A method for online monitoring and evaluation of municipal road cleaning quality

Abstract

The application relates to the technical field of municipal road management, in particular to an online monitoring and evaluation method for municipal road cleaning and sanitation quality, which comprises the following steps: obtaining three types of operation data of target road video monitoring, vehicle trajectory and road surface cleanliness sensor, and calculating an initial sanitation quality score; marking a road section as a key attention road section if the score is lower than a first quality threshold value, obtaining road section attribute information, analyzing a quality calibration factor and correcting the initial score to obtain a target score; and generating an improvement instruction if the target score still does not meet the standard. The application helps to overcome the defects of low monitoring efficiency, inaccurate evaluation and neglect of road section differences in the prior art.

Description

Technical Field

[0001] This application relates to the field of municipal road management technology, and in particular to an online monitoring and evaluation method for the quality of municipal road cleaning and maintenance. Background Technology

[0002] As the core carrier of urban transportation and citizens' travel, the quality of municipal road cleaning and maintenance directly affects the city's appearance, public environmental sanitation, and citizens' living experience. Currently, the monitoring and evaluation of municipal road cleaning and maintenance quality largely relies on manual inspections, which suffers from problems such as low efficiency, limited coverage, inconsistent evaluation standards, and strong subjectivity, making it difficult to achieve real-time, comprehensive, and accurate monitoring of road cleaning quality. Furthermore, different road sections have different attributes (such as commercial districts, residential areas, and road structures), making them susceptible to different pollution sources. Existing evaluation methods do not fully consider these differences, resulting in less objective and accurate evaluation results, failing to provide a scientific basis for optimizing and rectifying cleaning and maintenance operations. Therefore, there is an urgent need for a method for monitoring and evaluating the quality of municipal road cleaning and maintenance that can achieve online real-time monitoring, take into account differences between road sections, and provide accurate evaluations, thereby addressing the shortcomings of existing technologies. Summary of the Invention

[0003] To help overcome the shortcomings of existing technologies in monitoring the quality of municipal road cleaning and maintenance, such as low efficiency, inaccurate evaluation, and failure to take into account differences between road sections, this application provides an online monitoring and evaluation method for the quality of municipal road cleaning and maintenance.

[0004] This application provides an online monitoring and evaluation method for the quality of municipal road cleaning and maintenance, which adopts the following technical solution: An online monitoring and evaluation method for the quality of municipal road cleaning and maintenance includes: Acquire sweeping and cleaning operation data of the target road, including video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data; An initial cleaning quality score is obtained based on the video surveillance data, the vehicle trajectory data, and the road surface cleanliness sensor data. If the initial cleaning quality score is lower than the first quality threshold, the target road will be marked as a key concern section. Obtain the road segment attribute information corresponding to the key road segments of interest; Based on the road segment attribute information, the cleaning and maintenance operation data are analyzed to obtain a quality calibration factor; Based on the initial cleaning quality score and the quality calibration factor, the target cleaning quality score is obtained; If the target cleaning quality score is lower than the first quality threshold, a road section rectification instruction will be generated.

[0005] Optionally, obtaining the initial cleaning quality score based on the video surveillance data, the vehicle trajectory data, and the road surface cleanliness sensor data includes: Based on the video surveillance data, extract continuous video frames; The continuous video frames are segmented to obtain road surface area images, and road surface area images with exposed garbage or accumulated dust and mud are used as defect images. Based on the continuous video frames and the defect image, obtain the defect coverage probability value; Based on the road surface cleanliness sensor data, the road surface dust load value is obtained; Based on the vehicle trajectory data, the number of cleaning coverage times and the cleaning speed are obtained; An initial cleaning quality score is obtained based on the defect coverage probability value, the road surface dust load value, the number of sweeping coverages, and the sweeping speed.

[0006] Optionally, the step of analyzing the cleaning and maintenance operation data based on the road segment attribute information to obtain the quality calibration factor includes: Based on the road segment attribute information, determine whether there are any sources of pollution. If the pollution source exists, then obtain the pollution emission data corresponding to the pollution source; If the pollution emission data exceeds the corresponding emission data threshold, then based on the video surveillance data, it is determined whether there are cleaning and maintenance personnel or cleaning vehicles in the key road section of concern. If there are cleaning and maintenance personnel or cleaning vehicles in the key monitored road sections, information on their work activities will be obtained. Based on the aforementioned work behavior information, determine whether quality calibration is required; If quality calibration is required, obtain the quality calibration factor.

[0007] Optionally, determining whether a pollution source exists based on the road segment attribute information includes: Obtain the historical cleaning records corresponding to the key road sections under focus; Based on the historical cleaning records, the causes of pollution for each instance of substandard cleaning are identified. Obtain the frequency percentage of the causes corresponding to the pollution causes; If the frequency of the aforementioned cause exceeds a preset frequency threshold, then the presence of the easily polluting source is determined. If the frequency of the aforementioned reasons does not exceed the preset frequency threshold, then based on the road segment attribute information, it is determined that the key focus road segment is a commercial street segment or a residential area segment. If the key focus area is the commercial street section, then obtain the business type of the shops and the time period for garbage disposal; If the business type of the shop and the time period for garbage disposal affect the cleaning and sanitation operation data, then it is determined that there is a source of pollution. If the key road segment is the residential area road segment, then obtain the distribution of surrounding facilities corresponding to the residential area road segment; Based on the distribution of surrounding facilities, the potential pollution risk value corresponding to the cleaning and maintenance operation data is obtained; If the potential pollution risk value exceeds the second quality threshold, it is determined that there is a source of pollution.

[0008] Optionally, determining whether quality calibration is needed based on the work behavior information includes: Based on the video surveillance data, the location of the garbage accumulation and the focus of the workers' eyes are obtained; If the line of sight focuses on the location of the garbage, then it is determined whether the duration of the operation is less than the first duration threshold. If the duration of the operation is less than the first duration threshold, then it is determined whether there is a cleaning action sound in the operation audio; If the cleaning sound is present, then it is determined that no quality calibration is required; If the cleaning sound is not present, a quality calibration is required.

[0009] Optionally, determining whether quality calibration is needed based on the work behavior information further includes: Based on the video surveillance data, the brush contact status and brush speed of the sweeping vehicle are obtained; If the sweeping brush is in a ground contact state and the sweeping brush speed is greater than the speed threshold, then the dwell time of the sweeping vehicle in the key concern road section is obtained; If the dwell time is greater than the second duration threshold, it is determined that no quality calibration is required; If the dwell time is less than or equal to the second dwell time threshold, the change in road dust load during the dwell period is obtained based on the road surface cleanliness sensor data. If the change in road surface dust load is less than the change threshold, then a quality calibration is required. If the change in road surface dust load is greater than or equal to the change threshold, then it is determined that no quality calibration is required.

[0010] Optionally, determining whether quality calibration is needed based on the work behavior information further includes: Obtain information on the different operational behaviors of multiple cleaning and maintenance personnel or cleaning vehicles within the same key road section; Determine whether there is a contradictory relationship between different job behavior information. The contradictory relationship is that the job behavior information has opposite quality implications. The quality implications are implications used to characterize whether the cleaning has been completed to the required standard. If there is no conflict between different job behavior information, then it is determined whether quality calibration is required based on preset job rules; If there is a conflict between different job behavior information, then the conflict is determined to be an internal conflict within the same job unit or a conflict between different job units. If the conflicting relationship is an internal conflict within the same work unit, then obtain the proportion of conflicting behaviors corresponding to the conflicting relationship. If the proportion of contradictory behaviors is greater than a preset proportion threshold, it is determined that no quality calibration is required. If the proportion of contradictory behaviors is less than the preset proportion threshold, it is determined that quality calibration is required. If the proportion of contradictory behaviors is equal to the preset proportion threshold, then the priority level of different work behavior information is obtained; If the priority level satisfies the preset priority relationship, it is determined that no quality calibration is required; If the priority level does not meet the preset priority relationship, it is determined that quality calibration is required.

