Wagon balance settlement accumulation prediction method and system

Abstract

The invention provides a weighbridge settlement accumulation prediction method and system, and relates to the technical field of weighbridge settlement prediction, and the method comprises the steps: carrying out the real-time data collection of a target weighbridge region; feature analysis is carried out based on the multi-source real-time data set, and a settlement accumulation prediction channel is constructed; obtaining a settlement prediction value of the first time node through a short-term settlement prediction branch; obtaining a settlement prediction value of the first time slice through the long-term settlement prediction branch; performing geographically weighted regression analysis to generate a regional settlement difference coefficient; and carrying out space-time dynamic analysis on wagon balance settlement, carrying out settlement distribution abnormity analysis on a target wagon balance area, and triggering settlement early warning. Through the method and the device, the technical problem that the prediction result is not accurate enough due to the fact that settlement prediction in a certain specific time period is emphasized and the characteristics of short-term and long-term settlement are not fully combined in the prior art can be solved, and the accuracy of wagon balance settlement prediction is improved through combination of short-term and long-term prediction branches.

Description

Technical Field

[0001] The present application relates to the technical field of weighbridge settlement prediction, and in particular to a weighbridge settlement accumulation prediction method and system. Background Art

[0002] A weighbridge (also called a ground scale or ground weighing scale) is a device commonly used to weigh heavy objects. It is widely used in the fields of industry, transportation, logistics, etc., especially in the fields of cargo transportation, mineral mining, and building materials. However, during the long-term use of the weighbridge, affected by load changes, environmental factors, and aging infrastructure, the weighbridge may settle, resulting in reduced weighing accuracy and even equipment failure and safety hazards. Weighbridge settlement refers to the phenomenon that the position of the weighbridge gradually decreases due to factors such as compaction of the ground or foundation and load action. Settlement can be local or global. The accumulation of settlement will affect the accuracy of the weighbridge, especially after long-term operation, the settlement effect gradually accumulates, which may lead to an increase in weighing error, thereby affecting the efficiency of production and transportation. Short-term settlement may be affected by sudden factors such as load changes and climate change, while long-term settlement is mainly affected by foundation compaction and geological changes. Existing methods for predicting the accumulation of weighbridge settlement often focus on short-term settlement prediction or long-term trend prediction, and fail to fully consider the relationship and characteristics between short-term fluctuations and long-term settlement, resulting in inaccurate prediction results.

[0003] In summary, the existing technology has a technical problem that the prediction results are not accurate enough because it focuses on the settlement prediction in a specific time period and fails to fully combine the characteristics of short-term and long-term settlement. Summary of the invention

[0004] The purpose of this application is to provide a method and system for predicting the cumulative settlement of a weighbridge, so as to solve the technical problem that the existing technology focuses on the settlement prediction of a specific time period and fails to fully combine the characteristics of short-term and long-term settlement, resulting in inaccurate prediction results.

[0005] In view of the above problems, the present application provides a method and system for predicting the accumulation of weighbridge settlement.

[0006] In a first aspect, the present application provides a method for predicting the settlement accumulation of a weighing scale, which is implemented by a weighing scale settlement accumulation prediction system, wherein the method for predicting the settlement accumulation of a weighing scale comprises: collecting real-time data on a target weighing scale area according to multiple data sources to generate a multi-source real-time data set; performing feature analysis based on the multi-source real-time data set to extract multiple weighing scale features, and constructing a settlement accumulation prediction channel based on the multiple weighing scale features, wherein the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; training the short-term settlement prediction branch in combination with the multi-source real-time data set to obtain a settlement prediction value for a first time node, wherein the first time node is a real the next adjacent time node of the time point; through the long-term settlement prediction branch combined with the settlement prediction value of the first time node, a stabilization process is performed according to the settlement time series to obtain the settlement prediction value of the first time segment, and the first time segment includes multiple time nodes; the settlement prediction value of the first time node is subjected to geographical weighted regression analysis with the settlement prediction value of the first time segment to generate a regional settlement difference coefficient; traverse the target weighing scale area to perform spatiotemporal dynamic analysis on the weighing scale settlement in combination with the regional settlement difference coefficient, draw a regional settlement cumulative prediction trend chart, perform settlement distribution anomaly analysis on the target weighing scale area according to the regional settlement cumulative prediction trend chart, and generate settlement cumulative distribution anomaly data to trigger a settlement warning.

[0007] In the second aspect, the present application also provides a weighbridge settlement accumulation prediction system for executing a weighbridge settlement accumulation prediction method as described in the first aspect, wherein the weighbridge settlement accumulation prediction system includes: a data acquisition module, the data acquisition module is used to perform real-time data acquisition on the target weighbridge area according to multiple data sources to generate a multi-source real-time data set; a prediction channel construction module, the prediction channel construction module is used to perform feature analysis based on the multi-source real-time data set, extract multiple weighbridge features, and construct a settlement accumulation prediction channel based on the multiple weighbridge features, the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; a settlement prediction module, the settlement prediction module is used to train through the short-term settlement prediction branch in combination with the multi-source real-time data set to obtain a settlement prediction value at a first time node, the first time node being a real-time time point the next adjacent time node; a stabilization processing module, the stabilization processing module is used to perform stabilization processing according to the settlement time sequence through the long-term settlement prediction branch in combination with the settlement prediction value of the first time node, and obtain the settlement prediction value of the first time segment, the first time segment includes multiple time nodes; a weighted regression analysis module, the weighted regression analysis module is used to perform geographical weighted regression analysis on the settlement prediction value of the first time node and the settlement prediction value of the first time segment, and generate a regional settlement difference coefficient; an anomaly analysis module, the anomaly analysis module is used to traverse the target weighing scale area and combine the regional settlement difference coefficient to perform spatiotemporal dynamic analysis on the weighing scale settlement, draw a regional settlement cumulative prediction trend chart, perform settlement distribution anomaly analysis on the target weighing scale area according to the regional settlement cumulative prediction trend chart, and generate settlement cumulative distribution anomaly data to trigger a settlement warning.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: A multi-source real-time data set is generated by collecting real-time data from a target weighbridge area according to multiple data sources; feature analysis is performed based on the multi-source real-time data set to extract multiple weighbridge features, and a settlement accumulation prediction channel is constructed based on the multiple weighbridge features, wherein the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; training is performed by combining the short-term settlement prediction branch with the multi-source real-time data set to obtain a settlement prediction value for a first time node, wherein the first time node is the next adjacent time node of the real-time time point; training is performed by combining the long-term settlement prediction branch with the first time node The settlement prediction value is stabilized according to the settlement time series to obtain the settlement prediction value of the first time segment, which includes multiple time nodes; the settlement prediction value of the first time node is subjected to geographically weighted regression analysis with the settlement prediction value of the first time segment to generate a regional settlement difference coefficient; the target weighing scale area is traversed to perform spatiotemporal dynamic analysis on the weighing scale settlement in combination with the regional settlement difference coefficient, a regional settlement cumulative prediction trend chart is drawn, and settlement distribution anomaly analysis is performed on the target weighing scale area according to the regional settlement cumulative prediction trend chart, and settlement cumulative distribution anomaly data is generated to trigger a settlement warning. In other words, by integrating multi-source real-time data, combining short-term and long-term settlement prediction branches, and geographically weighted regression analysis, the settlement trend of different areas of the weighing scale is comprehensively predicted, thereby improving the accuracy of the weighing scale settlement prediction.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0011] Figure 1 The present invention is a flowchart of a method for predicting the accumulation of weighbridge settlement in the present invention.

[0012] Figure 2 This is a structural schematic diagram of a weighbridge settlement accumulation prediction system for this application.

[0013] Explanation of the reference numerals: data acquisition module 11 , prediction channel construction module 12 , sedimentation prediction module 13 , stabilization processing module 14 , weighted regression analysis module 15 , abnormality analysis module 16 . DETAILED DESCRIPTION

[0014] This application provides a method and system for predicting the cumulative settlement of a weighbridge, which solves the technical problem that the existing technology focuses on the settlement prediction of a specific time period and fails to fully combine the characteristics of short-term and long-term settlement, resulting in inaccurate prediction results. By integrating multi-source real-time data, combining short-term and long-term settlement prediction branches, and geographically weighted regression analysis, the settlement trend of different areas of the weighbridge is comprehensively predicted, thereby improving the accuracy of the settlement prediction of the weighbridge.

