Punching quality evaluation method for bridge construction

Abstract

The invention relates to the technical field of water conservancy construction monitoring, in particular to a punching quality evaluation method for bridge construction, which comprises the following steps: acquiring operation data in a water conservancy construction punching process; performing STL decomposition on the operation data to obtain trend term data and residual term data, and obtaining an initial time sequence weight of each period; according to the difference between the distribution conditions of the trend term data in each period and the adjacent period, the dispersion degree of the trend term data and the difference of the trend term data, obtaining the trend anomaly degree of each period; obtaining the residual anomaly degree of each period according to the discrete degree of the residual item data in each period and the difference between the discrete degrees of the residual item data; and obtaining the optimal time sequence weight of each period, predicting the operation data by using the optimal time sequence weight, and monitoring the water conservancy construction punching process based on a prediction result. The data prediction result is more accurate, so that the precision of the punching quality monitoring result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy construction monitoring, and particularly relates to a method for evaluating the drilling quality for bridge construction. Background Art

[0002] Hydrogeological exploration is a science that studies the formation, distribution, recharge, and movement laws of water in water conservancy construction. Its purpose is to provide a scientific basis for the rational development and utilization of water resources in water conservancy construction. The drilling operation in water conservancy construction refers to creating holes in soil, rock, or other geological materials by using specific equipment such as drills and punchers. Hydrogeological exploration provides necessary geological and hydrological information for water conservancy construction drilling, ensuring the safety, feasibility, and effectiveness of the project. By monitoring the operating parameters during the water conservancy construction drilling process, the construction quality can be ensured.

[0003] The existing method for monitoring the drilling quality in water conservancy construction is to predict the operating parameters during the drilling process through traditional prediction algorithms to predict the changes in the operating parameters in advance and achieve the purpose of early warning. However, the traditional prediction algorithms do not consider the abnormal operation problems of the drilling device and the influence of noise data during prediction, resulting in relatively inaccurate data prediction results, and thus relatively low accuracy in evaluating the drilling quality. Summary of the Invention

[0004] In order to solve the technical problem that the data prediction results obtained by using the existing method are relatively inaccurate, resulting in relatively low accuracy in evaluating the drilling quality, the purpose of the present invention is to provide a method for evaluating the drilling quality for bridge construction, and the specific technical solution adopted is as follows: In the first aspect, the present invention provides a method for evaluating the drilling quality for bridge construction, and the method includes the following steps: Obtain the operating data during the water conservancy construction drilling process within a set time period before the current moment; Perform STL decomposition on the operating data to obtain the trend item data and residual item data in each period; according to the time interval between each period and the current moment, obtain the initial time series weight of each period; According to the difference between the distribution of the trend item data in each period and the distribution of the trend item data in the adjacent period, the dispersion degree of the trend item data in each period, and the difference between the trend item data in each period and the trend item data in the adjacent period, obtain the trend anomaly degree of each period; According to the dispersion degree of the residual item data in each period and the difference between the dispersion degree of the residual item data in each period and the dispersion degree of the residual item data in other periods, obtain the residual anomaly degree of each period; Based on the degree of trend anomaly, the degree of residual anomaly, and the initial time series weight, the optimal time series weight for each period is obtained, and the operation data is predicted using the optimal time series weight. The monitoring of the water conservancy construction drilling process is carried out based on the prediction results.

[0005] Preferably, the degree of trend anomaly for each period is obtained according to the difference between the distribution of trend item data in each period and the distribution of trend item data in the adjacent period, the degree of dispersion of trend item data in each period, and the difference between trend item data in each period and the adjacent period, specifically including: Denote any one period as the target period. According to the difference between each trend item data in the target period and the adjacent trend item data, the data fluctuation degree of each trend item data in the target period is obtained, and the variance of the data fluctuation degrees of all trend item data in the target period is calculated to obtain the first characteristic coefficient; Calculate the difference between the mean and the median of all trend item data in the target period as the trend distribution degree of the target period; According to the difference between the trend distribution degrees of the target period and the adjacent period, and the difference between the mean values of trend item data between the target period and other periods, the second characteristic coefficient is obtained; The degree of trend anomaly of the target period is obtained according to the first characteristic coefficient and the second characteristic coefficient, and both the first characteristic coefficient and the second characteristic coefficient are positively correlated with the degree of trend anomaly.