[0011] Optionally, after determining whether a conflicting relationship exists between different job behavior information and whether it is an internal conflict within the same job unit or a conflict between different job units, the method further includes: If the conflicting relationship between the tasks is a conflict between different task units, then based on the video surveillance data, the accurate historical records of the tasks of different task units are obtained; Based on the accurate historical records of the tasks, the historical compliance rate corresponding to the task behavior information is obtained; If the historical compliance rate of the work unit is the highest and the distance from the work location to the garbage accumulation location is the smallest, then it is determined whether quality calibration is needed based on the work behavior information of the work unit. If there is no work unit with the highest historical compliance rate and the shortest distance from the garbage accumulation location, then the comprehensive score of different work units is obtained based on the historical compliance rate and the work distance. Based on the operational behavior information of the work unit with the highest comprehensive score, it is determined whether quality calibration is required.

[0012] Optionally, determining whether a pollution source exists based on the road segment attribute information further includes: Obtain multiple records of substandard cleaning of the key road sections within a preset historical period, with each record of substandard cleaning corresponding to a specific time of substandard cleaning. Centered on each time point where the standard is not met, extract the road segment image data within the forward and backward time windows; Target identification is performed on the image data of the road section, and non-fixed pollution events occurring within each time window are extracted; If the same non-stationary pollution event occurs repeatedly within multiple time windows of non-compliance, the location of the non-stationary pollution event is marked as a source of pollution. If the non-fixed pollution event does not recur within the time window of multiple non-compliance times, then the pavement structure information of the key concern road section is obtained; Based on the road surface structure information, determine whether there are structural defects. If the location of the structural defect exists, the rate of garbage or silt accumulation at the location of the structural defect under dry and humid weather conditions is obtained. If the accumulation rate is lower than a first velocity threshold in dry weather and higher than a second velocity threshold in humid weather, then the location of the structural defect is determined to be a source of easy contamination. If the accumulation rate is not lower than the first speed threshold in dry weather, or not higher than the second speed threshold in humid weather, then the location of the structural defect is determined not to be a source of easy contamination.

[0013] In summary, this application includes the following beneficial technical effects: By acquiring three core operational data sources—video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data—the system achieves comprehensive coverage of the cleaning and maintenance process and the cleanliness status of the road surface, eliminating reliance on manual inspections and improving the real-time nature and comprehensiveness of monitoring. Initial cleaning quality scores are calculated through multi-dimensional data fusion, and by combining road segment attribute information to identify pollution sources and analyze operational behavior information, quality calibration factors are obtained to calibrate the initial scores. This fully considers the attribute differences and pollution influencing factors of different road segments, solving the problems of strong subjectivity and neglect of road segment differences in existing evaluation methods, making the target cleaning quality scores more objective and accurate. Rectification instructions are generated for key monitoring road segments that fail to meet the scores, providing a clear scientific basis for optimizing cleaning and maintenance operations, personnel scheduling, and rectification implementation. This effectively improves the quality and efficiency of municipal road cleaning and maintenance, and enhances the city's appearance and public environmental sanitation. Attached Figure Description

[0014] Figure 1 This is a flowchart of the main process of an online monitoring and evaluation method for the quality of municipal road cleaning and maintenance, according to an embodiment of this application. Detailed Implementation

[0015] This application discloses an online monitoring and evaluation method for the quality of municipal road cleaning and maintenance.

[0016] Reference Figure 1 An online monitoring and evaluation method for the quality of municipal road cleaning and maintenance includes steps S101 to S107: Step S101: Obtain cleaning and maintenance operation data of the target road. The cleaning and maintenance operation data includes video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data.

[0017] Specifically, target roads refer to municipal roads that require quality monitoring and evaluation of cleaning and maintenance. These can be selected as single road sections, multiple road sections, or entire areas, based on the urban road management divisions. Cleaning and maintenance operation data are core data reflecting the cleaning process and road surface cleanliness. Video surveillance data is collected through high-definition surveillance cameras deployed along the target roads to capture real-time information on road debris, dust, and mud accumulation, as well as the operational status of cleaning personnel and vehicles. Vehicle trajectory data is collected through GPS positioning modules installed on cleaning vehicles, recording their travel paths, speeds, and operating times to determine cleaning coverage and efficiency. Road surface cleanliness sensor data is collected through dust sensors and particulate matter sensors installed on the road surface or cleaning vehicles to quantify cleanliness indicators such as dust content and debris residue, providing a quantitative basis for quality scoring. These three types of data are collected synchronously and transmitted in real-time to ensure data timeliness and completeness.

[0018] Step S102: Obtain an initial cleaning quality score based on video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data.

[0019] Specifically, the initial cleaning quality score is a preliminary quality evaluation result calculated based on three types of core operational data through a pre-set scoring model. It is used to initially determine whether the cleaning quality of the target road meets the standards. The scoring model adopts a weighted summation method, assigning different weights to each data indicator based on its importance. Among them, road surface cleanliness-related indicators (such as dust load value) have the highest weight, while cleaning operation process indicators (such as cleaning coverage times and cleaning speed) and road surface defect indicators (such as defect coverage probability value) are assigned corresponding weights in turn, ensuring that the score can comprehensively reflect the core influencing factors of cleaning quality.

[0020] Step S103: If the initial cleaning quality score is lower than the first quality threshold, the target road is marked as a key concern section.

[0021] Specifically, the first quality threshold is a pre-set critical value for compliance based on municipal road cleaning and maintenance quality standards. It is determined by the urban road management department, taking into account local environmental sanitation requirements and road classification (such as main roads, secondary roads, and branch roads). Different first quality thresholds can be set for roads of different classifications. When the initial cleaning quality score is lower than this threshold, it indicates that the cleaning quality of the target road has not met the basic requirements. This road should be marked as a key area of ​​concern for further detailed analysis and calibration to ensure the accuracy of the evaluation results and avoid misjudgments due to single data deviations.

[0022] Step S104: Obtain the road segment attribute information corresponding to the key road segments.

[0023] Specifically, road segment attribute information refers to basic and environmental information related to key road segments, mainly including road segment type (commercial street segment, residential area segment, industrial area segment, main road, secondary road, etc.), pavement structure information (drainage outlet location, curbstone integrity, pavement joint condition, etc.), distribution of surrounding facilities (shops, schools, residential areas, construction sites, etc.), historical cleaning records, and distribution of potential pollution sources. This information can be obtained from urban road management databases and geographic information systems (GIS) and is used for subsequent analysis of the root causes of substandard road segment cleaning quality, providing a basis for quality calibration.

[0024] Step S105: Analyze the cleaning and maintenance operation data based on the road segment attribute information to obtain the quality calibration factor.

[0025] Specifically, the quality calibration factor is a parameter used to correct the initial cleaning quality score. Its core function is to eliminate the impact of differences in road segment attributes on the cleaning quality evaluation. Because different road segments have different attributes, their pollution levels, pollution sources, and cleaning difficulties vary. For example, commercial street segments generate a large amount of garbage and experience frequent pollution, while residential street segments have garbage disposal concentrated at specific times. These factors can lead to different cleaning quality performances under the same cleaning operations. By analyzing road segment attribute information, identifying pollution sources, and judging the rationality of cleaning operations, the quality calibration factor is determined, making the calibrated target cleaning quality score more closely reflect the actual situation of the road segment.

[0026] Step S106: Obtain the target cleaning quality score based on the initial cleaning quality score and the quality calibration factor.

