Efficient transmission method of engineering test detection data based on 5G technology

Abstract

The present application relates to the technical field of digital information transmission, and in particular to an efficient transmission method for engineering test detection data based on 5G technology, which comprises the following steps: obtaining engineering test detection data of a target road section of a target road; performing noise analysis on the actual road surface flatness of the target road section to determine the overall noise of the actual road surface flatness of the target road section; adjusting the initial learning rate of the optimizer of the long short-term memory network model according to the overall noise to obtain a target long short-term memory network model; predicting the predicted road surface flatness of the target road section by using the target long short-term memory network model; determining a Huffman tree based on the actual road surface flatness and the predicted road surface flatness of the target road section; compressing the actual road surface flatness of the target road section by using the Huffman tree to obtain compressed data, and transmitting the compressed data based on 5G technology. The present application optimizes the compression effect of data compression and improves the data transmission efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital information transmission, and in particular to an efficient transmission method for engineering test detection data based on 5G technology. BACKGROUND

[0002] In traffic engineering, the engineering test detection data of the road is crucial to the safety, comfort and durability of the road, and the engineering test detection data of the road can include the road roughness. In order to ensure that the monitoring system can integrate and analyze a large amount of road roughness data in real time, optimize resource allocation, and improve the accuracy and efficiency of road infrastructure management, it is necessary to compress the detected road roughness in traffic engineering and then transmit it to the database.

[0003] In some scenarios, Huffman coding algorithm is often used to compress the road roughness, and then 5th-Generation Mobile Communication Technology (5G) is used for transmission. Due to factors such as vehicle bumping caused by tire wear and accumulation of mud layer, the detected road roughness may have too much noise. When Huffman coding algorithm is used to compress the road roughness with distorted data, the coding length of the compressed coding data is too long, and the compression effect is poor. Therefore, in order to improve the compression effect of Huffman coding algorithm for data compression, a Long Short-Term Memory (LSTM) model is used to predict the road roughness using historical road roughness sequence, and then the difference between the road roughness predicted by the LSTM model and the actual road roughness is used to optimize the Huffman coding algorithm, thereby improving the compression effect. However, when the LSTM model is used to predict the road roughness, not all historical road roughness at the time step is real road roughness, but it may also be noise data. If there is too much noise data at the time step, the predicted value predicted by the LSTM model will be inaccurate, the optimization effect of the Huffman coding algorithm will be poor, and the compression effect of the Huffman coding algorithm for data compression will be poor. The coding length of the compressed coding data is too long, the bandwidth requirement and data transmission time are high during 5G transmission, and the data transmission efficiency is low. SUMMARY

[0004] In order to solve the technical problems of poor compression effect of Huffman coding algorithm for data compression and low data transmission efficiency, the purpose of the present application is to provide an efficient transmission method for engineering test detection data based on 5G technology, and the technical solution adopted is as follows:

[0005] The embodiment of the present application provides a kind of based on 5G technology's engineering test detection data efficient transmission method, comprising: obtaining the engineering test detection data of the target road section of target road, engineering test detection data includes the actual road flatness of target road section, actual road flatness is determined by the height value between the multiple measuring points of target road section and measuring equipment;The actual road flatness of target road section is analyzed for noise, determine the overall noise of the actual road flatness of target road section, overall noise indicates the possibility that the actual road flatness of target road section belongs to noise;According to the overall noise of the actual road flatness of target road section, the initial learning rate of optimizer of long short-term memory network model is adjusted, to obtain target long short-term memory network model;The predicted road flatness of target road section is predicted using target long short-term memory network model;Based on the actual road flatness and predicted road flatness of target road section, determine the Huffman tree;The actual road flatness of target road section is compressed using Huffman tree, to obtain compressed data, and transmit compressed data based on 5G technology.

[0006] Optionally, the actual road flatness of target road section is analyzed for noise, to determine the overall noise of the actual road flatness of target road section, comprising: obtaining the height value between the multiple measuring points of target road section and measuring equipment;According to the difference between each adjacent height value, determine the availability of the actual road flatness of target road section, adjacent height value is the height value of adjacent acquisition time in time sequence, availability indicates the continuity of the actual road flatness of target road section;The first actual road flatness of first preset days and the first availability of first actual road flatness are obtained, and the second actual road flatness of second preset days and the second availability of second actual road flatness are obtained, first preset days are at least one day before the actual road flatness of target road section corresponding time, second preset days are at least one day after the actual road flatness of target road section corresponding time;According to the availability of the actual road flatness of target road section, first actual road flatness, first availability, second actual road flatness and second availability, determine the probability that the actual road flatness of target road section belongs to mutation point;According to the regression model of the actual road flatness of target road section, determine the single-dimensional noise of the actual road flatness of target road section, regression model is constructed based on the probability that the actual road flatness of target road section belongs to mutation point, change rate parameter and flatness upper limit value;According to the regression model of the actual road flatness of target road section and the regression model of adjacent road section on both sides of target road section, determine the overall trend similarity between the actual road flatness of target road section and the actual road flatness of adjacent road section;According to the single-dimensional noise of adjacent road section on both sides of target road section, the single-dimensional noise of target road section and overall trend similarity, determine the overall noise of the actual road flatness of target road section.

[0007] Optionally, before determining the single-dimensional noise of the actual road surface flatness of the target road section according to the regression model of the actual road surface flatness of the target road section, the method further comprises: obtaining historical actual road surface flatness of a plurality of road sections in a third preset number of days; determining a probability that each historical actual road surface flatness belongs to a mutation point and a desirability of each historical actual road surface flatness; determining the regression model according to the last day in the third preset number of days, a change rate parameter, an upper limit value of flatness, the probability that each historical actual road surface flatness belongs to a mutation point, and the day in the third preset number of days in which each historical actual road surface flatness is located; determining a correction error function according to the desirability of each historical actual road surface flatness, the regression model, and each historical actual road surface flatness; and solving the change rate parameter and the upper limit value of flatness of the regression model by a gradient descent method with the goal of minimizing the correction error function.

[0008] Optionally, determining the overall trend similarity between the actual road surface flatness of the target road section and the actual road surface flatness of the adjacent road sections according to the regression model of the actual road surface flatness of the target road section and the regression model of the adjacent road sections located on both sides of the target road section comprises: obtaining a first probability that each actual road surface flatness of the target road section belongs to a mutation point and a second probability that the actual road surface flatness of the adjacent road sections located on both sides of the target road section belongs to a mutation point in a fourth preset number of days; determining a first regression model of each actual road surface flatness of the target road section according to each first probability and a second regression model of each actual road surface flatness of the adjacent road sections according to each second probability; and determining the overall trend similarity by using the first regression model and the second regression model.

[0009] Optionally, determining the desirability of the actual road surface flatness of the target road section according to the difference between each adjacent height value comprises: determining an absolute value of the difference between each adjacent height value and an average value of each difference; calculating a square of the difference between the absolute value of each difference and the average value to obtain a square value, and superimposing each square value to obtain a first superimposed value; and performing negative correlation processing on the first superimposed value by using an exponential function to obtain the desirability of the actual road surface flatness of the target road section.

