Railway roadbed filling process automatic identification method, system and equipment and storage medium

Abstract

The invention relates to the technical field of process identification, and discloses an automatic identification method, system and device for a railroad bed filling process and a storage medium, and the method comprises the steps: constructing a process structure according to the construction data of railroad bed filling; according to the construction progress and the procedure structure, the front procedure is initialized; collecting a construction image, and performing target identification on the construction image to obtain an identification target; acquiring equipment sensor data of the construction equipment, and performing data analysis on the equipment sensor data according to the auto-encoder model to obtain an equipment state; and determining the current three-stage process according to the identification target and the equipment state, and updating the previous process according to the three-stage process. Through image recognition, data analysis and process reasoning, accurate recognition and automatic updating of the construction process can be realized, so that the management efficiency and the project progress of railway roadbed filling construction are improved, and the unmanned management degree of construction is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of process identification, and in particular to a method, system, equipment and storage medium for automatically identifying railway roadbed filling processes. Background Art

[0002] Traditional railway roadbed filling with manned construction faces many limitations of human conditions. The construction of railway roadbed filling in remote areas is bound to have huge economic costs and low efficiency. In recent years, unmanned construction has great application potential due to its characteristics of efficient project progress management, lower construction safety risks and costs, and strong adaptability to harsh environments. In unmanned construction, the control of construction progress and the organization and management of on-site equipment are inseparable from the accurate identification of construction processes.

[0003] The existing automatic identification method of construction process mainly relies on the image semantic information of the equipment to judge the process. However, since most of the construction equipment in railway roadbed filling construction travels in a straight line, there is no significant change in the equipment posture during the construction process, and there is no significant difference in the filling construction of each layer, it is difficult to identify the railway roadbed filling process using image semantics, and it is impossible to achieve accurate and effective process identification. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method, system, equipment and storage medium for automatic identification of railway roadbed filling procedures, so as to solve the problem that the existing image semantic analysis cannot accurately identify the construction procedures, so as to achieve the effect of accurately identifying the construction procedures of railway roadbed filling and improving the management efficiency of railway roadbed filling construction.

[0005] In a first aspect, the present invention provides a method for automatically identifying a railway roadbed filling process, the method comprising:

[0006] According to the construction data of railway roadbed filling, a process structure based on the hierarchical relationship of the construction process is constructed, and the process structure includes a primary process structure, a secondary process structure and a tertiary process structure;

[0007] Initialize the pre-process according to the construction progress and the process structure, the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process;

[0008] Collecting construction images of railway roadbed filling, and performing target recognition on the construction images to obtain recognized targets, wherein the recognized targets include construction personnel and construction equipment;

[0009] Acquire equipment sensor data of the construction equipment, and perform data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain an equipment status of the construction equipment;

[0010] The current third-level process is determined according to the identification target and the equipment status, and the preceding process is updated according to the current third-level process.

[0011] Furthermore, the step of constructing a process structure based on the hierarchical relationship of the construction process according to the construction data of the railway roadbed filling includes:

[0012] Taking each construction object of railway roadbed filling as a primary process, taking the construction sequence of each construction object as a primary process sequence, and constructing a primary process structure according to the primary process and the primary process sequence;

[0013] Taking each construction stage of the construction object as a secondary process, taking the stage sequence of each construction stage as a secondary process sequence, and constructing a secondary process structure according to the secondary process and the secondary process sequence;

[0014] Taking each construction action included in the construction stage as a three-level process, taking the action sequence of each construction action as a three-level process sequence, and constructing a three-level process structure according to the three-level processes and the three-level process sequence;

[0015] According to the hierarchical relationship between the primary process structure, the secondary process structure and the tertiary process structure, a process structure of the railway roadbed filling construction process is constructed.

[0016] Furthermore, the step of acquiring the equipment sensor data of the construction equipment and performing data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain the equipment status of the construction equipment includes:

[0017] According to the equipment type of each equipment in the construction area, a corresponding equipment sensor is pre-installed on each equipment, and the equipment sensor data includes positioning sensor data and state sensor data, and the state sensor data includes speed sensor data, acceleration sensor data and inclinometer sensor data;

[0018] Acquire the equipment sensor data of each equipment, and filter out the state sensor data of the construction equipment according to the positioning sensor data of the construction equipment;

[0019] Performing data analysis on the state sensor data of the construction equipment according to a pre-built autoencoder model to obtain the sensor state of each state sensor on the construction equipment, wherein the sensor state includes a normal state and an abnormal state;

[0020] The equipment status of the construction equipment is obtained according to the number of sensors whose sensor status is normal.