[0006] Preferably, the specific calculation formula of the second characteristic coefficient is: ; where, represents the second characteristic coefficient of the target period, t represents the t-th period, represents the trend distribution degree of the t-th period, represents the trend distribution degree of the (t - 1)-th period, represents the (t + 1)-th trend distribution degree, N represents the total number of periods, represents the mean value of all trend item data in the t-th period, represents the mean value of all trend item data in the r-th period other than the t-th period.

[0007] Preferably, the specific calculation formula of the data fluctuation degree is: ; where, represents the data fluctuation degree of the i-th trend item data in the t-th period, represents the i-th trend item data in the t-th period, It represents the (i - 1)-th trend item data in the t-th cycle. It represents the (i + 1)-th trend item data in the t-th cycle.

[0008] Preferably, obtaining the preferred timing weight of each cycle according to the trend anomaly degree, the residual anomaly degree, and the initial timing weight specifically includes: For any cycle, using the residual anomaly degree in this cycle to correct the trend anomaly degree to obtain the corrected anomaly degree of this cycle; obtaining the weight adjustment degree of this cycle according to the absolute value of the difference between the corrected anomaly degree and the trend anomaly degree, and there is a positive correlation between the absolute value of the difference and the weight adjustment degree; Normalizing the difference between the initial timing weight of this cycle and the weight adjustment degree to obtain the preferred timing weight of this cycle.

[0009] Preferably, using the residual anomaly degree in this cycle to correct the trend anomaly degree to obtain the corrected anomaly degree of this cycle specifically includes: ; Wherein, represents the corrected anomaly degree of the t-th cycle, represents the trend anomaly degree of the t-th cycle, represents the residual anomaly degree of the t-th cycle, represents the residual anomaly degree of the r-th cycle other than the t-th cycle, N represents the total number of cycles, and exp( ) represents the exponential function with the natural constant e as the base.

[0010] Preferably, obtaining the residual anomaly degree of each cycle according to the dispersion degree of the residual term data in each cycle and the difference situation between the dispersion degree of the residual term data in each cycle and that in other cycles specifically includes: Denote any cycle as the selected cycle, and obtain the residual data characteristics of the selected cycle according to the number of residual term data included in the selected cycle, the mean value of all residual term data, and the difference between each residual term data and the mean value of all residual term data; obtain the residual anomaly degree of the selected cycle according to the balance degree of the difference situation between the residual data characteristics of the selected cycle and those of each other cycle.

[0011] Preferably, obtaining the residual anomaly degree of the selected cycle according to the balance degree of the difference situation between the residual data characteristics of the selected cycle and those of each other cycle specifically includes: Calculate the ratio between the residual data characteristics of the selected period and those of any other period, and take the absolute value of the difference between 1 and the ratio as the difference characteristic between the selected period and the any other period; calculate the mean value between the difference characteristics of the selected period and each other period to obtain the residual anomaly degree of the selected period.

[0012] Preferably, the calculation formula of the residual data characteristics of the selected period is specifically: ; Wherein, represents the residual data characteristics of the selected period, s represents the s-th period, represents the total number of residual term data included in the s-th period, represents the mean value of all residual term data in the s-th period, represents the u-th residual term data in the s-th period.

[0013] Preferably, obtaining the initial timing weight of each period according to the time interval between each period and the current moment specifically includes: For any period, obtain the shortest time interval between the period and the current moment, and perform positive correlation normalization processing on the shortest time interval to obtain the initial timing weight of the period.

[0014] The embodiments of the present invention have at least the following beneficial effects: The present invention first performs STL decomposition operation on the collected operation data to obtain data with different characterization properties, that is, the trend term data and residual term data under each period, providing a data basis for subsequent reasonable and comprehensive analysis of the abnormal situation of operation data. And, the initial timing weight is obtained through the time distribution of the time series data, and the initial timing weight preliminarily characterizes the data proportion of the operation data in each period for predictive analysis. Then, analyze the trend term data under each period, fully considering the difference and dispersion degree of the trend term data between adjacent periods, so that the obtained trend anomaly degree can reflect the comprehensive abnormal situation of the operation data in each period in terms of data change trend. Further, analyze the residual term data under each period, fully considering the dispersion degree of the residual term data in each period and the difference between the residual term data of adjacent periods, so that the obtained residual anomaly degree can reflect the abnormal situation of the operation data in each period in terms of residual terms, and characterize the possible degree of data anomaly caused by environmental factors in the operation data of each period. Finally, combining the analysis results of the data anomaly degrees in the two aspects to correct the initial timing weight can fully consider the influence of the punching out-of-row problem and noise data, making the data prediction result more accurate, and thus improving the accuracy of the punching quality detection result. Description of the Drawings

