A device and method for remotely sorting construction site earthwork

Abstract

The application discloses a device and method for remotely classifying and sorting earthwork of a construction site. The device collects the hydraulic pressure of the oil pipe of the excavator in real time, obtains the hydraulic peak value P0 of each excavation, takes a photo of the earthwork to obtain a camera photo, and records the shooting time. The camera photo is subjected to similarity analysis with the earthwork photo in the earthwork database, and the photos with a similarity of S or above are selected as the to-be-confirmed photos. According to the hydraulic peak value P1 of the earthwork category in the to-be-confirmed photo, the earthwork category with the hydraulic peak value P1 closest to the hydraulic peak value P0 of the on-site is selected as the earthwork category in the camera photo. The hydraulic peak value P0 of the on-site is stored into the earthwork database as the hydraulic peak value P1 of the camera photo.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of geotechnical engineering, and in particular to a device and method for remotely classifying and sorting earthwork at a construction site. BACKGROUND

[0002] Classifying earthwork at a construction site can more conveniently and quickly screen usable earthwork from engineering spoil, thereby promoting the implementation of green engineering.

[0003] The existing method of classifying and identifying earthwork at a construction site is to transport excavated earthwork to a designated site for inspection and then classification. This method not only increases transportation costs and time costs, but also requires the construction of a detection center, greatly increasing costs, and cannot quickly classify earthwork at a construction site.

[0004] The existing technology lacks an integrated system from excavation to classification to transportation command, and the existing technology lacks a device or method for identifying the types of soil and rock during excavation, which reduces the utilization rate of soil and rock and causes waste of earthwork resources, cannot timely transport earthwork, and slows down the construction speed of the project. SUMMARY

[0005] The present application aims to provide a device that can remotely classify and then transport earthwork at a construction site, greatly speeding up the classification and transportation of earthwork, saving costs, making earthwork applicable, and promoting the development of green engineering. The present application preliminarily matches the parameters of the pictures in the database with the pictures taken on site to find one or several types of soil with high similarity; further determines the type of earthwork by the hydraulic pressure of the oil pipe of the excavator, thereby improving the accuracy of earthwork type identification. The technical solution adopted by the present application is as follows:

[0006] The present application provides a device for remotely classifying and then transporting earthwork at a construction site, which comprises an excavator hydraulic monitoring system, a camera system, a classification system, a command system, a transportation system, and a data storage system.

[0007] The excavator hydraulic monitoring system comprises a hydraulic monitor of a hydraulic pipeline and a signal transmission device. The hydraulic monitor collects the on-site hydraulic pressure of the hydraulic pipeline, such as the oil pipe, of the excavator in real time, and sends the hydraulic collection time and the on-site hydraulic pressure to the classification system by the signal transmission device.

[0008] The classification system analyzes the dynamic change process of the on-site hydraulic pressure of the excavator according to the data provided by the hydraulic monitor, and obtains the on-site hydraulic peak value P0 for each excavation.

[0009] The camera system comprises cameras arranged at the perimeter of the earthwork excavation, the cameras being high-definition cameras; the cameras perform real-time shooting at the earthwork excavation site, ensuring the authenticity and accuracy of the obtained photos, obtaining camera photos; the camera photos include first category cameras and second category cameras, the first category cameras shoot photos of the earthwork directly below the excavator bucket to obtain first category camera photos; the second category cameras shoot the area in front of the excavator to obtain second category camera photos; the picture overlap rate of the second category camera photos and the first category camera photos is more than 5% and less than 50%, preferably 10%, or 15%, or 20%; the actual area of the earthwork in the first category camera photos and / or the second category camera photos is more than 1 square meter and less than 20 square meters, preferably 2 square meters, or 3 square meters, or 4 square meters, or 8 square meters, or 12 square meters, or 16 square meters; the horizontal plane of the camera is higher than the horizontal plane of the excavation perimeter, so that the high-definition camera can shoot the excavated earthwork without dead angles. The camera system sends the camera photos and the shooting time to the classification system.

[0010] The classification system confirms the on-site hydraulic peak value P0 corresponding to each camera photo according to the hydraulic collection time and the shooting time.