[0027] Specifically, the target cleaning quality score is the final score after the initial cleaning quality score has been corrected by a quality calibration factor. The calculation method is to multiply the initial cleaning quality score by the quality calibration factor (when the road section is difficult to clean and heavily polluted, the calibration factor is greater than 1, and the initial score is appropriately increased; when the road section is easy to clean and lightly polluted, the calibration factor is less than 1, and the initial score is appropriately decreased). This ensures that the final score can objectively reflect the actual cleaning quality of the road section and avoid unfair evaluation due to differences in road section attributes.

[0028] Step S107: If the target cleaning quality score is lower than the first quality threshold, a road section rectification instruction is generated.

[0029] Specifically, the road section rectification instruction is used to guide cleaning and sanitation companies in optimizing their operations and improving quality. It includes the specific location of the road section to be focused on, the specific problems with substandard cleaning quality (such as garbage accumulation, severe dust buildup, insufficient cleaning coverage, etc.), rectification requirements (such as rectification deadlines, rectification standards, required increase in cleaning frequency or personnel), and the acceptance standards after rectification. Once generated, the instruction is sent to the corresponding cleaning and sanitation company and management personnel through the system platform to ensure the orderly progress of rectification work and timely improvement of road section cleaning quality.

[0030] In one embodiment of this example, step S102, based on video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data, obtains an initial cleaning quality score, including steps S201 to S206: Step S201: Extract continuous video frames based on video surveillance data.

[0031] Specifically, video surveillance data consists of continuous dynamic images. To facilitate the analysis of road conditions and operational status, it is necessary to extract video frames at fixed time intervals (e.g., one frame every 10 seconds) from the continuous video to form a continuous sequence of static images. The extracted video frames must be clear enough to clearly identify road debris, dust accumulation, and the operational status of cleaning personnel and vehicles, providing a foundation for subsequent image segmentation and defect identification.

[0032] Step S202: Perform image segmentation on consecutive video frames to obtain road surface area images, and take the road surface area images with exposed garbage or accumulated dust and mud as defect images.

[0033] Specifically, image segmentation uses computer vision algorithms (such as semantic segmentation) to separate the road surface area from non-road surface areas (such as sidewalks, green belts, buildings, vehicles, etc.) in a video frame, retaining only the road surface area image and eliminating the interference of non-road surface areas on the sanitation quality evaluation. Exposed litter refers to visible solid waste on the road surface (such as plastic bags, paper scraps, fruit peels, etc.), while accumulated dust and mud refer to pollutants such as dust and mud accumulated on the road surface. Image recognition algorithms detect the presence of these pollutants in the road surface area image; if present, the image is marked as a defect image, which is used to subsequently calculate the defect coverage probability value.

[0034] Step S203: Obtain the defect coverage probability value based on continuous video frames and defect images.

[0035] Specifically, the defect coverage probability value refers to the proportion of defective images in all extracted consecutive video frames, calculated by combining the proportion of the defective area in the road surface area image. It is used to quantify the degree of road surface contaminant coverage. The calculation method is: Defect Coverage Probability Value = (Number of Defective Images / Total Number of Video Frames) × (Average Area of ​​Defective Area / Total Area of ​​Road Surface Area). The larger this value, the wider the coverage of road surface contaminants, the more severe the pollution, and the greater the negative impact on the initial cleaning quality score.

[0036] Step S204: Obtain the road dust load value based on the road surface cleanliness sensor data.

[0037] Specifically, the road surface cleanliness sensor data includes real-time data on road dust concentration and particulate matter content. The road dust load value is a quantitative indicator derived from statistical analysis of these data, reflecting the degree of road dust accumulation. The data acquisition cycle is consistent with the video frame extraction cycle, and the average value of sensor data within the same time period is taken as the road dust load value, with units of mg / m³. 2 The higher the value, the lower the cleanliness of the road surface and the worse the cleaning quality.

[0038] Step S205: Based on vehicle trajectory data, obtain the number of cleaning coverage times and cleaning speed.

[0039] Specifically, the sweeping coverage count refers to the number of effective sweeps performed by the sweeping vehicle on the target road. By analyzing vehicle trajectory data, it is determined whether the vehicle completely covers all lanes of the target road. A single complete coverage is counted as one effective sweep. Repeated coverage of the same area does not count as an additional sweep. This reflects the completeness of the sweeping operation's coverage. Sweeping speed refers to the average speed of the sweeping vehicle during operation, calculated from the travel distance and time in the vehicle trajectory data, in km / h. Excessive sweeping speed leads to incomplete cleaning, while excessively slow speed affects operational efficiency. Therefore, a reasonable speed range needs to be set as an important reference for scoring.

[0040] Step S206: Obtain an initial cleaning quality score based on the defect coverage probability value, road surface dust load value, number of sweeping coverage times, and sweeping speed.

[0041] Specifically, a weighted summation method is used to calculate the initial cleaning quality score, with preset weights for each indicator (which can be adjusted according to actual management needs): road dust load value 40%, defect coverage probability value 30%, sweeping coverage frequency 15%, and sweeping speed 15%. First, each indicator is standardized (converting indicator values ​​to scores from 0 to 100). Road dust load value and defect coverage probability value are negatively correlated with the score (the higher the indicator value, the lower the standardized score), while sweeping coverage frequency and sweeping speed are positively correlated with the score (the more reasonable the indicator value, the higher the standardized score). Then, the standardized score of each indicator is multiplied by its corresponding weight, and the summation yields the initial cleaning quality score (out of 100). A higher score indicates better initial cleaning quality.

[0042] In one embodiment of this example, step S105, based on road segment attribute information, analyzes the cleaning and maintenance operation data to obtain quality calibration factors, including steps S301 to S306: Step S301: Based on the road segment attribute information, determine whether there are any sources of pollution.

[0043] Specifically, pollution sources refer to areas or factors that easily generate pollutants and lead to a decline in road surface cleaning quality. These include fixed pollution sources (such as areas with concentrated shops, construction site entrances and exits, and areas around garbage transfer stations) and non-fixed pollution sources (such as temporary construction waste dumping sites and vendor operating areas). By analyzing road segment attribute information such as road segment type, distribution of surrounding facilities, and historical cleaning records, it is determined whether the key road segments under focus have the aforementioned pollution sources. The presence of pollution sources increases the difficulty of cleaning and needs to be considered during quality calibration.

[0044] Step S302: If there is a source of pollution, obtain the pollution emission data corresponding to the source of pollution.

[0045] Specifically, pollution emission data refers to data on the types, quantities, and frequencies of pollutants generated by easily polluting sources, such as the amount of household waste generated by shops, the amount of mud and sand carried out at construction site entrances and exits, and the amount of waste generated by street vendors. This data can be obtained through video surveillance data analysis, sensor data collection, and historical record queries, and is used to determine the degree of impact of easily polluting sources on the quality of road cleaning.

[0046] Step S303: If the pollution emission data exceeds the corresponding emission data threshold, then based on the video surveillance data, determine whether there are cleaning and maintenance personnel or cleaning vehicles in the key road sections.

[0047] Specifically, emission data thresholds are pre-set critical values ​​for pollutant emissions based on the types of pollution sources, such as thresholds for household waste emissions from shops and thresholds for sediment carried out from construction sites. Exceeding these thresholds indicates that the pollution sources have a significant impact on road surface pollution, requiring further assessment of whether cleaning operations are being carried out in a timely manner. Through real-time analysis of video surveillance data, it is possible to identify whether cleaning personnel (such as sanitation workers) or cleaning vehicles are operating in key monitored road sections. If so, it indicates that cleaning operations have been carried out; if not, it indicates that cleaning operations are lagging behind, and the calibration factor needs to be adjusted during quality calibration.