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0016] Figure 1 It is a schematic flowchart of a method for evaluating the drilling quality for bridge construction provided by the present invention; Figure 2 It is a distribution curve graph of operation data within a set time period provided by an embodiment of the present invention; Figure 3 It is the seasonal term of the operation data provided by an embodiment of the present invention; Figure 4 It is the trend term of the operation data provided by an embodiment of the present invention; Figure 5 It is the residual term of the operation data provided by an embodiment of the present invention; Figure 6 It is the prediction result of the operation data provided by an embodiment of the present invention. Detailed implementation manners

[0017] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features, and effects of a method for evaluating the drilling quality for bridge construction proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0019] Such as Figure 1 It is a schematic flowchart of a method for evaluating the drilling quality for bridge construction.

[0020] Step 1: Perform STL decomposition on the operation data to obtain the trend term data and residual term data for each period; according to the time interval between each period and the current moment, obtain the initial time series weight for each period.

[0021] Due to the influence of the geological conditions at the drilling site, abnormal operations may occur during the water conservancy construction drilling process, such as abnormal vibration. In severe cases, the direction of the drill bit may deviate, affecting the quality of the drilled holes. However, when the existing prediction algorithms predict the vibration data during the water conservancy construction drilling process, there may be abnormal data in the collected historical vibration data. Some of these abnormal data may be generated when the drilling equipment itself malfunctions, and this part of the data has certain value for predicting the operating conditions. While the other part may be abnormal data under the influence of the external environment, that is, noise data caused by interference from objective environment and other factors. Using this part of the data to predict the operating conditions has no reference value.

[0022] In order to be able to more effectively process the changing trends and abnormal conditions of the operating data at multiple levels within a set time period, first, the operating data is decomposed by STL to obtain the trend item data and residual item data for each period. It should be noted that in this embodiment, the vibration data is used as the operating data and is described by taking the vibration data as an example. The STL decomposition algorithm is a well-known technology. By performing the STL decomposition operation on all the operating data within a set time period, the seasonal item, trend item, and residual item of the operating data can be obtained, as Figure 2 shown in the distribution schematic diagram of the operating data within a set time period, as Figure 3 shown as the seasonal item of the operating data, that is, the periodic item, as Figure 4 shown as the trend item of the operating data, as Figure 5 shown as the residual item of the operating data. By analyzing the seasonal item, the period of the operating data can be obtained, and then the corresponding trend item and residual item within the time period of each period can be obtained. Through Figure 3 it can be understood that the period in the example diagram is 0.1.

[0023] Considering that during the drilling process of the water conservancy construction drilling device, there is also a certain degree of volatility in the operating data under normal operating conditions, and this data fluctuation phenomenon is allowed, random, and unpredictable. Since the original data to be collected needs to have a certain degree of stability during data prediction, that is, the data fluctuation is small. In order to improve the stability degree of the operating data, in this embodiment, by setting the time series weight, the collected operating data is weighted.

[0024] Based on this, first, a preliminary analysis is performed on the time series distribution characteristics of the operating data within a set time period, and a basic initial time series weight can be obtained, that is, according to the time interval between each period and the current moment, the initial time series weight of each period is obtained. Specifically, for any period, the shortest time interval between the period and the current moment is obtained, and the shortest time interval is subjected to positive correlation normalization processing to obtain the initial time series weight of the period.

[0025] In this embodiment, taking any one cycle as an example for illustration, the shortest time interval between the t-th cycle and the current moment can be the time interval between the moment corresponding to the last running data in the t-th cycle and the current moment. Furthermore, the calculation formula for the initial timing weight of the t-th cycle can be expressed as: ; Among them, represents the initial timing weight of the t-th cycle, represents the current moment, represents the moment closest to the current moment in the t-th cycle, represents the shortest time interval between the t-th cycle and the current moment, and exp( ) represents the exponential function with the natural constant e as the base.

[0026] The shortest time interval between the t-th cycle and the current moment reflects the distance in time sequence between the t-th cycle and the current moment. The closer the time sequence distance, the more valuable the prediction of the running data at the next moment of the current moment by the t-th cycle is. The data change trends between the two are more continuous in time sequence, and thus the value of the corresponding initial timing weight is larger.