[0011] The earthwork database in the data storage system contains photos and characteristic parameters of various types of soil and rock, and the types and parameters of these soil and rock are constructed in advance by photographing during geological drilling sampling and indoor experimental results. The relationship between the elastic modulus of the soil and / or rock and the hydraulic peak value of the excavator oil pipe is established, i.e., an elastic modulus-hydraulic peak value BP neural network prediction model is established. The input value of the model is the elastic modulus of a certain type of earthwork, and the output value is the predicted hydraulic peak value of the earthwork, which is used as the picture hydraulic peak value P1; the training method of the elastic modulus-hydraulic peak value BP neural network prediction model adopts conventional model establishment and training methods in the field, including performing experiments in advance to obtain a training set and a prediction set, which are used for model training and verification, respectively, and both the training set and the prediction set contain multiple groups of elastic modulus and hydraulic peak value data. The model establishment and training methods in the prior art are included in the present application, and will not be described here.

[0012] The classification system performs similarity analysis on the first type of camera photos and earthwork photos in the earthwork database, and determines that the critical standard for similarity screening is similarity S; the pictures with similarity reaching S and above are screened from the earthwork database as the pictures to be confirmed, wherein 60%≤S≤100%, and S is preferably 80%. When the number of the pictures to be confirmed is 1, the earthwork category in the picture to be confirmed is taken as the earthwork category in the first type of camera photos; when the number of the pictures to be confirmed is greater than 1, the earthwork category corresponding to the picture hydraulic peak value P1 closest to the field hydraulic peak value P0 is selected according to the picture hydraulic peak value P1 of the earthwork category in each picture to be confirmed, and the earthwork category corresponding to the picture hydraulic peak value P1 is taken as the earthwork category in the first type of camera photos, that is, the earthwork category corresponding to the picture hydraulic peak value P1 closest to the field hydraulic peak value P0 is selected as the earthwork category in the camera photos. Preferably, the picture hydraulic peak value P1 satisfies 0.8P0≤P1≤1.2P0 at the same time.

[0013] Preferably, the classification system screens the pictures with similarity reaching S and above from the earthwork database as the pictures to be confirmed; and selects the earthwork category corresponding to the picture hydraulic peak value P1 closest to the field hydraulic peak value P0 as the earthwork category in the camera photos according to the picture hydraulic peak value P1 of the earthwork category in the picture to be confirmed.

[0014] Preferably, when the picture hydraulic peak value P1 of a certain picture to be confirmed is not stored in the earthwork database, the predicted hydraulic peak value obtained by the pre-trained elastic modulus-hydraulic peak value BP neural network prediction model is taken as the picture hydraulic peak value P1.

[0015] The classification system confirms the earthwork category of the second type of camera photos in the same way as confirming the earthwork category of the first type of camera photos.

[0016] Preferably, the function of the classification system is implemented by a computer, and the collected earthwork parameters are calculated by artificial intelligence classification and recognition. The database of the computer contains parameters of various types of soil and rock, such as color parameters, texture parameters, shape parameters of soil and rock, and some characteristic parameters unique to some types of soil and rock. The types and parameters of these soil and rock are constructed by photographing during geological drilling sampling and indoor experimental results. Each type of soil in the database is related to the modulus of soil and rock and the hydraulic pressure of the oil pipe of the excavator.

[0017] Preferably, the similarity analysis includes feature recognition of the photographed image, obtaining the basic parameters of the earthwork in the image, including one or more of color, texture, shape, glossiness, and then preliminarily matching the parameters of the image in the earthwork database, giving each basic parameter the same or different weight, and using Intel Movidius neural computing stick to calculate the similarity S. The above calculation method greatly improves the calculation efficiency of the rock classification and recognition artificial intelligence algorithm.

[0018] The classification system sends the earthwork category of the photographed image to the command system.

[0019] The command system commands the transportation system to transport the earthwork classified to the recycling point for processing and utilization according to the earthwork category of the photographed image.

[0020] Preferably, the command system is a program written by Python programming, which contains the name of the required earthwork. When the classification system transmits the earthwork category, the command system judges whether the earthwork of each earthwork transport vehicle reaches the utilization value according to the proportion and volume of each type of soil and rock, and outputs "yes" or "no" instructions.