[0015] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0016] For example, please refer to the attached Figure 1 The present application provides a method for predicting the cumulative settlement of a weighbridge, wherein the method is applied to a system for predicting the cumulative settlement of a weighbridge, and the method specifically comprises the following steps: Step 1: Collect real-time data of the target scale area according to multiple data sources to generate a multi-source real-time data set.

[0017] Specifically, multiple data sources refer to real-time information from different channels or types, which are related to the settlement prediction of weighbridges, including weighbridge sensor data, meteorological data, soil monitoring data, vehicle transportation data, geological data, etc. The target weighbridge area refers to a specific area where the weighbridge is installed. The settlement changes in this area need to be monitored and predicted in real time. It may be a factory, warehouse, or mining area, etc., usually covering one or more pieces of land where the weighbridge equipment is located. Through various sensors and monitoring equipment, such as ground settlement meters, GPS positioning systems, strain gauges, inclinometers, etc., real-time data of the target weighbridge area is collected, including meteorological data (such as temperature, humidity, precipitation), soil data (such as humidity, density), foundation data (such as groundwater level, soil type), and load data (such as vehicle weight, frequency, etc.). For example, suppose that in a certain weighbridge area, a load sensor is installed to detect the weight of passing vehicles. Every time a vehicle passes, the sensor will record and transmit its weight data. In addition, the weather station records the temperature and precipitation every hour. At the same time, the soil sensor monitors the soil moisture in real time, records the data at the same time point, and forms a multi-source real-time data set. By collecting multi-dimensional real-time data, it is possible to comprehensively reflect the multi-factor background of settlement changes (such as load, climate, soil conditions, etc.), realize real-time monitoring of settlement conditions in the target scale area, and timely capture potential abnormal changes.

[0018] Step 2: Perform feature analysis based on the multi-source real-time data set, extract multiple weighbridge features, and construct a settlement accumulation prediction channel based on the multiple weighbridge features, wherein the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch.

[0019] Specifically, multi-source real-time data sets are different types of data sets collected by multiple sensors and monitoring equipment, such as real-time data from weighbridge sensors, meteorological monitoring, soil sensors, vehicle transportation records, etc. Based on the multi-source real-time data sets, factors that may affect settlement are identified, and statistical methods (such as Pearson correlation coefficient, regression analysis, machine learning models) are used to quantify the degree of influence of each influencing factor on settlement, and the importance score of each influencing factor is automatically calculated. Factors with higher scores have greater influence on settlement. Based on the multiple importance scores calculated, feature analysis is performed on the multi-source real-time data sets, that is, key features related to settlement are extracted from the multi-source real-time data sets, that is, multiple weighbridge features, including settlement rate (settlement per unit time), environmental impact (such as the impact of temperature and humidity on settlement), geological condition characteristics (such as soil type, rock depth), load distribution (such as settlement differences that may be caused by uneven load), etc. Weighbridge features refer to various factors that affect the settlement of weighbridges, including the settlement rate of weighbridges, environmental changes (such as temperature and humidity), geological conditions, load, etc.

[0020] Based on the extracted features of multiple weighbridges, a settlement accumulation prediction channel is constructed, including short-term settlement prediction branches and long-term settlement prediction branches. The short-term settlement prediction branch predicts the settlement changes of the weighbridge in the short term based on real-time data in a short period of time (such as hourly or daily data). The long-term settlement prediction branch performs long-term prediction of settlement accumulation based on data in a long period of time (such as historical settlement data), and predicts the settlement changes of the weighbridge in a long period of time (such as a few months or a few years).

[0021] The short-term settlement prediction branch uses the long short-term memory network (LSTM) as the core algorithm, combining the scale features in multi-source real-time data sets and time series data for dynamic data capture and modeling. LSTM is a deep learning model specifically designed for processing time series data, which can effectively handle the complexity and nonlinear changes of short-term data. Multiple scale features are used as input data in time series, using windowed time series data, training at short time nodes, and continuously adjusting model parameters in combination with loss functions (such as mean square error) to obtain the short-term settlement prediction branch. For example, compare the error between the predicted value and the true value, and adjust the parameters of the model, including the number of hidden layer units, learning rate, Epoch number (number of training rounds), etc.

[0022] The long-term settlement prediction branch uses the autoregressive integrated moving average model (ARIMA) as the core algorithm, combined with the long-term trend data extracted from the weighbridge features, to perform time series modeling and prediction. The ARIMA model is good at capturing long-term trends and cyclical changes in time series. Multiple weighbridge features are used as input data, and the data is differentially processed and tested for stationarity to ensure the applicability of the time series model. The model parameters (such as autoregressive coefficients and moving average terms) are fitted by input data, and the long-term trend prediction is performed using the future time series prediction results to obtain the long-term settlement prediction branch. The short-term settlement prediction branch uses the LSTM model, inputs the settlement rate characteristics, temperature change rate, etc., to predict the settlement of the weighbridge in the next month; the long-term settlement prediction branch uses the autoregressive model, inputs the settlement rate characteristics, geological condition characteristics, etc., to predict the settlement trend of the weighbridge in the next year. For example, if the short-term prediction results show that the settlement rate will accelerate in the next month, it may be necessary to check whether the supporting structure of the weighbridge is firm and carry out necessary reinforcement; if the long-term prediction results show that the settlement amount in the next year is large, it may be necessary to consider replacing the weighbridge or strengthening the foundation. By combining short-term and long-term settlement predictions, multi-level predictions of scale settlement can be achieved, from short-term changes to long-term trends, greatly improving the accuracy of the prediction.

[0023] Step three: training the short-term settlement prediction branch in combination with the multi-source real-time data set to obtain a settlement prediction value of a first time node, where the first time node is the next adjacent time node of the real-time time point.

[0024] Specifically, the short-term settlement prediction branch is part of the prediction channel, focusing on capturing the dynamic changes of the settlement of the scale within a short time range. Using the LSTM model as the prediction core, the feature sequence of the multi-source real-time data set is input to train the model to predict the settlement trend in the short term. The multi-source real-time data set is aligned and serialized according to the timestamp of data acquisition to obtain a time series data set. According to the current time node, the adjacent time node of the next time is determined. Using the recursive characteristics of LSTM, the serialized data is input and combined with the environmental condition characteristics at the current moment to gradually capture the change law in the time dimension. The time series data set and multiple scale features (such as settlement rate, environmental changes, geological conditions, etc.) are input into the prediction model for training, and the loss function (such as MSE) is introduced to calculate the error between the predicted value and the true value to evaluate the prediction error of the model. Through the supervised prediction training method, the model parameters are continuously optimized until the model converges and reaches the expected accuracy. After the model training converges, the latest real-time data is input to predict the settlement prediction value at the next time point. The first time node refers to the next adjacent time node of the real-time time point, which is the target time point that the model needs to predict. By using the short-term settlement prediction branch and multi-source real-time data sets to train the model and predict the settlement at the next time node after the real-time time point, it helps to timely understand the settlement of the weighbridge in the future period of time, so as to take corresponding preventive measures.

[0025] Step 4: The long-term settlement prediction branch is combined with the settlement prediction value of the first time node to perform stabilization processing according to the settlement time sequence to obtain the settlement prediction value of the first time segment, where the first time segment includes multiple time nodes.