[0027] is a preset influence factor. In this embodiment, the influence factor is determined by using the time length within a set time period and is used to adjust the shortest time interval. The influence factor is specifically expressed as , represents the time length of the set time period. When the time length of the overall running data is relatively long, the data change rate should be slower to ensure that all the collected running data within the set time period can participate in the prediction analysis process, that is, to a certain extent, the stability degree of the data is improved. Furthermore, the more data participates, the more accurate the overall prediction result is. When the time length corresponding to the overall running data is relatively short, the fluctuation degree of the data itself is relatively small, and the degree of adjustment using the time length can be appropriately reduced.

[0028] The setting of the influence factor can better reflect the characteristic that the closer to the current moment, the closer the similarity of the normal fluctuations of the data is, and then the higher the overall prediction accuracy is. The initial timing weight reflects the weight proportion corresponding to the data of each cycle from the time distribution of each cycle, and thus reflects the practical value of the data within the cycle for the prediction analysis process.

[0029] Step 2: Based on the differences between the distribution of the trend item data in each period and the distribution of the trend item data in the adjacent period, the degree of dispersion of the trend item data in each period, and the differences between the trend item data in each period and the adjacent period, obtain the degree of trend anomaly for each period.

[0030] During the actual water conservancy construction drilling process, the collected operation data inevitably contains abnormal operation data. When only considering the time distribution to perform weighted prediction on the operation data of each period, the influence of abnormal data is not considered, which may result in relatively low accuracy of the predicted data. In this implementation, by analyzing the distribution of the trend item data and the residual item data of each period respectively, and then evaluating the degree of data anomaly of each period from two different characteristic data aspects, finally, combined with the initial time series weights, it is possible to maximize the removal of noise interference, retain the change characteristics of the data itself, and make the prediction result more accurate.

[0031] First, the trend item data of each period reflects the change trend of the operation data within each period. The data change trend of each period may have a certain degree of anomaly, or may show anomalies due to the influence of noise. In order to analyze the anomalies more accurately, the trend item data of each period is analyzed from multiple aspects specifically. The first aspect is to analyze the degree of dispersion of the trend item data of each period. The second aspect is to analyze the differences between the distribution of the trend item data of each period and the distribution of the trend item data of the adjacent period. The third aspect is to analyze the differences between the trend item data of each period and the adjacent period. Combining these three aspects, obtain the degree of trend anomaly for each period.

[0032] Based on this, first analyze the first aspect. Denote any period as the target period. According to the differences between each trend item data in the target period and the adjacent trend item data, obtain the data fluctuation degree of each trend item data in the target period, and calculate the variance of the data fluctuation degrees of all trend item data in the target period to obtain the first characteristic coefficient.

[0033] In this embodiment, take the t-th period as the target period, and take any trend item data in the t-th period as an example for illustration. Then the calculation formula for the data fluctuation degree of the i-th trend item data in the target period can be expressed as: ; where represents the data fluctuation degree of the i-th trend item data in the t-th period, represents the i-th trend item data in the t-th period, represents the (i - 1)-th trend item data in the t-th period, Represents the (i + 1)-th trend item data in the t-th period.

[0034] Reflects the data ratio between the i-th trend item data and its adjacent previous trend item data. Reflects the data ratio between the i-th trend item data and its adjacent next trend item data. Furthermore, the degree of data fluctuation reflects the difference change between the i-th trend item data and its two adjacent previous and next trend item data. The larger its value, the greater the difference between the i-th trend item data and the adjacent trend item data.

[0035] Then, calculate the variance of the data fluctuation degrees of all trend item data in the target period to obtain the first characteristic coefficient. The first characteristic coefficient characterizes the fluctuation of the difference changes of all trend item data corresponding to the target period within the adjacent data range. The larger the value of the first characteristic coefficient, the greater the fluctuation of the change trend of the trend item data in the target period and the greater the degree of dispersion.

[0036] Then, conduct a joint analysis on the second aspect and the third aspect. Specifically, calculate the difference between the mean value and the median value of all trend item data in the target period as the trend distribution degree of the target period; based on the difference situation between the trend distribution degrees of the target period and the adjacent period, as well as the difference situation of the mean values of the trend item data between the target period and other periods, obtain the second characteristic coefficient.

[0037] In this embodiment, the calculation formula for the second characteristic coefficient of the t-th period, which is also the second characteristic coefficient of the target period, can be expressed as: ; Wherein, Represents the second characteristic coefficient of the target period, t represents the t-th period, Represents the trend distribution degree of the t-th period, Represents the trend distribution degree of the (t - 1)-th period, Represents the (t + 1)-th trend distribution degree, N represents the total number of periods, Represents the mean value of all trend item data within the t-th period, Represents the mean value of all trend item data within the r-th period other than the t-th period.

