Fire-fighting equipment intelligent inspection path planning method based on internet of things
By identifying and classifying target deceleration points, and combining track structure and health monitoring data, the weights are dynamically adjusted and the inspection path is optimized. This solves the problems of insufficient inspection efficiency and data quality in existing technologies, and achieves efficient and reliable inspection results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG JINQIAO SECURITY EQUIP CO LTD
- Filing Date
- 2026-03-30
- Publication Date
- 2026-07-10
Smart Images
Figure CN121960920B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of inspection path planning technology, and more specifically, relates to an intelligent inspection path planning method for fire equipment based on the Internet of Things. Background Technology
[0002] Fire safety is of paramount importance in key locations such as oilfield stations, chemical industrial parks, and substations. Regular inspections of fire-fighting facilities and potential fire hazards are crucial for ensuring safety. The selection of inspection routes directly determines the timeliness of hazard detection, thus highlighting the importance and necessity of route planning.
[0003] Current path planning for inspection robots mainly focuses on obstacle avoidance and shortest path optimization. Existing technologies, such as the Chinese invention patent application with application number 202210066355.5, disclose an autonomous driving inspection method for fire-fighting robots. This method uses LiDAR combined with the LOAM algorithm for mapping and localization, and integrates global planning and dynamic window algorithm (DWA) to achieve local obstacle avoidance, thus solving the problem that fire-fighting robots have difficulty in autonomous navigation and real-time obstacle avoidance in complex environments.
[0004] Existing technologies, such as the Chinese invention patent application with application number 202411304718.X, disclose a method and system for fire protection facility safety inspection based on intelligent robots. This method collects data by establishing a three-dimensional model to plan the path and constructing a support vector machine (SVM) model to intelligently analyze and predict the facility status. This solves the problems of traditional manual inspection relying on visual judgment, being unable to predict facility failures in advance, and lacking dynamic adaptive capabilities in inspection tasks.
[0005] Based on existing technologies, current path planning methods primarily focus on the shortest time / shortest path as the sole optimization objective, failing to adequately consider the unique operating conditions of overhead rail or utility tunnel inspection robots. These robots operate on high-altitude tracks or in confined spaces, where factors such as equipment loosening and structural degradation directly impact inspection efficiency and timeliness. Furthermore, existing technologies lack a unified approach to critical locations requiring deceleration, such as curves, ramps, and potential hazard points, thus affecting data acquisition quality and compromising stability during the data collection process.
[0006] Furthermore, the existing technology separates structural feature data from health monitoring data, making it difficult to balance inspection continuity and data reliability, thus hindering further improvements in inspection timeliness and overall inspection effectiveness. Summary of the Invention
[0007] In view of this, in order to solve the above problems, a method for intelligent inspection path planning of fire equipment based on the Internet of Things is proposed.
[0008] The objective of this invention can be achieved through the following technical solution: This invention provides an intelligent inspection path planning method for fire equipment based on the Internet of Things. The method includes: acquiring track structure feature data, track health monitoring data and historical fire inspection logs for each candidate inspection path, and identifying target deceleration points on each candidate path. The target deceleration points include past fire hazard points, track fixed deceleration points and track status hazard points.
[0009] Risk levels are determined for each target deceleration point. Based on these risk levels, a benchmark traffic speed is matched. The fire inspection task completion efficiency index and inspection data quality index for each candidate inspection path are calculated in combination with the benchmark traffic speed.
[0010] The weights of inspection data quality indicators are dynamically adjusted based on the severity and marking time of past fire hazards recorded in historical fire inspection logs.
[0011] Based on the aforementioned inspection efficiency index, inspection data quality index, and dynamic weights, a comprehensive recommendation index for each candidate inspection path is calculated through weighted fusion.
[0012] The candidate inspection path with the highest comprehensive recommendation index is determined as the target inspection path.
[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention identifies three types of target deceleration points by combining track structure feature data, track health monitoring data and historical fire inspection logs, which solves the problem of the current track structure feature data and health monitoring data being separated. This provides a complete input for subsequent risk classification and path optimization, so that the final selected target path can prioritize the data collection quality of high-risk areas while meeting the basic inspection timeliness.
[0014] (2) This invention classifies the risk of each target deceleration point, thereby unifying the risk level of different types of deceleration points into quantifiable level information. This classification process breaks the current limitation that health monitoring data is only used for post-fault diagnosis and is not incorporated into real-time path planning as a dynamic constraint. It enables path planning to comprehensively consider physical structural constraints, structural health status and fire hazards under a unified risk metric, providing a basis for the calculation of subsequent efficiency and quality indicators.