[0021] Preferably, the command system makes correct judgments according to the classification of the classification system, and issues instructions to transport the usable earthwork to the designated place for processing, and the unusable earthwork is transported to another place for processing.

[0022] The data storage system is divided into three storage modules. The earthwork database is stored in module one. Preferably, the photographed image, the shooting time, the excavation date, the excavation site, the on-site hydraulic pressure, the hydraulic monitoring time, the on-site hydraulic peak value are stored in the earthwork database of module one of the data storage system at the same time, and the value of the on-site hydraulic peak value is stored as the picture hydraulic peak value P1 in the earthwork database. Preferably, according to the category and volume of soil and rock excavated by the excavator each time, the proportion and volume of each type of soil and rock of each earthwork transport vehicle are determined and synchronized to module two of the data storage system. Preferably, the final transportation site of the earthwork is synchronized to module three of the data storage system.

[0023] The application also provides a method for remotely classifying and sorting earthwork at a construction site, comprising the following steps:

[0024] Step one: real-time acquisition of the on-site hydraulic pressure of the excavator oil pipe;

[0025] Step two: obtaining the on-site hydraulic peak value P0 of each excavation according to the on-site hydraulic pressure and the acquisition time thereof;

[0026] Step three: taking a picture of the on-site earthwork to obtain a photographed image, and recording the shooting time;

[0027] Step four: Similarity analysis is performed on the photographed picture and the earthwork picture in the earthwork database to determine the critical standard of similarity screening, which is similarity S, S satisfies 60%≤S≤100%; the pictures with similarity reaching S or above are screened out from the earthwork database as the pictures to be confirmed; the earthwork database contains various earthwork pictures and characteristic parameters of earthwork;

[0028] Step five: according to the picture hydraulic peak P1 of the earthwork category in the picture to be confirmed, the earthwork category of the picture hydraulic peak P1 closest to the value of the field hydraulic peak P0 is selected as the earthwork category in the photographed picture;

[0029] Step six: the field hydraulic peak P0 is stored into the earthwork database as the picture hydraulic peak P1 along with the photographed picture;

[0030] Step seven: the earthwork category obtained according to the classification system is used to guide the transportation work.

[0031] Step eight: steps one to six are repeated every time a new earthwork is excavated.

[0032] Preferably, an elastic modulus-hydraulic peak BP neural network prediction model is established; the input value of the prediction model is the elastic modulus of the earthwork category, and the output value of the prediction model is the predicted hydraulic peak; when the picture hydraulic peak P1 of the picture to be confirmed is not stored in the earthwork database, the predicted hydraulic peak is used as the picture hydraulic peak P1.

[0033] The present application has the beneficial effects including:

[0034] 1. The present application installs the excavator hydraulic monitoring system on the excavator, can implement monitoring the change of the excavator hydraulic, timely transmits the hydraulic information, can help to indirectly judge the modulus of the excavated earthwork, the present application not only utilizes the picture comparison when determining the earthwork category, but also utilizes the hydraulic peak monitored by the excavator hydraulic monitoring system, the two are combined to greatly improve the accuracy of the earthwork category identification.

[0035] 2. The camera system of the present application includes a first category camera and a second category camera, the first category camera photographs the picture of the earthwork below the excavator bucket to obtain the first category photographed picture; the second category camera photographs the area in front of the excavator to obtain the second category photographed picture; the picture overlap rate of the second category photographed picture and the first category photographed picture is above 5% and below 50%; the actual area of the earthwork in the first category photographed picture and / or the second category photographed picture is above 1 square meter and below 20 square meters. The camera arrangement thus set improves the accuracy of identification, and the identification work of the second category photographed picture prepares for the next transportation work.

[0036] 3. The application establishes a data storage system, which systematically manages data, strengthens the management of projects, and facilitates the subsequent learning and research.