[0038] In this embodiment, for the t-th cycle, the absolute value of the difference between the mean and the median of all the trend item data in the t-th cycle is used as the trend distribution degree of the t-th cycle. The difference between the mean and the median of all the data in one cycle is used to reflect the data distribution trend in that cycle. When the difference between the mean and the median is smaller, it indicates that the trend item data in the cycle is approximately distributed near the median, and further indicates that the trend item data in that cycle is more normal. When the difference between the mean and the median is larger, the data distribution is more discrete, indicating that the trend item data distribution in the cycle is more abnormal.

[0039] It represents the difference in the trend distribution degree between the target cycle and the previous adjacent cycle. It represents the difference in the trend distribution degree between the target cycle and the next adjacent cycle. The closer the ratio of the two is to 1, the closer the trend distribution change of the target cycle in the adjacent cycles is, and further the smaller the possibility of the target cycle being abnormal, and the smaller the value of the corresponding second characteristic coefficient.

[0040] It represents taking the mean of all the trend item data in each cycle as the data unit and performing a discrete calculation with the mean of the trend item data corresponding to the t-th cycle as the standard, which reflects the fluctuation between the mean of the trend item data in the t-th cycle and the means of the trend item data in all other cycles. The larger its value, the greater the degree of dispersion of the trend item data in the t-th cycle, and the greater the possibility of being abnormal, and the larger the value of the second characteristic coefficient.

[0041] The first characteristic coefficient reflects the possibility of the t-th cycle being abnormal from the aspect of the fluctuation degree of the data trend distribution, and the second characteristic coefficient reflects the possibility of the t-th cycle being abnormal from the aspect of the data difference between adjacent cycles and the degree of dispersion of the target cycle. The abnormality of the trend item data is evaluated by combining the characteristics of the trend item data of the target cycle in these aspects, that is, the trend abnormality degree of the target cycle is obtained according to the first characteristic coefficient and the second characteristic coefficient. Both the first characteristic coefficient and the second characteristic coefficient are positively correlated with the trend abnormality degree. In this embodiment, the normalized value of the product of the first characteristic coefficient and the second characteristic coefficient of the target cycle is used as the trend abnormality degree of the target cycle. The trend abnormality degree of the target cycle fully analyzes the distribution and change of the trend item data under the target cycle, and reflects the possibility of the operation data in the target cycle being abnormal.

[0042] It should be noted that in this embodiment, the change situation of the trend item data between each cycle and the two adjacent cycles before and after it is analyzed. At the same time, the change situation between each trend item data in each cycle and the two adjacent trend item data before and after it is also considered. Considering that there may be cases where the adjacent previous or adjacent subsequent data or cycles cannot be obtained, this embodiment does not consider the cycles where adjacent data cannot be obtained, nor does it consider the trend item data where adjacent data cannot be obtained within each cycle.

[0043] Step 3: Obtain the residual abnormality degree of each cycle according to the dispersion degree of the residual item data under each cycle and the difference between the dispersion degree of the residual item data under each cycle and that under other cycles.

[0044] The residual item data under each cycle obtained by STL decomposition represents the volatility and irregularity in the original data that cannot be captured by the trend and seasonality, which contains more random data in the operation data. In most cases, it is the random change of data affected by the external environment, and its overall distribution is similar to the Gaussian white noise distribution type. Based on this feature, this embodiment quantifies the characteristics of the residual item data within each cycle to describe the degree to which the operation data within each cycle is affected by external environmental factors.

[0045] Due to the unpredictability of the residual item data, the comprehensive discreteness calculation process of the residual item data of each cycle and the total number of the residual item data can be used to characterize the data characteristics of the residual item corresponding to each cycle. Furthermore, the similarity degree and difference situation of the data characteristics of the residual item between each cycle and other cycles are determined to determine the abnormal situation of the residual item of each cycle.

[0046] Based on this, the residual abnormality degree of each cycle is obtained according to the dispersion degree of the residual item data under each cycle and the difference between the dispersion degree of the residual item data under each cycle and that under other cycles.

[0047] Specifically, first, any cycle is denoted as the selected cycle. According to the number of residual item data included in the selected cycle, the mean value of all residual item data, and the difference between each residual item data and the mean value of all residual item data, the residual data characteristics of the selected cycle are obtained.

[0048] In this embodiment, when the s-th cycle is used as the selected cycle, the calculation formula for the residual data characteristics of the selected cycle can be expressed as: ; where represents the residual data characteristics of the selected cycle, s represents the s-th cycle, represents the total number of residual item data included in the s-th cycle, represents the mean of all residual term data within the s-th period, represents the u-th residual term data within the s-th period.