[0026] Specifically, according to the settlement prediction value of the first time node, the settlement values ​​of the first N time nodes of the real-time time node are analyzed to obtain the settlement value for a period of time. According to the time series data set combined with multiple weighbridge features, a sequence of settlement value data arranged in chronological order, i.e., a time series, is obtained. The settlement values ​​for a period of time are stabilized to remove trend or seasonal fluctuations, usually by performing differential processing on the settlement prediction values. If the data after the initial differential is still not stable, a secondary differential may be required until the data becomes stable. The settlement prediction value of the first time segment is finally obtained through the settlement time series data after the stabilization processing, which usually includes the prediction values ​​of multiple time nodes, for example, starting from the first time node S(t+1), S(t+2), S(t+3), S(t+4)...S(t+k) are obtained after the stabilization processing, and the settlement prediction values ​​of multiple time nodes are integrated according to the settlement sequence to obtain the settlement prediction value of the first time segment, and the first time segment covers multiple time nodes. By combining the long-term settlement prediction branch with the settlement prediction value of the first time node and then stabilizing it, the non-stationarity in the data can be effectively eliminated, providing accurate and stable results for settlement prediction in the future.

[0027] Step 5: Perform geographically weighted regression analysis on the predicted settlement value of the first time node and the predicted settlement value of the first time segment to generate a regional settlement difference coefficient.

[0028] Specifically, GPS or other positioning technologies are used to obtain the longitude and latitude coordinate data of each geographical location point in the target weighbridge area. The settlement prediction value of the first time node and the settlement prediction value of the first time segment are integrated and identified with the corresponding geographical coordinate data, that is, the settlement prediction value is associated with different geographical locations in the target area (such as the coordinates of each scale) to form a multidimensional settlement data set. The weighted value of each geographical location point is weighted according to factors such as geographical distance and settlement change. The closer the distance between geographical locations, the greater the weight in the regression analysis and the more significant the impact. Geographically weighted regression analysis is a spatial data analysis method used to explore and model how the relationship between variables changes with changes in geographical location. Performing geographically weighted regression analysis on the settlement prediction value of the first time node and the settlement prediction value of the first time segment means that different regression coefficients will be applied at different geographical locations to analyze the relationship between the two settlement prediction values, in order to accurately capture the local characteristics of spatial data, so as to better understand and predict the settlement of the weighbridge area.

[0029] Based on the geographic coordinate data, the Gaussian kernel function is used to calculate the distance between different weighbridges and construct a weighted matrix. Based on the weighted matrix, the regression coefficient of each geographical location is calculated through regression analysis, and the regression coefficient of each geographical location describes the settlement trend of the point. By comparing the regression coefficients of time points and time periods, the settlement difference coefficient in the region is calculated to indicate the spatial difference in settlement changes in the region. The regional settlement difference coefficient reflects the difference between the settlement prediction value at the first time node and the settlement prediction value at the first time segment. By generating the regional settlement difference coefficient, the settlement of the weighbridge is accurately predicted, which not only reflects the relationship between the settlement prediction value at the first time node and the settlement prediction value at the first time segment, but also helps to evaluate the influence of different geographical locations on settlement prediction, thereby improving the accuracy and reliability of settlement prediction.

[0030] Step six: traverse the target weighing scale area and conduct spatiotemporal dynamic analysis on the weighing scale settlement in combination with the regional settlement difference coefficient, draw a regional settlement cumulative prediction trend chart, conduct settlement distribution anomaly analysis on the target weighing scale area based on the regional settlement cumulative prediction trend chart, generate settlement cumulative distribution anomaly data, and trigger a settlement warning.

[0031] Specifically, the target weighbridge area is traversed according to the geographic coordinate data and the regional settlement difference coefficient, and the settlement prediction value of each geographic location is combined with the corresponding regional settlement difference coefficient to obtain the settlement space data set. According to the settlement time sequence and the regional settlement difference coefficient, the target weighbridge area is traversed, and the settlement prediction value of each geographic location at different time nodes is combined with the regional settlement difference coefficient to obtain the settlement time data set. Using the settlement space data set and the settlement time data set, a spatiotemporal dynamic analysis of the weighbridge settlement is performed, and the settlement changes and spatial differences of each weighbridge at different time points are analyzed, and the impact of the spatial distribution and time changes of the weighbridge area on the settlement is considered in a timely manner. The settlement values ​​of each geographic location at different time nodes are analyzed, and the cumulative settlement values ​​are calculated to understand the settlement accumulation in the weighbridge area.

[0032] The plane coordinate system is constructed by taking the settlement time series as the horizontal axis and the cumulative settlement values ​​of multiple geographical locations as the vertical axis. The cumulative settlement values ​​of each geographical location are visualized in chronological order, and each data point represents the cumulative settlement value of a geographical location at a specific time node. The data in the plane coordinate system are plotted into a graph through visualization tools (such as Excel, Matplotlib library in Python, etc.), and a regional settlement cumulative prediction trend graph is generated, which clearly shows the settlement change trend of each scale at different time nodes, which helps to predict possible settlement risks in the future.

[0033] According to the regional settlement accumulation prediction trend chart, further abnormal analysis is carried out to mark the geographical locations where the settlement rate is abnormally fast, the settlement amount is abnormally large, or the settlement pattern is significantly different from the surrounding areas. The identified abnormal points and their settlement data are sorted into settlement cumulative distribution abnormal data. According to the safety standards of the weighbridge and historical settlement data, the threshold of the settlement warning is set. The settlement cumulative distribution abnormal data is compared with the warning threshold. When the abnormal data exceeds the warning threshold, the settlement warning is triggered. The content of the warning usually includes the location, time, settlement value and possible impact range of the abnormality. The warning information is conveyed to the relevant staff through system notifications, text messages, emails, etc., and on-site inspections are carried out in time or other measures are taken according to the warning information, such as further monitoring and strengthening safety protection. By traversing the target weighbridge area and combining the regional settlement difference coefficient for spatiotemporal dynamic analysis, the abnormal situation of settlement distribution can be accurately identified, the potential settlement risk can be discovered in time, and the settlement warning can be triggered, so as to take necessary measures to ensure the safe operation of the weighbridge and prevent possible damage.

[0034] Further, step 2 of this application includes: Perform a settlement impact analysis on the multi-source real-time data set to determine multiple influencing factors, each of which has an importance score; perform a settlement accumulation analysis on the multi-source real-time data set based on the multiple importance scores to determine settlement rate characteristics; perform an environmental change analysis on the multi-source real-time data set based on the multiple importance scores to determine environmental impact characteristics; perform a geological change analysis on the multi-source real-time data set based on the multiple importance scores to determine geological condition characteristics; perform a load change analysis on the multi-source real-time data set based on the multiple importance scores to determine load balancing characteristics; associate and integrate the settlement rate characteristics, the environmental impact characteristics, the geological condition characteristics, and the load balancing characteristics to determine the multiple weighbridge characteristics.

[0035] Specifically, the settlement impact analysis is performed on multi-source real-time data sets, that is, the degree of influence of different factors (such as environmental changes, meteorological data, geological data, load, etc.) on the settlement of the scale is analyzed. Influencing factors refer to factors that may cause changes in the settlement of the scale, such as soil moisture, temperature changes, vehicle load, etc. Statistical analysis methods (such as Pearson correlation coefficient, regression analysis, etc.) are used to quantify the contribution of each influencing factor to the settlement change, and the importance score of each influencing factor is obtained, that is, multiple importance scores, which reflect the relative importance of the factor on the settlement. The greater the impact, the higher the importance score.

[0036] The Pearson correlation coefficient is an indicator that measures the degree of linear correlation between two variables, and its value ranges from -1 to 1. If the Pearson correlation coefficient is close to 1, it means that there is a strong positive correlation between the two variables; if it is close to -1, it means a strong negative correlation; if it is close to 0, it means that there is no significant linear relationship. For each pair of variables (such as temperature and settlement, humidity and settlement, etc.), the Pearson correlation coefficient is calculated. Regression analysis is a statistical method used to study the relationship between a dependent variable (such as settlement) and one or more independent variables (such as temperature, humidity, etc.). Linear regression or multiple regression analysis (depending on the number of factors) is used to establish a mathematical model between settlement and each influencing factor. The regression coefficient indicates the degree of influence of each factor on settlement. The impact score of each factor is calculated by combining the regression coefficient, Pearson correlation coefficient and other methods, and a weighted score is assigned to each factor. The higher the score, the greater the impact on settlement.