[0021] Furthermore, the step of performing data analysis on the state sensor data of the construction equipment according to the pre-built autoencoder model to obtain the sensor state of each state sensor on the construction equipment includes:

[0022] Inputting the state sensor data of the construction equipment into a trained autoencoder model, and calculating an error index between the reconstructed data output by the autoencoder model and preset normal sensor data;

[0023] The error indicator is compared with a preset error threshold, and the sensor state of the corresponding state sensor is obtained according to the comparison result.

[0024] Furthermore, the loss function of the autoencoder model is expressed by the following formula:

[0025]

[0026] The error index is expressed by the following formula:

[0027]

[0028] In the formula, represents the i-th reconstructed data, e i represents the i-th normal sensor data, and n represents the total number of data.

[0029] Furthermore, the step of obtaining the equipment status of the construction equipment according to the number of sensors whose sensor status is normal includes:

[0030] Determine whether the number of sensors in the normal state is greater than a threshold number. If so, determine that the equipment state of the construction equipment is a normal working state. Otherwise, determine that the equipment state of the construction equipment is a non-working state.

[0031] Further, the step of determining the current third-level process according to the identification target and the equipment status, and updating the preceding process according to the current third-level process includes:

[0032] In response to the equipment being in a normal working state, performing process reasoning on the identified target to determine a current third-level process;

[0033] Determine whether the current three-level process is the same as the previous three-level process, and if they are the same, keep the previous process unchanged;

[0034] If they are different, determining whether the three-level process sequence of the preceding three-level process is the last of the three-level processes;

[0035] If it is not the last one of the three-level processes, the preceding three-level process is updated according to the current three-level process, and the preceding two-level process and the preceding one-level process are kept unchanged;

[0036] If it is the last of the third-level process, determine whether the second-level process sequence of the preceding second-level process is the last of the second-level process;

[0037] If it is not the last of the secondary processes, the preceding third-level process is updated according to the current third-level process, and the preceding second-level process is updated to the second-level process of the next order according to the sequence of the secondary processes, and the preceding first-level process is kept unchanged;

[0038] If it is the last second-level process, the preceding third-level process is updated according to the current third-level process, and the preceding first-level process is updated to the first-level process of the next priority according to the first-level process sequence, and the preceding second-level process is updated to the second-level process of the first priority according to the second-level process sequence corresponding to the updated preceding first-level process.

[0039] In a second aspect, the present invention provides a railway roadbed filling process automatic identification system, the system comprising:

[0040] A process structure building module is used to build a process structure based on the hierarchical relationship of the construction process according to the construction data of the railway roadbed filling, and the process structure includes a primary process structure, a secondary process structure and a tertiary process structure;

[0041] A process initialization module is used to initialize the pre-process according to the construction progress and the process structure, wherein the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process;

[0042] A target recognition module is used to collect construction images of railway roadbed filling, and perform target recognition on the construction images to obtain recognized targets, wherein the recognized targets include construction personnel and construction equipment;

[0043] A state recognition module is used to obtain equipment sensor data of the construction equipment, and perform data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain the equipment state of the construction equipment;

[0044] The process reasoning and updating module is used to determine the current third-level process according to the identification target and the equipment status, and to update the preceding process according to the current third-level process.

[0045] In a third aspect, an embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0046] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0047] The present invention provides a method, system, device and storage medium for automatically identifying railway roadbed filling procedures. The present invention establishes a procedure structure of the hierarchical relationship of construction procedures based on the construction process of railway roadbed filling, determines the personnel and equipment at the construction site based on image recognition and data processing, and determines the current procedure by reasoning about the personnel and equipment on the basis of the procedure structure, and updates the preceding procedure according to the current procedure, so as to realize accurate identification of the construction procedure, thereby improving the construction management efficiency and engineering progress of railway roadbed filling, and further improving the degree of unmanned management of construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 1 is a schematic diagram of a flow chart of a method for automatically identifying a railway roadbed filling process according to an embodiment of the present invention;

[0049] Figure 2 Schematic diagram of railway roadbed filling in an embodiment of the present invention;

[0050] Figure 3 It is a process structure diagram of the railway roadbed filling construction process in an embodiment of the present invention;

[0051] Figure 4 2 is a schematic diagram of the structure of the railway roadbed filling process automatic identification system according to an embodiment of the present invention;

[0052] Figure 5 It is a diagram of the internal structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] See also Figure 1 The first embodiment of the present invention provides a method for automatically identifying a railway roadbed filling process, which includes steps S10 to S50:

[0055] Step S10, constructing a process structure based on a hierarchical relationship of construction processes according to the construction data of railway roadbed filling, wherein the process structure includes a primary process structure, a secondary process structure and a tertiary process structure;

[0056] Step S20, initializing the pre-process according to the construction progress and the process structure, wherein the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process;

[0057] Step S30, collecting a construction image of railway roadbed filling, and performing target recognition on the construction image to obtain a recognition target, wherein the recognition target includes construction personnel and construction equipment;

[0058] Step S40, acquiring equipment sensor data of the construction equipment, and performing data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain an equipment status of the construction equipment;

[0059] Step S50, determining the current third-level process according to the identification target and the equipment status, and updating the preceding process according to the current third-level process.