[0015] (3) This invention calculates the efficiency index of fire inspection task completion and the quality index of inspection data, so that the path planning can simultaneously consider the inspection timeliness and data collection effect. This avoids the one-sidedness of using only the shortest time or the shortest path as the optimization goal. While ensuring the operational stability of the suspended rail or pipe gallery inspection robot during the inspection process, it also takes into account the data collection quality and inspection safety.
[0016] (4) This invention dynamically adjusts the weight of inspection data quality indicators based on historical fire inspection logs, so that paths with higher historical risks can obtain higher weights in the comprehensive recommendation index, thus making them more likely to be selected in the balance between efficiency and quality. This solves the problem that historical logs are only used for post-event statistics and cannot affect the selection of current paths, ensuring that the final selected inspection path achieves a balance between efficiency and quality. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a schematic diagram of the overall implementation process of the present invention.
[0019] Figure 2 This is a schematic diagram of the track condition hazard identification process of the present invention.
[0020] Figure 3 This is a schematic diagram of the loosening tendency judgment process of the present invention.
[0021] Figure 4 This is a schematic diagram of the vibration sudden change trend judgment process of the present invention. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0023] Current path planning methods in industrial inspection scenarios often focus solely on the shortest time or shortest path as the optimization objective, lacking a unified consideration for critical locations requiring deceleration, such as curves, slopes, and potential hazard points. This affects operational stability and data collection quality, making it difficult to strike a balance between inspection efficiency and data reliability.
[0024] On the other hand, structural features such as curve curvature and health monitoring data such as vibration frequency are fragmented in practical applications. Curvature is only used to preset speed limits, and vibration frequency is only used for post-fault diagnosis. They are not used as input for real-time path planning, which restricts the further improvement of the overall efficiency of the inspection system.
[0025] Based on this, this invention discloses an intelligent inspection path planning method for fire equipment based on the Internet of Things, which can further improve the overall efficiency of the inspection system and prioritize the data collection quality of high-risk areas while ensuring inspection efficiency.
[0026] Please refer to the details. Figure 1 As shown, Figure 1 This is a flowchart illustrating an IoT-based intelligent inspection path planning method for fire equipment, which is provided in an embodiment of the present invention. The method specifically includes the following steps: S1. Based on the location distribution of fire equipment, a path search algorithm is used to generate multiple candidate inspection paths from the starting point to the ending point, covering all fire equipment that must be inspected.
[0027] This step provides a basis for comparison in subsequent path selection. In practice, fire equipment locations are designated as essential nodes. A K-shortest path algorithm (such as Yen's algorithm or Eppstein's algorithm) based on essential node constraints is used to find multiple shortest paths from the starting point to the ending point, requiring each path to pass through all essential nodes. By changing the order of visits to essential nodes or using a path splicing strategy, multiple candidate paths that meet the coverage requirements can be obtained. For example, in a simple network containing two essential nodes C and D, with the starting point A and the ending point B, paths can be generated that start from starting point A, pass through nodes C and D in sequence to reach ending point B, and paths that start from starting point A, pass through nodes D and C in sequence to reach ending point B.
[0028] For large-scale track networks, the shortest sub-paths between essential nodes can be calculated first, and then candidate paths can be generated by splicing and combining them. Finally, the generated paths are filtered based on the actual connectivity of the tracks, and the passable paths are retained as candidate inspection paths.
[0029] After obtaining the candidate inspection path according to the above steps, continue to perform the following steps.
[0030] S2. Obtain track structure feature data, track health monitoring data and historical fire inspection logs for each candidate inspection path, and identify target deceleration points on each candidate path. Target deceleration points include past fire hazard points, track fixed deceleration points and track condition hazard points.
[0031] In the intelligent inspection path planning of fire equipment, the robot will encounter various locations that require deceleration when running along the track. These locations include curves and ramps that must be decelerated due to physical structural limitations, abnormal vibration points that require slow passage due to track degradation, and areas that require key re-inspection due to historical fire hazards.
[0032] Existing technologies typically handle these deceleration requirements in a fragmented manner; for example, vibration anomalies are only used for post-incident maintenance. This prevents robots from predicting the full deceleration requirements during the planning phase, potentially leading to safety accidents or data acquisition failures when passing through potential hazard points at high speeds. To comprehensively acquire deceleration requirements along the path, this embodiment first identifies three types of target deceleration points based on track structure feature data, track health monitoring data, and historical fire inspection logs.