[0010] Optionally, determining the probability that the actual road surface flatness of the target road section belongs to a mutation point according to the desirability of the actual road surface flatness of the target road section, the first actual road surface flatness, the first desirability, the second actual road surface flatness, and the second desirability comprises: calculating a second superimposed value of a first product of each first actual road surface flatness and the first desirability in a first preset number of days and a third superimposed value of a second product of each second actual road surface flatness and the second desirability in a second preset number of days; calculating a third product of an absolute value of the difference between the second superimposed value and the third superimposed value and an inverse of the desirability of the actual road surface flatness; and performing normalization on the third product by using a maximum-minimum value normalization function to obtain the probability that the actual road surface flatness of the target road section belongs to a mutation point.

[0011] Optionally, determining the single-dimensional noise of the actual road surface flatness of the target road section according to the regression model of the actual road surface flatness of the target road section comprises: determining a target calculation formula in the regression model according to the probability that the actual road surface flatness of the target road section belongs to a mutation point; and determining the predicted road surface flatness of the target road section by using the target calculation formula.

[0012] Optionally, determining the single-dimensional noise of the actual road surface flatness of the target road section according to the regression model of the actual road surface flatness of the target road section comprises: determining a target calculation formula in the regression model according to the probability that the actual road surface flatness of the target road section belongs to a mutation point; and determining the predicted road surface flatness of the target road section by using the target calculation formula.

[0013] Optionally, determining the overall noise of the actual road surface flatness of the target road section according to the single-dimensional noise of the adjacent road sections on both sides of the target road section, the single-dimensional noise of the target road section, and the overall trend similarity comprises: calculating an average single-dimensional noise of the single-dimensional noise of the adjacent road sections on both sides of the target road section; multiplying the average single-dimensional noise after negative correlation processing by an exponential function and the single-dimensional noise of the target road section to obtain a fourth product; and determining the product of the fourth product and the overall trend similarity as the overall noise of the actual road surface flatness of the target road section.

[0014] Optionally, adjusting the initial learning rate of the optimizer of the long short-term memory network model according to the overall noise of the actual road surface flatness of the target road section to obtain a target long short-term memory network model comprises: obtaining the initial learning rate of the optimizer of the long short-term memory network model and the overall noise of the actual road surface flatness of the target road section at a time step; calculating the difference between a preset value and each actual road surface flatness of the target road section, and calculating a fourth superimposed value of the difference between each actual road surface flatness of the target road section and the ratio of the number of actual road surface flatness of the target road section at the time step; taking the absolute value of the fourth superimposed value; determining the product of the absolute value and the initial learning rate as the optimized learning rate; and determining the target long short-term memory network model according to the optimized learning rate.

[0015] Optionally, determining the Huffman tree based on the actual road surface flatness and the predicted road surface flatness of the target road section comprises: determining a normalized difference of the actual road surface flatness of the target road section according to the predicted road surface flatness and the actual road surface flatness of the target road section; determining the forward nature of the node formed by the actual road surface flatness of the target road section based on the normalized difference and the number of occurrences of the actual road surface flatness of the target road section; and determining the Huffman tree by using the forward nature of the node formed by the actual road surface flatness of the target road section.

[0016] The application has the following beneficial effects: first, the engineering test detection data of the target section of the target road is obtained, the engineering test detection data including the actual pavement flatness of the target section, the actual pavement flatness being determined by the height value between the plurality of measurement points of the target section and the measuring device; then, the actual pavement flatness of the target section is analyzed for noise to determine the overall noise of the actual pavement flatness of the target section, the overall noise representing the possibility that the actual pavement flatness of the target section belongs to noise; second, the initial learning rate of the optimizer of the long short-term memory network model is adjusted according to the overall noise of the actual pavement flatness of the target section, to obtain a target long short-term memory network model; the target long short-term memory network model is used to predict the predicted pavement flatness of the target section; and the Huffman tree is determined based on the actual pavement flatness and the predicted pavement flatness of the target section; finally, the actual pavement flatness is compressed by using the Huffman tree to obtain compressed data, and the compressed data is transmitted based on the 5G technology.

[0017] In this way, the initial learning rate of the optimizer of the long short-term memory network model is adjusted according to the overall noise of the actual pavement flatness of the target road obtained by measurement, the accuracy of the predicted value of the LSTM model is improved, the deviation between the predicted value and the actual pavement flatness is reduced, the more accurate predicted pavement flatness predicted based on the LSTM model is used to improve the compression effect of the Huffman coding algorithm, and the encoded data obtained by compression can be transmitted by 5G with a shorter encoding length, thereby reducing the bandwidth requirement and data transmission time in the 5G transmission process and improving the efficiency of data transmission. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art and the advantages thereof, a brief introduction will be given below to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without any creative effort.

[0019] Figure 1 A flowchart of the engineering test detection data efficient transmission method based on the 5G technology provided by the embodiments of the present application;

[0020] Figure 2 A module composition schematic diagram of the engineering test detection data efficient transmission system based on the 5G technology provided by the embodiments of the present application;

[0021] Figure 3 A structure schematic diagram of the engineering test detection data efficient transmission system based on the 5G technology for realizing the embodiments of the present application. DETAILED DESCRIPTION

[0022] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the following describes in detail the specific implementation, structure, features and effects of a 5G technology-based engineering test detection data efficient transmission method according to the present application, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0024] The specific scheme of the 5G technology-based engineering test detection data efficient transmission method provided by the present application is described in detail below in combination with the accompanying drawings.

[0025] Embodiment one:

[0026] Please refer to Figure 1 which shows the flowchart of the 5G technology-based engineering test detection data efficient transmission method provided by one embodiment of the present application, including:

[0027] S101, obtaining engineering test detection data of a target road section of a target road.

[0028] The engineering test detection data includes the actual road surface flatness of the target road section.

[0029] Specifically, the target road refers to a road that needs to be detected for flatness. The target road section refers to a section of the target road. The actual road surface flatness of the target road section can be measured by a measuring device, which includes a laser range finder and a road flatness detection vehicle. The laser range finder can be installed in the center of the vehicle body of the road flatness detection vehicle, and is used to measure the distance between the center of the vehicle body and the ground surface of the target road section. There are multiple measurement points in the target road section, and the actual road surface flatness of the target road section is represented by the variance of the height values between these measurement points and the center of the vehicle body. For example, when the front of the road flatness detection vehicle reaches the boundary of the target road section, the laser range finder starts to collect the distance between the center of the vehicle body and the ground surface of the measurement point, which is the height value between the measurement point and the measuring device. At a frequency of x frames per second, the measurement is ended when the tail of the road flatness detection vehicle drives off the target road section, and a total of n height values of the center of the vehicle body to the ground are obtained. The actual road surface flatness of the target road section is represented by the variance of the n height values.

[0030] Further, for the target road section, two road sections are further set as adjacent road sections at a preset distance on both sides of the target road section, and actual road flatness on the two adjacent road sections is obtained by using the measuring device, so as to facilitate subsequent embodiment analysis of the trend similarity of road flatness between road sections. The preset distance can be determined according to actual conditions, and in the embodiment of the present application, the value is 1 km, and the two sides of the target road section refer to two road sections adjacent to the left side and the right side of the target road section in the direction in which the measuring device advances. Of course, according to actual conditions, the two sides of the target road section can also be other conditions, which are not limited in the embodiment of the present application.

[0031] S102, the actual road flatness of the target road section is analyzed for noise, and the overall noise of the actual road flatness of the target road section is determined.

[0032] The overall noise represents the possibility that the actual road flatness of the target road section belongs to noise.