[0060] The present invention provides a method for automatically identifying the process of railway roadbed filling. Before describing the method, the process of railway roadbed filling is briefly introduced. Figure 2 The layers that need to be filled in the railway roadbed can be divided into cushion layer, bottom layer, top layer and surface layer from bottom to top. The cushion layer is divided into three layers: bottom, middle and top. The bottom layer and top layer have n layers of filling, i.e., bottom n layers to bottom 1 layer, top n layers to top 1 layer. The specific size of n is determined according to the road grade and local construction requirements. The surface layer is divided into membrane layer and surface layer. Before filling each layer, the surface of the roadbed needs to be surveyed and mapped, and the grid lines need to be marked to guide the filling range. The main process of filling construction is to spread and roll the filling materials according to the grid line requirements.

[0061] Based on the above filling process, in this embodiment, a construction process hierarchy structure of railway roadbed filling is first established, and its specific construction steps include:

[0062] Taking each construction object of railway roadbed filling as a primary process, taking the construction sequence of each construction object as a primary process sequence, and constructing a primary process structure according to the primary process and the primary process sequence;

[0063] Taking each construction stage of the construction object as a secondary process, taking the stage sequence of each construction stage as a secondary process sequence, and constructing a secondary process structure according to the secondary process and the secondary process sequence;

[0064] Taking each construction action included in the construction stage as a three-level process, taking the action sequence of each construction action as a three-level process sequence, and constructing a three-level process structure according to the three-level processes and the three-level process sequence;

[0065] According to the hierarchical relationship among the primary process structure, the secondary process structure and the tertiary process structure, a process structure of the railway roadbed filling construction process is constructed.

[0066] In this embodiment, the hierarchical structure of the construction process is composed of multiple levels of processes, and the number of levels is related to the division accuracy of the construction process characteristics. The more detailed the construction process division is, the more levels there are, and there is an inclusion relationship between the upper and lower levels. For the construction process of railway roadbed filling, in this embodiment, the process structure can be set to three layers, from top to bottom, respectively, a first-level process structure, a second-level process structure, and a third-level process structure, wherein the first-level process structure should contain multiple second-level process structures, and each second-level process structure contains multiple third-level process structures.

[0067] In this embodiment, the construction object of filling is taken as the first-level process, and the construction sequence of each construction object is taken as the first-level process sequence. The first-level process structure is composed of each first-level process and the first-level process sequence. If the first-level process is recorded as a i , represents the i-th primary process. According to the construction sequence, the primary process structure can be expressed as [a 1 , ..., a i ], according to the above filling process, the first-level process, that is, the construction object, includes cushion layer, bottom layer, top layer and surface layer. The first-level process sequence, that is, the construction sequence, is constructed in sequence. Then the first-level process structure is [cushion layer, bottom layer, top layer, surface layer].

[0068] For each primary process a i , that is, the construction progress of each construction object can be divided into multiple construction stages, and each construction stage is regarded as a secondary process, recorded as It represents the jth construction stage of the i-th construction object. According to the stage sequence, the secondary process structure of the i-th construction object can be expressed as In this embodiment, the construction stage of the cushion layer can be divided into lower layer surveying and setting out, lower layer construction, middle layer surveying and setting out, middle layer construction, upper layer surveying and setting out, and upper layer construction according to the order of construction stages. The secondary process structure of the cushion layer for the construction object can be expressed as [lower layer surveying and setting out, lower layer construction, middle layer surveying and setting out, middle layer construction, upper layer surveying and setting out, and upper layer construction]; similarly, for the construction stage of the bottom layer, it can be divided into bottom layer surveying and setting out and bottom layer construction according to the order of construction stages. If the bottom layer includes n layers, it can be divided into bottom n layer surveying and setting out, bottom n layer construction, ..., bottom 1 layer surveying and setting out, and bottom 1 layer construction. Assuming n is 1, the secondary process structure of the bottom layer The process structure is expressed as [bottom layer surveying and laying out, bottom layer construction]; for the construction stage of the top layer, it can be divided into top layer surveying and laying out and top layer construction according to the order of construction stages. If the top layer includes n layers, it can be divided into top n layer surveying and laying out, top n layer construction, ..., top 1 layer surveying and laying out and top 1 layer construction. Assuming n is 1, the secondary process structure of the top layer is expressed as [top layer surveying and laying out, top layer construction]; for the construction stage of the surface layer, it can be divided into membrane layer surveying and laying out, membrane layer construction, surface layer surveying and laying out and surface layer construction according to the order of construction stages. The secondary process structure of the surface layer is expressed as [membrane layer surveying and laying out, membrane layer construction, surface layer surveying and laying out, surface layer construction].

