Road preventive maintenance method based on disease development prediction

By constructing a disease development prediction model, we can identify the significant influencing factors of highway diseases and take targeted maintenance measures. This solves the shortcomings of traditional maintenance models, achieves comprehensive prediction and prevention of road diseases, and improves maintenance efficiency and scientific rigor.

CN118982340APending Publication Date: 2024-11-19ZHENGZHOU UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411018658.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-19

Smart Images

  • Figure CN118982340A_ABST
    Figure CN118982340A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of road maintenance, and discloses a road preventive maintenance method based on disease development prediction, which comprises the following steps of: acquiring related data information of a road and performing data conversion preprocessing; constructing a disease severity regression analysis model by taking the road surface disease severity as an explained variable and taking other road related data information except the road surface disease as explaining variables, screening out significant influence factors of the road surface disease severity, and taking corresponding road preventive maintenance measures; the method comprises the following steps: constructing a disease quantity regression analysis model by taking a pavement disease quantity as an explained variable and taking other road related data information except the pavement diseases as explaining variables, screening out significant influence factors of the pavement disease quantity, and calculating a predicted value of the pavement disease quantity under the influence of the significant influence factors of the crack quantity; and corresponding highway preventive maintenance measures are taken. According to the method, the disease development prediction model is constructed, so that a scientific basis is provided for road disease treatment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of road maintenance, and relates to a highway preventive maintenance method based on disease development prediction. Background Art

[0002] Highway transportation is the most widely used mode of transportation with the widest coverage, the largest population, and the largest population. At this stage, my country has basically completed the task of highway construction. Highway maintenance will become the primary task of the highway transportation industry. The traditional preventive maintenance model will also be continuously improved on the original basis to adapt to the rapid development of highway maintenance needs.

[0003] Traditional highway maintenance plans mostly use manual surveys, and on this basis, road maintenance decisions are made based on previous road maintenance experience. This method is mostly used in specific situations and specific periods. Although the decision is simple and efficient, subjective decision-making has always been its disadvantage, which easily leads to untimely maintenance and unreasonable maintenance decisions. It often causes a waste of funds and does not use the funds where they are needed. With the rapid development of highway construction, the demand for road maintenance is also increasing rapidly. The traditional maintenance model can no longer meet the speed and accuracy requirements of highway maintenance.

[0004] Although many models have been proposed by domestic and foreign scholars in the study of road performance prediction, they basically use the relationship between different pathogenic factors and road diseases to predict the possibility and severity of road diseases at different levels of pathogenic factors, and rarely consider the impact of different pathogenic factors on historical maintenance content. Therefore, it is impossible to comprehensively predict the possible damage or disease of the road, which will hinder the progress of road maintenance work to a certain extent, and it is impossible to take targeted preventive maintenance measures. Summary of the invention

[0005] In response to the above technical problems, the present invention provides a highway preventive maintenance method based on disease development prediction. By constructing a disease development prediction model, significant influencing factors of the number of pavement diseases can be screened out, and corresponding highway preventive maintenance measures can be taken to reduce and prevent the occurrence and further deterioration of road diseases, which has important guiding significance for road maintenance practice.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a highway preventive maintenance method based on disease development prediction, comprising the following steps:

[0008] S1: Obtaining highway-related data information and performing data conversion preprocessing, wherein the highway-related data information includes road design information, geographic location information, historical maintenance content, road surface diseases, weather data, and traffic data;

[0009] S2: Construct a regression analysis model of pavement disease severity with the severity of pavement disease as the explained variable and other highway-related data information other than pavement disease as the explanatory variables, screen out the significant influencing factors of pavement disease severity and take corresponding highway preventive maintenance measures;

[0010] S3: A pavement disease quantity regression analysis model is constructed with the number of pavement diseases as the explained variable and other highway-related data information except pavement diseases as the explanatory variables. The significant influencing factors of the pavement disease quantity are screened out, the predicted value of the pavement disease quantity under the influence of the significant influencing factors of the crack quantity is calculated, and corresponding highway preventive maintenance measures are taken.