[0033] The specific implementation process for identifying the three types of target deceleration points is as follows: All points marked with fire hazard indicators are extracted from historical fire inspection logs and recorded as past fire hazard points. From the track structure characteristic data, curve curvature exceeding a preset curvature threshold, slope exceeding a preset slope threshold, and track structure joint locations are marked as fixed deceleration points. The preset curvature threshold and preset slope threshold are set according to the requirements of the industrial pipe gallery track safety operation specifications. For example, the preset curvature threshold is typically set to 0.02 radians per second, and the preset slope threshold is typically set to 5°.
[0034] Please see Figures 2 to 4 As shown, for potential track condition hazards, dynamic judgment is made based on health monitoring data. The specific judgment process is as follows: R1. Obtain the vibration frequency detected by each track monitoring point during multiple inspections within a preset historical time window (e.g., the most recent 30 days), construct a vibration frequency time series, and perform linear regression fitting to obtain the slope of the vibration frequency change over time.
[0035] R2. If the slope of the change is negative and its absolute value exceeds the first preset slope threshold, and the goodness of fit of the linear regression exceeds the preset fit threshold, then it is determined that the track monitoring point has a loosening tendency.
[0036] Understandably, the first preset slope threshold is set as follows: during the initial stabilization period after the track is put into normal operation (usually the 3rd to 6th month after it is put into use), the inspection vibration frequency data of each monitoring point are continuously collected, the statistical distribution of the daily frequency change is calculated, and the upper limit of its 95% confidence interval is taken as the first preset slope threshold. The preset fitting threshold is usually set to 0.6.
[0037] R3. Use the real-time fixed vibration frequency of each track monitoring point as its current reference value, and obtain the detection vibration frequency when the most recent inspection passed as the current passing value.
[0038] R4. Calculate the difference between the current passing value and the average value of the vibration frequency detected by multiple historical inspections. Take the absolute value of the difference as the vibration frequency difference. Also, calculate the ratio of the current passing value to the current reference value to obtain the current vibration deviation.
[0039] It should be noted that the vibration frequency difference is used to measure the absolute difference between the vibration response of the track during the current inspection and the long-term historical normal level of that point.
[0040] If the difference is positive, it indicates that the current vibration frequency is higher than the historical normal value, and there may have been an abnormal impact or a change in stiffness; if the difference is negative, it indicates that the vibration frequency is lower than the historical normal value, and there may have been loosening or a decrease in stiffness.
[0041] It should also be noted that the current vibration deviation is used to measure the change factor of the track's vibration response relative to the current no-load reference state during the current inspection.
[0042] When the vibration deviation approaches 1, it indicates that the vibration response under load is basically consistent with the unloaded reference, and the structure is normal. When the vibration deviation is greater than 1, it indicates that the load response is amplified, which may be due to instantaneous impact or abnormal structural stiffness. When the vibration deviation is less than 1, the load response is attenuated, which may be due to energy dissipation or loose connection.
[0043] R5. If the detected vibration frequency difference at a certain track monitoring point exceeds the preset vibration frequency difference value and the current vibration deviation exceeds the preset deviation value, then it is determined that the track monitoring point has a sudden vibration trend.
[0044] R6. Mark track monitoring points that show signs of loosening or sudden vibration as potential track condition hazards.
[0045] Under normal operating conditions, the vibration frequency of the track fluctuates between approximately 2 Hz and 3 Hz, with a ratio between approximately 0.8 and 1.2. Preferably, the present invention can set the preset vibration frequency difference to 3 Hz and the preset deviation to 1.5.
[0046] It should be noted that the specific values given in this embodiment are for illustrative purposes only and are not intended to impose any specific limitations. Implementers can adjust them according to the type of track in the actual scenario.
[0047] Ultimately, past fire hazard points, fixed deceleration points on the track, and potential track condition hazards were all considered as target deceleration points on the candidate path.
[0048] Through the above steps, this embodiment integrates previously scattered track structure feature data, track health monitoring data, and historical fire inspection logs into the front-end input of path planning. This enables path planning to comprehensively predict deceleration requirements throughout the entire process from three dimensions: physical structure, real-time status, and historical experience. In particular, for potential track condition hazards, by distinguishing between loosening tendencies and sudden vibration trends, it can accurately identify long-term cumulative damage and occasional disturbances, providing a reliable data foundation for subsequent risk classification. This avoids misjudgments and missed alarms under traditional single-threshold alarm methods, ensuring that the robot can predict all deceleration locations during the planning phase.