[0049] It reflects the difference between each residual term data and the mean of the residual term data within the selected period, and characterizes the fluctuation and dispersion degree of the residual term data in the selected period. Considering that the distribution equilibrium of the residual term data in each period is different, directly using variance to evaluate the dispersion degree is relatively absolute. In this embodiment, the difference between the fluctuation situation and the equilibrium situation of the residual term data is reflected by calculating the ratio of the dispersion degree to the mean. At the same time, considering that the number of residual term data in each period may be different, the number of residual term data is used as a weight for further product calculation. The greater the corresponding data dispersion degree and the more the number of residual term data, the greater the value of the corresponding residual data feature, indicating that the data feature of the residual term in the selected period is more in line with the noise distribution. The residual data feature characterizes the degree to which the feature distribution of the residual term data in each period satisfies the noise distribution.

[0050] Furthermore, according to the equilibrium degree of the difference between the residual data feature of the selected period and that of each other period, the residual abnormality degree of the selected period is obtained. More specifically, by calculating the ratio between the residual data feature of the selected period and that of any other period, the absolute value of the difference between 1 and the ratio is used as the difference feature between the selected period and the any other period; the mean value of the difference features between the selected period and each other period is calculated to obtain the residual abnormality degree of the selected period.

[0051] In this embodiment, the calculation formula for the residual abnormality degree of the selected period can be expressed as: ; where, represents the residual abnormality degree of the selected period, s represents the s-th period, N represents the total number of periods, represents the residual data feature of the s-th period, represents the residual data feature of the v-th period other than the s-th period.

[0052] represents the ratio between the residual data feature of the selected period and that of other periods, is the difference feature between the selected period and the v-th other period. When the ratio is closer to 1, the value of the difference feature is smaller, indicating that the residual data features between the selected period and other periods are more similar, and the possibility of abnormality in the residual term of the selected period is smaller, and the value of the residual abnormality degree is smaller.

[0053] The degree of residual abnormality in each period reflects the likelihood of abnormal residual term data within the corresponding period. The larger the value of the degree of residual abnormality, the greater the likelihood of abnormal residual term data within the corresponding period, and thus the greater the likelihood that the abnormal situation of the operation data within that period is due to noise data caused by external environmental factors.

[0054] Step 4, obtain the preferred time series weight for each period based on the degree of trend abnormality, the degree of residual abnormality, and the initial time series weight, use the preferred time series weight to predict the operation data, and monitor the water conservancy construction drilling process based on the prediction results.

[0055] The degree of trend abnormality in each period can reflect the abnormal situation of the operation data in terms of trend within each period, and the degree of residual abnormality in each period can reflect the abnormal situation of the residual term data of the operation data within each period. When the value of the degree of residual abnormality is larger, it indicates that the greater the likelihood that the abnormal situation of the operation data within the period in terms of trend is due to abnormalities caused by external environmental factors, and thus it indicates that there may be noise data in the operation data within that period, and then a smaller weight needs to be assigned to that period for data prediction analysis. When the value of the degree of residual abnormality is smaller, it indicates that the greater the likelihood that the abnormal situation of the operation data within the period in terms of trend is due to abnormalities during the drilling process, and then a larger weight needs to be assigned to that period for data prediction analysis.

[0056] Based on this, it is necessary to finally obtain the preferred time series weight for each period by combining the degree of trend abnormality, the degree of residual abnormality, and the initial time series weight.

[0057] Specifically, first, use the degree of residual abnormality in each period to correct the degree of trend abnormality so that the corrected degree of trend abnormality excludes the interference of noise and can truly represent the abnormal situation of the operation data in terms of trend within the period. That is, for any period, use the degree of residual abnormality in that period to correct the degree of trend abnormality to obtain the corrected degree of abnormality for that period.

[0058] In this embodiment, taking the t-th period as an example for illustration, the calculation formula for the corrected degree of abnormality of the t-th period can be expressed as: ; Wherein, represents the corrected degree of abnormality of the t-th period, represents the degree of trend abnormality of the t-th period, represents the degree of residual abnormality of the t-th period, represents the degree of residual abnormality of the r-th period other than the t-th period, N represents the total number of periods, and exp( ) represents the exponential function with the natural constant e as the base.

[0059] It reflects the difference between the abnormal degree of residuals in the t-th cycle and that in other cycles. The larger its value, the more discrete the abnormal degree of residuals in the t-th cycle, the greater the difference from other cycles, and the greater the possibility that there are noise data in the operation data in the t-th cycle. Therefore, it is necessary to adjust the abnormal degree of the trend to obtain a greater abnormal degree.