[0048] Step S304: If there are cleaning and maintenance personnel or cleaning vehicles in the key road section, obtain the operation behavior information.

[0049] Specifically, the operational behavior information reflects the quality of the cleaning and maintenance work performed by personnel or vehicles, and is used to determine whether the work is standardized and thorough. Operational actions are identified through video surveillance data, such as the sweeping and litter-picking actions of cleaning personnel, and the brush rotation and water spraying actions of cleaning vehicles. Operational dwell time refers to the duration of time that cleaning personnel or vehicles remain in key monitored areas, used to determine whether the work is sufficient. Operational audio is acquired through the microphones built into the surveillance cameras or separately deployed audio acquisition equipment, including sounds produced by cleaning actions (such as sweeping sounds and brush rotation sounds) and the communication voices of the workers, used to further determine whether the work was actually carried out.

[0050] Step S305: Based on the work behavior information, determine whether quality calibration is required.

[0051] Specifically, the core of determining whether quality calibration is needed is to analyze whether the cleaning operation is standardized and thorough. If the operation is standardized and can effectively address the pollution impact of easily polluting sources, it means that the initial cleaning quality score can reflect the actual situation and no calibration is needed. If the operation is not standardized and does not effectively address the pollution impact, it means that the initial score is too low, which may be due to inadequate operation. It needs to be corrected through quality calibration factors to avoid misjudging the cleaning quality of the road section.

[0052] Step S306: If quality calibration is required, obtain the quality calibration factor.

[0053] Specifically, the quality calibration factor ranges from 0.8 to 1.2, determined based on the degree of non-standard operation and the impact of potential pollution sources. For example, if the operation is slightly non-standard and the impact of potential pollution sources is small, the calibration factor is 1.05; if the operation is severely non-standard and the impact of potential pollution sources is large, the calibration factor is 1.15; and if the operation is basically standard and the impact of potential pollution sources is small, the calibration factor is 1.0. The specific value of the calibration factor can be calculated through a preset calibration model, combined with the operation behavior score and the potential pollution source impact score, to ensure that the calibrated score can objectively reflect the actual cleaning quality of the road section.

[0054] In one embodiment of this example, step S301, based on road segment attribute information, determines whether there is a pollution source, including steps S401 to S410: Step S401: Obtain the historical cleaning records corresponding to the key road sections.

[0055] Specifically, historical cleaning records refer to the cleaning quality evaluation records, cleaning operation records, and rectification records of key road sections over a period of time (such as the past 3 months or 6 months). These records include information such as the compliance status of each cleaning operation, the reasons for non-compliance, the cleaning frequency, and the personnel involved. They are obtained from the urban road cleaning management database and are used to analyze the patterns and main reasons for non-compliance in road cleaning.

[0056] Step S402: Based on historical cleaning records, obtain the cause of pollution for each instance of substandard cleaning.

[0057] Specifically, the cause of pollution refers to the specific factors that lead to substandard road cleaning, mainly including garbage accumulation, dust and mud accumulation, insufficient sweeping and covering, and pollution from easily polluting sources. By reviewing the explanations of substandard cleaning in historical cleaning records, the specific cause of pollution corresponding to each substandard cleaning can be extracted. For example, if substandard cleaning is caused by garbage disposal by shops multiple times within a certain period, then the cause of pollution is garbage disposal by shops.

[0058] Step S403: Obtain the frequency percentage of each cause corresponding to a pollution cause.

[0059] Specifically, the frequency percentage of a cause refers to the proportion of a particular pollution cause appearing in all records of non-compliance with cleaning standards out of the total number of non-compliance records. The calculation method is: Frequency percentage of a cause = (Number of occurrences of a particular pollution cause / Total number of non-compliance records) × 100%. For example, if a section of road experienced 20 instances of non-compliance with cleaning standards in the past 6 months, and 12 of these instances were due to garbage disposal by shops, then the frequency percentage of that pollution cause is 60%.

[0060] Step S404: If the frequency of a cause exceeds a preset frequency threshold, it is determined that there is a source of pollution.

[0061] Specifically, the preset frequency threshold is a pre-set critical value for judging potential pollution sources, usually set at 50% (which can be adjusted according to actual conditions). If the frequency percentage of a certain pollution cause exceeds this threshold, it indicates that the pollution cause is the main factor leading to substandard road cleaning, and the corresponding pollution source (such as shops or construction sites) is a potential pollution source. For example, if the frequency percentage of garbage disposal by shops is 60%, exceeding 50%, then it is determined that there is a potential pollution source centered on shops in that road section.

[0062] Step S405: If the frequency of no cause exceeds the preset frequency threshold, then based on the road segment attribute information, determine whether the key road segment to focus on is a commercial street segment or a residential street segment.

[0063] Specifically, if the frequency percentage of all pollution causes does not exceed the preset frequency threshold, it indicates that the substandard road cleaning is the result of multiple factors working together. In this case, further analysis is needed in conjunction with the type of road section, focusing on whether the road section is a commercial street or a residential area, because the pollution characteristics and cleaning difficulty of these two types of road sections are significantly different from other road sections (such as industrial area road sections and main roads), and are prone to specific types of pollution.

[0064] Step S406: If the key focus area is a commercial street, obtain the business type of the shops and the time of garbage disposal.

[0065] Specifically, commercial street sections refer to sections with concentrated shops, high pedestrian traffic, and frequent commercial activities. Shop types include catering, retail, and entertainment, and different types of shops generate different types and amounts of waste (e.g., catering shops generate a large amount of kitchen waste, while retail shops generate a large amount of packaging waste). Waste disposal times refer to the peak times when shops and pedestrians dispose of waste (e.g., breakfast time, dinner time, and after business hours). This information is obtained through street section attribute information queries and video surveillance data analysis.

[0066] Step S407: If the business type of the shop and the time of garbage disposal affect the cleaning and maintenance operation data, it is determined that there is a source of pollution.

[0067] Specifically, assessing the impact of shop operation type and garbage disposal time on cleaning and maintenance data mainly involves examining whether it leads to a surge in street litter volume, increased cleaning difficulty, and consequently, a decline in cleaning quality. For example, in commercial streets with a high concentration of restaurants, a large amount of food waste is generated during dinner hours, resulting in increased street litter accumulation and a higher probability of defective coverage. If this impact persists and leads to substandard cleaning, then the commercial street section is considered to have a source of pollution (i.e., an area with a high concentration of restaurants).

[0068] Step S408: If the key road segment is a residential area road segment, then obtain the distribution of surrounding facilities corresponding to the residential area road segment.

[0069] Specifically, residential road sections refer to municipal roads surrounding residential communities. The distribution of surrounding facilities includes community entrances and exits, garbage collection points, kindergartens, convenience stores, etc. The distribution of these facilities can affect the pollution of the road surface. For example, household waste is easily generated near community entrances and exits, garbage is easily scattered around garbage collection points, and children's toys, snack packaging, and other garbage are easily generated around kindergartens.

[0070] Step S409: Based on the distribution of surrounding facilities, obtain the potential pollution risk value corresponding to the cleaning and maintenance operation data.

[0071] Specifically, the potential pollution risk value is an indicator used to quantify the degree of pollution impact of surrounding facilities on a road section. It is calculated based on factors such as the type, number, and distance of the surrounding facilities from the road section. For example, the closer the garbage collection points are to the road section and the more numerous they are, the higher the potential pollution risk value; the more facilities such as kindergartens and convenience stores there are, the higher the potential pollution risk value will be. The calculation of the potential pollution risk value adopts a graded assignment method, with different types of facilities corresponding to different base scores, and the scores are adjusted in combination with the distance factor to finally obtain a comprehensive potential pollution risk value (0-100 points).