[0049] S3. Risk classification is performed on each target deceleration point to obtain the risk level. Based on the risk level, a benchmark passage speed is matched, and the fire inspection task completion efficiency index and inspection data quality index of each candidate inspection path are calculated in combination with the benchmark passage speed.
[0050] While inspection robots need to slow down when passing through fire hazard points, fixed deceleration points on the track, and potential track condition hazards, their impact on inspection safety and data collection quality varies. Existing path planning methods that aim for the shortest time ignore this difference, resulting in an inability to reasonably balance passage efficiency and data reliability.
[0051] Based on this, this step introduces risk classification to quantify the degree of danger at different deceleration points, laying the foundation for subsequent matching of differentiated benchmark traffic speeds and dynamic adjustment of the weights of inspection data quality indicators, ultimately achieving the optimal path selection that balances safety, efficiency, and data quality.
[0052] The specific implementation method for risk classification is as follows: For each target deceleration point, first determine whether it belongs to a single category, that is, only one of the following: past fire hazard points, track fixed deceleration points, or track condition hazard points. If it is a single category, then determine its risk level according to the following sub-steps: If the point is a past fire hazard point, then extract the number of historical hazard markers for that point from the historical fire inspection log. In addition to the time and severity of each hazard marker, based on the time of each marker, the time interval between each hazard marker and the start time of the current inspection task is obtained, denoted as . Simultaneously, the severity of each hazard marker is converted into a severity score. The severity score is positively correlated with the severity level. For example, assuming severity is categorized as mild, moderate, and severe, a severity score of 1 is assigned to mild, 2 to moderate, and 3 to severe. This yields the severity score for each hazard marker, denoted as [score here]. , Indicates the sequence number of the hazard marker. The fire hazard index is calculated based on the severity score and time interval using the following formula. : .
[0053] In the formula, It is a natural constant. The preset time decay factor, The larger the value, the faster the historical data decays, and the higher the weight of recent hidden dangers on current risks. The time interval described in this invention is in days. When the unit is days, Typically, the value is between 0.3 and 0.8, but in this invention, it is preferably set to 0.5. This indicates that a high concentration of potential risks affects the weighting. Indicates taking The maximum value in the index is used to normalize the time interval, so that the value of the exponent is not affected by the absolute time length.
[0054] This is a time decay term, used to reflect the timeliness of potential hazards; the more recent the hazard occurred (…), the more likely it is to decay. The smaller the value, the closer the time decay term is to 1, the greater its contribution to the fire hazard index, and the longer the hazard has been present. The larger the value, the closer the time decay term is to 0, and the smaller its contribution to the fire hazard index. Dividing by Normalization is performed to make the decay rate adapt to the time span of the hidden danger record, so as to avoid all decay terms approaching zero due to excessive absolute time.
[0055] Hidden dangers have a significant impact on weight. This is used to correct the impact of the concentration of potential hazards over time on risk. If a hazard erupts in a concentrated period (e.g., three times within a month), its urgency for re-inspection should be higher than that of a hazard with the same number of occurrences but spread over several years. The more densely distributed the hazards, the greater the urgency. The larger the value, the greater the fire hazard index.
[0056] This index comprehensively reflects the severity, timing, and frequency of historical hazards at the location; the higher the value, the more urgent the need for re-inspection at that location.
[0057] It should be added that the method for setting the weight of the impact of densely packed hidden dangers is as follows: calculate the average of the time intervals between adjacent hidden danger markers. and standard deviation Thus, the coefficient of variation is obtained. , .
[0058] Calculate the impact weight of densely populated hidden dangers based on the mean and coefficient of variation. ; .
[0059] in, To set a reference time interval, it can be set according to the inspection cycle. For example, if the regular inspection cycle is once a month, then... The value is 30 days.
[0060] This represents the ratio of the actual hazard interval to the baseline interval. This item reflects the frequency of potential hazards. The smaller, The larger the value, the denser the potential hazards, indicating that the hazards are more frequent. The impact weight of dense hidden dangers is negatively correlated.
[0061] This item reflects the evenness of the distribution of potential hazards. The larger the value (the more uneven the distribution), the smaller this term becomes. The weight of the dense distribution of hidden dangers is negatively correlated with that of fire-fighting equipment. This is because, under the same average interval, even intervals (such as once every 10 days) are more periodic than intervals with large fluctuations (such as once every 1 day or once every 19 days). This means that there are periodic problems with fire-fighting equipment. Therefore, this item is set to give higher weight to even distribution.