[0011] The road design information in the technical solution of the present invention includes road surface slope, road surface flatness, road surface anti-skid performance index, road surface structure strength index, road surface driving quality index and road surface damage condition index.

[0012] The historical maintenance contents described in the technical solution of the present invention include roadbed damage, facility damage and unclean road surface.

[0013] The weather factor data described in the technical solution of the present invention include the highest temperature, the lowest temperature, the maximum wind force level, the precipitation and the snowfall of each month.

[0014] The traffic data described in the technical solution of the present invention includes traffic flow, vehicle type and truck ratio.

[0015] The pavement damage described in the technical solution of the present invention includes rutting depth and the number of cracks.

[0016] The specific steps of the above step S2 in the technical solution of the present invention are as follows:

[0017] S201: The rutting depth is used to represent the severity of the disease, and the severity of the disease is divided into primary rutting depth and secondary rutting depth. That is, the rutting depth is used as the explained variable, and other road-related data information except the rutting depth and the number of cracks are used as the explanatory variables. The disease severity regression analysis model is a binomial Logit regression model. When the disease severity is primary rutting depth, the disease severity y=0, and the probability of primary rutting is p. When the disease severity is secondary rutting depth, the disease severity y=1, and the probability of secondary rutting is 1-p. Then:

[0018]

[0019] Where: x i is the explanatory variable; i is the estimated coefficient of the explanatory variable, ε i is the error term;

[0020] S202: Goodness of fit test of disease severity regression analysis model;

[0021] S203: Parameter estimation of the disease severity regression analysis model is performed to screen out the significant influencing factors of the rutting depth;

[0022] S204: Take corresponding highway preventive maintenance measures according to the significant influencing factors of the rutting depth.

[0023] The specific steps of the above step S3 in the technical solution of the present invention are as follows:

[0024] S301: The number of cracks is used to represent the number of defects, that is, the number of cracks is used as the explained variable, and other road-related data information except rutting depth and number of cracks is used as the explanatory variable. The defect quantity analysis regression model is a negative binomial regression model. Since the number of cracks Y i The probability distribution function of is:

[0025] Y i ~Negbin(λ i ,γ),

[0026] Where: i is the expected value of the number of the i-th crack, γ is the dispersion coefficient;

[0027] λ i Perform logarithmic processing to establish a linear relationship with the explanatory variables:

[0028]

[0029] Where: X ik The kth explanatory variable representing the number of the i-th crack, β0 and β k is the coefficient to be estimated;

[0030] S302: Perform goodness of fit test on disease quantity regression analysis model;

[0031] S303: Estimating parameters of the disease quantity regression analysis model and screening out significant influencing factors of the crack quantity;

[0032] S304: Calculate the predicted value of the number of cracks under the influence of the significant influencing factors of the number of cracks, and take corresponding highway preventive maintenance measures.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] The present invention integrates various data sources and constructs a more complete indicator set. It then takes the severity of pavement diseases and the number of pavement diseases as explained variables, and takes other highway-related data information except pavement diseases as explanatory variables to construct a disease severity regression analysis model and a disease number regression analysis model respectively. The present invention can screen out the significant influencing factors of the number of pavement diseases, and take corresponding highway preventive maintenance measures, thereby reducing and preventing the occurrence and further deterioration of road diseases. It has important guiding significance for road maintenance practice, and provides a scientific basis and technical support for improving the level of road disease control. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a comparison chart of the actual value and predicted value of the number of cracks in the embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention. Unless otherwise specified, the technical means used in the examples are conventional means known to those skilled in the art. The test methods in the following examples are conventional methods unless otherwise specified.