[0037] 4. The application strengthens the utilization of construction site earthwork, reduces project spoil, promotes green engineering, and through the above system, information transmission is convenient and fast, construction time is shortened, and construction cost is saved, saving time and effort. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0039] Figure 1 is a layout diagram for the construction site camera system;

[0040] Figure 2 is a soil and stone intelligent identification and classification landfill map;

[0041] Figure 3 is an example diagram of the implementation process of the system of the present application;

[0042] Figure 4 is an example of the analysis system similarity comparison result presentation, Figure 4a is a camera photo taken by the camera of the present application; Figure 4b is talc in the soil and stone database; Figure 4c is calcite in the soil and stone database; Figure 4d is quartz in the soil and stone database; Figure 4e is orthoclase in the soil and stone database.

[0043] In the figure: 1-high definition camera; 2-camera support; 3-soil and stone being excavated; 4-soil and stone classification computer; 5-computer required by the command system; 6-transport vehicle; 7-soil and stone needing to be transported; 8-computer for data storage. Specific implementation method

[0044] In order to make the purpose, technical solutions and advantages of the present application more clear, the technical solutions of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0045] Example 1

[0046] As shown in Figure 1 , in the construction site of a certain highway, a high-definition camera is installed in front of the excavation face, and the content captured by the adjacent camera has more than 10% overlap, which facilitates image collection of the earthwork in the construction site; Figure 2 The left graph in Figure 2 is a schematic diagram of the camera installation position, The right graph in

[0047] is a panoramic view of the excavation site from a distance.

[0048] Start the shooting work to obtain one or more groups of earthwork JPG format photos; at least two cameras are used to shoot the earthwork section at the same time, the first type of camera shoots the photos of the earthwork directly below the excavator bucket to obtain the first type of camera photos; the second type of camera shoots the area in front of the excavator to obtain the second type of camera photos; the camera performs real-time shooting, and 3 first type of camera photos and 3 second type of camera photos are selected at 3 time points; the computer software compares the picture overlap rates of the second type of camera photos and the first type of camera photos, which are 12%, 18.4%, and 37% respectively; the actual shooting areas of the first type of camera photos are 2.5 square meters, 6.5 square meters, and 12 square meters respectively; the actual shooting areas of the first type of camera photos are 2.8 square meters, 7.3 square meters, and 13 square meters respectively; the camera system sends the camera photos and the shooting time to the classification system.

[0049] Example 2

[0050] As shown in Figure 2 , the device of the application is composed of an excavator hydraulic monitoring system, a camera system, a classification system, a command system, a transportation system, and a data storage system;

[0051] As shown in Figure 3 , the screening idea of the embodiment of the application is to perform picture preliminary screening from the earthwork database according to the camera photos, and further screening according to the hydraulic peak value monitored by the hydraulic monitoring system; the 8 earthwork photos in the figure are used to represent the earthwork database, but the earthwork photos in the earthwork database of the application are much more than Figure 3 8 in Figure 3 The data quantity of the content of the earthwork database of the application is not limited.

[0052] The hydraulic monitoring system of the excavator comprises a hydraulic monitor and a signal transmission device, the hydraulic monitor collects the on-site hydraulic pressure of the hydraulic pipeline of the excavator, such as the oil pipeline, in real time, and sends the hydraulic collection time and the on-site hydraulic pressure to the classification system through the signal transmission device.

[0053] The classification system analyzes the dynamic change process of the on-site hydraulic pressure of the excavator according to the data provided by the hydraulic monitor, and obtains the on-site hydraulic peak value P0 for each excavation.

[0054] The earthwork database in the data storage system contains photos and characteristic parameters of various types of soil and rock, wherein the types of soil include gravel soil, sand soil, silt, clay, special soil, and the special soil is further divided into loess, expansive soil, soft soil, frozen soil, red clay, saline soil and fill; the types of rock include (1) magmatic rock, including granite, tuff, andesite, peridotite, syenite, rhyolite, syenite porphyry, basalt, trachyte, granite porphyry, granodiorite, granodiorite porphyry, dacite, pyroxenite, diorite, diorite porphyrite, and porphyritic granite; (2) sedimentary rock, including mudstone, limestone, coal, dolomite, sandstone, siltstone, conglomerate, siliceous rock, breccia, shale, etc.; (3) metamorphic rock, including greisen, phyllite, marble, slate, schist, gneiss, quartzite, skarn, cataclastic rock, mylonite, serpentine, hornstone, and greenstone; the characteristic parameters include their color Y, texture W, shape X and some types of soil and rock specific mechanical characteristics T, which are divided into four dimensions of Y, W, X and T. The types and parameters of these soil and rock are constructed by taking photos during geological exploration drilling sampling and indoor experimental results.