[0062] It should be noted that when When the sample size is sufficient, the mean and standard deviation can be effectively calculated. At this point, the frequency and evenness of potential hazards can be assessed simultaneously. Therefore, multiplying these two values reflects the impact of density on the need for re-inspection. When the sample size is insufficient, only one sample is used. The item reflects the density, abandoning uniformity correction to ensure the effectiveness of the weights.
[0063] The calculated fire hazard index is matched with a preset risk level mapping relationship to obtain the corresponding risk level.
[0064] Understandably, the risk level mapping relationship is established as follows: The fire hazard index of all past fire hazard points within the inspection area corresponding to the candidate inspection path is collected during each historical inspection, forming a fire hazard index sample set. This sample set is then sorted by value from smallest to largest. The lower third of this set is used as the boundary between low and medium risk, and the upper third is used as the boundary between medium and high risk. The risk level corresponding to the fire hazard index is matched based on these boundaries.
[0065] If the location is a fixed deceleration point on the track, its risk level is determined according to the curve curvature and slope, based on a preset risk level classification rule. For example, a high risk can be defined as a curvature exceeding 0.02 radians per second and a slope exceeding 5°, a medium risk as a curvature exceeding only the threshold or a slope exceeding only the threshold, and a low risk as the rest.
[0066] If the location is a potential safety hazard point for the track, it will be judged based on whether there is a tendency to loosen and a sudden change in vibration: if both tendencies to loosen and sudden changes in vibration are present, it will be marked as high risk; if only a sudden change in vibration is present, it will be marked as low risk; if only a tendency to loosen is present, it will be marked as medium risk.
[0067] Understandably, a loosening trend reflects long-term cumulative damage to the track structure, and its development is continuous and irreversible, posing a persistent threat to operational safety. In contrast, a sudden vibration trend may be caused by accidental factors (such as temporary overloads or external impacts), and has a certain degree of recoverability, thus its risk level is relatively low. When both occur simultaneously, it indicates that the structure has both accumulated damage and a sudden anomaly, a precursor to impending failure, and is therefore classified as high-risk.
[0068] It should be noted that if the same monitoring point repeatedly experiences sudden vibration changes during multiple inspections (such as more than 3 times in a row), it is necessary to reassess whether there is a hidden loosening trend, or directly change the risk level to medium risk.
[0069] If the target deceleration point belongs to multiple categories, such as being both a past fire hazard point and a track condition hazard point, then the risk level of each category shall be assessed according to the method of the single category mentioned above, and then the highest level shall be taken as the final risk level of the target deceleration point.
[0070] Through the aforementioned risk grading process, this embodiment transforms the three types of target deceleration points into a unified and quantifiable risk level, making structural constraints, health status, and historical fire safety conditions comparable under the same risk scale. Furthermore, it unifies the risk levels of different types of deceleration points into quantifiable level information. This grading process breaks through the current limitation that health monitoring data is only used for post-event fault diagnosis and not incorporated into real-time path planning as a dynamic constraint. It enables path planning to comprehensively consider physical structural constraints, structural health status, and fire hazards under a unified risk metric, providing a basis for subsequent calculations of efficiency and quality indicators.
[0071] Furthermore, the baseline passage speed based on the risk level can be determined according to the following rules: if the risk level is low, the default initial inspection speed is directly used as the baseline passage speed; if the risk level is medium, 50% of the default initial inspection speed is used as the baseline passage speed; if the risk level is high, a preset minimum inspection speed is used as the baseline passage speed. The default initial inspection speed is usually preset according to the inspection task requirements. For example, on a straight track without deceleration requirements, the robot operates at an economical cruising speed, which can be determined according to the equipment manual; in this embodiment, it is set to 1 meter per second. The preset minimum inspection speed can be set according to the track safety speed limit, for example, 0.2 meters per second.
[0072] Considering that high-risk deceleration points require longer deceleration sections and lower passing speeds, their impact on total time is far greater than that of low-risk points. However, the total path length alone cannot accurately reflect the inspection time of different paths. Therefore, it is necessary to convert the risk level into specific path time calculations so that efficiency indicators can truly reflect the actual inspection efficiency of different paths while taking into account both safety and data quality.
[0073] Based on this, the present invention converts the matched benchmark traffic speed into path time and calculates the fire inspection task completion efficiency index of each candidate inspection path according to the path time. The specific calculation process is as follows: First, the total length of each candidate inspection path is calculated, and the risk level of all target deceleration points on the path is obtained. Based on the risk level of each deceleration point, the length of the affected section of each target deceleration point is determined.