[0033] Specifically, the overall noise represents the possibility that the actual road flatness belongs to noise, that is, the authenticity of the actual road flatness, the higher the overall noise, the greater the possibility that the actual road flatness belongs to noise, and the lower the authenticity of the actual road flatness. In determining the overall noise of the actual road flatness, first, the noise of the actual road flatness of each target road section on each day is determined, and the single data point is analyzed first. For each target road section, since each target road section changes with time and presents a Logistic function model (regression model), it is necessary to minimize the error sum of squares between the predicted value and the actual value for function fitting. However, if the actual road flatness is not acceptable data and has a large contribution rate to the minimized error sum of squares function, the fitting result becomes extremely inaccurate, so the acceptability of each actual road flatness is calculated first by the characteristics of the continuous change of the target road section. Secondly, because the Logistic function model has unknown mutation points in the initial parameters, the mutation points are determined by the left and right fluctuations of the actual road flatness under the acceptability, the optimal function is determined by the mutation points, the error and the function, and the single-dimensional noise of the actual road flatness is determined by the difference between the actual road flatness and the function. Then, the adjacent road sections on both sides of the current target road section are analyzed, and the overall noise of the actual road flatness of the target road section is determined by the overall trend similarity between the actual road flatness of the target road section and the actual road flatness of the adjacent road sections.

[0034] Further, in determining the overall noise of the actual road surface flatness, as an optional embodiment of the present application, first, the height values between the plurality of measuring points of the target road section and the measuring device are obtained. Then, the acceptability of the actual road surface flatness of the target road section is determined according to the difference between each adjacent height value, the adjacent height values being the height values of the adjacent collection times in time sequence, and the acceptability representing the continuity of the actual road surface flatness of the target road section. Then, the first actual road surface flatness of the first preset number of days and the first acceptability of the first actual road surface flatness, and the second actual road surface flatness of the second preset number of days and the second acceptability of the second actual road surface flatness are obtained, the first preset number of days being at least one day before the corresponding time of the actual road surface flatness of the target road section, and the second preset number of days being at least one day after the corresponding time of the actual road surface flatness of the target road section. Next, the probability that the actual road surface flatness of the target road section belongs to the mutation point is determined according to the acceptability of the actual road surface flatness of the target road section, the first actual road surface flatness, the first acceptability, the second actual road surface flatness and the second acceptability. Then, the single-dimensional noise of the actual road surface flatness of the target road section is determined according to the regression model of the actual road surface flatness of the target road section, the regression model being constructed based on the probability that the actual road surface flatness of the target road section belongs to the mutation point, the change rate parameter and the flatness upper limit value. Then, the overall trend similarity between the actual road surface flatness of the target road section and the actual road surface flatness of the adjacent road sections on both sides of the target road section is determined according to the regression model of the actual road surface flatness of the target road section and the regression models of the adjacent road sections on both sides of the target road section. Finally, the overall noise of the actual road surface flatness of the target road section is determined according to the single-dimensional noise of the adjacent road sections on both sides of the target road section, the single-dimensional noise of the target road section and the overall trend similarity.

[0035] Specifically, the acceptability represents the continuity of the actual road surface flatness of the target road section. Since the road surface is usually composed of continuous materials (such as soil, concrete, paving materials, etc.), which are continuous in physical and mechanical properties, the change of the actual road surface flatness of the road should also be continuous, and the higher the continuity of the actual road surface flatness of the target road section, the higher the acceptability. In determining the acceptability of the actual road surface flatness of the target road section, first, the difference between each adjacent height value is determined, and the average value of each difference is determined. Then, the square of the difference between each difference and the average value is calculated to obtain a square value, and each square value is superimposed to obtain a first superimposed value. Finally, the first superimposed value is negatively correlated by using an exponential function to obtain the acceptability of the actual road surface flatness of the target road section.

[0036] In the above formula, P

[0037]

[0038] In the above formula, P QThis indicates the feasibility of the actual road surface smoothness Q of the target road section. exp represents an exponential function with base e, used to evaluate the smoothness of the road surface. Negative correlation processing is performed. n represents the height values ​​between the center of the vehicle and the ground at the measurement points of the target road segment when obtaining the actual road surface smoothness Q of the target road segment using the measuring equipment. This represents the height value between the center of the vehicle and the ground where the measurement point of the target road section is located, collected by the measuring equipment when obtaining the actual road surface smoothness Q of the target road section. This represents the height value between the center of the vehicle body and the ground where the measurement point of the target road section is located, collected by the measuring equipment when obtaining the actual road surface smoothness Q of the target road section. This represents the absolute value of the height difference between the (i+1)th and ithth vehicle center points and the ground level of the measurement point on the target road segment. u represents the average of the differences between two adjacent height values ​​among the n height values ​​collected using the measuring equipment when obtaining the actual road surface smoothness Q of the target road segment.

[0039] Furthermore, since road surfaces are typically composed of continuous materials (such as soil, concrete, paving materials, etc.), and these materials are continuous in physical and mechanical properties, the variation in road surface smoothness should also be continuous. Analysis of the data acquisition process for obtaining the actual road surface smoothness Q of the target road section reveals that, because it involves continuously collecting the height values ​​of each measurement point, the differences in adjacent height values ​​between the center of the vehicle and the ground where the measurement points of the target road section are located are similar. Therefore, the variation between adjacent height values... variance To determine the similarity of the differences between adjacent height values, the larger the variance, the smaller the similarity. Q The smaller.

[0040] Furthermore, after determining the feasibility of the actual road surface smoothness of the target road segment, the probability that the actual road surface smoothness of the target road segment belongs to a sudden change point is then determined. To determine this probability, the following steps are taken: first, the first actual road surface smoothness and the first feasibility of the first actual road surface smoothness are obtained for the first preset number of days; second, the second actual road surface smoothness and the second feasibility of the second actual road surface smoothness are obtained for the second preset number of days. Then, the second superposition value of the first product of each first actual road surface smoothness and the first feasibility in the first preset number of days, and the third superposition value of the second product of each second actual road surface smoothness and the second feasibility in the second preset number of days are calculated. Next, the absolute value of the difference between the second superposition value and the third superposition value is calculated as the third product of the reciprocal of the feasibility of the actual road surface smoothness. Finally, the third product is normalized using a maximum-minimum value normalization function to obtain the probability that the actual road surface smoothness of the target road segment belongs to a sudden change point.

[0041] Specifically, the first actual road flatness and the second actual road flatness can be obtained according to the manner described in step S101 of the above embodiment, the first acceptability and the second acceptability can be calculated according to the calculation formula of the acceptability described in the above embodiment of the present application, and the values of the first preset number of days and the second preset number of days can be determined according to actual conditions, which can be the same or different. In the embodiment of the present application, the values of the first preset number of days and the second preset number of days are both 7 days.

[0042] Further, the probability that the actual road flatness of the target road section belongs to a mutation point can be calculated using the following formula:

[0043]

[0044] In the above formula, T Q represents the probability that the actual road flatness Q of the target road section belongs to a mutation point. F represents a maximum-minimum value normalization function, which is used to normalize . v represents that a total of v days are selected before and after the time corresponding to the actual road flatness Q of the target road section. P Q ′ i represents the first acceptability of the first actual road flatness of the i-th day selected before the time corresponding to the actual road flatness Q of the target road section. Z ′ Qi represents the first actual road flatness of the i-th day selected before the time corresponding to the actual road flatness Q of the target road section. P Q i represents the second acceptability of the second actual road flatness of the i-th day selected after the time corresponding to the actual road flatness Q of the target road section. Z ′ Q ′ i represents the second actual road flatness of the i-th day selected after the time corresponding to the actual road flatness Q of the target road section, P Q represents the acceptability of the actual road flatness Q of the target road section.