[0072] Step S410: If the potential pollution risk value exceeds the second quality threshold, it is determined that there is a source of pollution.

[0073] Specifically, the second quality threshold is a preset potential pollution risk threshold, usually set at 60 points (which can be adjusted according to the actual situation). If the potential pollution risk value exceeds this threshold, it indicates that the surrounding facilities have a significant impact on the pollution of the road section, which may easily lead to substandard road cleaning quality. At this time, it is determined that there are sources of pollution in the residential area (such as around garbage collection points or near the entrances and exits of the community).

[0074] In one embodiment of this example, step S305, based on the work behavior information, determines whether quality calibration is required, including steps S501 to S505: Step S501: Based on video surveillance data, obtain the location of garbage accumulation and the focus of the workers' line of sight.

[0075] Specifically, the location of the garbage accumulation refers to the exact coordinates of the garbage accumulation on the road surface. This is extracted using image recognition algorithms from video surveillance data to accurately locate the lane and area where the garbage is located. The location of the worker's line of sight refers to the location of the road surface corresponding to the direction the worker's eyes are looking at. This is obtained using facial recognition algorithms and eye-tracking algorithms to determine whether the worker has discovered the garbage accumulation location.

[0076] Step S502: If the focus of the line of sight matches the location where the garbage is stuck, then determine whether the duration of the operation is less than the first duration threshold.

[0077] Specifically, if the focus of the worker's gaze matches the location of the litter, it indicates that the worker has discovered the litter on the road. At this point, it is necessary to determine whether the worker has processed the litter. The first time threshold is a preset minimum time (e.g., 30 seconds) required for the worker to process a single piece of litter. If the processing time is less than this threshold, it indicates that the worker may not have thoroughly processed the litter and has only stayed briefly before leaving. If the processing time is greater than or equal to this threshold, it indicates that the worker may have completed the litter cleanup.

[0078] Step S503: If the duration of the operation is less than the first duration threshold, determine whether there is a cleaning action sound in the operation audio.

[0079] Specifically, the cleaning sounds in the audio recording include sweeping sounds, sounds of picking up trash, and the rubbing of trash bags. An audio recognition algorithm determines the presence of these sounds. If cleaning sounds are present, it indicates that the worker did indeed clean up trash during their brief stop, but the stop was short due to the small amount of trash and the fast cleaning speed. If no cleaning sounds are present, it means that although the worker discovered trash, they did not perform any cleaning work, which is considered a non-standard operating procedure.

[0080] Step S504: If there is a cleaning sound, it is determined that no quality calibration is required.

[0081] Specifically, if workers find litter, pause briefly, and make cleaning noises, it indicates that their work is in accordance with regulations and they can handle litter on the road in a timely manner. The initial cleaning quality score can reflect the actual cleaning situation, so there is no need for quality calibration.

[0082] Step S505: If there is no cleaning sound, it is determined that a quality calibration is required.

[0083] Specifically, if workers find trash but stay for a short time without taking any cleaning action, it indicates that the work behavior is not standardized and the trash on the road is not dealt with in a timely manner, resulting in a low initial cleaning quality score. In this case, quality calibration is required to correct the initial score through calibration factors to avoid misjudging the cleaning quality of the road section due to non-standard work practices.

[0084] In one embodiment of this example, step S305, based on the work behavior information, determines whether quality calibration is required, and further includes steps S601 to S606: Step S601: Based on video surveillance data, obtain the sweeping brush contact status and sweeping brush speed of the sweeping vehicle.

[0085] Specifically, the brush contact status refers to whether the sweeper's brush is in contact with the road surface, which is divided into contact status and off-ground status. The contact status indicates that the sweeper is performing sweeping operations, while the off-ground status indicates that it is not performing sweeping operations. The position status of the brush is determined by image recognition algorithms based on video monitoring data. The brush rotation speed refers to the rotation speed of the sweeper's brush, which is calculated from the number of frames of brush rotation in the video monitoring data, and the unit is revolutions per minute (r / min). Whether the brush rotation speed meets the standard directly affects the sweeping effect.

[0086] Step S602: If the sweeping brush is in the ground contact state and the sweeping brush speed is greater than the speed threshold, then obtain the dwell time of the sweeping vehicle in the key road section.

[0087] Specifically, the speed threshold is a preset critical value for the effective sweeping speed of the brush (e.g., 150 r / min). If the brush touches the ground and the speed is greater than this threshold, it means that the sweeping vehicle is carrying out effective sweeping operations. The dwell time refers to the sum of the driving time and the dwell time of the sweeping vehicle in the key road section. It is used to judge the sufficiency of the sweeping vehicle's operation in that road section. The longer the dwell time, the more thorough the sweeping.

[0088] Step S603: If the dwell time is greater than the second duration threshold, it is determined that no quality calibration is required.

[0089] Specifically, the second duration threshold is a preset minimum operating time for sweeping vehicles in key monitored road sections, set according to the length of the road section and the degree of pollution (e.g., the second duration threshold is 5 minutes for a 1-kilometer road section). If the dwell time exceeds this threshold, it indicates that the sweeping vehicle has carried out sufficient sweeping operations in that road section, effectively removing pollutants from the road surface. The initial cleaning quality score reflects the actual situation, and no quality calibration is required.

[0090] Step S604: If the dwell time is less than or equal to the second dwell time threshold, then based on the road surface cleanliness sensor data, obtain the change in road surface dust load during the dwell period.

[0091] Specifically, the change in road dust load during the stay refers to the difference between the road dust load value before the sweeping vehicle enters the key concern section and the road dust load value after leaving the section. A positive difference indicates an increase in road dust load (increased pollution), while a negative difference indicates a decrease in road dust load (effective sweeping). The change is calculated by obtaining dust load values ​​at different time points using road cleanliness sensors, and is used to judge the actual effectiveness of the sweeping operation.

[0092] Step S605: If the change in road dust load is less than the change threshold, it is determined that quality calibration is required.

[0093] Specifically, the change threshold is a preset critical value for the effective cleaning operation (e.g., -5mg / m³). 2 A negative value indicates a decrease in dust load. If the change in road dust load is less than this threshold (e.g., a change of -3 mg / m³), then the change is considered to be negative. 2 This indicates that although the sweeping vehicle carried out its work, the sweeping effect was poor and it failed to effectively reduce the dust load on the road surface. The low initial cleaning quality score may be due to the poor sweeping effect, and quality calibration is required.

[0094] Step S606: If the change in road dust load is greater than or equal to the change threshold, then it is determined that no quality calibration is required.

[0095] Specifically, if the change in road surface dust load is greater than or equal to the change threshold (e.g., a change of -6 mg / m³), 2 This indicates that the sweeping operation of the sweeping vehicle is effective and can significantly reduce the dust load on the road surface. The initial cleaning quality score can reflect the actual cleaning effect and no quality calibration is required.

[0096] In one embodiment of this example, step S305, based on the work behavior information, determines whether quality calibration is required, and further includes steps S701 to S710: Step S701: Obtain information on the different operational behaviors of multiple cleaning and maintenance personnel or cleaning vehicles within the same key road section.

[0097] Specifically, multiple cleaning and maintenance personnel and multiple cleaning vehicles may be working simultaneously or sequentially on the same key road section. The working behaviors of different personnel and vehicles may vary. Therefore, it is necessary to obtain information on the working behaviors of all personnel and vehicles, including their respective working actions, dwell time, working audio, sweeping status, etc., for comprehensive analysis of work quality.

[0098] Step S702: Determine whether there is a conflicting relationship between different work behavior information. A conflicting relationship is a relationship in which work behavior information has opposite quality implications. A quality implication is an implication used to characterize whether the cleaning has been completed to the required standard.