[0074] In this embodiment, the length of the affected section is positively correlated with the risk level and also positively proportional to the total track length. For example, a high-risk level corresponds to 3% to 5% of the total track length, a medium-risk level corresponds to 1% to 3%, and a low-risk level corresponds to 0.5% to 1%. To avoid calculation distortion due to excessively short or long sections, a lower limit of 1 meter is set for each. The upper limits for the affected section lengths corresponding to low, medium, and high risks are 5 meters, 10 meters, and 20 meters, respectively. This embodiment is merely an example; implementers can adjust the above proportions and boundary values according to this rule and in conjunction with actual scenarios.
[0075] Then, based on the affected areas of each deceleration point, the candidate path is divided into multiple continuous road segments. The division method is as follows: the starting point of the path, the starting and ending points of each affected area, and the ending point of the path are used as dividing points, and the interval between adjacent dividing points is regarded as an independent road segment, thereby dividing the candidate path into multiple continuous road segments.
[0076] For each segment after division, if the segment contains one or more target deceleration points, the benchmark speed corresponding to the highest risk level among these deceleration points is taken as the passing speed of the segment. If the segment does not contain any target deceleration points, the preset default initial inspection speed is used as the passing speed.
[0077] Next, based on the length of each road segment and its corresponding speed, the passage time of each road segment is calculated. The passage time is the ratio of the road segment length to the speed. The passage times of each road segment are then summed to obtain the estimated total time for the candidate path.
[0078] Finally, the estimated total time for all candidate paths is normalized to obtain the fire inspection task completion efficiency index for each path. The normalization method can be min-max normalization. .
[0079] in, Indicates the first The efficiency index for completing fire inspection tasks for each candidate inspection path. and These represent the maximum and minimum estimated total time among all candidate paths, respectively. No. The estimated total time for each candidate inspection path, Indicates the candidate inspection path number. .
[0080] Through this implementation process, this embodiment transforms the risk level of each target deceleration point into a quantified path time calculation, enabling the efficiency index to truly reflect the actual inspection efficiency of different paths while taking into account both safety and data quality, thus providing a reliable efficiency basis for subsequent multi-target path selection.
[0081] Furthermore, considering that the robot's operating speed at deceleration points directly affects the quality of data acquisition during fire inspections—lower speeds result in clearer images and more accurate vibration data from the sensors, which is more conducive to the early detection of fire hazards—it is necessary to convert the baseline speed at each deceleration point into a quantifiable data quality score to assess the data acquisition quality of the entire path in ensuring key inspection areas. This allows for the synergistic optimization of efficiency and quality in path planning.
[0082] In this embodiment, the specific calculation method for the inspection data quality index is as follows: For each target deceleration point on the candidate path, obtain its baseline passage speed based on risk level matching. , Indicates the target deceleration point number. .
[0083] Calculate the single-point data quality score for each target deceleration point based on the baseline traffic speed. ; .
[0084] in, The default initial inspection speed. The minimum inspection speed is preset. The lower the speed, the higher the data quality score. When the baseline speed is the minimum inspection speed, the single-point data quality score is 1, indicating the highest quality. When the baseline speed is the default initial inspection speed, the single-point data quality score is 0, indicating the lowest quality.
[0085] If there are no target deceleration points on a candidate path, the inspection data quality index of that path is directly set to a preset default value, such as 0. Otherwise, the arithmetic mean of the single-point data quality scores of all target deceleration points is taken to obtain the inspection data quality index of the candidate path.
[0086] This embodiment converts the passing speed at each deceleration point into a uniform data quality score, and reflects the overall data collection effect of the entire path in the form of an average value. This indicator, together with the efficiency indicator, forms the basis of multi-objective optimization, enabling path selection to prioritize data collection quality in high-risk areas while ensuring basic inspection efficiency.
[0087] S4. Based on the severity and marking time of past fire hazards recorded in historical fire inspection logs, dynamically adjust the weight of inspection data quality indicators.
[0088] Since the weighting of data quality indicators and efficiency indicators directly affects the tendency of route selection, and currently historical fire inspection logs are only used for post-event statistics and are not involved in real-time route planning, areas with historically high risks do not receive priority in route selection.
[0089] Based on this, this step dynamically adjusts the weight of data quality indicators based on the fire hazard index of past fire hazard points on each candidate path, so that the data quality indicators of paths with higher historical risks receive higher quality weights, making them more likely to be selected in the trade-off between efficiency and quality.