[0045] Further, for the mutation point of the regression model (such as the Logistic function model) to be constructed, the acceptability of the actual road flatness Q of the target road section is a prerequisite. If the road flatness within the preset number of days before the time corresponding to the actual road flatness Q of the target road section is significantly different from the road flatness within the preset number of days after the time corresponding to the actual road flatness Q of the target road section , the smaller the actual road flatness Q of the target road section, i.e., the larger the reciprocal of the actual road flatness Q of the target road section, the greater the probability that the actual road flatness Q of the target road section belongs to a mutation point.​

[0046] Further, the regression model can be a Logistic function model used to represent the change of the road flatness over time. Before the regression model is applied to determine the actual road flatness of the single dimension noise, the regression model needs to be constructed. When constructing the regression model, first, the historical actual road flatness of a plurality of road sections in a third preset number of days is obtained; then, the probability that each historical actual road flatness belongs to a mutation point and the acceptability of each historical actual road flatness are determined; and then, the regression model is determined according to the last day in the third preset number of days, the change rate parameter, the flatness upper limit value, the probability that each historical actual road flatness belongs to a mutation point, and the day in the third preset number of days in which each historical actual road flatness is located. Next, the correction error function is determined according to the acceptability of each historical actual road flatness, the regression model, and each historical actual road flatness. Finally, the change rate parameter and the flatness upper limit value of the regression model are solved by the gradient descent method with the minimization of the correction error function as the goal.

[0047] Specifically, the third preset number of days can be determined according to actual conditions, which is not limited in the embodiments of the present application. The probability that the historical actual road flatness belongs to a mutation point and the acceptability of each historical actual road flatness can be calculated by the calculation formula described in the embodiments of the present application. In the embodiments of the present application, the regression model is taken as an example of the Logistic function model, which is expressed by the following formula:

[0048]

[0049] In the above formula, Z(t) represents the Logistic function model, L represents the flatness upper limit value, and k represents the change rate parameter, which is used to control the steepness of the sharp decline of the road flatness. A t represents the day in the third preset number of days in which the historical actual road flatness A is located. t represents the third preset number of days. t1 represents the last day in the third preset number of days in which the historical actual road flatness is obtained. A e represents the natural constant.

[0050] Further, the correction error function can be expressed by the following formula:

[0051]

[0052] In the above formula, E represents the correction error function. m represents that m days of historical actual road flatness are obtained. P i represents the acceptability of the historical actual road flatness on the i-th day in m. Z(i) represents the predicted road flatness obtained by substituting the probability that the historical actual road flatness on the i-th day belongs to a mutation point into the above Logistic function model.i represents the historical actual road roughness on the i-th day.

[0053] Further, because the minimum value of the error of the predicted road roughness and the historical actual road roughness is used to determine the parameter L and the parameter k of the Logistic function model, the embodiment of the present application makes the contribution of the historical actual road roughness with lower desirability to the error smaller, and the contribution of the historical actual road roughness with higher desirability to the error higher, so that the parameter L and the parameter k of the Logistic function model are obtained most accurately. The parameter L and the parameter k of the Logistic function model can be obtained by the gradient descent method, and thus the regression model is constructed.

[0054] Further, after the regression model is constructed, the one-dimensional noise of the actual road roughness of the target road section can be determined by using the regression model. In determining the one-dimensional noise of the actual road roughness of the target road section, as an optional embodiment of the present application, the target calculation formula in the regression model is determined according to the probability that the actual road roughness of the target road section belongs to the abrupt point; the predicted road roughness of the target road section is determined by using the target calculation formula; and the absolute value of the difference between the predicted road roughness and the actual road roughness of the target road section is determined as the one-dimensional noise of the actual road roughness of the target road section.

[0055] Specifically, as described in the above embodiment of the present application, the regression model includes two parts. The first part is that when the probability of the abrupt point is greater than 0.7, the predicted road roughness of the target road section is calculated by using the formula The second part is that when the probability of the abrupt point is not greater than 0.7, the predicted road roughness of the target road section is calculated by using the formula Therefore, the probability that the actual road roughness of the target road section belongs to the abrupt point is compared with 0.7 to determine the target calculation formula. After the target calculation formula is determined, the predicted road roughness of the target road section is calculated according to the formula.

[0056] Further, the one-dimensional noise of the actual road roughness of the target road section can be expressed by the following formula:

[0057] W Q = |Z(Q) - Z Q |;

[0058] In the above formula, W Q represents the one-dimensional noise of the actual road roughness Q of the target road section. Z(Q) represents the predicted road roughness of the target road section. Z Q represents the actual road roughness Q of the target road section.

[0059] Further, because the actual road evenness of different road sections of the target road presents the law shown by the regression model in time sequence, the greater the |Z(Q)-Z Q | is, the greater the single-dimensional noise of the actual road evenness of the target road section is.

[0060] Further, because the adjacent road sections adjacent to the target road section use similar materials and construction methods in the construction process, and the traffic flow and vehicle types borne by the adjacent road sections are similar, the wear and deformation trends of the adjacent road sections are similar, so the overall trends should be similar. The overall trend similarity can be determined by the integral error of the regression model. Therefore, as an optional embodiment of the present application, when determining the overall trend similarity, first, the first probability that each actual road evenness of the target road section in the fourth preset number of days belongs to a mutation point, and the second probability that the actual road evenness of the adjacent road section on both sides of the target road section belongs to a mutation point are obtained; then, the first regression model of each actual road evenness of the target road section is determined according to the first probability, and the second regression model of each actual road evenness of the adjacent road section is determined according to the second probability; finally, the overall trend similarity is determined by using the first regression model and the second regression model.

[0061] Specifically, the fourth preset number of days can be a plurality of consecutive days, and the first probability that each actual road evenness of the target road section in the fourth preset number of days belongs to a mutation point and the second probability that each actual road evenness of the adjacent road section in the fourth preset number of days belongs to a mutation point can be calculated by the formula recorded in the above embodiments of the present application. After obtaining each first probability and second probability, they are compared with 0.7, so as to obtain the first regression model and the second regression model. The overall trend similarity can be calculated by the following formula:

[0062]

[0063] In the above formula, S represents the overall trend similarity between the actual road evenness of the target road section and the actual road evenness of the adjacent road section. exp represents the exponential function with e as the base, which is used for negative correlation processing of . t0 represents the initial number of days when the fourth preset number of days is observed in total. t1 represents the end number of days when the fourth preset number of days is observed in total. Z ′ (t) represents the Logistic function model (the second regression model) of the actual road evenness of the adjacent road section adjacent to one side of the target road section. Z(t) represents the Logistic function model (the first regression model) of the actual road evenness of the target road section. Z″(t) represents the Logistic function model (the second regression model) of the actual road evenness of the adjacent road section adjacent to the other side of the target road section. represents the negative correlation processing of |Z ′(t) is integrated from t0 to t1. represents the integral of |Z"(t)-Z(t)| from t0 to t1. ′ (t) represents the difference between the second regression model of the actual road roughness of the adjacent road section adjacent to one side of the target road section and the first regression model of the actual road roughness of the target road section, i.e. the error between the predicted road roughness of the adjacent road section adjacent to one side of the target road section and the predicted road roughness of the target road section. Z"(t)-Z(t) represents the difference between the second regression model of the actual road roughness of the adjacent road section adjacent to the other side of the target road section and the first regression model of the actual road roughness of the target road section, i.e. the error between the predicted road roughness of the adjacent road section adjacent to the other side of the target road section and the predicted road roughness of the target road section.