[0037] Embodiment 1

[0038] The pavement damage rate refers to the comprehensive index after conversion of various types of damage (roadbed damage, facility damage, and unclean pavement), that is, the area of ​​various damages occurring in the evaluated section is multiplied by the corresponding weight and converted into the percentage of the sum of the serious cracks or potholes in the total area of ​​the section. The multifunctional road condition rapid detection system is used to detect the pavement damage rate of the entire line and all lanes. The slope refers to the degree of steepness of the surface unit. The ratio of the vertical height of the slope to the horizontal distance is usually called the slope. The pavement flatness refers to the vertical deviation slope of the pavement surface relative to the ideal plane. Cracks are mainly caused by the long-term occurrence of conditions such as reduced asphalt adhesion, heavy traffic, and vehicle overload. Rutting is mainly caused by repeated driving loads and external environmental effects. The internal reason is the poor high temperature stability of asphalt concrete and the insufficient resistance to plastic deformation. The pavement produces cumulative permanent depressions, which mostly occur on both sides of the wheel track, bridge deck pavement, and curved sections.

[0039] This example takes the G107 line reconstruction section in Xinxiang as an example. The G107 line passes through Weihui, Yanjin and Yuanyang in Xinxiang, and has been fully opened to traffic since September 27, 2021. Combining field surveys, Amap open platform, field collection and video observation and literature search, five types of data were obtained, namely: (1) Design data and inspection reports of the G107 line: design standards of the G107 line, main standards, procedures and specifications adopted, route starting and ending points, total length, overview of major towns and projects along the line, pavement damage condition index (PCI), pavement ride quality index (RQI), pavement skid resistance index (SRI) and pavement structure strength index (PSSI); (2) Geographical location information: administrative divisions of Xinxiang City and G107 route information; (3) Historical information Historical maintenance content: the time and location of historical maintenance content (damage to the roadbed, damage to facilities, and unclean road surface) of G107; (4) Typical road diseases: the time and location of typical road diseases (rutting depth and number of cracks) of G107; (5) Weather factor data: monthly maximum and minimum temperatures, wind force level, and precipitation; (6) Traffic data: a video observation survey method was used to survey three intersections, including the intersections of G107 and Provincial Highway 310, Provincial Highway 311, and Provincial Highway 227, to obtain information on hourly traffic flow, vehicle type, and truck ratio.

[0040] The acquired data is preprocessed to organize the disordered raw data into a consistent data set suitable for modeling, laying the foundation for subsequent data feature analysis and model building. The specific operations are as follows:

[0041] (1) Duplicate value screening: Use the duplicate value search function of Excel to select the same type of damage, such as road surface damage, damage to facilities along the line, and roadbed damage. Different types of typical damages are placed together to facilitate classification and subsequent spatiotemporal feature analysis, ensuring the uniqueness and accuracy of the data.

[0042] (2) Outlier processing: Use Excel's search function to detect outliers in the disease data, such as dates with normal inspection conditions and locations where the number of pavement joints is zero. Further verification is performed on the detected outliers, and they are corrected or deleted according to the actual situation. Because outliers may have an adverse effect on the modeling results, they need to be handled with caution to ensure the accuracy of the final results.

[0043] (3) Data conversion: First, all specific disease descriptions and inspection conditions were converted to facilitate subsequent analysis and modeling. For example, after the original inspection data was imported into Excel, the accumulated grain, scattered garbage and gravel on the K6-K8 road surface were uniformly converted into the problem of unclean road surface in the historical maintenance type, and the damaged guardrail, 4 curbs, 7 missing anti-glare panels, and 1 damaged sign were uniformly converted into the problem of damaged facilities in the historical maintenance type, which facilitated the subsequent data processing and modeling analysis.

[0044] In order to further screen out the significant influencing factors of rutting depth and crack number and take corresponding highway preventive maintenance measures, the specific steps are as follows:

[0045] (1) The rutting depth is used to represent the severity of the disease, and the severity of the disease is divided into primary rutting depth and secondary rutting depth. That is, the rutting depth is used as the explained variable, and other highway-related data information other than rutting depth and crack number is used as the explanatory variable to construct the disease severity analysis model. Since there may be multiple key influencing factors of rutting depth when analyzing different highways (such as peak hour traffic volume, truck ratio, anti-skid performance and pavement damage rate), the disease severity regression analysis model conforms to the binomial logit regression model.