[0055] The relationship between the elastic modulus of soil and / or rock and the hydraulic peak value of the oil pipeline of the excavator is established, that is, an elastic modulus-hydraulic peak value BP neural network prediction model is established. The input value of the model is the elastic modulus of a certain type of earthwork, and the output value is the predicted hydraulic peak value of the earthwork, which is used as the picture hydraulic peak value P1; the training method of the elastic modulus-hydraulic peak value BP neural network prediction model adopts the conventional model establishment and training method in the field, including obtaining the training set and the prediction set by experiments in advance, which are used for model training and verification respectively, and each of the training set and the prediction set contains multiple groups of elastic modulus and hydraulic peak value data. The model establishment and training method in the prior art is included in the present application, and will not be described here.

[0056] The classification system confirms the on-site hydraulic peak value P0 corresponding to each camera photo according to the hydraulic collection time and the shooting time.

[0057] The classification system performs similarity analysis on the first type of camera photos and earthwork photos in the earthwork database, and determines that the critical standard for similarity screening is similarity S; the pictures with similarity reaching S or above are screened from the earthwork database as the pictures to be confirmed, wherein 60%≤S≤100%, in some implementation scenarios, S is 60%; in some implementation scenarios, S is 99.9%; and generally, S is set to 80%.

[0058] In some implementation scenarios, the number of pictures to be confirmed is 1, and the earthwork category in the picture to be confirmed is taken as the earthwork category in the first type of camera photos.

[0059] In some implementation scenarios of the present application, the number of pictures to be confirmed is greater than 1, and according to the picture hydraulic peak value P1 of the earthwork category in each picture to be confirmed, the picture hydraulic peak value P1 closest to the field hydraulic peak value P0 is selected, and the earthwork category corresponding to the picture hydraulic peak value P1 is taken as the earthwork category of the first type of camera photos; the picture hydraulic peak value P1 simultaneously satisfies 0.8P0≤P1≤1.2P0, in some embodiments, P1=1.1P0, in some embodiments, P1=0.8P0; and in some embodiments, P1=1.2P0.

[0060] In some implementation scenarios of the present application, the picture hydraulic peak value P1 of the picture to be confirmed is not stored in the earthwork database, and the picture hydraulic peak value P1 of the picture to be confirmed is obtained through a pre-trained elastic modulus-hydraulic peak BP neural network prediction model, and then compared with the field hydraulic peak value P0.

[0061] The classification system confirms the earthwork category of the second type of camera photos according to the same method as confirming the earthwork category of the first type of camera photos.

[0062] In some implementation scenarios of the present application, the similarity analysis includes feature recognition of the camera photos by using a computer to obtain basic parameters of the earthwork in the photos, including one or more of color, texture, shape, and mechanical characteristics, and then performing preliminary matching of the parameters with the parameters of the pictures in the earthwork database, giving each basic parameter the same or different weight, and calculating the similarity S by using an Intel Movidius neural computing stick, wherein the mechanical characteristics of the earthwork are deduced from the hydraulic changes of the excavator. The basic parameters of the earthwork are one-to-one corresponding to the four parameters of color Y, texture W, shape X, and mechanical characteristics T in the original data of the computer, and the weights of the four parameters are respectively assigned as 3, 4, 2, and 1. According to the similarity and weight of each of the four parameters, the similarity S is calculated.

[0063] According to the earthwork category and quantity of each excavation, the proportion of each type of earthwork of each earthwork truck is statistically analyzed.

[0064] The available earthwork condition is set in the command system computer program, for example, the granite, greisen and marble account for more than 50% and then have available value, that is, economic value. The result obtained by the classification system is transmitted to the command system through the network, and the computer judges whether the earthwork is needed, for example, when the earthwork of the vehicle has available value, the result outputs "yes", and the earthwork is determined as available, then the command system gives the transportation system an instruction one, and transports the earthwork to the designated processing plant for earthwork recycling. When the earthwork of the vehicle does not have available value, the result outputs "no", and the earthwork is determined as unavailable, then the command system gives the transportation system an instruction two, and transports the earthwork to the designated location for backfilling or stacking.