[0064] Further, because the target road section and its two adjacent road sections have their own Logistic function models in terms of the probability of the actual road roughness being a sudden change point, the overall trend similarity can be determined based on the coincidence degree of the Logistic function models of the target road section and its two adjacent road sections, which in the above formula is represented by the smaller the area enclosed between the integral curve of the target road section and the integral curve of the adjacent road section, the higher the S, i.e. the higher the overall trend similarity between the actual road roughness of the target road section and the actual road roughness of the adjacent road section.

[0065] Further, after obtaining the overall trend similarity and the single-dimensional noise of the target road section, the overall noise of the actual road roughness of the target road section can be determined in combination with the single-dimensional noise of the adjacent road section. As an optional embodiment of the present application, when determining the overall noise of the actual road roughness of the target road section, first, the average single-dimensional noise of the single-dimensional noise of the adjacent road sections on both sides of the target road section is calculated; then the average single-dimensional noise is negatively correlated by using an exponential function and multiplied by the single-dimensional noise of the target road section to obtain a fourth product; finally, the product of the fourth product and the overall trend similarity is determined as the overall noise of the actual road roughness of the target road section.

[0066] Specifically, the overall noise of the actual road roughness of the target road section can be represented by the following formula:

[0067]

[0068] In the above formula, M Q represents the overall noise of the actual road roughness Q of the target road section. exp represents the exponential function with e as the base, which is used to negatively correlate W Q represents the single-dimensional noise of the actual road roughness Q of the target road section. W ′Q represents the single-dimensional noise of the actual pavement evenness of the adjacent road section adjacent to one side of the target road section on the same day. W ′ Q ′ represents the single-dimensional noise of the actual pavement evenness of the adjacent road section adjacent to the other side of the target road section on the same day. S represents the overall trend similarity between the actual pavement evenness of the target road section and the actual pavement evenness of the adjacent road section.

[0069] S103, adjusting the initial learning rate of the optimizer of the long short-term memory network model according to the overall noise of the actual pavement evenness of the target road section, to obtain a target long short-term memory network model.

[0070] Specifically, if there is a lot of noise data on the time step of the long short-term memory network model, it will greatly affect the prediction result of the long short-term memory network model, so the learning rate of the optimizer in the long short-term memory network model is adjusted through the overall noise of all the actual pavement evenness on the time step, to weaken the sensitivity of the long short-term memory network model to noise, so that the prediction result of the long short-term memory network model is more accurate. The optimizer of the long short-term memory network model can be an Adam optimizer.

[0071] Further, when adjusting the initial learning rate of the optimizer of the long short-term memory network model, as an optional embodiment of the present application, first, the initial learning rate of the optimizer of the long short-term memory network model and the overall noise of the plurality of actual pavement evenness of the target road section on the time step are obtained; then the difference between the preset value and each actual pavement evenness of the target road section is calculated, and the fourth superimposed value of the ratio of each difference to the number of actual pavement evenness of the target road section on the time step is calculated; the absolute value of the fourth superimposed value is taken; secondly, the product of the absolute value and the initial learning rate is determined as the optimized learning rate; finally, the target long short-term memory network model is determined according to the optimized learning rate.

[0072] Specifically, the plurality of actual pavement evenness of the target road section can be the historical actual pavement evenness of the target road section collected for consecutive h days. The preset value can be valued according to the actual situation, and the value in the embodiment of the present application is 1. The optimizer can be determined according to the actual scene, and the Adam optimizer is used in the embodiment of the present application. Further, the optimized learning rate can be expressed as follows:

[0073]

[0074] In the above formula, represents the optimized learning rate of the Adam optimizer for all the historical actual pavement evenness on the time step. h represents that there are h historical actual pavement evenness of the target road section on the time step. M irepresents the average non-noise rate of all historical actual road roughnesses on the time step. X represents the initial learning rate of the Adam optimizer for all historical actual road roughnesses on the time step. represents the average non-noise rate of all historical actual road roughnesses on the time step. X represents the initial learning rate of the Adam optimizer for all historical actual road roughnesses on the time step.

[0075] Further, in the long short-term memory network model, a large amount of noise in the data will reduce the accuracy of the prediction result. At this time, the average non-noise rate of the historical actual road roughness on the time step is used to optimize the learning rate of the Adam optimizer. Optimizing the learning rate of the Adam optimizer can reduce the fluctuation of the long short-term memory network model in the error caused by noise, can greatly reduce the overreaction made on the noisy data, make each weight update more robust, help the long short-term memory network model to adjust smoothly in the training process, reduce the sensitivity to noise, thereby improving the stability and prediction accuracy of the long short-term memory network model.

[0076] In addition, the historical actual road roughness of other road sections of the target road can also be used to optimize the learning rate of the optimizer of the long short-term memory network model, or the historical actual road roughness of different road sections of other roads can be used to optimize the learning rate of the optimizer of the long short-term memory network model. The embodiments of the present application are not limited here.

[0077] S104, predicting the predicted road roughness of the target section by using the target long short-term memory network model.

[0078] Specifically, the learning rate of the Adam optimizer in the target long short-term memory network model has been optimized, so that the accuracy of predicting the predicted road roughness of the target section by using the target long short-term memory network model is higher. When predicting the predicted road roughness of the target section, at least one day of historical actual road roughness of the target section can be input into the target long short-term memory network model, and the target long short-term memory network model is used to predict the road roughness of the target section to obtain the predicted road roughness.

[0079] S105, determining the Huffman tree based on the actual road roughness and the predicted road roughness of the target section.

[0080] In particular, since the road flatness has time series characteristics, for the road flatness, the optimized target long short-term memory network model is used to predict the road flatness of the target road section, so as to obtain the predicted road flatness corresponding to the actual road flatness of the target road. If the difference between the predicted road flatness of the target long short-term memory network model and the actual road flatness is large, it may be caused by vehicle bumping due to tire wear, deviation caused by occasional soil accumulation, and the like, which is more likely to be noise data and is not the data that needs to be focused on. Therefore, the actual road flatness and the predicted road flatness of the target road section are used to optimize the Huffman tree. The actual road flatness with a large difference from the predicted road flatness is compressed by using the optimized Huffman tree, so that the Huffman coding for this part is longer (low forwardness), and the actual road flatness with a small difference from the predicted road flatness is compressed by using the optimized Huffman tree, so that the Huffman coding is shorter (high forwardness), so that the actual road flatness with a small difference from the predicted road flatness is better compressed and better protected.

[0081] Further, in determining the Huffman tree, as an optional embodiment of the present application, first, the normalized difference of the actual road flatness of the target road section is determined according to the predicted road flatness and the actual road flatness of the target road section; then the forwardness of the node formed by the actual road flatness of the target road section is determined based on the normalized difference and the number of times of occurrence of the actual road flatness of the target road section; and finally, the Huffman tree is determined by using the forwardness of the node formed by the actual road flatness of the target road section.