[0090] Specifically, the execution steps for dynamically adjusting the weights of inspection data quality indicators are as follows: Extract the fire hazard index of each target deceleration point on each candidate inspection path, sum them to obtain the fire hazard index of each candidate inspection path, denoted as... .
[0091] Fire hazard index based on each candidate inspection path Dynamically set the weights of inspection data quality indicators in each candidate inspection path. ; .
[0092] in, This represents the initial default weights for the quality indicators of the inspection data. Ensure that the adjusted weight does not exceed 1, to avoid a single indicator dominating the subsequent comprehensive recommendation index.
[0093] The item represents the proportion of the fire hazard index of each candidate inspection route to the sum of the fire hazard indices of all candidate inspection routes. This proportion reflects the degree of historical risk concentration of that route relative to other routes; the more concentrated the risk, the higher the risk concentration. The larger the value, the greater the amplification of the initial default weight; conversely, the smaller the value, the closer it is to the initial default weight. Through this adjustment, paths with higher fire hazard indices receive higher inspection data quality weights, meaning that when calculating the comprehensive recommendation index for subsequent candidate inspection paths, the inspection data quality index is more likely to be referenced.
[0094] Understandably, The value can be set according to the actual task requirements; for example, it can be set in a fire-specific inspection. The value is 0.7, which can be set during routine inspections. The value is 0.3. This embodiment is only an example and is not intended to impose any specific limitations.
[0095] This step dynamically adjusts the weights of inspection data quality indicators based on historical fire inspection logs, giving higher quality weights to routes with higher historical risks in the comprehensive recommendation index. This makes them more likely to be selected in the balance between efficiency and quality, solving the problem that historical logs are only used for post-event statistics and cannot influence current route selection. This ensures that the final selected inspection route achieves a balance between efficiency and quality.
[0096] S5. Based on the inspection efficiency index, inspection data quality index and dynamic weight, the comprehensive recommendation index of each candidate inspection path is calculated by linear weighted fusion. The linear weighted calculation is a common existing technical method, and the formula will not be shown here.
[0097] S6. The candidate inspection path with the highest comprehensive recommendation index is determined as the target inspection path.
[0098] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.
Claims
1. A method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things, characterized in that, include: The track structure feature data, track health monitoring data, and historical fire inspection logs of each candidate inspection path are obtained. Target deceleration points on each candidate path are identified, and the specific identification process is as follows: Extract all locations that have been marked with fire hazards from historical fire inspection logs and record them as past fire hazard points; Mark the curve curvature exceeding the preset curvature threshold, the slope exceeding the preset slope threshold, and the location of the track structure joint in the track structure feature data as fixed deceleration points of the track. The real-time fixed vibration frequency and the detected vibration frequency during inspection are extracted from the track health monitoring data for each track monitoring point. The fixed vibration frequency and the detected vibration frequency are combined to determine whether there is a loosening trend or a sudden vibration trend. Track monitoring points with a loosening trend or a sudden vibration trend are marked as track condition hazard points. Past fire hazard points, fixed deceleration points on the track, and potential track condition hazard points are selected as target deceleration points on the candidate path. Risk classification is performed on each target deceleration point to obtain the risk level. Based on the risk level, a benchmark passage speed is matched, and the fire inspection task completion efficiency index and inspection data quality index of each candidate inspection path are calculated in combination with the benchmark passage speed. The specific calculation process for the efficiency index of fire inspection task completion is as follows: The total length of each candidate inspection path is calculated, and the risk level of each target deceleration point in the candidate inspection path is obtained. Based on the risk level of each target deceleration point, the length of the affected section of each target deceleration point is determined, and based on the length of the affected section, the candidate path is divided into multiple continuous road segments; For each road segment, if the road segment contains a target deceleration point, the benchmark speed corresponding to the highest risk level of each target deceleration point in the road segment shall be used as the passing speed of the road segment; if the road segment does not contain a target deceleration point, the preset default initial inspection speed shall be used as the passing speed. Based on the length of each road segment and its corresponding speed, the transit time of each road segment is calculated, and the total estimated time of the candidate route is obtained by summing them up. The estimated total time is normalized to obtain the efficiency index for completing the fire inspection task; The specific calculation method for the inspection data quality indicators is as follows: The baseline travel speed at each target deceleration point in the candidate inspection path is denoted as . Calculate the single-point data quality score for each target deceleration point. , , The default initial inspection speed. This is the preset minimum inspection speed. Indicates the target deceleration point number. ; If there is no target deceleration point on the candidate path, the inspection data quality index is set to the preset default value; otherwise, the average of the single-point data quality scores of all target deceleration points is taken to obtain the inspection data quality index of the candidate path. Based on the severity and marking time of past fire hazards recorded in historical fire inspection logs, the weights of inspection data quality indicators are dynamically adjusted. The specific dynamic adjustment method is as follows: Extract the fire hazard index of each target deceleration point on each candidate inspection path, sum them up to obtain the fire hazard index of each candidate inspection path, denoted as . , Indicates the candidate inspection path number. ; Based on the fire hazard index of each candidate inspection path, the weights of the inspection data quality indicators in each candidate inspection path are dynamically set. , , This represents the initial default weights for the quality indicators of the inspection data; Based on inspection efficiency indicators, inspection data quality indicators, and dynamic weights, a comprehensive recommendation index for each candidate inspection path is calculated through weighted fusion. The candidate inspection path with the highest comprehensive recommendation index is determined as the target inspection path.