[0099] Specifically, quality indications refer to signals reflecting whether cleaning operations have met standards, as reflected in work behaviors. For example, the quality indication corresponding to "workers picking up and cleaning garbage" is "cleaning meets standards," while the quality indication corresponding to "workers not performing any cleaning actions and merely observing" is "cleaning does not meet standards." A contradictory work relationship refers to situations where the quality indications of different work behaviors are opposite. For example, if one worker's behavior indicates "cleaning meets standards," while another worker's behavior indicates "cleaning does not meet standards," then a contradictory work relationship exists.

[0100] Step S703: If there is no conflict between different work behavior information, determine whether quality calibration is required based on the preset work rules.

[0101] Specifically, if the quality indicators corresponding to the work behavior information of all personnel and vehicles are consistent (both are "cleaning meets standards" or all are "cleaning does not meet standards"), then there is no conflict in the work. In this case, the need for calibration is determined according to the preset work rules. The preset work rules refer to the judgment criteria set based on whether the work behavior meets the specifications. For example, if all work behaviors are compliant (quality indicator is "cleaning meets standards"), then no calibration is needed; if all work behaviors are non-compliant (quality indicator is "cleaning does not meet standards"), then calibration is needed.

[0102] Step S704: If there is a conflict between different work behavior information, then determine whether the conflict is an internal conflict within the same work unit or a conflict between different work units.

[0103] Specifically, conflicts within the same work unit refer to situations where the conflicting personnel or vehicles belong to the same cleaning and maintenance company, while conflicts between different work units refer to situations where the conflicting personnel or vehicles belong to different cleaning and maintenance companies. The purpose of distinguishing between conflict types is to determine whether calibration is needed based on different rules, because work standards within the same unit are usually consistent, while work standards between different units may differ.

[0104] Step S705: If the conflict relationship between tasks is an internal conflict within the same task unit, then obtain the proportion of conflict behaviors corresponding to the conflict relationship between tasks.

[0105] Specifically, the contradictory behavior ratio refers to the proportion of work behaviors with a quality notice of "cleaning substandard" out of all contradictory work behaviors. The calculation method is: Contradictory behavior ratio = (Number of substandard work behaviors / Total number of contradictory work behaviors) × 100%. For example, if there are 3 workers in the same unit, and one worker's behavior notice is "substandard" while the other two are "up to standard," then the contradictory behavior ratio is 33.3%.

[0106] Step S706: If the proportion of contradictory behaviors is greater than the preset proportion threshold, it is determined that no quality calibration is required.

[0107] Specifically, the preset percentage threshold is a pre-set critical value for judging the degree of influence of contradictory behaviors (e.g., 50%). If the percentage of contradictory behaviors is greater than this threshold, it means that most of the contradictory work behaviors are "substandard". In this case, the initial cleaning quality score is low, which is consistent with the actual situation and no quality calibration is required.

[0108] Step S707: If the proportion of contradictory behaviors is less than the preset proportion threshold, it is determined that quality calibration is required.

[0109] Specifically, if the proportion of contradictory behaviors is less than the preset threshold, it means that most of the contradictory work behaviors are "up to standard" and only a few are "not up to standard". In this case, the initial cleaning quality score is low, which may be due to a few non-standard work behaviors. Quality calibration is required to correct the initial score.

[0110] Step S708: If the proportion of contradictory behaviors is equal to the preset proportion threshold, then obtain the priority level of different work behavior information.

[0111] Specifically, the priority level is set based on the reliability of the work behavior information. For example, the work behavior information of sweeping vehicles (such as brush speed and dust load changes) is more reliable than the work behavior information of operators, and therefore has a higher priority level; the audio information of operators is more reliable than video action information, and therefore has a higher priority level. The priority level is preset by the system and sorted according to the type and reliability of the work behavior information.

[0112] Step S709: If the priority level meets the preset priority relationship, it is determined that no quality calibration is required.

[0113] Specifically, the preset priority relationship means that the quality indication corresponding to the high priority level of the operation behavior information is "up to standard". If this relationship is met, it means that the more reliable operation behavior information reflects that the cleaning operation is up to standard. The initial cleaning quality score is low, which may be due to the deviation of the low priority level operation behavior information. No quality calibration is required.

[0114] Step S710: If the priority level does not meet the preset priority relationship, it is determined that quality calibration is required.

[0115] Specifically, if the priority level does not meet the preset priority relationship (i.e., the quality indication corresponding to the operation behavior information with the higher priority level is "not up to standard"), it means that the more reliable operation behavior information reflects that the cleaning operation is not up to standard. In this case, it is necessary to perform quality calibration in combination with the actual situation to ensure that the scoring is objective.

[0116] In one embodiment of this example, if a conflicting relationship exists between different work behavior information in step S704, then after determining that the conflicting relationship is an internal conflict within the same work unit or a conflict between different work units, steps S801 to S805 are further included: Step S801: If the conflict is between different work units, then obtain the accurate historical records of the work of different work units based on the video surveillance data.

[0117] Specifically, accurate historical records of operations refer to the records of different work units regarding the compliance status, rectification completion rate, and quality evaluation level of their cleaning operations over a past period (such as the past 3 months or 6 months). These records reflect the stability and accuracy of the work quality of each work unit. These records can be obtained from the urban road cleaning management database and include information such as the cleaning scores of the road sections under the responsibility of each work unit, rectification status of non-compliance, and inspection feedback results, providing a basis for subsequently judging the reliability of operational behavior information.

[0118] Step S802: Based on the accurate historical records of the task, obtain the historical compliance rate corresponding to the task behavior information.

[0119] Specifically, the historical compliance rate refers to the proportion of times each work unit achieved the preset cleaning quality standard out of the total number of operations. It is calculated as: Historical Compliance Rate = (Number of Compliance Operations / Total Number of Operations) × 100%. A higher historical compliance rate indicates more stable work quality and more standardized work practices by the work unit, and thus higher reliability of its work behavior information. Conversely, a lower historical compliance rate indicates lower reliability of the work behavior information. For example, if work unit A has a historical compliance rate of 90% and work unit B has a historical compliance rate of 70%, then work unit A's work behavior information is more valuable for reference.

[0120] Step S803: If there is a work unit with the highest historical compliance rate and the shortest distance between the work unit and the garbage accumulation location, then determine whether quality calibration is needed based on the work behavior information of the work unit.

[0121] Specifically, the distance from the work site to the garbage accumulation location refers to the straight-line distance between the work position of the personnel or cleaning vehicle and the garbage accumulation location. The smaller the distance, the closer the personnel or vehicle is to the contaminated area, and the higher the reference value of their work behavior information for judging the garbage cleanup status. If a certain work unit has the highest historical compliance rate among all work units with conflicting relationships, and its work distance from the garbage accumulation location is the smallest, then the work behavior information of that unit is the most reliable. In this case, based on the work behavior information of that unit, and according to the previously preset judgment rules (such as work actions, dwell time, cleaning effect, etc.), it is determined whether quality calibration is needed. For example, if work unit A has the highest historical compliance rate (90%) and its work distance from the garbage accumulation location is the closest (5 meters), then the calibration requirement is determined based on the work behavior information of work unit A.

[0122] Step S804: If there is no work unit with the highest historical compliance rate and the shortest distance from the garbage accumulation location, then obtain the comprehensive score of different work units based on the historical compliance rate and the work distance.