[0065] The data storage system is divided into three storage modules, the earthwork database is stored in module one, after one flow cycle, the camera photos, shooting time, excavation date, excavation location, on-site hydraulic pressure, hydraulic monitoring time, on-site hydraulic peak value are stored in the earthwork database of module one of the data storage system at the same time, and the value of the on-site hydraulic peak value is stored as the picture hydraulic peak value P1 in the earthwork database; according to the category and volume of the earth and rock excavated by the excavator each time, the proportion and volume of each type of earth and rock of each earthwork transport vehicle are determined and synchronized to module two of the data storage system; the final transportation site of the earthwork is synchronized to module three of the data storage system.

[0066] The data storage system stores the data of each link throughout the entire implementation process, so as to facilitate computer self-learning next time.

[0067] Example 3

[0068] The actual recycling situation encountered in the excavation process involved in the present application includes:

[0069] Each vehicle is mainly composed of granite, and after screening, the granite is made into concrete aggregate and directly applied to engineering;

[0070] Each vehicle is mainly composed of round gravel clay, which is used as roadbed filler; but in other implementation cases of the present application, the engineering involved does not need roadbed filler, so the utilization of round gravel clay is not economical, and the round gravel clay cannot be utilized and is transported to the waste dump.

[0071] Each vehicle is mainly composed of clay, and since the utilization cost of clay is relatively high, and there is no special situation, the clay is classified as non-usable soil and transported to the waste dump; in other implementation cases of the present application, there are brick factories or economic transportation is relatively developed around, and the profit point of producing bricks is high, so the clay is made into non-burning bricks or ceramsite for further utilization.

[0072] In various embodiments of the present application, whether the excavated earthwork can be recycled is determined according to the scale, use, and actual situation of the recycling project and the local excavation industry.

[0073] Embodiment 4

[0074] As shown in FIG. 4, according to the result graph in the embodiment of Embodiment 2, Figure 4a The photographed image is a camera image photographed by the camera of the present application; Figures 4b-4e The image with a similarity greater than 60% screened from the earthwork database is a to-be-confirmed image, and the detailed information is shown in the following table:

[0075]

[0076] Since the on-site hydraulic peak value P0 is 2.5 MPa, which is closest to the hydraulic peak value of calcite, the earthwork type in the camera image is finally confirmed as calcite.

[0077] The above specific embodiments are used to explain and illustrate the present application, rather than limit the present application, and any modification and change made to the present application within the spirit and protection scope of the claims of the present application shall fall into the protection scope of the present application.

Claims

1. A device for remotely sorting and selecting earth and rock at a construction site, the device comprising an excavator hydraulic monitoring system, a camera system, a sorting system, a command system, a transportation system, and a data storage system; The data storage system includes an earthwork database containing various earthwork photos and characteristic parameters of earthwork. The excavator hydraulic monitoring system includes a hydraulic monitor installed in the excavator's hydraulic lines. The hydraulic monitor collects the on-site hydraulic pressure of the excavator's oil lines in real time and sends the collection time and the on-site hydraulic pressure to a classification system. The classification system obtains the peak hydraulic pressure P0 for each excavation based on data provided by the hydraulic monitor. The camera system is used to take photos of the earthwork at the site to obtain video photos; the camera system sends the video photos and the shooting time to the classification system; The classification system performs similarity analysis between the video photos and earthwork photos in the earthwork database, and determines the similarity screening threshold as similarity S, where S satisfies 60% ≤ S ≤ 100%. Images with similarity of S or higher are selected from the earthwork database as images to be confirmed. Based on the hydraulic peak value P1 of the earthwork category in the images to be confirmed, the earthwork category with the hydraulic peak value P1 that is closest to the on-site hydraulic peak value P0 is selected as the earthwork category in the video photos. The command system simultaneously stores the on-site hydraulic peak value P0 as the image hydraulic peak value P1 along with the video photo into the earthwork database. The command system directs the transportation system to carry out transportation work based on the earthwork categories derived from the classification system.