[0069] For each secondary process That is, each construction stage corresponds to different construction actions, and each construction action is regarded as a three-level process, recorded as The kth construction action of the jth construction stage of the i-th construction object can be expressed as the three-level process structure of the jth construction stage of the i-th construction object according to the construction action sequence: In this embodiment, for each construction stage, for the surveying and layout stage, the construction actions include measurement and layout, and for the layer construction stage, the construction actions include paving and rolling. Taking the lower layer construction stage of the cushion layer construction as an example, its three-level process structure can be expressed as [paving, rolling].

[0070] Finally, according to the hierarchical relationship between the process structures at all levels, the process structure of railway roadbed filling is constructed, such as Figure 3 As shown, the bottom layer and the top layer are both taken as 1 layer as an example. Figure 3 The process structure of the railway roadbed filling process represents the upper and lower hierarchical relationship and the front and back order relationship. Based on this process structure, the next step of process identification can be carried out.

[0071] In the process identification process of this embodiment, the preceding process is first initialized according to the current construction progress. The preceding process is represented by a process structure, that is, the current construction processes at all levels are determined according to the current construction progress, thereby obtaining [pre-first-level process, pre-second-level process, pre-third-level process]. For example, the current construction progress is to pave the lower construction stage of the cushion layer, and the preceding process is initialized to [cushion layer, lower construction, paving]. Then, the current construction process is inferred through image recognition and data processing, and it is determined whether the preceding process needs to be updated based on the inferred construction process, thereby obtaining the current construction process, that is, the latest preceding process is the current construction process.

[0072] In this embodiment, the automatic identification of the current process can be divided into three parts: image recognition, data processing and process reasoning. Among them, image recognition refers to collecting construction pictures of the construction area by means of a camera device arranged in the railway roadbed construction area that can fully cover the construction area, and then performing target recognition on the construction picture to determine the targets in the current construction area. The identified targets include construction personnel and construction equipment. The construction equipment includes bulldozers, graders, rollers, levels and total stations. It should be noted here that when laying out the lines, no specific instruments and equipment are required, but the appearance of grid lines on the ground is used as the judgment. Therefore, in this embodiment, the identified targets include not only the construction equipment but also the grid lines on the ground. In order to unify the description, the grid lines are also classified into the construction equipment.

[0073] In a preferred embodiment, the present invention uses the YOLO model to perform target recognition on the construction image. Of course, other neural network models can also be used to achieve target recognition, and no excessive restrictions are made here.

[0074] After the construction personnel and construction equipment in the current construction area are determined by image recognition, the equipment status of the construction equipment is determined by analyzing the sensor data of the sensors installed on the construction equipment. The specific steps include:

[0075] According to the equipment type of each equipment in the construction area, a corresponding equipment sensor is pre-installed on each equipment, and the equipment sensor data includes positioning sensor data and state sensor data, and the state sensor data includes speed sensor data, acceleration sensor data and inclinometer sensor data;

[0076] Acquire the equipment sensor data of each equipment, and filter out the state sensor data of the construction equipment according to the positioning sensor data of the construction equipment;

[0077] Performing data analysis on the state sensor data of the construction equipment according to a pre-built autoencoder model to obtain the sensor state of each state sensor on the construction equipment, wherein the sensor state includes a normal state and an abnormal state;

[0078] The equipment status of the construction equipment is obtained according to the number of sensors whose sensor status is normal.

[0079] In this embodiment, firstly, according to the equipment type of each equipment, the corresponding sensor is pre-installed on each equipment to collect data, wherein the sensor includes a positioning sensor and a state sensor, and the state sensor includes a speed sensor, an acceleration sensor and an inclinometer sensor. Specifically, for the railway roadbed filling construction, the equipment used includes a bulldozer, a grader, a roller, a level and a total station, etc., and a positioning sensor is installed on each equipment, and a speed sensor is also installed on the bulldozer, the grader and the roller. In addition, an inclinometer sensor is also installed on the bulldozer and the grader, and an acceleration sensor is also installed on the roller; the positioning sensor is used to feedback the area where the construction equipment is located, the speed sensor data is used to feedback the speed state of the equipment, and judge whether the equipment is acting in accordance with the normal working mode, and the acceleration sensor is used to feedback whether the roller of the roller is in a working state or a non-working state; the inclinometer data is used to feedback whether the bucket posture of the grader or the bulldozer meets the characteristics of the normal working state.