[0037] The settlement rate feature is an indicator that describes the settlement speed of the weighbridge, usually the amount of settlement per unit time. Using multiple importance scores, the settlement accumulation analysis is performed on multi-source real-time data sets. By accumulating and analyzing the settlement data of the weighbridge over a period of time, the overall change trend of the settlement is calculated. The settlement data at each moment is calculated to obtain the settlement at each time point. The settlement change per unit time is calculated, that is, the change in settlement in a continuous time period is divided by the time difference to obtain the settlement rate. By calculating the settlement rate in different time periods and combining it with the importance score, the settlement rate feature is extracted to reflect the settlement trend of the weighbridge area and help predict future settlement changes. The settlement rate feature is usually calculated based on a time window. For example, the data from the past hour, day, or week can be used to estimate the settlement rate.

[0038] Environmental change analysis refers to evaluating the impact on the settlement of the weighbridge by analyzing the changes in environmental factors, including temperature, humidity, precipitation, air pressure, etc. The changes in the collected environmental data are calculated, that is, the difference or percentage change in the environmental data within a certain period of time, to evaluate the fluctuation of environmental factors. According to the changes in environmental factors and the importance score, the environmental features with strong influence are extracted, which can be the relationship between the environmental change and the settlement rate, or the impact of the change pattern of a specific environmental factor on the settlement. Environmental impact characteristics refer to the relevant features extracted from environmental data that can affect the settlement of the weighbridge, and the correlation with the settlement process is quantified in this step. Similarly, geological change analysis refers to evaluating the impact on the settlement of the weighbridge by analyzing the changes in geological conditions, including groundwater level, soil type, etc. Geological condition characteristics refer to characteristics related to geological factors such as soil type, groundwater level, soil moisture, etc. Load change analysis refers to analyzing the impact of changes in load (such as vehicle weight, traffic flow) on the settlement of the weighbridge to obtain load balancing characteristics.

[0039] The settlement rate characteristics, environmental impact characteristics, geological condition characteristics and load balance characteristics are correlated and integrated, and the relationship between each characteristic is analyzed to identify which characteristics have a greater impact on settlement and which characteristics have interactive effects. All extracted features are integrated to obtain multiple weighbridge characteristics, and multiple influencing factors (settlement rate, environment, geology, load, etc.) are taken into consideration to avoid the limitations of single factor prediction and provide a more comprehensive settlement prediction. Through the calculation of multi-source data sets and importance scores, the various factors affecting settlement are comprehensively analyzed to ensure the accuracy and comprehensiveness of the settlement prediction model, extract and analyze the settlement rate, accurately identify the change trend of settlement, predict future settlement rates, and combine environmental factors, geological conditions, loads and other multi-dimensional factors to optimize the settlement prediction model, which can effectively respond to settlement changes under different conditions and provide early warning of settlement risks.

[0040] Further, step three of this application includes: The multi-source real-time data sets are aligned according to the data acquisition timestamp to obtain a multi-source aligned data set; serialization processing is performed based on the data acquisition timestamp in combination with the multi-source aligned data set to construct a time series data set; an adjacent search is performed on the time series data set according to the real-time time node to determine the first time node, and the first time node is the next adjacent time node of the real-time time point; the short-term settlement prediction branch uses a long short-term memory network to capture data according to the time series data set in combination with the multiple weighbridge features to generate short-term dynamic change data; a loss function is introduced, and an error analysis is performed based on the short-term dynamic change data in combination with the loss function to draw a short-term settlement loss change curve; the short-term settlement loss change curve is traversed using the first time node as an index to determine the settlement loss value of the first time node; the settlement loss value of the first time node is used as supervision data, and supervised prediction training is performed on the short-term dynamic change data to generate an initial training result, and when it is verified that the initial training result tends to converge, a settlement prediction value of the first time node is generated.

[0041] Specifically, multi-source data usually comes from different devices or systems, and the collection time may be inconsistent. The multi-source real-time data sets are aligned according to the data collection timestamp. The data collection timestamp refers to the specific collection time of each set of data during the multi-source data collection process. It is used to mark the time sequence of the data to ensure the synchronization and accuracy of multi-source data. After timestamp alignment, data from different sources are integrated into the same time node to form a synchronized data set, that is, a multi-source aligned data set. For example, sedimentation data is collected every 1 hour, and environmental data is collected every 30 minutes. Through timestamp alignment, all data are mapped to the same time axis, and missing data is supplemented by interpolation or filling methods.

[0042] The aligned multi-source aligned datasets are sorted in chronological order, that is, the multi-source real-time datasets are aligned with the data acquisition timestamp to ensure that each data point has a corresponding time stamp. The data is serialized according to the data acquisition timestamp, that is, the data is arranged in chronological order to construct a time series dataset. The time series dataset contains multiple observations arranged in chronological order, which can reflect the dynamic change trend of the data. The data at each time point contains multiple variables or features, such as sedimentation rate, temperature, and load.

[0043] According to the real-time time node (current time point), the time node adjacent to the real-time time node is searched in the time series data set through adjacent search to determine the next adjacent time point, that is, the first time node. The adjacent time node refers to the time point that follows the real-time time point in time. The LSTM model is selected as the short-term sedimentation prediction model because the long short-term memory network (LSTM) is a special recurrent neural network (RNN) that is very suitable for processing and predicting time series data. The LSTM structure includes an input gate, a forget gate, and an output gate, which captures the time dependency of the data through memory units.

[0044] The time series data set and multiple weighbridge features are input into the LSTM network to capture the dynamic change characteristics of the settlement in a short time. LSTM updates the weights through the back-propagation algorithm, and optimizes the prediction results in combination with the loss function (such as mean square error) to obtain short-term dynamic change data. Through the powerful time-dependent capture ability of the LSTM network, the short-term settlement prediction branch can generate accurate short-term dynamic change data. During the training process, the error between the predicted value and the true value is compared by calculating the loss function (such as mean square error), and the loss curve is drawn to evaluate the convergence of the model. The mean square error (MSE) is the average of the squares of the difference between the predicted value and the actual value. The data is predicted using the mean square error, and then the difference between the predicted value and the actual value is calculated and squared. The squared errors of all data points are summed and divided by the number of data points to obtain the MSE value. The smaller the MSE value, the smaller the prediction error of the model and the better the performance of the model. During the training process, the MSE value of each iteration step is continuously recorded, and these MSE values ​​are plotted into a short-term settlement loss change curve, with the horizontal axis representing the number of iterations or time and the vertical axis representing the MSE value. By observing the MSE curve, we can determine the performance changes of the model during the training process and whether it has converged. If the MSE curve decreases with the increase of iterations and tends to be stable, it means that the performance of the model is improving and has converged to a better solution. If the MSE curve rises or fluctuates, the parameters or structure of the model need to be adjusted to improve performance.

[0045] Taking the first time node as the index, traverse the short-term settlement loss change curve to determine the settlement loss value of the corresponding first time node. Using the settlement loss value of the first time node as the supervision data, supervised training is performed on the short-term dynamic change data to make the predicted value approach the true value. The model continuously tries to learn how to generate outputs that match the supervision data based on the input data. During the training process, the model will continuously adjust its parameters to minimize the difference between the predicted value and the supervision data, such as adjusting the learning rate, the number of iterations, etc. Verify the initial training results and check whether the model training converges. The verification standard is that the loss value decreases and tends to be stable. When the training results of the model tend to converge, that is, the loss value no longer decreases significantly, it is considered that the model has learned enough information to generate a prediction value and obtain the settlement prediction value of the first time node. By utilizing the time memory characteristics of the LSTM network, the short-term dynamics of settlement changes are accurately captured, and the performance of the prediction model is gradually improved by combining supervised training and loss curve analysis.

