enabling remote demolition control, and performing ventilation and smoke extraction and redundant data collection and verification.

Images