2. The apparatus according to claim 1, wherein the earthwork photographs include photographs of soil and / or stone, and the characteristic parameters of the earthwork include, but are not limited to, one or more of the following: color, texture, shape, mechanical characteristics, and gloss of the soil and / or rock.

3. The apparatus according to claim 1, wherein the camera system comprises a first type of camera and a second type of camera, the first type of camera taking pictures of the earth and rock below the excavator bucket to obtain a first type of video photo; the second type of camera taking pictures of the area in front of the excavator to obtain a second type of video photo; the overlap rate between the second type of video photo and the first type of video photo is more than 5% and less than 50%; the actual area occupied by the earth and rock in the first type of video photo and / or the second type of video photo is more than 1 square meter and less than 20 square meters; the classification system confirms the on-site hydraulic peak value P0 corresponding to the first type of video photo and the second type of video photo respectively, based on the hydraulic acquisition time and the shooting time.

4. The apparatus according to claim 1, wherein the classification system includes an elastic modulus-hydraulic peak value BP neural network prediction model; the input value of the prediction model is the elastic modulus of the earthwork category, and the output value of the prediction model is the predicted hydraulic peak value; when the earthwork database does not store the image hydraulic peak value P1 of the image to be confirmed, the predicted hydraulic peak value is used as the image hydraulic peak value P1.

5. The device according to claim 1, wherein the peak hydraulic pressure P1 of the image simultaneously satisfies 0.8P0≤P1≤1.2P0.

6. The apparatus according to claim 2, wherein the similarity analysis includes performing feature recognition on the photograph to obtain basic parameters of the earthwork in the photograph, including one or more of color, texture, shape, mechanical characteristics, and gloss, and then performing preliminary matching with feature parameters in the earthwork database, assigning the same or different weights to each feature parameter, and calculating the similarity S using the Intel Movidius neural computing stick.

7. The apparatus according to claim 1, wherein the classification system sends the earthwork category of the photographic images to the command system; the command system directs the transportation work of the transportation system according to the earthwork category of the photographic images.

8. The device according to claim 1, wherein the video photos, shooting time, excavation date, excavation location, on-site hydraulic pressure, hydraulic monitoring time, and on-site hydraulic peak value are synchronously stored in the earthwork database; the proportion and volume of each type of soil and rock in each earthwork transport vehicle are determined according to the type and volume of soil and rock excavated by the excavator each time, and synchronized to the data storage system; and the final transport location of the earthwork is synchronized to the data storage system.

9. A method for remotely sorting and selecting earth and rock at a construction site, comprising the following steps: Step 1: Real-time acquisition of on-site hydraulic pressure in the excavator's oil pipes; Step 2: Obtain the peak hydraulic pressure P0 for each excavation based on the on-site hydraulic pressure and the acquisition time; Step 3: Take photos of the earthwork at the site to obtain video photos, and record the shooting time; Step 4: Perform similarity analysis between the video photos and the earthwork photos in the earthwork database, and determine the similarity screening threshold as similarity S, where S satisfies 60% ≤ S ≤ 100%; select images with a similarity of S or higher from the earthwork database as images to be confirmed; the earthwork database contains various earthwork photos and earthwork feature parameters. Step 5: Based on the image hydraulic peak value P1 of the earthwork category in the image to be confirmed, select the earthwork category of the image hydraulic peak value P1 that is closest to the on-site hydraulic peak value P0 as the earthwork category in the photograph. Step Six: Simultaneously store the on-site hydraulic peak value P0 as the image hydraulic peak value P1 along with the video photo into the earthwork database; Step 7: Direct the transportation work based on the earthwork category obtained from the classification system.

10. According to the method of claim 9, an elastic modulus-hydraulic peak value BP neural network prediction model is established; the input value of the prediction model is the elastic modulus of the earthwork category, and the output value of the prediction model is the predicted hydraulic peak value; when the earthwork database does not store the image hydraulic peak value P1 of the image to be confirmed, the predicted hydraulic peak value is used as the image hydraulic peak value P1.

Images