2. The method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things as described in claim 1, characterized in that: The specific steps for determining whether there is a tendency to loosen or a sudden change in vibration are as follows: The vibration frequencies detected by each track monitoring point during multiple inspections within a preset historical time window are obtained, forming a vibration frequency time series and performing linear regression fitting to obtain the slope of vibration frequency change over time. If the slope of the change is negative and its absolute value exceeds the first preset slope threshold, and the goodness of fit of the linear regression exceeds the preset fit threshold, then it is determined that the track monitoring point has a loosening tendency. The real-time fixed vibration frequency of each track monitoring point is used as its current reference value, and the vibration frequency detected during the most recent inspection is used as the current passing value. Calculate the difference between the current passing value and the average value of the vibration frequency detected by multiple inspections, and take the absolute value of the difference as the vibration frequency difference. At the same time, calculate the ratio of the current passing value to the current reference value to obtain the current vibration deviation. If the detected vibration frequency difference at a certain track monitoring point exceeds the preset vibration frequency difference value and the current vibration deviation of the track monitoring point exceeds the preset deviation value, then a sudden vibration trend is identified.
3. The method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things as described in claim 1, characterized in that: The specific implementation method for risk classification of each target deceleration point is as follows: For each target deceleration point, determine whether it belongs to a single category. If it belongs to a single category, perform the following sub-steps: If the category is a past fire hazard point, extract the number of historical hazard markings for that point from the historical fire inspection log, as well as the time and severity of each hazard marking, calculate the fire hazard index for that point, and match the risk level corresponding to the fire hazard index. If the category is a fixed deceleration point on the track, its risk level is determined according to the curve curvature and gradient, and in accordance with the preset risk level classification rules. If the category is a potential safety hazard point in the track condition, and if the potential hazard point has both a tendency to loosen and a sudden change in vibration, then the risk level of the potential hazard point is marked as high risk; if the potential hazard point only has a sudden change in vibration, then the risk level of the potential hazard point is marked as low risk; if the potential hazard point only has a tendency to loosen, then the risk level of the potential hazard point is marked as medium risk. If a target belongs to multiple categories, the risk level of each category is assessed using the method for a single category, and the highest level is taken as the final risk level of the target deceleration point.
4. The method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things as described in claim 3, characterized in that: The specific calculation steps for the fire hazard index are as follows: The severity of each hazard marker is converted into a severity score, and marked as follows: , Indicates the sequence number of the hazard marker. ; The weight of the impact of densely packed hazards is set based on the time interval between adjacent hazard markers, denoted as . ; Calculating the fire hazard index , In the formula, Indicates the first The time interval between the time of marking the next hazard and the start time of the current inspection task. It is a natural constant. Mark the number of times historical hidden dangers are identified. This is the preset time decay factor.
5. The method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things as described in claim 4, characterized in that: The specific method for setting the weight of the impact of densely populated hidden dangers is as follows: Calculate the mean time interval between adjacent hazard markers and standard deviation Thus, the coefficient of variation is obtained. ; Calculate the impact weight of densely populated hidden dangers based on the mean and coefficient of variation. , In the formula, To set a reference time interval.
6. The method for intelligent inspection path planning of fire-fighting equipment based on the Internet of Things as described in claim 1, characterized in that: The determination of the influence zone length of each target deceleration point based on the comprehensive risk level of each target deceleration point includes: a first length corresponding to a high risk level, a second length corresponding to a medium risk level, and a third length corresponding to a low risk level, with the first length > the second length > the third length.