[0123] Specifically, if there is no single work unit with both the highest historical compliance rate and the shortest work distance (e.g., multiple units have the same historical compliance rate, or the unit with the highest compliance rate does not have the shortest work distance), then the operational reliability of each work unit is quantified through a comprehensive score. The comprehensive score is calculated using a weighted summation method, with a preset weight of 60% for historical compliance rate and 40% for work distance; where a higher historical compliance rate results in a higher score, and a shorter work distance also results in a higher score. First, the historical compliance rate (0-100%) and work distance (standardized to 0-100 points, with shorter distances resulting in higher scores) are standardized, then multiplied by their respective weights, and summed to obtain the comprehensive score for each work unit. The higher the comprehensive score, the stronger the reliability of the operational behavior information.

[0124] Step S805: Based on the work behavior information of the work unit with the highest comprehensive score, determine whether quality calibration is required.

[0125] Specifically, the work unit with the highest overall score best reflects the actual cleaning situation of the key monitoring sections. Therefore, this unit's work behavior information is the core judgment criterion, combined with the work behavior judgment rules mentioned above (such as whether the workers handled the garbage, whether the cleaning effect of the sweeping vehicles met the standards, etc.), to determine whether quality calibration is needed. For example, if work unit C with the highest overall score shows that its workers cleaned up the garbage in a standardized manner and the cleaning effect met the standards, then calibration is not required; if its work behavior information shows that the work was not standardized and pollution was not effectively treated, then calibration is required.

[0126] In one embodiment of this invention, step S301, based on road segment attribute information, determines whether a pollution source exists, and further includes steps S901 to S909: Step S901: Obtain multiple records of non-compliance in cleaning of key road sections within a preset historical period. Each record of non-compliance corresponds to a time of non-compliance.

[0127] Specifically, the preset historical period can be set according to actual management needs (such as the past 1 month or 3 months). Records of substandard cleaning are obtained from the urban road cleaning management database, including information such as the specific location of the substandard road section, the time of substandard cleaning, and the degree of substandard cleaning. The time of substandard cleaning refers to the specific point in time (accurate to the minute) when the system detects that the road surface cleaning quality is lower than the first quality threshold, providing a time benchmark for subsequent analysis of the correlation between pollution events and substandard cleaning.

[0128] Step S902: Using each non-compliant moment as the center, extract the road segment image data within the forward and backward time windows.

[0129] Specifically, the time window is preset to 30 minutes before and after each non-compliance moment (adjustable according to actual conditions), forming a 60-minute time window. Road segment image data is acquired through high-definition surveillance cameras along the target road, ensuring continuous and clear image data within the time window, and fully capturing changes in road surface pollution and potential pollution events before and after the non-compliance moment.

[0130] Step S903: Perform target identification on the road segment image data and extract non-fixed pollution events that occur within each time window.

[0131] Specifically, computer vision target recognition algorithms (such as the YOLO algorithm) are used to analyze road segment image data to identify non-stationary pollution events. Non-stationary pollution events differ from stationary pollution sources (such as shops and garbage transfer stations) in that their occurrence time and location are random. They mainly include temporary dumping of construction waste (such as sand and bricks carelessly dumped by construction companies), waste left by street vendors (such as food packaging and fruit peels from mobile vendors), and mud carried onto the road by construction vehicles (such as mud and sand scattered on the road surface by construction vehicles leaving the site). Through target recognition, information such as the type, location, and duration of non-stationary pollution events occurring within each time window is extracted.

[0132] Step S904: If the same non-stationary pollution event occurs repeatedly within multiple time windows of non-compliance, the location of the non-stationary pollution event is marked as a source of pollution.

[0133] Specifically, to determine whether the same non-fixed pollution event recurs, we mainly look at whether the type of pollution event and its location are consistent (e.g., temporary dumping of construction waste at the same location multiple times). If a non-fixed pollution event recurs within multiple time windows of substandard conditions, it indicates that this type of pollution event is prone to occur at that location, and that this pollution event is a significant cause of substandard road cleaning. Therefore, this location should be marked as a source of pollution, requiring focused monitoring and enhanced cleaning in the future.

[0134] Step S905: If a non-fixed pollution event does not recur within the time window of multiple non-compliance times, obtain the pavement structure information of the key road section.

[0135] Specifically, if non-recurring pollution incidents do not recur, it indicates that substandard road cleaning may be related to the road surface's structure. In this case, road surface structure information should be obtained. This information is acquired from the Urban Road Geographic Information System (GIS) and includes the location of drainage outlets, the integrity of curbs, and the condition of road joints. The location of drainage outlets is used to determine if there is water accumulation leading to sediment buildup; the integrity of curbs is used to determine if there is garbage blockage or sediment buildup; and the condition of road joints is used to determine if there are problems such as dust or mud accumulation in gaps.

[0136] Step S906: Based on the road surface structure information, determine whether there are structural defects.

[0137] Specifically, by analyzing pavement structure information, structural defects are identified. These defects include areas with blocked drainage outlets, damaged curbs, and cracked pavement joints. Blocked drainage outlets refer to areas where drainage outlets are clogged with garbage and silt, preventing rainwater from draining in a timely manner and easily leading to the accumulation of silt and garbage. Damaged curbs refer to areas where curbs are broken or missing, easily causing garbage to get stuck and silt to accumulate. Cracked pavement joints refer to cracks in pavement expansion joints and construction joints, where dust, silt, and small pieces of garbage easily accumulate. These structural defect locations easily become areas where pollutants accumulate, leading to substandard pavement cleaning.

[0138] Step S907: If there is a structural defect location, obtain the rate of garbage or silt accumulation at the structural defect location under dry and humid weather conditions.

[0139] Specifically, accumulation rate refers to the amount of debris or silt accumulated at the location of a structural defect per unit time, measured in g / h. By deploying road surface cleanliness sensors near the structural defect locations, accumulation data was collected under both dry and humid weather conditions, and the accumulation rate under different weather conditions was calculated. Under dry weather conditions, the accumulated material mainly consists of dust and small pieces of debris; under humid weather conditions, the accumulated material mainly consists of silt and sticky debris. The difference in accumulation rate under different weather conditions reflects the degree of impact of the structural defect on pollution.

[0140] Step S908: If the accumulation rate is lower than the first velocity threshold in dry weather and higher than the second velocity threshold in humid weather, then the location of the structural defect is determined to be a source of easy contamination.

[0141] Specifically, the first speed threshold (dry weather) and the second speed threshold (wet weather) are preset critical values ​​for accumulation speed, set according to the road surface structure type and cleaning standards (e.g., the first speed threshold is 5g / h, and the second speed threshold is 15g / h). If the accumulation speed is low in dry weather and high in wet weather, it indicates that the structural defect location is prone to rapid accumulation of pollutants in wet weather and is difficult to clean. This is a significant factor leading to substandard road cleaning and is therefore identified as a source of pollution. The cleaning plan for this location needs to be optimized (e.g., increasing the cleaning frequency in wet weather).

[0142] Step S909: If the accumulation rate is not lower than the first velocity threshold in dry weather, or not higher than the second velocity threshold in humid weather, then the location of the structural defect is determined not to be a source of easy contamination.

[0143] Specifically, if the accumulation rate is not lower than the first speed threshold in dry weather, it indicates that pollutants will accumulate rapidly at the location of the structural defect even in dry weather, possibly due to other stationary pollution sources rather than the structural defect itself. If the accumulation rate is not higher than the second speed threshold in humid weather, it indicates that pollutants are not easily accumulated at the location of the structural defect in easily polluted humid weather, and the impact on road surface cleaning quality is minimal. Therefore, in both of the above cases, the location of the structural defect is determined not to be a source of pollution.