[0080] In this embodiment, for bulldozers, graders, rollers and other equipment that need to work dynamically, the corresponding sensor data collected by the sensors installed on these equipment are analyzed by the auto-encoding model to determine whether the sensor data is normal, thereby judging whether these equipment are in normal working state. For static working equipment such as levels and total stations, there is no need to judge the equipment status.

[0081] Specifically, for the construction equipment that needs to work dynamically in the identification target, firstly, the construction equipment in the construction image is screened out through the positioning data, and the corresponding sensor data is extracted. Then, the sensor data is analyzed through the trained autoencoder model to determine whether the sensor data is normal. Among them, the autoencoder is a neural network, and its input and output are consistent. The autoencoder is based on the idea of ​​sparse coding and realizes sample reconstruction by recombining high-order features. Its training goal is to minimize the reconstruction error. In this embodiment, the sensor data of the equipment in normal working state is collected as a data set for model training to master the data characteristics of the normal state of the equipment:

[0082] h=f(e)=φ(We+b)

[0083] Wherein, h represents data features, f represents encoder, φ represents encoder activation function, W represents encoder weight parameter, b represents encoder bias difference, and e represents input device sensor data. In this embodiment, the average of multiple samplings of sensor data under normal working state through sliding window is used as input data e.

[0084] Then use the decoder to restore the normal sensor data features:

[0085]

[0086] In the formula, Represents the reconstructed data, g represents the decoder, φ` represents the decoder activation function, W` represents the decoder weight parameter, and b` represents the decoder bias difference.

[0087] The loss function of the autoencoder is expressed as:

[0088]

[0089] In the formula, represents the i-th reconstructed data, e i represents the i-th normal sensor data, and n represents the total number of data.

[0090] Then, the sensor data of the identified construction equipment is reconstructed through the trained autoencoder, and the error between the reconstructed data and the normal data is used to determine whether the sensor data is normal. The specific steps include:

[0091] Inputting the state sensor data of the construction equipment into a trained autoencoder model, and calculating an error index between the reconstructed data output by the autoencoder model and preset normal sensor data;

[0092] The error indicator is compared with a preset error threshold, and the sensor state of the corresponding state sensor is obtained according to the comparison result.

[0093] In this embodiment, the error index is constructed by the error between the reconstructed data and the sensor data under normal working conditions, and the error index can be expressed as:

[0094]

[0095] The normal sensor data here can be characterized by a stable value obtained through multiple acquisitions or an average value obtained through multiple acquisitions, etc., which can represent the data of the sensor in a normal working state. Then, the data state of the sensor data is judged by comparing the error index with a preset error threshold. In this embodiment, the error threshold is the MAE upper limit value MAE of the training data set. maxTo determine, take MAE max *(1 / λ) is used as the error threshold, where λ represents the correction coefficient and takes the value [0,1], which is used to characterize the integrity and representativeness of the data set. When the training data set can fully represent all situations of normal construction status, λ takes 1.

[0096] When the error index is less than the error threshold, the group of sensor data is judged to be normal, otherwise it is judged to be abnormal. Considering that in the actual construction environment, the construction actions of the equipment are obtained after being processed by different sensors, and a certain equipment may have only one action at a time, therefore, in this embodiment, when the equipment status of the construction equipment is determined by the sensor format of the sensor data being in a normal state, it is preferred that when at least one sensor on the equipment is judged to be normal, the equipment status of the equipment is in a normal working state.

[0097] Since the training data set is the sensor data of the training equipment in normal working state, when the equipment is in non-working state, the sensor data generated will be judged as abnormal, and the same applies to other types of sensors except positioning sensor data. Therefore, normal and abnormal here refer to the results of sensor data classification by the trained autoencoder during the construction process, and finally output the equipment status.