[0046] Further, step 4 of this application includes: Based on the settlement prediction value of the first time node, the settlement values ​​of the N time nodes before the real-time time node are extracted, wherein N is an integer greater than 1; based on the time series data set combined with the multiple scale characteristics, the settlement time sequence is determined, and the settlement values ​​of the first N time nodes are differentially processed according to the settlement time sequence to generate multiple differential processing results; the multiple differential processing results are subjected to a stationarity test to generate multiple data stationarity coefficients; it is determined whether the multiple data stationarity coefficients meet the expected stationarity threshold, if the multiple data stationarity coefficients do not meet the expected stationarity threshold, threshold, then according to the data smoothness coefficient that does not meet the expected smoothness threshold, the settlement value of the ith time node is extracted for secondary difference, until the multiple data smoothness coefficients meet the expected smoothness threshold, i is an integer greater than 0; if the multiple data smoothness coefficients meet the expected smoothness threshold, the settlement values ​​of the first N time nodes are combined with the settlement prediction value of the first time node for autoregressive fitting calculation to determine the settlement prediction values ​​of multiple time nodes; the settlement prediction values ​​of the multiple time nodes are correlated and integrated according to the settlement time series to obtain the settlement prediction value of the first time segment.

[0047] Specifically, according to the settlement prediction value of the first time node, the time range of the historical settlement data to be extracted is determined. The settlement values ​​of the first N time nodes of the real-time time node are extracted, that is, the settlement values ​​of a period of time close to the current time node. Using the time series data set and multiple weighbridge features, the time series characteristics of the settlement are determined, including identifying the trend changes and seasonal fluctuations of the settlement data. The settlement values ​​of the first N time nodes are differentially processed according to the settlement time series. Differential processing is a statistical technique used to calculate the amount of data change at consecutive time points. It is a common method in time series analysis. The main purpose is to convert non-stationary time series into stationary series. The trend is eliminated by calculating the difference between each time point and the previous time point. For each of the first N time nodes, the differential results are calculated separately to obtain multiple differential processing results. Different differential processing results can reveal different trend components, such as long-term trends, seasonal fluctuations, etc., reflecting the dynamic characteristics of settlement changes. For example, if the settlement values ​​at time points T1 and T2 are S1 and S2 respectively, then the differential result ΔS=S2-S1.

[0048] Perform a stationarity test on multiple difference processing results to verify whether the data after difference has met the stationarity requirements. The stationarity test is used to detect the stationarity of the data series, that is, whether the statistical characteristics of the data change over time. The performance of a stationary time series is stable, and its trend will not change drastically in the past, present or future. In time series analysis, non-stationary data needs to be differenced to make it stationary. Stationarity tests usually include unit root tests, such as ADF tests, which are used to determine whether the time series has a unit root, that is, whether there is non-stationarity. Apply a stationarity test method (such as ADF, KPSS or PP test) to each difference series, and calculate the stationarity coefficient of each series to obtain multiple data stationarity coefficients. The stationarity test of each difference processing result will produce a stationarity coefficient, indicating whether the difference series is stationary.

[0049] A desired stability threshold is defined in advance to determine whether the data is stable. If multiple data stability coefficients do not meet the desired stability threshold, the settlement value of the i-th time node is extracted for the data stability coefficient that does not meet the expected stability threshold, that is, the settlement value of the corresponding time node is subjected to secondary difference processing until the data is stable. The existing differential data is differentiated again to make the data more stable by further removing trends or periodic changes. The corresponding settlement value is extracted for the unsteady differential data, and secondary difference is performed until the data is stable. This process is recursive until all data stability coefficients meet the expected stability threshold.

[0050] When the stationary coefficients of multiple data meet the expected stationary threshold, the settlement values ​​of the first N time nodes and the settlement prediction value of the first time node are input into the autoregressive model for fitting. The autoregressive model is a statistical model that predicts the current and future settlement values ​​through the settlement values ​​of the first few time nodes. On the basis of the autoregressive model, the moving average (MA) term is introduced, that is, the error term of the time series is considered in the model at the same time, which is usually called the ARMA model (autoregressive moving average model) to better capture the long-term trend and fluctuation of the data. After autoregressive fitting, the settlement prediction values ​​of multiple time nodes are calculated, including the future settlement change trend, which are usually derived step by step through the autoregressive model based on the weights calculated in the formula. For example, based on the settlement values ​​of the first N time nodes and the settlement prediction value of the first time node, S(t+2), S(t+3), S(t+4)…S(t+k) are calculated, and so on, until the settlement prediction values ​​of multiple time nodes are obtained.

[0051] The settlement time series refers to the time sequence of the settlement prediction values ​​of multiple time nodes, which ensures the order of the prediction values ​​and makes them logically related in time. The settlement prediction values ​​of multiple time nodes are integrated according to the settlement time series to ensure that the prediction values ​​at different time points can be smoothly connected to form a continuous settlement prediction trend. Ensure that the settlement prediction values ​​of multiple time nodes are arranged in the correct time sequence, connect the settlement prediction values ​​of multiple time nodes according to the time sequence, and ensure that they are continuous in the time series. Association integration usually refers to organizing the settlement prediction values ​​of different time nodes in time sequence to form a continuous time series prediction. The result of the integration is the settlement change trend within a given time period, that is, the settlement prediction value of the first time segment. Through differential processing, the trend component in the settlement data is removed, making the data more suitable for model fitting and improving the accuracy of prediction. The autoregressive model is used to capture the long-term trend in the time series data to improve the accuracy of prediction. The settlement prediction values ​​of multiple time nodes are associated and integrated to obtain a more comprehensive time segment settlement prediction result.

[0052] Furthermore, step five of this application includes: Traverse the target weighing scale area to perform geographic information sensing and obtain geographic coordinate data of the target weighing scale area; integrate and identify the settlement prediction value of the first time node and the settlement prediction value of the first time segment according to the geographic coordinate data to generate a multidimensional settlement data set; traverse the multidimensional settlement data set to perform weighting and construct a weighted matrix; perform regression analysis based on the weighted matrix and the geographic coordinate data to generate multiple regression coefficients for multiple geographic location points; perform settlement identification based on the multiple regression coefficients of the multiple geographic location points and determine the settlement difference coefficient of the region.

[0053] Specifically, the target weighbridge area is scanned through specific sensing equipment (such as GPS, GIS system, etc.) to obtain the geographic coordinate data of the target weighbridge area, including the specific location of the geographic location (such as longitude, latitude, altitude information, etc.). The settlement prediction value of the first time node and the settlement prediction value of the first time segment are integrated according to the corresponding geographic coordinate data, and a geographic location information is added to the settlement prediction value of each time node to obtain a multidimensional settlement data set. The multidimensional settlement data set not only contains the settlement prediction value in the time series, but also combines the geographic location corresponding to each time node, realizing the spatial and temporal dual display of the settlement value.

[0054] Traverse the multidimensional settlement data set, combine the Gaussian kernel function to calculate the weight, calculate the similarity between it and other location points, and obtain the influence weight of all geographical location points. The calculated multiple influence weight values ​​are mapped and arranged according to the data points in the multidimensional settlement data set to form a weighted matrix, which contains the weights of different geographical locations. Each element w(i,j) represents the influence weight of the i-th data point on the j-th data point. For each geographical location point, the weighted matrix is ​​combined with the settlement prediction value of the first time node, and regression analysis is performed to calculate the first regression coefficient, which represents the degree of influence of the geographical location on the settlement prediction value. Traverse the geographical coordinate data, combine the settlement prediction value of the first time segment for regression calculation, and obtain the second regression coefficient, which represents the relationship between the settlement prediction value and the geographical coordinates in the time segment. Integrate the first regression coefficient and the second regression coefficient of each geographical location point to obtain multiple regression coefficients of multiple geographical location points, including short-term and long-term regression coefficients, to reveal the settlement change rules between different locations.

[0055] Based on multiple regression coefficients of multiple geographic locations, the settlement changes at different locations are marked, thereby identifying the settlement patterns of certain areas. Areas with large regression coefficients (such as 0.7) may indicate that the settlement changes in the area are large, while areas with small regression coefficients (such as 0.5) may indicate that the settlement changes are small. Based on the differences in regression coefficients between time points and time periods at each geographic location, the spatial heterogeneity of settlement is determined, and the mean, standard deviation, or maximum difference of the regression coefficient differences at all locations is calculated to quantify the differences in settlement within the region. If the calculated regional settlement difference coefficient is large (such as close to 1 or greater), it means that the settlement changes within the region are quite different, and the settlement changes at some locations may be significant, while the changes at other locations are small. If the calculated regional settlement difference coefficient is small (such as close to 0), it means that the settlement changes within the region are relatively uniform, and the differences in settlement changes between different locations are small. Through geographically weighted regression analysis, the settlement differences between different geographic locations can be accurately reflected, so as to better understand the distribution characteristics of scale settlement.