[0046] This embodiment uses the kilometer average rutting depth to represent the severity of the disease, including 96 kilometer average rutting depths in both the up and down directions. The median of the kilometer average rutting depth is 4.08, and the data is divided into two categories. The first 50% is the first-level rutting, and the second 50% is the second-level rutting. There are 49 second-level rutting and 47 first-level rutting. Therefore, the explained variables are selected as first-level rutting and second-level rutting, that is, when the severity of the disease is the first-level rutting depth, the severity of the disease y=0, and the probability of the occurrence of the first-level rutting is p; when the severity of the disease is the second-level rutting depth, the severity of the disease y=1, and the probability of the occurrence of the second-level rutting is 1-p, then:

[0047]

[0048] Where: x i is the explanatory variable; i is the estimated coefficient of the explanatory variable, ε i is the error term.

[0049] Likelihood ratio test: Under the condition of significance level α = 0.05, the chi2 test was performed on the disease severity analysis model: LR chi2(4) = 55.02, (Prob>chi2) = 0.000<0.001, that is, compared with the null model containing only the intercept term, the model has a better goodness of fit overall.

[0050] Goodness of fit index: Likelihood ratio index McFadden's PseudoR 2 =0.4135, the explanatory power of the disease severity analysis model prediction variables is increased by 41.35% compared with the null model, indicating that the model is effective and has a high degree of fit.

[0051] The parameter estimation results of the disease severity analysis model are shown in Table 1.

[0052] Table 1 Parameter estimation results of disease severity analysis model

[0053]

[0054]

[0055] Prob>chi2=0.0000, which is less than the confidence level of 0.01, indicating that the model will not fall into the small probability interval, and the overall model is statistically significant at the level of 0.01. The peak hour traffic volume p=0.005<0.01, indicating that it is correlated at the significance level of 0.01 and is positively correlated with the severity of the disease; the truck ratio p=0.046<0.05, indicating that it is correlated at the significance level of 0.05 and is negatively correlated with the severity of the disease; the anti-skid performance p=0.831>0.1, indicating that it cannot pass the t test, is unqualified in the model, and has no significant effect on the severity of the disease; the pavement damage rate p=0.004<0.01, indicating that it is correlated at the significance level of 0.01 and is positively correlated with the severity of the disease. The results show that the significant factors affecting the severity of the disease are peak hour traffic volume, truck ratio and road damage rate, which are positively correlated, negatively correlated and positively correlated with the severity of the disease respectively.

[0056] Take the following preventive measures based on the above significant factors: (1) Set up corresponding checkpoints or monitoring on sections with heavy traffic to prevent vehicles from speeding or overloading. (2) Pay attention to existing rutting problems on sections with high pavement damage rates and strengthen treatment and repair work. (3) Establish tidal lanes and adjust the use direction of the lanes according to the different traffic conditions in the morning and evening to improve rutting sections, relieve traffic pressure on severely worn sections, widen lanes to improve wear, etc. (4) For local minor rutting or wear-type rutting, take measures such as roughening and cleaning, tack oil treatment, surface milling, leveling and cleaning.

[0057] (2) When the number of cracks is used to represent the number of defects, that is, when the defect quantity analysis model is constructed with the number of cracks as the explained variable and other highway-related data information except rutting depth and number of cracks as the explanatory variables, there are multiple different numbers of cracks, the number of explanatory variables is large (such as peak hour traffic volume, truck ratio, road surface flatness, slope and road damage rate) and they are over-dispersed. Therefore, the defect quantity analysis regression model conforms to the negative binomial regression model.