[0098] Through the above steps, the construction personnel, construction equipment and equipment status of the current construction site can be obtained. Then, by reasoning these data, the current third-level process being executed at the current construction site can be determined, and by reasoning and deciding between the current third-level process and the preceding process, the preceding process can be updated, thereby realizing automatic identification of the construction process. The specific steps include:

[0099] In response to the equipment being in a normal working state, performing process reasoning on the identified target to determine a current third-level process;

[0100] Determine whether the current three-level process is the same as the previous three-level process, and if they are the same, keep the previous process unchanged;

[0101] If they are different, determining whether the three-level process sequence of the preceding three-level process is the last of the three-level processes;

[0102] If it is not the last one of the three-level processes, the preceding three-level process is updated according to the current three-level process, and the preceding two-level process and the preceding one-level process are kept unchanged;

[0103] If it is the last of the third-level process, determine whether the second-level process sequence of the preceding second-level process is the last of the second-level process;

[0104] If it is not the last of the secondary processes, the preceding third-level process is updated according to the current third-level process, and the preceding second-level process is updated to the second-level process of the next order according to the sequence of the secondary processes, and the preceding first-level process is kept unchanged;

[0105] If it is the last second-level process, the preceding third-level process is updated according to the current third-level process, and the preceding first-level process is updated to the first-level process of the next priority according to the first-level process sequence, and the preceding second-level process is updated to the second-level process of the first priority according to the second-level process sequence corresponding to the updated preceding first-level process.

[0106] In this embodiment, the current three-level process is first inferred based on the identification target and the equipment status. According to the above filling steps, the construction actions in the three-level process structure mainly include measurement, layout, paving and rolling. Among them, measurement is mainly completed by construction personnel through a level or a total station. Therefore, when the identification target is a construction personnel and a level or a total station, it can be judged that the current three-level process is measurement; layout is mainly drawn by construction personnel. Therefore, when the identification target is a construction personnel and a grid line, it can be judged that the current three-level process is layout; paving is mainly completed by a bulldozer or a grader. Since the driver is in the machine, when the identification target is a bulldozer or a grader and the equipment is in normal working condition, it can be judged that the current three-level process is paving. Rolling is mainly completed by a roller. Similarly, when the identification target is a roller and the equipment is in normal working condition, it can be judged that the current three-level process is rolling. According to the above logic, the inference rules are constructed, and the inference decision algorithm is used to build the inference model. The identification target and equipment status are input into the inference model to obtain the current three-level process.

[0107] Then, the process update is also realized based on the inference decision algorithm. Figure 3 Take the process structure as an example to illustrate the specific steps of reasoning and decision-making. First, determine whether the current three-level process is the same as the previous three-level process. Take the initialized previous process as [base layer, lower layer construction, paving] as an example. If the current three-level process is paving, the current three-level process is the same as the previous three-level process, which means that the previous process is still being executed. At this time, keep the previous process unchanged.

[0108] If the current third-level process is rolling, that is, the current third-level process is different from the preceding third-level process, then determine whether the preceding third-level process is the last in its three-level process structure. If it is not the last, then use the current third-level process to update the preceding third-level process, and the preceding second-level process and the preceding first-level process remain unchanged. For example, in the preceding process [base layer, lower layer construction, paving], paving belongs to the first position in the three-level process structure [paving, rolling] corresponding to the lower layer construction, that is, it is not the last. At this time, use rolling to replace paving, and other processes remain unchanged. The updated preceding process is [base layer, lower layer construction, rolling].

[0109] If the preceding three-level process is the last at this time, for example, the preceding process is [pad layer, middle layer construction, rolling], and the current three-level process is measurement, then it is necessary to determine whether the preceding two-level process is the last in its process structure. If it is not the last, for example, the middle layer construction is the fourth in its process structure instead of the last, it can be seen that since rolling is the last action of the middle layer construction, measurement is actually the first action of the next construction stage. Therefore, in this case, the preceding two-level process is updated to the next-ranked process in its process structure, that is, updated to the upper-layer surveying and layout, and the current three-level process is used to update the preceding three-level process, and the preceding first-level process remains unchanged. At this time, the preceding process is changed to [pad layer, upper-layer surveying and layout, measurement].

[0110] If the preceding secondary process is the last in its process structure, for example, the preceding process is [bed layer, upper layer construction, rolling], and the current third-level process is measurement, similarly, the current third-level process is actually the first action of the next construction stage, and the next construction stage actually does not belong to the current construction target, but to the first construction stage of the next construction target. At this time, the preceding first-level process is updated to the first-level process of the next sequence in its process structure, the preceding second-level process is updated to the first-sequence second-level process in the second-level process structure corresponding to the preceding first-level process, and the preceding third-level process is updated to the current third-level process, that is, the updated preceding process is expressed as [bottom layer, bottom layer surveying and laying out, measurement]. Then, according to the updated preceding process, iterative judgment is continued, thereby realizing the automatic identification of the railway roadbed filling process. In this embodiment, based on the above-mentioned reasoning logic, a decision tree model is preferably used to construct a process reasoning model, thereby realizing accurate, timely and effective updating of the construction process.