[0060] The adjusted abnormal degree It reflects the size of the true noise abnormality after the cycle is adjusted by the abnormal degree of residuals. The larger its value, the greater the proportion of noise abnormality in the data abnormality degree in the t-th cycle, and the smaller the proportion of the true abnormal situation of the corresponding operation data. Therefore, an exponential function is used for negative correlation mapping to obtain the corrected abnormal degree.

[0061] The corrected abnormal degree of each cycle characterizes the true abnormal operation situation of the operation data in each cycle that is not affected by other external factors. Furthermore, by comparing the difference between the corrected abnormal degree and the trend abnormal degree, it can be known whether there is noise in the data in each cycle. Specifically, the weight adjustment degree of the cycle is obtained according to the absolute value of the difference between the corrected abnormal degree and the trend abnormal degree, and there is a positive correlation between the absolute value of the difference and the weight adjustment degree.

[0062] In this embodiment, the calculation formula for the weight adjustment degree of the t-th cycle can be expressed as: where represents the weight adjustment degree of the t-th cycle, represents the corrected abnormal degree of the t-th cycle, represents the trend abnormal degree of the t-th cycle. represents the difference between the corrected abnormal degree and the trend abnormal degree of the t-th cycle, reflecting the difference between the true abnormal degree of the t-th cycle not affected by other factors and the comprehensive abnormal degree. The smaller this difference, the closer the abnormal degree of the change trend of the operation data in this cycle is before and after excluding the interference of external factors, and thus the more it indicates that the abnormal situation in this cycle is a real abnormal phenomenon, and the smaller the value of the corresponding weight adjustment degree, that is, there is no need to make a large adjustment operation on the initial time series weight of this cycle.

[0063] The larger the value of ,

[0064] Finally, the initial time series weight of each period is adjusted by using the weight adjustment degree, that is, the difference between the initial time series weight of the period and the weight adjustment degree is normalized to obtain the preferred time series weight of the period.

[0065] For the initial time series weight of each period, a reduction operation is performed on the basis of the initial time series weight according to the weight adjustment degree, so that the weight corresponding to the period with noise interference is lower, and the weight corresponding to the period without noise interference is higher. Finally, using the preferred time series weight of each period, predictive analysis is performed on the operation data within each period to obtain the prediction result of the operation data. In this embodiment, a weighted moving average prediction algorithm is used to perform predictive analysis on the operation data within each period. This algorithm is a well-known technology and will not be introduced in detail here. As Figure 6 shown is the prediction result of the operation data, where since the time length of the horizontal axis from 0 to 1 is the time length of the set time period, the horizontal axis starts from 1 as the predicted data.

[0066] Furthermore, the water conservancy construction drilling process is monitored based on the prediction result. In this embodiment, by setting a vibration threshold, when the operation data in the prediction result is greater than or equal to the preset vibration threshold, it indicates that an abnormal situation is about to occur during the drilling process, and relevant staff can be reminded through early warning to conduct inspections. Among them, the vibration threshold can be the average value of all vibration frequencies of the drilling equipment during a period of normal operation, or the standard vibration frequency set by relevant staff according to work experience for each different drilling operation.

[0067] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A drilling quality assessment method for bridge construction, characterized in that: The method comprises the following steps: Obtain the operation data of the drilling process of water conservancy construction within a set time period before the current moment; Perform STL decomposition on the operation data to obtain trend item data and residual item data in each cycle; obtain the initial time series weight of each cycle according to the time interval between each cycle and the current moment; According to the difference between the distribution of the trend item data in each period and the distribution of the trend item data in the adjacent period, the discrete degree of the trend item data in each period and the difference between the trend item data in each period and the adjacent period, the trend abnormality degree of each period is obtained; According to the discrete degree of the residual data in each period and the difference between the discrete degree of the residual data in each period and other periods, the residual abnormality of each period is obtained; The preferred time series weight of each cycle is obtained according to the trend abnormality degree, residual abnormality degree and initial time series weight, the operation data is predicted using the preferred time series weight, and the drilling process of water conservancy construction is monitored based on the prediction result.