[0082] In particular, the normalized difference of the actual road flatness of the target road section can be calculated by the following formula:

[0083]

[0084] In the above formula, C represents the normalized difference between the predicted value and the actual value of each actual road flatness. Y represents the predicted value of each actual road flatness, i.e. the predicted road flatness. Z represents the actual value of each actual road flatness. |Y-Z| represents the difference between the predicted value and the actual value. C max represents the maximum value of the difference between all predicted values and actual values of the target road section. C min represents the minimum value of the difference between all predicted values and actual values of the target road section.

[0085] Further, the forwardness of the node formed by the actual road flatness can be represented by the following formula:

[0086]

[0087] In the above formula, G represents the forwardness of each actual road surface roughness constituting node. exp represents the exponential function with e as the base, which is used for negative correlation processing of a represents the number of times of the same value of the actual road surface roughness. C i represents the normalized difference between the predicted value and the actual value when the same value of the actual road surface roughness appears the ith time. The higher the forwardness of the actual road surface roughness constituting node, the higher the importance of the actual road surface roughness, and the shorter the Huffman code when the optimized Huffman tree compresses the actual road surface roughness with high forwardness, so that it is better compressed. The lower the forwardness of the actual road surface roughness constituting node, the lower the importance of the actual road surface roughness, which can be removed as noise data, so as to retain more important data

[0088] The forwardness of the actual road surface roughness constituting node has been determined through the above embodiment, and next, the actual road surface roughness with a forwardness lower than a threshold value can be removed as noise data, and the remaining actual road surface roughness with a forwardness higher than the threshold value can be constructed into a Huffman tree based on the method of constructing a Huffman tree in the prior art, that is, the forwardness corresponding nodes of the two smallest nodes are always merged to form a new node, and then merged with the forwardness corresponding node of the remaining smallest node, to construct a Huffman tree based on each actual road surface roughness.

[0089] In S106, the actual road surface roughness of the target road section is compressed using the Huffman tree to obtain compressed data, and the compressed data is transmitted based on the 5G technology.

[0090] Specifically, the compression of the actual road surface roughness of the target road section using the Huffman tree can be replacing each character in the actual road surface roughness with the corresponding Huffman code to obtain the compressed data. For the principle of compressing data using the Huffman tree, please refer to the prior art, which will not be described here in detail.

[0091] The initial learning rate of the optimizer of the long short-term memory network model is adjusted by the overall noise of the actual road surface roughness of the target road measured, the accuracy of the predicted value of the LSTM model is improved, and the deviation between the predicted value and the actual road surface roughness is reduced. The more accurate predicted road surface roughness predicted based on the LSTM model is used to improve the compression effect of the Huffman coding algorithm, and the encoded data obtained by compression can be transmitted in a shorter encoding length, thereby reducing the bandwidth demand and data transmission time in the 5G transmission process, and improving the efficiency of data transmission. In addition, since the volume of the compressed data is reduced, the probability of encountering errors in the 5G transmission process is also reduced, thereby improving the stability, security and reliability of the data transmission process.

[0092] Embodiment two:

[0093] Corresponding to the efficient transmission method for engineering test and inspection data based on 5G technology provided in the above embodiments, based on the same technical concept, this invention also provides an efficient transmission system for engineering test and inspection data based on 5G technology, such as... Figure 2 As shown, Figure 2 This invention provides a schematic diagram of the module composition of a high-efficiency transmission system for engineering test and inspection data based on 5G technology. The system 200 includes: an acquisition module 201, used to acquire engineering test and inspection data of a target road segment, including the actual road surface smoothness of the target road segment, which is determined by the height values ​​between multiple measurement points on the target road segment and the measuring equipment; and an analysis module 202, used to perform noise analysis on the actual road surface smoothness of the target road segment, determining the overall noise level of the actual road surface smoothness of the target road segment, whereby the overall noise level characterizes the target road surface smoothness. The actual road surface smoothness of the target road segment is likely to be noise; adjustment module 203 is used to adjust the initial learning rate of the optimizer of the Long Short-Term Memory Network model according to the overall noise of the actual road surface smoothness of the target road segment to obtain the target Long Short-Term Memory Network model; prediction module 204 is used to predict the predicted road surface smoothness of the target road segment using the target Long Short-Term Memory Network model; determination module 205 is used to determine the Huffman tree based on the actual road surface smoothness and predicted road surface smoothness of the target road segment; compression module 206 is used to compress the actual road surface smoothness of the target road segment using the Huffman tree to obtain compressed data, and transmit the compressed data based on 5G technology.

[0094] This invention adjusts the initial learning rate of the optimizer of a Long Short-Term Memory (LSTM) network model by measuring the overall noise level of the actual road surface smoothness of the target road. This improves the accuracy of the LSTM model's predictions and reduces the deviation between the predicted values ​​and the actual road surface smoothness. Based on the more accurate road surface smoothness predictions from the LSTM model, the compression effect of the Huffman coding algorithm is enhanced. The compressed coded data can be transmitted via 5G with a shorter code length, thereby reducing bandwidth requirements and data transmission time during 5G transmission and improving data transmission efficiency. Furthermore, the reduced data size after compression lowers the probability of errors during 5G transmission, improving the stability, security, and reliability of data transmission.

[0095] Optionally, the analysis module 202 is further configured to acquire height values between the plurality of measuring points and the measuring device of the target road section; determine the acceptability of the actual road surface flatness of the target road section according to differences between adjacent height values, the adjacent height values being height values of adjacent collection times in time sequence, the acceptability representing continuity of the actual road surface flatness of the target road section; acquire the first actual road surface flatness of the first preset number of days and the first acceptability of the first actual road surface flatness, and the second actual road surface flatness of the second preset number of days and the second acceptability of the second actual road surface flatness, the first preset number of days being at least one day before the actual road surface flatness corresponding time of the target road section, the second preset number of days being at least one day after the actual road surface flatness corresponding time of the target road section; determine the probability that the actual road surface flatness of the target road section belongs to a mutation point according to the acceptability of the actual road surface flatness of the target road section, the first actual road surface flatness, the first acceptability, the second actual road surface flatness and the second acceptability; determine the single-dimensional noise of the actual road surface flatness of the target road section according to the regression model of the actual road surface flatness of the target road section, the regression model being constructed based on the probability that the actual road surface flatness of the target road section belongs to a mutation point, the change rate parameter and the flatness upper limit value; determine the overall trend similarity between the actual road surface flatness of the target road section and the actual road surface flatness of the adjacent road sections on both sides of the target road section according to the regression model of the actual road surface flatness of the target road section and the regression models of the adjacent road sections on both sides of the target road section; and determine the overall noise of the actual road surface flatness of the target road section according to the single-dimensional noise of the adjacent road sections on both sides of the target road section, the single-dimensional noise of the target road section and the overall trend similarity.

[0096] Optionally, the determination module 205 is further configured to acquire historical actual road surface flatnesses of a plurality of road sections in a third preset number of days; determine the probability that each historical actual road surface flatness belongs to a mutation point and the acceptability of each historical actual road surface flatness;

[0097] determine the regression model according to the last day in the third preset number of days, the change rate parameter, the flatness upper limit value, the probability that each historical actual road surface flatness belongs to a mutation point and the day of each historical actual road surface flatness in the third preset number of days; determine the correction error function according to the acceptability of each historical actual road surface flatness, the regression model and each historical actual road surface flatness; and solve the change rate parameter and the flatness upper limit value of the regression model by the gradient descent method with the goal of minimizing the correction error function.