[0058] In this embodiment, the number of cracks per 100 meters is used to represent the number of defects, including 750 cracks in both the upward and downward directions. Therefore, the explained variable is the number of cracks per 100 meters, and the explanatory variables are the selected key influencing factors of the number of cracks (such as peak hour traffic volume, truck ratio, road surface flatness, slope and road surface damage rate).

[0059] Since the number of cracks Y i The probability distribution function of is:

[0060] Y i ~Negbin(λ i ,γ),

[0061] Where: i is the expected value of the number of the i-th crack, and γ is the dispersion coefficient.

[0062] λ i Perform logarithmic processing to establish a linear relationship with the explanatory variables:

[0063]

[0064] Where: X ik The kth explanatory variable representing the number of the i-th crack, β0 and β k is the coefficient to be estimated.

[0065] The mean absolute deviation (MAD) and mean squared predictive error (MSPE) were used to evaluate the goodness of fit of the negative binomial regression model. The goodness of fit index results are shown in Table 2.

[0066] MAD and MSPE are calculated as follows:

[0067]

[0068] Where Y is the actual value of the number of the i-th crack, Y i is the predicted value of the number of the i-th crack.

[0069] Table 2 Fit index results

[0070]

[0071] Since the smaller the MAD and MSPE values, the more accurate the model prediction is, this shows that the negative binomial regression model has a higher data fit and stronger accident prediction ability.

[0072] The parameter estimation results of the disease quantity regression analysis model are shown in Table 3.

[0073] Table 3 Parameter estimation results of disease quantity regression analysis model

[0074]

[0075]

[0076] Prob>chi2=0.0000, which is less than the confidence level of 0.01, indicating that the model will not fall into the small probability interval, and the model as a whole is statistically significant at the level of 0.01. The p-value of peak hour traffic volume=0.449>0.1, indicating that it cannot pass the t-test, is unqualified in the model, and has no significant effect on the number of diseases; the p-value of truck proportion=0.094<0.1, indicating that it is correlated at the significance level of 0.1, and is negatively correlated with the number of diseases; the p-value of road surface flatness=0.622>0.1, indicating that it cannot pass the t-test, is unqualified in the model, and has no significant effect on the number of diseases; the p-value of slope=0.072<0.1, indicating that it is correlated at the significance level of 0.1, and is positively correlated with the number of diseases; the p-value of road surface damage rate=0.000<0.01, indicating that it is correlated at the significance level of 0.01, and is positively correlated with the number of diseases. The results show that the significant factors affecting the number of cracks are the proportion of trucks, slope and road damage rate, which are negatively correlated and positively correlated with the number of diseases, respectively. Therefore, slurry sealing technology, micro-surfacing technology, reinforcement technology and thin-layer overlay technology are used in these sections to prevent the occurrence of cracks.

[0077] There is a certain randomness in the number of defects. Simply using the actual number of accidents or accident rates to evaluate the number of defects will result in certain deviations. The expected number of defects describes the average level of the number of cracks and can better reflect the actual number of defects on the road. This embodiment estimates the expected number of cracks produced every 100 meters of K13-K52 based on the parameter estimation results of the defect number regression analysis model, and selects the top 15 of the highest cracks as important targets for disease prevention and maintenance (as shown in Table 4). Therefore, slurry sealing technology, micro-surfacing technology, reinforcement technology and thin-layer overlay technology are used in these sections to prevent the occurrence of cracks for preventive maintenance.

[0078] The comparison between the actual value and the predicted value of the number of cracks in this embodiment is shown in the figure below. Figure 1 It can be seen that the predicted values ​​and actual values ​​of the number of cracks in sections K32+500-K32+600, K14+600-K14+700, K39+600-K39+700, K25+400-K25+500, K24+800-K24+900, K25+700-K25+800 and K45+200-K45+300 are quite different, indicating that more cracks may occur in these sections in the future, and preventive maintenance is a top priority.