2. A drilling quality assessment method for bridge construction according to claim 1, characterized in that: The abnormal trend degree of each cycle is obtained according to the difference between the distribution of the trend item data in each cycle and the distribution of the trend item data in the adjacent cycle, the discrete degree of the trend item data in each cycle, and the difference between the trend item data in each cycle and the adjacent cycle, specifically including: Record any cycle as the target cycle, obtain the data fluctuation degree of each trend item data under the target cycle according to the difference between each trend item data under the target cycle and the adjacent trend item data, and calculate the variance of the data fluctuation degree of all trend item data under the target cycle to obtain the first characteristic coefficient; Calculate the difference between the mean and median of all trend item data in the target period as the trend distribution degree of the target period; According to the difference between the trend distribution degree of the target period and the adjacent period, and the difference between the mean value of the trend item data of the target period and other periods, the second characteristic coefficient is obtained; The degree of abnormal trend of the target period is obtained according to the first characteristic coefficient and the second characteristic coefficient, and both the first characteristic coefficient and the second characteristic coefficient are positively correlated with the degree of abnormal trend.

3. A drilling quality assessment method for bridge construction according to claim 2, characterized in that: The calculation formula of the second characteristic coefficient is specifically: ； in, represents the second characteristic coefficient of the target cycle, t represents the tth cycle, Indicates the trend distribution degree of the tth period, Indicates the trend distribution degree of the t-1th period, represents the distribution degree of the t+1th trend, N represents the total number of cycles, represents the mean value of all trend item data in the tth period, It represents the mean value of all trend item data in the rth period except the tth period.

4. A drilling quality assessment method for bridge construction according to claim 2, characterized in that: The calculation formula of the data fluctuation degree is specifically: ； in, Indicates the data fluctuation degree of the i-th trend item data in the t-th period, represents the i-th trend item data in the t-th period, represents the i-1th trend item data in the tth period, Represents the i+1th trend item data in the tth period.

5. The method for evaluating the drilling quality of bridge construction according to claim 1, characterized in that: The preferred time series weight of each cycle is obtained according to the trend abnormality degree, the residual abnormality degree and the initial time series weight, specifically including: For any period, the trend abnormality degree is corrected by using the residual abnormality degree under the period to obtain the corrected abnormality degree of the period; the weight adjustment degree of the period is obtained according to the absolute value of the difference between the corrected abnormality degree and the trend abnormality degree, and there is a positive correlation between the absolute value of the difference and the weight adjustment degree; The difference between the initial timing weight of the cycle and the weight adjustment degree is normalized to obtain the preferred timing weight of the cycle.

6. A drilling quality assessment method for bridge construction according to claim 5, characterized in that: The method of correcting the trend abnormality by using the residual abnormality under the period to obtain the corrected abnormality of the period specifically includes: ； in, represents the degree of correction anomaly in the tth period, Indicates the degree of trend anomaly in the tth period, represents the residual abnormality of the tth period, represents the residual abnormality of the r-th period except the t-th period, N represents the total number of periods, and exp( ) represents an exponential function with the natural constant e as the base.

7. A drilling quality assessment method for bridge construction according to claim 1, characterized in that: The residual abnormality of each period is obtained according to the discreteness of the residual item data in each period and the difference between the discreteness of the residual item data in each period and other periods, specifically including: Any period is recorded as a selected period, and the residual data characteristics of the selected period are obtained according to the number of residual item data contained in the selected period, the mean of all residual item data, and the difference between each residual item data and the mean of all residual item data; The residual abnormality of the selected period is obtained according to the balance degree of the difference between the residual data characteristics of the selected period and each other period.

8. A drilling quality assessment method for bridge construction according to claim 7, characterized in that: The method of obtaining the residual abnormality of the selected period according to the balance degree of the difference between the residual data characteristics of the selected period and each other period specifically includes: Calculate the ratio between the residual data features of the selected period and any other period, and take the absolute value of the difference between 1 and the ratio as the difference feature between the selected period and any other period; calculate the mean value between the difference features of the selected period and each other period to obtain the residual abnormality of the selected period.

9. A drilling quality assessment method for bridge construction according to claim 8, characterized in that: The calculation formula of the residual data characteristics of the selected period is specifically: ； in, represents the residual data characteristics of the selected period, s represents the sth period, Represents the total number of residual data contained in the sth period, represents the mean of all residual data in the sth period, Represents the u-th residual data in the s-th period.

10. A drilling quality assessment method for bridge construction according to claim 1, characterized in that: The initial timing weight of each cycle is obtained according to the time interval between each cycle and the current moment, specifically including: For any cycle, the shortest time interval between the cycle and the current moment is obtained, and the shortest time interval is subjected to positive correlation normalization processing to obtain the initial timing weight of the cycle.

Images