[0098] Optionally, the determining module 205 is further configured to acquire a first probability that each actual road surface flatness of the target road section in the fourth preset number of days belongs to the abrupt point, and a second probability that the actual road surface flatness of the adjacent road sections on both sides of the target road section belongs to the abrupt point; determine a first regression model of each actual road surface flatness of the target road section according to the first probabilities, and determine a second regression model of each actual road surface flatness of the adjacent road sections according to the second probabilities; and determine the overall trend similarity by using the first regression model and the second regression model.

[0099] Optionally, the determining module 205 is further configured to determine absolute values of differences between each adjacent height value, and an average value of each difference; calculate squares of differences between each absolute value and the average value to obtain square values, and superimpose each square value to obtain a first superimposed value; and obtain the acceptability of the actual road surface flatness of the target road section by using an exponential function to perform a negative correlation processing on the first superimposed value.

[0100] Optionally, the determining module 205 is further configured to calculate a second superimposed value of a first product of each first actual road surface flatness and the first acceptability in the first preset number of days, and a third superimposed value of a second product of each second actual road surface flatness and the second acceptability in the second preset number of days; calculate a third product of an absolute value of a difference between the second superimposed value and the third superimposed value and a reciprocal of the acceptability of the actual road surface flatness; and obtain the probability that the actual road surface flatness of the target road section belongs to the abrupt point by using a maximum-minimum value normalization function to normalize the third product.

[0101] Optionally, the determining module 205 is further configured to determine a target calculation formula in the regression model according to the probability that the actual road surface flatness of the target road section belongs to the abrupt point; determine the predicted road surface flatness of the target road section by using the target calculation formula; and determine the absolute value of a difference between the predicted road surface flatness and the actual road surface flatness of the target road section as the single-dimensional noise of the actual road surface flatness of the target road section.

[0102] Optionally, the determining module 205 is further configured to calculate an average single-dimensional noise of the single-dimensional noises of the adjacent road sections on both sides of the target road section; multiply the average single-dimensional noise by the single-dimensional noise of the target road section after performing a negative correlation processing on the average single-dimensional noise by using an exponential function to obtain a fourth product; and determine the overall noise of the actual road surface flatness of the target road section as a product of the fourth product and the overall trend similarity.

[0103] Optionally, the adjustment module 203 is further configured to obtain the initial learning rate and the time step of the optimizer of the long short-term memory network model, and the overall noise of the plurality of actual road surface flatnesses of the target road section; calculate the difference between the preset value and each actual road surface flatness of the target road section, and calculate the fourth superposition value of the difference between each actual road surface flatness and the number of the actual road surface flatness of the target road section at the time step; take the absolute value of the fourth superposition value; determine the product of the absolute value and the initial learning rate as the optimized learning rate; and determine the target long short-term memory network model according to the optimized learning rate.

[0104] Optionally, the determination module 205 is further configured to determine the normalized difference of the actual road surface flatness of the target road section according to the predicted road surface flatness and the actual road surface flatness of the target road section; determine the forward nature of the node formed by the actual road surface flatness of the target road section based on the normalized difference and the number of occurrences of the actual road surface flatness of the target road section; and determine the Huffman tree by using the forward nature of the node formed by the actual road surface flatness of the target road section.

[0105] Embodiment three:

[0106] Corresponding to the above-mentioned embodiment, based on the same technical concept, the embodiment of the present application also provides another kind of based on 5G technology's engineering test detection data high efficiency transmission system, this based on 5G technology's engineering test detection data high efficiency transmission system is used to execute the above-mentioned based on 5G technology's engineering test detection data high efficiency transmission method, Figure 3 To realize the structure of one kind of based on 5G technology's engineering test detection data high efficiency transmission system of the embodiment of the present application, as Figure 3 Shown. Based on 5G technology's engineering test detection data high efficiency transmission system can be configured or performance difference and produce relatively big, can include one or more than one processor 301 and memory 302, memory 302 is used to store the computer program that can be run on the processor, processor 301 is used to execute the program stored on memory 302, realizes the various steps in the method embodiment in the above Figure 1 Memory 302 can be temporary storage or persistent storage. The application stored in the memory 302 can include one or more modules (not shown in the figure), each module can include a series of computer executable instructions in the based on 5G technology's engineering test detection data high efficiency transmission system.

[0107] Further, the processor 301 can be configured to communicate with the memory 302, and execute a series of computer-executable instructions in the memory 302 to implement the 5G technology-based engineering test detection data efficient transmission system. The 5G technology-based engineering test detection data efficient transmission system can further include one or more power supplies 303, one or more wired or wireless network interfaces 304, one or more input / output interfaces 305, and one or more keyboards 306.

[0108] In particular, in the embodiment, the 5G technology-based engineering test detection data efficient transmission system includes a processor, a communication interface, a memory, and a communication bus; the processor, the communication interface, and the memory communicate with each other through the bus; the memory is used to store a computer program; the processor is used to execute the program stored in the memory to implement the steps of the method embodiments and has the beneficial effects of the method embodiments. For the sake of brevity, the embodiment will not be described here. Figure 1

[0109] It should be noted that the 5G technology-based engineering test detection data efficient transmission system provided by the embodiment of the present application is based on the same inventive concept as the 5G technology-based engineering test detection data efficient transmission method provided by the embodiment of the present application. Therefore, the specific implementation of this embodiment can refer to the foregoing implementation of the 5G technology-based engineering test detection data efficient transmission method, and has the same or similar beneficial effects. Repetitive parts will not be described here.

[0110] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0111] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts of each of the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.​

Claims

1. An efficient transmission method for engineering test detection data based on 5G technology, characterized by, The method comprises the following steps: Obtain the engineering test detection data of the target road section, which includes the actual road surface roughness of the target road section determined by the height values between the multiple measuring points and the measuring device of the target road section; Perform noise analysis on the actual road surface roughness of the target road section to determine the overall noise of the actual road surface roughness of the target road section, which represents the possibility that the actual road surface roughness of the target road section belongs to noise; Adjust the initial learning rate of the optimizer of the long short-term memory network model according to the overall noise of the actual road surface roughness of the target road section to obtain a target long short-term memory network model; Use the target long short-term memory network model to predict the predicted road surface roughness of the target road section; Determine the Huffman tree based on the actual road surface roughness and the predicted road surface roughness of the target road section; Compress the actual road surface roughness of the target road section using the Huffman tree to obtain compressed data, and transmit the compressed data based on 5G technology; Determine the Huffman tree based on the actual road surface roughness and the predicted road surface roughness of the target road section, which comprises the following steps: Determine the normalized difference of the actual road surface roughness of the target road section according to the predicted road surface roughness and the actual road surface roughness of the target road section, and calculate it using the following formula: wherein C represents the normalized difference between the predicted value and the actual value of each actual road roughness; Y represents the predicted value of each actual road roughness, i.e. the predicted road roughness; Z represents the actual value of each actual road roughness; |Y-Z| represents the difference between the predicted value and the actual value; C max represents the maximum value of the difference between all predicted values and actual values of the target section; C min represents the minimum value of the difference between all predicted values and actual values of the target section. Determine the forward property of the actual road surface roughness node of the target road section based on the normalized difference and the number of occurrences of the actual road surface roughness, and express it using the following formula: wherein G represents the forwardness of each actual road roughness constituting node; a represents the number of times of appearance of the same value of actual road roughness; C i represents the normalized difference between the predicted value and the actual value when the same value of actual road roughness appears the i-th time. Determine the Huffman tree using the forward property of the actual road surface roughness node of the target road section, which comprises the following steps: Merge the nodes corresponding to the forward properties of the two smallest nodes to form a new node, and then merge it with the node corresponding to the forward property of the smallest remaining node to construct the Huffman tree based on each actual road surface roughness.