[0144] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for online monitoring and evaluation of the quality of municipal road sweeping and sanitation, characterized in that, include: Acquire sweeping and cleaning operation data of the target road, including video surveillance data, vehicle trajectory data, and road surface cleanliness sensor data; An initial cleaning quality score is obtained based on the video surveillance data, the vehicle trajectory data, and the road surface cleanliness sensor data. If the initial cleaning quality score is lower than the first quality threshold, the target road will be marked as a key concern section. Obtain the road segment attribute information corresponding to the key road segments of interest; Based on the road segment attribute information, the cleaning and maintenance operation data are analyzed to obtain a quality calibration factor; Based on the initial cleaning quality score and the quality calibration factor, the target cleaning quality score is obtained; If the target cleaning quality score is lower than the first quality threshold, a road section rectification instruction will be generated.

2. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 1, characterized in that, The process of obtaining an initial cleaning quality score based on the video surveillance data, the vehicle trajectory data, and the road surface cleanliness sensor data includes: Based on the video surveillance data, extract continuous video frames; The continuous video frames are segmented to obtain road surface area images, and road surface area images with exposed garbage or accumulated dust and mud are used as defect images. Based on the continuous video frames and the defect image, obtain the defect coverage probability value; Based on the road surface cleanliness sensor data, the road surface dust load value is obtained; Based on the vehicle trajectory data, the number of cleaning coverage times and the cleaning speed are obtained; An initial cleaning quality score is obtained based on the defect coverage probability value, the road surface dust load value, the number of sweeping coverages, and the sweeping speed.

3. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 1, characterized in that, The process of analyzing the cleaning and maintenance operation data based on the road segment attribute information to obtain the quality calibration factor includes: Based on the road segment attribute information, determine whether there are any sources of pollution. If the pollution source exists, then obtain the pollution emission data corresponding to the pollution source; If the pollution emission data exceeds the corresponding emission data threshold, then based on the video surveillance data, it is determined whether there are cleaning and maintenance personnel or cleaning vehicles in the key road section of concern. If there are cleaning and maintenance personnel or cleaning vehicles in the key monitored road sections, information on their work activities will be obtained. Based on the aforementioned work behavior information, determine whether quality calibration is required; If quality calibration is required, obtain the quality calibration factor.

4. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 3, characterized in that, The determination of whether a pollution source exists based on the road segment attribute information includes: Obtain the historical cleaning records corresponding to the key road sections under focus; Based on the historical cleaning records, the causes of pollution for each instance of substandard cleaning are identified. Obtain the frequency percentage of the causes corresponding to the pollution causes; If the frequency of the aforementioned cause exceeds a preset frequency threshold, then the source of pollution is determined to exist. If the frequency of the aforementioned reasons does not exceed the preset frequency threshold, then based on the road segment attribute information, it is determined that the key focus road segment is a commercial street segment or a residential area segment. If the key focus area is the commercial street section, then obtain the business type of the shops and the time period for garbage disposal; If the business type of the shop and the time period for garbage disposal affect the cleaning and sanitation operation data, then it is determined that there is a source of pollution. If the key road segment is the residential area road segment, then obtain the distribution of surrounding facilities corresponding to the residential area road segment; Based on the distribution of surrounding facilities, the potential pollution risk value corresponding to the cleaning and maintenance operation data is obtained; If the potential pollution risk value exceeds the second quality threshold, it is determined that there is a source of pollution.

5. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 3, characterized in that, The step of determining whether quality calibration is needed based on the work behavior information includes: Based on the video surveillance data, the location of the garbage accumulation and the focus of the workers' eyes are obtained; If the line of sight focuses on the location of the garbage, then it is determined whether the duration of the operation is less than the first duration threshold. If the duration of the operation is less than the first duration threshold, then it is determined whether there is a cleaning action sound in the operation audio; If the cleaning sound is present, then it is determined that no quality calibration is required; If the cleaning sound is not present, a quality calibration is required.

6. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 3, characterized in that, The step of determining whether quality calibration is needed based on the work behavior information also includes: Based on the video surveillance data, the brush contact status and brush speed of the sweeping vehicle are obtained; If the sweeping brush is in a ground contact state and the sweeping brush speed is greater than the speed threshold, then the dwell time of the sweeping vehicle in the key concern road section is obtained; If the dwell time is greater than the second duration threshold, it is determined that no quality calibration is required; If the dwell time is less than or equal to the second dwell time threshold, the change in road dust load during the dwell period is obtained based on the road surface cleanliness sensor data. If the change in road surface dust load is less than the change threshold, then a quality calibration is required. If the change in road surface dust load is greater than or equal to the change threshold, then it is determined that no quality calibration is required.

7. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 3, characterized in that, The step of determining whether quality calibration is needed based on the work behavior information also includes: Obtain information on the different operational behaviors of multiple cleaning and maintenance personnel or cleaning vehicles within the same key road section; Determine whether there is a contradictory relationship between different job behavior information. The contradictory relationship is that the job behavior information has opposite quality implications. The quality implications are implications used to characterize whether the cleaning has been completed to the required standard. If there is no conflict between different job behavior information, then it is determined whether quality calibration is required based on preset job rules; If there is a conflict between different job behavior information, then the conflict is determined to be an internal conflict within the same job unit or a conflict between different job units. If the conflicting relationship is an internal conflict within the same work unit, then obtain the proportion of conflicting behaviors corresponding to the conflicting relationship. If the proportion of contradictory behaviors is greater than a preset proportion threshold, it is determined that no quality calibration is required. If the proportion of contradictory behaviors is less than the preset proportion threshold, it is determined that quality calibration is required. If the proportion of contradictory behaviors is equal to the preset proportion threshold, then the priority level of different work behavior information is obtained; If the priority level satisfies the preset priority relationship, it is determined that no quality calibration is required; If the priority level does not meet the preset priority relationship, it is determined that quality calibration is required.

8. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 7, characterized in that, If there is a conflicting relationship between different job behavior information, then after determining whether the conflicting relationship is an internal conflict within the same job unit or a conflict between different job units, the following steps are also included: If the conflicting relationship is between different work units, then based on the video surveillance data, obtain the accurate historical records of the work of different work units; Based on the accurate historical records of the tasks, the historical compliance rate corresponding to the task behavior information is obtained; If the historical compliance rate of the work unit is the highest and the distance from the work location to the garbage accumulation location is the smallest, then it is determined whether quality calibration is needed based on the work behavior information of the work unit. If there is no work unit with the highest historical compliance rate and the shortest distance from the garbage accumulation location, then the comprehensive score of different work units is obtained based on the historical compliance rate and the work distance. Based on the operational behavior information of the work unit with the highest comprehensive score, it is determined whether quality calibration is required.

9. The online monitoring and evaluation method for the quality of municipal road sweeping and sanitation according to claim 3, characterized in that, The determination of whether a pollution source exists based on the road segment attribute information also includes: Obtain multiple records of substandard cleaning of the key road sections within a preset historical period, with each record of substandard cleaning corresponding to a specific time of substandard cleaning. Centered on each time point where the standard is not met, extract the road segment image data within the forward and backward time windows; Target identification is performed on the image data of the road section to extract non-fixed pollution events that occur within each time window; If the same non-stationary pollution event occurs repeatedly within multiple time windows of non-compliance, the location of the non-stationary pollution event is marked as a source of pollution. If the non-fixed pollution event does not recur within the time window of multiple non-compliance times, then the pavement structure information of the key concern road section is obtained; Based on the road surface structure information, determine whether there are structural defects. If the location of the structural defect exists, the rate of garbage or silt accumulation at the location of the structural defect under dry and humid weather conditions is obtained. If the accumulation rate is lower than a first velocity threshold in dry weather and higher than a second velocity threshold in humid weather, then the location of the structural defect is determined to be a source of easy contamination. If the accumulation rate is not lower than the first speed threshold in dry weather, or not higher than the second speed threshold in humid weather, then the location of the structural defect is determined not to be a source of easy contamination.

Images