[0056] Furthermore, the present application also includes the following steps: The multidimensional settlement data set is traversed, and weight calculation is performed in combination with the Gaussian kernel function to determine multiple impact weight values; mapping and arranging are performed according to the multiple impact weight values ​​in combination with the multidimensional settlement data set to construct a weighted matrix; the geographic coordinate data is traversed, and regression calculation is performed according to the weighted matrix in combination with the settlement prediction value of the first time node to generate a first regression coefficient; the geographic coordinate data is traversed, and regression calculation is performed according to the weighted matrix in combination with the settlement prediction value of the first time segment to generate a second regression coefficient; the first regression coefficient and the second regression coefficient are added to the multiple regression coefficients.

[0057] Specifically, the multidimensional settlement data set is traversed and each data point is processed. For each data point, the Gaussian kernel function is used to calculate its weight. The Gaussian kernel function is a commonly used kernel function that can measure the similarity between data points. The weight is calculated based on the distance between two data points, and multiple influence weight values ​​are determined. The closer the distance between the data points, the greater the weight; conversely, the smaller the weight. The influence weight value is determined based on the similarity between the geographical location points, and is used to reflect the importance of each geographical location point in settlement prediction. Through the Gaussian kernel function calculation, the weight value can reflect the impact of a certain location on the settlement changes in the surrounding area. For example, for location A and location B, the distance between the two is calculated and the Gaussian kernel function is applied to obtain the influence weight.

[0058] After calculating the influence weights of all geographical locations, the calculated influence weight values ​​are mapped and arranged according to the data points in the multidimensional settlement data set to establish a weighted matrix. Each element in the matrix represents the degree of influence between different locations. The weighted matrix is ​​a matrix composed of influence weight values, which is used to weight each data point in the regression calculation. The weighted matrix is ​​generated by mapping the weights calculated by the Gaussian kernel function to the settlement data set. For each geographical coordinate data point, the weighted matrix and the settlement prediction value of the first time node are used for regression calculation to generate the first regression coefficient. The first regression coefficient is calculated using the weighted matrix and the geographical coordinate data, which represents the degree of influence of the geographical location on the settlement prediction value.

[0059] The regression coefficient represents the relationship between the predicted settlement value and each geographical location point, reflecting the contribution of different location points to the settlement prediction. Regression calculation uses regression analysis method to combine geographical coordinate data, weighted matrix and predicted settlement value to generate regression coefficient. Traverse the geographical coordinate data, combine the predicted settlement value of the first time segment to perform regression calculation, and obtain the second regression coefficient through regression analysis, which represents the relationship between the predicted settlement value and the geographical coordinates in the time segment. Add the generated first regression coefficient and the second regression coefficient to the obtained regression coefficient set to form the final regression coefficient set. Through weighted matrix and regression analysis, the influence of different geographical locations on settlement prediction can be accurately reflected, and the prediction accuracy can be improved.

[0060] Furthermore, step six of this application includes: According to the geographic coordinate data, the target weighing scale area is traversed in combination with the regional settlement difference coefficient to generate a settlement space data set; according to the settlement time sequence, the target weighing scale area is traversed in combination with the regional settlement difference coefficient to generate a settlement time data set; based on the settlement space data set and the settlement time data set, a spatiotemporal dynamic analysis of the weighing scale settlement is performed to determine the cumulative settlement values ​​of multiple geographical locations; the horizontal axis is set according to the settlement time sequence, and the vertical axis is set according to the cumulative settlement values ​​of the multiple geographical locations, and a plane coordinate system is constructed according to the horizontal axis and the vertical axis for visualization processing to construct the regional settlement cumulative prediction trend chart.

[0061] Specifically, the geographic coordinate data of the target weighbridge area is combined with the regional settlement difference coefficient, and the settlement spatial data set is generated by traversing the geographic coordinate data. The coordinate data of each geographical location point (each point in the weighbridge area) is traversed, and the settlement change of each geographical location is identified by combining the settlement difference coefficient of each geographical location to generate a spatial data set. The coordinates of each geographical location point are combined with its corresponding settlement difference coefficient to form a settlement spatial data set, which includes multiple geographical location points and corresponding settlement difference coefficients.

[0062] The settlement time series is determined by analyzing the time series data set combined with multiple weighbridge features to determine the sequence of settlement data at different time nodes. The settlement time series describes the law of settlement changes over time, which can help analyze the dynamic change process of settlement. Combined with the settlement time series, the target weighbridge area is traversed to generate a time-related settlement data set. Analyze the settlement time series based on the time series data set. The time series data set is a multi-source aligned data set to ensure that the data is aligned by timestamp. At each time node in the time series, combined with the regional settlement difference coefficient, a settlement data set for each time node is generated, which contains the settlement prediction value and regional settlement difference coefficient for each geographic location over time.

[0063] Combined with the settlement space dataset and the settlement time dataset, the settlement of each geographical location is analyzed in a spatiotemporal dynamic manner, and the process of settlement change over time is analyzed, while the spatial differences of different geographical locations are also considered. Through spatiotemporal dynamic analysis, the settlement value of each geographical location at different time nodes is analyzed, the cumulative settlement value of each geographical location is calculated, and the settlement accumulation in the scale area is determined. According to the settlement time sequence, the horizontal axis is set to represent the time sequence. According to the cumulative settlement values ​​of different geographical locations, the vertical axis is set to mark the settlement amount. A plane coordinate system is constructed based on the horizontal and vertical axes. In the plane coordinate system, the cumulative settlement value of each geographical location is visualized in time sequence. Through visualization, a regional settlement accumulation prediction trend chart is constructed to show the settlement accumulation in the scale area over time. By comprehensively analyzing the settlement changes in the target scale area, both the changes in the time series and the spatial differences in geographical locations are considered.

[0064] In summary, the method for predicting the cumulative settlement of a weighbridge provided in this application has the following technical effects: A multi-source real-time data set is generated by collecting real-time data from a target weighbridge area according to multiple data sources; feature analysis is performed based on the multi-source real-time data set to extract multiple weighbridge features, and a settlement accumulation prediction channel is constructed based on the multiple weighbridge features, wherein the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; training is performed by combining the short-term settlement prediction branch with the multi-source real-time data set to obtain a settlement prediction value for a first time node, wherein the first time node is the next adjacent time node of the real-time time point; training is performed by combining the long-term settlement prediction branch with the first time node The settlement prediction value is stabilized according to the settlement time series to obtain the settlement prediction value of the first time segment, which includes multiple time nodes; the settlement prediction value of the first time node is subjected to geographically weighted regression analysis with the settlement prediction value of the first time segment to generate a regional settlement difference coefficient; the target weighing scale area is traversed to perform spatiotemporal dynamic analysis on the weighing scale settlement in combination with the regional settlement difference coefficient, a regional settlement cumulative prediction trend chart is drawn, and settlement distribution anomaly analysis is performed on the target weighing scale area according to the regional settlement cumulative prediction trend chart, and settlement cumulative distribution anomaly data is generated to trigger a settlement warning. In other words, by integrating multi-source real-time data, combining short-term and long-term settlement prediction branches, and geographically weighted regression analysis, the settlement trend of different areas of the weighing scale is comprehensively predicted, thereby improving the accuracy of the weighing scale settlement prediction.