2. The method of claim 1, wherein the method is a 5G technology-based method for efficiently transmitting engineering test detection data. The noise analysis on the actual road surface roughness of the target road section to determine the overall noise of the actual road surface roughness of the target road section comprises the following steps: Obtain the height values between the multiple measuring points and the measuring device of the target road section; Determine the acceptability of the actual road surface roughness of the target road section according to the difference between adjacent height values, which are the height values collected at adjacent time points, and the acceptability represents the continuity of the actual road surface roughness of the target road section; Obtain the first actual road surface roughness of the first preset number of days and the first acceptability of the first actual road surface roughness, and the second actual road surface roughness of the second preset number of days and the second acceptability of the second actual road surface roughness, wherein the first preset number of days is at least one day before the corresponding time of the actual road surface roughness of the target road section, and the second preset number of days is at least one day after the corresponding time of the actual road surface roughness of the target road section; determine a probability that the actual road surface flatness of the target road section belongs to a mutation point according to the desirability of the actual road surface flatness of the target road section, the first actual road surface flatness, the first desirability, the second actual road surface flatness, and the second desirability; determine a single-dimensional noise of the actual road surface flatness of the target road section according to a regression model of the actual road surface flatness of the target road section, the regression model being constructed based on the probability that the actual road surface flatness of the target road section belongs to a mutation point, a rate of change parameter, and an upper limit value of flatness; determine an overall trend similarity between the actual road surface flatness of the target road section and the actual road surface flatness of adjacent road sections on both sides of the target road section according to the regression model of the actual road surface flatness of the target road section and the regression model of the adjacent road sections; determine an overall noise of the actual road surface flatness of the target road section according to the single-dimensional noise of the target road section, the overall trend similarity, and the single-dimensional noise of the adjacent road sections.

3. The method of claim 2, wherein the method is a 5G technology-based method for efficiently transmitting engineering test detection data. Before the determining of the single-dimensional noise of the actual road surface flatness of the target road section according to the regression model of the actual road surface flatness of the target road section, the method further comprises: obtain historical actual road surface flatnesses of a plurality of road sections in a third preset number of days; determine a probability that each of the historical actual road surface flatnesses belongs to a mutation point and a desirability of each of the historical actual road surface flatnesses; determine the regression model according to a last day in the third preset number of days, a rate of change parameter, an upper limit value of flatness, the probability that each of the historical actual road surface flatnesses belongs to a mutation point, and a day in the third preset number of days on which each of the historical actual road surface flatnesses is located; determine a correction error function according to the desirability of each of the historical actual road surface flatnesses, the regression model, and each of the historical actual road surface flatnesses; minimize the correction error function as an objective, and solve the rate of change parameter and the upper limit value of flatness of the regression model by a gradient descent method.

4. The method of claim 2, wherein the method further comprises: The determining of the overall trend similarity between the actual road surface flatness of the target road section and the actual road surface flatness of the adjacent road sections on both sides of the target road section according to the regression model of the actual road surface flatness of the target road section and the regression model of the adjacent road sections comprises: obtain a first probability that each of the actual road surface flatnesses of the target road section belongs to a mutation point and a second probability that actual road surface flatnesses of the adjacent road sections on both sides of the target road section belong to a mutation point in a fourth preset number of days; determine a first regression model of each of the actual road surface flatnesses of the target road section according to each of the first probabilities, and determine a second regression model of each of the actual road surface flatnesses of the adjacent road sections according to each of the second probabilities; determine the overall trend similarity by using the first regression model and the second regression model.

5. The method of claim 2, wherein the method further comprises: The determining of the desirability of the actual road surface flatness of the target road section according to the difference between each of the adjacent height values comprises: determine an absolute value of the difference between each of the adjacent height values, and an average value of each of the differences; Square the difference between the absolute value of each of the difference values and the average value to obtain a square value, and superimpose each of the square values to obtain a first superimposed value; The target road surface roughness is obtained by using the exponential function to perform negative correlation processing on the first superimposed value.

6. The method of claim 2, wherein the method further comprises: The probability that the actual road surface roughness of the target road section belongs to a mutation point is determined according to the target road surface roughness, the first actual road surface roughness, the first acceptability, the second actual road surface roughness, and the second acceptability. A second superimposed value of the first product of each of the first actual road surface roughness and the first acceptability in the first preset number of days, and a third superimposed value of the second product of each of the second actual road surface roughness and the second acceptability in the second preset number of days are calculated. The third product of the absolute value of the difference between the second superimposed value and the third superimposed value and the inverse of the acceptability of the actual road surface roughness is calculated. The probability that the actual road surface roughness of the target road section belongs to a mutation point is obtained by using the maximum and minimum value normalization function to normalize the third product.

7. The method of claim 2, wherein the method further comprises: The single-dimensional noise of the actual road surface roughness of the target road section is determined according to the regression model of the actual road surface roughness of the target road section. A target calculation formula in the regression model is determined according to the probability that the actual road surface roughness of the target road section belongs to a mutation point; The predicted road surface roughness of the target road section is determined by using the target calculation formula; The absolute value of the difference between the predicted road surface roughness and the actual road surface roughness of the target road section is determined as the single-dimensional noise of the actual road surface roughness of the target road section.

8. The method of claim 2, wherein the method further comprises: The overall noise of the actual road surface roughness of the target road section is determined according to the single-dimensional noise of the adjacent road sections on both sides of the target road section, the single-dimensional noise of the target road section, and the overall trend similarity. The average single-dimensional noise of the single-dimensional noise of the adjacent road sections on both sides of the target road section is calculated. The fourth product is obtained by multiplying the average single-dimensional noise after negative correlation processing by using the exponential function and the single-dimensional noise of the target road section. The product of the fourth product and the overall trend similarity is determined as the overall noise of the actual road surface roughness of the target road section.

9. The method of claim 1-8, wherein, The initial learning rate of the optimizer of the long short-term memory network model is adjusted according to the overall noise of the actual road surface roughness of the target road section to obtain a target long short-term memory network model. The initial learning rate and the time step of the optimizer of the long short-term memory network model are obtained, and the overall noise of multiple actual road surface roughnesses of the target road section is obtained. The difference between a preset value and each of the actual road surface roughnesses of the target road section is calculated, and the fourth superimposed value of the ratio of the difference between each of the actual road surface roughnesses of the target road section and the number of actual road surface roughnesses of the target road section in the time step is calculated. The absolute value of the fourth superimposed value is taken. The product of the absolute value and the initial learning rate is determined as the optimized learning rate. The target long short-term memory network model is determined according to the optimized learning rate.

Images