[0079] Table 4. Expected number of cracks ranking

[0080]

[0081] The embodiments described above are only preferred embodiments of the present invention and are only used to explain the present invention, not to limit the scope of implementation of the present invention. For those skilled in the art, other implementation methods can certainly be easily made by replacement or modification based on the technical contents disclosed in this specification. Therefore, all changes and improvements made on the principles of the present invention should be included in the scope of the patent application of the present invention.

Claims

1. A highway preventive maintenance method based on disease development prediction, characterized in that: The following steps are involved: S1: Obtaining highway-related data information and performing data conversion preprocessing, wherein the highway-related data information includes road design information, geographic location information, historical maintenance content, road surface diseases, weather data, and traffic data; S2: Construct a regression analysis model of pavement disease severity with the severity of pavement disease as the explained variable and other highway-related data information other than pavement disease as the explanatory variables, screen out the significant influencing factors of pavement disease severity and take corresponding highway preventive maintenance measures; S3: A pavement disease quantity regression analysis model is constructed with the number of pavement diseases as the explained variable and other highway-related data information except pavement diseases as the explanatory variables. The significant influencing factors of the pavement disease quantity are screened out, the predicted value of the pavement disease quantity under the influence of the significant influencing factors of the crack quantity is calculated, and corresponding highway preventive maintenance measures are taken.

2. The highway preventive maintenance method according to claim 1, characterized in that: The road design information includes road surface slope, road surface flatness, road surface anti-skid performance index, road surface structure strength index, road surface ride quality index and road surface damage condition index.

3. The highway preventive maintenance method according to claim 1, characterized in that: The historical maintenance content includes roadbed damage, facility damage and unclean road surface.

4. The highway preventive maintenance method according to claim 1, characterized in that: The weather factor data include the monthly maximum temperature, minimum temperature, maximum wind level, precipitation and snowfall.

5. The highway preventive maintenance method according to claim 1, characterized in that: The traffic data includes traffic flow, vehicle types and truck ratio.

6. The highway preventive maintenance method according to claim 1, characterized in that: The pavement damage includes rutting depth and crack quantity.

7. The highway preventive maintenance method according to claim 1 or 6, characterized in that: The specific steps of step S2 are as follows: S201: The rutting depth is used to represent the severity of the disease, and the severity of the disease is divided into primary rutting depth and secondary rutting depth. That is, the rutting depth is used as the explained variable, and other road-related data information except the rutting depth and the number of cracks are used as the explanatory variables. The disease severity regression analysis model is a binomial Logit regression model. When the disease severity is primary rutting depth, the disease severity y=0, and the probability of primary rutting is p. When the disease severity is secondary rutting depth, the disease severity y=1, and the probability of secondary rutting is 1-p. Then: Where: x i is the explanatory variable; i is the estimated coefficient of the explanatory variable, ε i is the error term; S202: Goodness of fit test of disease severity regression analysis model; S203: Parameter estimation of the disease severity regression analysis model is performed to screen out the significant influencing factors of the rutting depth; S204: Take corresponding highway preventive maintenance measures according to the significant influencing factors of the rutting depth.

8. The highway preventive maintenance method according to claim 1 or 6, characterized in that: The specific steps of step S3 are as follows: S301: The number of cracks is used to represent the number of defects, that is, the number of cracks is used as the explained variable, and other road-related data information except rutting depth and number of cracks is used as the explanatory variable. The defect quantity analysis regression model is a negative binomial regression model. Since the number of cracks Y i The probability distribution function of is: Y i ~Negbin(l i ,c), Where: i is the expected value of the number of the i-th crack, γ is the dispersion coefficient; λ i Perform logarithmic processing to establish a linear relationship with the explanatory variables: Where: X ik The kth explanatory variable representing the number of the i-th crack, β0 and β k is the coefficient to be estimated; S302: Perform goodness of fit test on disease quantity regression analysis model; S303: Estimating parameters of the disease quantity regression analysis model and screening out significant influencing factors of the crack quantity; S304: Calculate the predicted value of the number of cracks under the influence of the significant influencing factors of the number of cracks, and take corresponding highway preventive maintenance measures.