[0111] The present embodiment provides a method for automatically identifying railway roadbed filling procedures. The present invention establishes a procedure structure of a hierarchical relationship of construction procedures based on the construction process of railway roadbed filling, determines the personnel and equipment at the construction site based on image recognition and data processing, and determines the current procedure by reasoning about the personnel and equipment on the basis of the procedure structure, and updates the preceding procedure according to the current procedure, thereby achieving accurate identification of the construction procedure, improving the construction management efficiency and project progress of railway roadbed filling, and improving the degree of unmanned management of the construction.

[0112] See also Figure 4 Based on the same inventive concept, a second embodiment of the present invention provides an automatic identification system for railway roadbed filling process, comprising:

[0113] A process structure construction module 10 is used to construct a process structure based on a hierarchical relationship of construction processes according to the construction data of railway roadbed filling, wherein the process structure includes a primary process structure, a secondary process structure and a tertiary process structure;

[0114] A process initialization module 20 is used to initialize the pre-process according to the construction progress and the process structure, wherein the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process;

[0115] The target recognition module 30 is used to collect the construction images of the railway roadbed filling, and perform target recognition on the construction images to obtain the recognized targets, wherein the recognized targets include construction personnel and construction equipment;

[0116] A state recognition module 40 is used to obtain equipment sensor data of the construction equipment, and perform data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain the equipment state of the construction equipment;

[0117] The process reasoning and updating module 50 is used to determine the current third-level process according to the identification target and the equipment status, and update the preceding process according to the current third-level process.

[0118] The technical features and technical effects of the railway roadbed filling process automatic identification system proposed in the embodiment of the present invention are the same as the method proposed in the embodiment of the present invention, and will not be repeated here. Each module in the above-mentioned railway roadbed filling process automatic identification system can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0119] In addition, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0120] See also Figure 5 , an internal structure diagram of a computer device in an embodiment, the computer device can specifically be a terminal or a server. The computer device includes a processor, a memory, a network interface, a display and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for automatically identifying a railway roadbed filling process is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0121] It can be understood by those skilled in the art that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have the same component arrangement.

[0122] In addition, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0123] In summary, an embodiment of the present invention proposes a method, system, device and storage medium for automatic identification of railway roadbed filling processes. The method constructs a process structure based on the hierarchical relationship of construction processes according to the construction data of railway roadbed filling, and the process structure includes a first-level process structure, a second-level process structure and a third-level process structure; initializes the preceding process according to the construction progress and the process structure, and the preceding process includes a first-level preceding process, a second-level preceding process and a third-level preceding process; collects construction images of railway roadbed filling, and performs target recognition on the construction images to obtain recognition targets, and the recognition targets include construction personnel and construction equipment; obtains equipment sensor data of the construction equipment, and performs data analysis on the equipment sensor data according to a pre-constructed autoencoder model to obtain the equipment status of the construction equipment; determines the current third-level process according to the recognition target and the equipment status, and updates the preceding process according to the current third-level process. The present invention establishes a process structure of a hierarchical relationship of construction processes based on the construction process of railway roadbed filling, determines the personnel and equipment at the construction site based on image recognition and data processing, and determines the current process by reasoning about the personnel and equipment on the basis of the process structure, and updates the preceding process according to the current process, thereby achieving accurate identification of the construction process, improving the construction management efficiency and project progress of railway roadbed filling, and improving the degree of unmanned management of the construction.

[0124] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above-mentioned embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The above-mentioned embodiments only express several preferred implementation modes of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in the technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be based on the protection scope of the claims.

Claims

1. A method for automatically identifying railway roadbed filling procedures, characterized in that: include: According to the construction data of railway roadbed filling, a process structure based on the hierarchical relationship of the construction process is constructed, and the process structure includes a primary process structure, a secondary process structure and a tertiary process structure; Initialize the pre-process according to the construction progress and the process structure, the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process; Collecting construction images of railway roadbed filling, and performing target recognition on the construction images to obtain recognized targets, wherein the recognized targets include construction personnel and construction equipment; Acquire equipment sensor data of the construction equipment, and perform data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain an equipment status of the construction equipment; The current third-level process is determined according to the identification target and the equipment status, and the preceding process is updated according to the current third-level process.