[0065] Embodiment 2: Based on the same inventive concept as the method for predicting the cumulative settlement of a scale in the aforementioned embodiment 1, the present application also provides a system for predicting the cumulative settlement of a scale, as shown in the attached Figure 2 The weighbridge settlement accumulation prediction system comprises: A data acquisition module 11, the data acquisition module 11 is used to collect real-time data from a target weighbridge area according to multiple data sources to generate a multi-source real-time data set; a prediction channel construction module 12, the prediction channel construction module 12 is used to perform feature analysis based on the multi-source real-time data set, extract multiple weighbridge features, and construct a settlement accumulation prediction channel based on the multiple weighbridge features, the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; a settlement prediction module 13, the settlement prediction module 13 is used to train through the short-term settlement prediction branch in combination with the multi-source real-time data set to obtain a settlement prediction value for a first time node, the first time node being the next adjacent time node of the real-time time point; a stabilization processing module 14, the stabilization processing module 14 is used to The long-term settlement prediction branch combines the settlement prediction value of the first time node and performs stabilization processing according to the settlement time sequence to obtain the settlement prediction value of the first time segment, and the first time segment includes multiple time nodes; the weighted regression analysis module 15, the weighted regression analysis module 15 is used to perform geographical weighted regression analysis on the settlement prediction value of the first time node and the settlement prediction value of the first time segment, and generate a regional settlement difference coefficient; the anomaly analysis module 16, the anomaly analysis module 16 is used to traverse the target weighing scale area and combine the regional settlement difference coefficient to perform spatiotemporal dynamic analysis on the weighing scale settlement, draw a regional settlement cumulative prediction trend chart, and perform settlement distribution anomaly analysis on the target weighing scale area according to the regional settlement cumulative prediction trend chart, generate settlement cumulative distribution anomaly data, and trigger a settlement warning.

[0066] Furthermore, the prediction channel construction module 12 in the weighbridge settlement accumulation prediction system is also used for: Perform a settlement impact analysis on the multi-source real-time data set to determine multiple influencing factors, each of which has an importance score; perform a settlement accumulation analysis on the multi-source real-time data set based on the multiple importance scores to determine settlement rate characteristics; perform an environmental change analysis on the multi-source real-time data set based on the multiple importance scores to determine environmental impact characteristics; perform a geological change analysis on the multi-source real-time data set based on the multiple importance scores to determine geological condition characteristics; perform a load change analysis on the multi-source real-time data set based on the multiple importance scores to determine load balancing characteristics; associate and integrate the settlement rate characteristics, the environmental impact characteristics, the geological condition characteristics, and the load balancing characteristics to determine the multiple weighbridge characteristics.

[0067] Furthermore, the settlement prediction module 13 in the weighbridge settlement accumulation prediction system is also used for: The multi-source real-time data sets are aligned according to the data acquisition timestamp to obtain a multi-source aligned data set; serialization processing is performed based on the data acquisition timestamp in combination with the multi-source aligned data set to construct a time series data set; an adjacent search is performed on the time series data set according to the real-time time node to determine the first time node, and the first time node is the next adjacent time node of the real-time time point; the short-term settlement prediction branch uses a long short-term memory network to capture data according to the time series data set in combination with the multiple weighbridge features to generate short-term dynamic change data; a loss function is introduced, and an error analysis is performed based on the short-term dynamic change data in combination with the loss function to draw a short-term settlement loss change curve; the short-term settlement loss change curve is traversed using the first time node as an index to determine the settlement loss value of the first time node; the settlement loss value of the first time node is used as supervision data, and supervised prediction training is performed on the short-term dynamic change data to generate an initial training result, and when it is verified that the initial training result tends to converge, a settlement prediction value of the first time node is generated.

[0068] Furthermore, the stabilization processing module 14 in the weighbridge settlement accumulation prediction system is also used for: Based on the settlement prediction value of the first time node, the settlement values ​​of the N time nodes before the real-time time node are extracted, wherein N is an integer greater than 1; based on the time series data set combined with the multiple scale characteristics, the settlement time sequence is determined, and the settlement values ​​of the first N time nodes are differentially processed according to the settlement time sequence to generate multiple differential processing results; the multiple differential processing results are subjected to a stationarity test to generate multiple data stationarity coefficients; it is determined whether the multiple data stationarity coefficients meet the expected stationarity threshold, if the multiple data stationarity coefficients do not meet the expected stationarity threshold, threshold, then according to the data smoothness coefficient that does not meet the expected smoothness threshold, the settlement value of the ith time node is extracted for secondary difference, until the multiple data smoothness coefficients meet the expected smoothness threshold, i is an integer greater than 0; if the multiple data smoothness coefficients meet the expected smoothness threshold, the settlement values ​​of the first N time nodes are combined with the settlement prediction value of the first time node for autoregressive fitting calculation to determine the settlement prediction values ​​of multiple time nodes; the settlement prediction values ​​of the multiple time nodes are correlated and integrated according to the settlement time series to obtain the settlement prediction value of the first time segment.

[0069] Furthermore, the weighted regression analysis module 15 in the weighbridge settlement accumulation prediction system is also used for: Traverse the target weighing scale area to perform geographic information sensing and obtain geographic coordinate data of the target weighing scale area; integrate and identify the settlement prediction value of the first time node and the settlement prediction value of the first time segment according to the geographic coordinate data to generate a multidimensional settlement data set; traverse the multidimensional settlement data set to perform weighting and construct a weighted matrix; perform regression analysis based on the weighted matrix and the geographic coordinate data to generate multiple regression coefficients for multiple geographic location points; perform settlement identification based on the multiple regression coefficients of the multiple geographic location points and determine the settlement difference coefficient of the region.

[0070] Furthermore, the weighted regression analysis module 15 in the weighbridge settlement accumulation prediction system is also used for: The multidimensional settlement data set is traversed, and weight calculation is performed in combination with the Gaussian kernel function to determine multiple impact weight values; mapping and arranging are performed according to the multiple impact weight values ​​in combination with the multidimensional settlement data set to construct a weighted matrix; the geographic coordinate data is traversed, and regression calculation is performed according to the weighted matrix in combination with the settlement prediction value of the first time node to generate a first regression coefficient; the geographic coordinate data is traversed, and regression calculation is performed according to the weighted matrix in combination with the settlement prediction value of the first time segment to generate a second regression coefficient; the first regression coefficient and the second regression coefficient are added to the multiple regression coefficients.

[0071] Furthermore, the abnormal analysis module 16 in the weighbridge settlement accumulation prediction system is also used for: According to the geographic coordinate data, the target weighing scale area is traversed in combination with the regional settlement difference coefficient to generate a settlement space data set; according to the settlement time sequence, the target weighing scale area is traversed in combination with the regional settlement difference coefficient to generate a settlement time data set; based on the settlement space data set and the settlement time data set, a spatiotemporal dynamic analysis of the weighing scale settlement is performed to determine the cumulative settlement values ​​of multiple geographical locations; the horizontal axis is set according to the settlement time sequence, and the vertical axis is set according to the cumulative settlement values ​​of the multiple geographical locations, and a plane coordinate system is constructed according to the horizontal axis and the vertical axis for visualization processing to construct the regional settlement cumulative prediction trend chart.

[0072] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1The method and specific examples of the cumulative prediction of the settlement of a weighbridge in the first embodiment are also applicable to the cumulative prediction system of the settlement of a weighbridge in the present embodiment. Through the above detailed description of the cumulative prediction method of the settlement of a weighbridge, those skilled in the art can clearly know the cumulative prediction system of the settlement of a weighbridge in the present embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0073] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0074] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A method for predicting the cumulative settlement of a weighbridge, characterized in that: include: Collect real-time data from multiple data sources to the target weighbridge area and generate multi-source real-time data sets; Performing feature analysis based on the multi-source real-time data set, extracting multiple weighbridge features, and constructing a settlement accumulation prediction channel based on the multiple weighbridge features, wherein the settlement accumulation prediction channel includes a short-term settlement prediction branch and a long-term settlement prediction branch; The short-term settlement prediction branch is trained in combination with the multi-source real-time data set to obtain a settlement prediction value of a first time node, where the first time node is a next adjacent time node of the real-time time point; By combining the long-term settlement prediction branch with the settlement prediction value of the first time node, a stabilization process is performed according to the settlement time sequence to obtain a settlement prediction value of a first time segment, where the first time segment includes multiple time nodes; Performing a geographically weighted regression analysis on the predicted settlement value at the first time node and the predicted settlement value at the first time segment to generate a regional settlement difference coefficient; Traverse the target weighbridge area and conduct spatiotemporal dynamic analysis on the weighbridge settlement in combination with the regional settlement difference coefficient, draw a regional settlement cumulative prediction trend chart, conduct settlement distribution anomaly analysis on the target weighbridge area based on the regional settlement cumulative prediction trend chart, and generate settlement cumulative distribution anomaly data to trigger a settlement warning.

2. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 1, characterized in that: Based on the multi-source real-time data set, feature analysis is performed to extract multiple weighbridge features, including: Performing a settlement impact analysis based on the multi-source real-time data set to determine a plurality of impact factors, each of the plurality of impact factors having an importance score; performing sedimentation accumulation analysis on the multi-source real-time data set based on a plurality of importance scores to determine sedimentation rate characteristics; Performing an environmental change analysis on the multi-source real-time data set based on the multiple importance scores to determine environmental impact characteristics; Performing geological change analysis on the multi-source real-time data set based on the multiple importance scores to determine geological condition characteristics; Performing load variation analysis on the multi-source real-time data sets based on the multiple importance scores to determine load balancing characteristics; The settlement rate characteristics, the environmental impact characteristics, the geological condition characteristics, and the load balancing characteristics are correlated and integrated to determine the multiple weighbridge characteristics.

3. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 1, characterized in that: The short-term settlement prediction branch is combined with the multi-source real-time data set for training to obtain a settlement prediction value at a first time node, including: Aligning the multi-source real-time data sets according to data acquisition timestamps to obtain a multi-source aligned data set; Perform serialization processing based on the data acquisition timestamp and the multi-source aligned data set to construct a time series data set; Performing an adjacent search on the time series data set according to the real-time time node to determine a first time node, where the first time node is a next adjacent time node of the real-time time point; The short-term settlement prediction branch uses a long short-term memory network to capture data according to the time series data set combined with the multiple weighbridge features to generate short-term dynamic change data; Introducing a loss function, performing error analysis based on the short-term dynamic change data combined with the loss function, and drawing a short-term settlement loss change curve; Using the first time node as an index to traverse the short-term settlement loss change curve, and determine the settlement loss value at the first time node; The settlement loss value at the first time node is used as supervision data, and supervised prediction training is performed on the short-term dynamic change data to generate an initial training result. When it is verified that the initial training result tends to converge, the settlement prediction value at the first time node is generated.

4. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 3, characterized in that: The long-term settlement prediction branch is combined with the settlement prediction value of the first time node to perform stabilization processing according to the settlement time sequence to obtain the settlement prediction value of the first time segment, including: Based on the predicted settlement value of the first time node, extracting the settlement values ​​of the N time nodes before the real-time time node, where N is an integer greater than 1; Analyze the time series data set in combination with the multiple weighbridge features to determine the settlement time sequence, perform differential processing on the settlement values ​​of the first N time nodes according to the settlement time sequence, and generate multiple differential processing results; Performing a stationarity test on the plurality of difference processing results to generate a plurality of data stationarity coefficients; Determine whether the multiple data stability coefficients meet the expected stability threshold. If the multiple data stability coefficients do not meet the expected stability threshold, extract the settlement value of the i-th time node according to the data stability coefficient that does not meet the expected stability threshold and perform secondary difference until the multiple data stability coefficients meet the expected stability threshold, where i is an integer greater than 0; If the multiple data stability coefficients meet the expected stability threshold, the settlement values ​​of the first N time nodes are combined with the settlement prediction value of the first time node to perform autoregressive fitting calculation to determine the settlement prediction values ​​of multiple time nodes; The settlement prediction values ​​of the multiple time nodes are associated and integrated according to the settlement time sequence to obtain the settlement prediction value of the first time segment.

5. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 1, characterized in that: Performing a geographically weighted regression analysis on the predicted settlement value of the first time node and the predicted settlement value of the first time segment to generate a regional settlement difference coefficient includes: Traverse the target weighbridge area to perform geographic information sensing and obtain the geographic coordinate data of the target weighbridge area; Integrate and identify the settlement prediction value of the first time node and the settlement prediction value of the first time segment according to the geographic coordinate data to generate a multidimensional settlement data set; Traversing the multidimensional settlement data set to perform weighting, constructing a weighting matrix, performing regression analysis based on the weighting matrix combined with the geographic coordinate data, and generating multiple regression coefficients for multiple geographic location points; The settlement identification is performed according to the multiple regression coefficients of the multiple geographical locations to determine the regional settlement difference coefficient.

6. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 5, characterized in that: Traversing the multidimensional settlement data set for weighting, constructing a weighting matrix, performing regression analysis based on the weighting matrix combined with the geographic coordinate data, and generating multiple regression coefficients for multiple geographic location points, including: Traversing the multidimensional settlement data set, performing weight calculation in combination with a Gaussian kernel function, and determining a plurality of impact weight values; Mapping and arranging the multiple impact weight values ​​in combination with the multidimensional settlement data set to construct a weighted matrix; Traversing the geographic coordinate data, performing regression calculation according to the weighted matrix combined with the settlement prediction value of the first time node, and generating a first regression coefficient; Traversing the geographic coordinate data, performing regression calculation according to the weighted matrix combined with the settlement prediction value of the first time segment, and generating a second regression coefficient; The first regression coefficient and the second regression coefficient are added to the plurality of regression coefficients.

7. A method for predicting the cumulative settlement of a weighbridge as claimed in claim 5, characterized in that: Traverse the target weighbridge area and conduct a spatiotemporal dynamic analysis of the weighbridge settlement in combination with the regional settlement difference coefficient, and draw a regional settlement cumulative prediction trend chart, including: Traversing the target weighbridge area in combination with the regional settlement difference coefficient according to the geographic coordinate data to generate a settlement spatial data set; Traversing the target weighbridge area in accordance with the settlement time sequence combined with the settlement difference coefficient of the area to generate a settlement time data set; Based on the settlement space data set and the settlement time data set, a spatiotemporal dynamic analysis is performed on the settlement of the scale to determine the accumulated settlement values ​​of multiple geographical locations; The horizontal axis is set according to the settlement time sequence, the vertical axis is set according to the cumulative settlement values ​​of the multiple geographical locations, a plane coordinate system is constructed based on the horizontal axis and the vertical axis for visualization, and the regional settlement cumulative prediction trend chart is constructed.

8. A weighbridge settlement accumulation prediction system, characterized in that: The steps for implementing the method for predicting the cumulative settlement of a weighbridge as described in any one of claims 1 to 7, the system for predicting the cumulative settlement of a weighbridge comprising: A data acquisition module, the data acquisition module is used to collect real-time data from a target weighbridge area according to multiple data sources to generate a multi-source real-time data set; A prediction channel construction module, the prediction channel construction module is used to perform feature analysis based on the multi-source real-time data set, extract multiple weighbridge features, and construct a settlement accumulation prediction channel based on the multiple weighbridge features, the settlement accumulation prediction channel including a short-term settlement prediction branch and a long-term settlement prediction branch; A settlement prediction module, the settlement prediction module is used to obtain a settlement prediction value of a first time node by training the short-term settlement prediction branch in combination with the multi-source real-time data set, where the first time node is a next adjacent time node of the real-time time point; A stabilization processing module, the stabilization processing module is used to perform stabilization processing according to the settlement time sequence by combining the long-term settlement prediction branch with the settlement prediction value of the first time node to obtain the settlement prediction value of the first time segment, wherein the first time segment includes multiple time nodes; A weighted regression analysis module, the weighted regression analysis module is used to perform a geographical weighted regression analysis on the settlement prediction value of the first time node and the settlement prediction value of the first time segment to generate a regional settlement difference coefficient; An abnormality analysis module is used to traverse the target weighing scale area and conduct spatiotemporal dynamic analysis on the weighing scale settlement in combination with the regional settlement difference coefficient, draw a regional settlement cumulative prediction trend chart, conduct settlement distribution abnormality analysis on the target weighing scale area according to the regional settlement cumulative prediction trend chart, and generate settlement cumulative distribution abnormality data to trigger a settlement warning.

Images