2. The method for automatically identifying railway roadbed filling process according to claim 1, characterized in that: The step of constructing a process structure based on the hierarchical relationship of construction processes according to the construction data of railway roadbed filling includes: Taking each construction object of railway roadbed filling as a primary process, taking the construction sequence of each construction object as a primary process sequence, and constructing a primary process structure according to the primary process and the primary process sequence; Taking each construction stage of the construction object as a secondary process, taking the stage sequence of each construction stage as a secondary process sequence, and constructing a secondary process structure according to the secondary process and the secondary process sequence; Taking each construction action included in the construction stage as a three-level process, taking the action sequence of each construction action as a three-level process sequence, and constructing a three-level process structure according to the three-level processes and the three-level process sequence; According to the hierarchical relationship among the primary process structure, the secondary process structure and the tertiary process structure, a process structure of the railway roadbed filling construction process is constructed.

3. The method for automatically identifying railway roadbed filling process according to claim 1, characterized in that: The step of acquiring the equipment sensor data of the construction equipment and performing data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain the equipment status of the construction equipment includes: According to the equipment type of each equipment in the construction area, a corresponding equipment sensor is pre-installed on each equipment, and the equipment sensor data includes positioning sensor data and state sensor data, and the state sensor data includes speed sensor data, acceleration sensor data and inclinometer sensor data; Acquire the equipment sensor data of each equipment, and filter out the state sensor data of the construction equipment according to the positioning sensor data of the construction equipment; Performing data analysis on the state sensor data of the construction equipment according to a pre-built autoencoder model to obtain the sensor state of each state sensor on the construction equipment, wherein the sensor state includes a normal state and an abnormal state; The equipment status of the construction equipment is obtained according to the number of sensors whose sensor status is normal.

4. The method for automatically identifying railway roadbed filling process according to claim 3, characterized in that: The step of performing data analysis on the state sensor data of the construction equipment according to the pre-built autoencoder model to obtain the sensor state of each state sensor on the construction equipment comprises: Inputting the state sensor data of the construction equipment into a trained autoencoder model, and calculating an error index between the reconstructed data output by the autoencoder model and preset normal sensor data; The error indicator is compared with a preset error threshold, and the sensor state of the corresponding state sensor is obtained according to the comparison result.

5. The method for automatically identifying railway roadbed filling process according to claim 4, characterized in that: The loss function of the autoencoder model is expressed as follows: The error index is expressed by the following formula: In the formula, represents the i-th reconstructed data, e i represents the i-th normal sensor data, and n represents the total number of data.

6. The method for automatically identifying railway roadbed filling process according to claim 3, characterized in that: The step of obtaining the equipment status of the construction equipment according to the number of sensors whose sensor status is normal includes: Determine whether the number of sensors in the normal state is greater than a threshold number. If so, determine that the equipment state of the construction equipment is a normal working state. Otherwise, determine that the equipment state of the construction equipment is a non-working state.

7. The method for automatically identifying railway roadbed filling process according to claim 2, characterized in that: The step of determining the current third-level process according to the identification target and the equipment state, and updating the preceding process according to the current third-level process includes: In response to the equipment being in a normal working state, performing process reasoning on the identified target to determine a current third-level process; Determine whether the current three-level process is the same as the previous three-level process, and if they are the same, keep the previous process unchanged; If they are different, determining whether the three-level process sequence of the preceding three-level process is the last of the three-level processes; If it is not the last one of the three-level processes, the preceding three-level process is updated according to the current three-level process, and the preceding two-level process and the preceding one-level process are kept unchanged; If it is the last of the third-level process, determine whether the second-level process sequence of the preceding second-level process is the last of the second-level process; If it is not the last of the secondary processes, the preceding third-level process is updated according to the current third-level process, and the preceding second-level process is updated to the second-level process of the next order according to the sequence of the secondary processes, and the preceding first-level process is kept unchanged; If it is the last second-level process, the preceding third-level process is updated according to the current third-level process, and the preceding first-level process is updated to the first-level process of the next priority according to the first-level process sequence, and the preceding second-level process is updated to the second-level process of the first priority according to the second-level process sequence corresponding to the updated preceding first-level process.

8. A railway roadbed filling process automatic identification system, characterized in that: include: A process structure building module is used to build a process structure based on the hierarchical relationship of the construction process according to the construction data of the railway roadbed filling, and the process structure includes a primary process structure, a secondary process structure and a tertiary process structure; A process initialization module is used to initialize the pre-process according to the construction progress and the process structure, wherein the pre-process includes a pre-level one process, a pre-level two process and a pre-level three process; A target recognition module is used to collect construction images of railway roadbed filling, and perform target recognition on the construction images to obtain recognized targets, wherein the recognized targets include construction personnel and construction equipment; A state recognition module is used to obtain equipment sensor data of the construction equipment, and perform data analysis on the equipment sensor data according to a pre-built autoencoder model to obtain the equipment state of the construction equipment; The process reasoning and updating module is used to determine the current third-level process according to the identification target and the equipment status, and to update the preceding process according to the current third-level process.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images