Concrete quality detection system and detection method based on machine vision

Abstract

The invention discloses a concrete slump real-time detection system and method based on machine vision. Video streams in the concrete discharging process are collected through a high-definition camera device installed at the front end of a concrete discharging opening; transmitting the video stream to a data center, analyzing the video stream in real time by using a pre-trained three-dimensional deep convolutional neural network model, extracting flow characteristics of the concrete mixture, and calculating a slump value; judging whether the slump numerical value exceeds a preset range or not, and if yes, sending alarm information to a central management platform through an alarm module; and adjusting the material ratio of the concrete by field operators according to the slump numerical value and the alarm information. According to the invention, the advanced deep learning algorithm is utilized to carry out real-time analysis on the concrete blanking process, the slump value is accurately calculated, and automatic and intelligent quality detection is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of concrete quality detection, and in particular to a machine vision-based concrete slump real-time detection system and detection method. Background Art

[0002] With the rapid development of the construction industry, concrete, as one of the most commonly used building materials, has a quality that directly affects the safety and service life of the project. Performance indicators such as concrete workability and fluidity are crucial to ensuring construction quality, among which slump is a key indicator for evaluating concrete fluidity and workability.

[0003] Traditionally, the determination of concrete slump mainly relies on manual sampling and testing, which requires manual operation by technicians on site and testing through a standard slump cylinder. This method has the following shortcomings:

[0004] 1. High labor intensity: The manual sampling and testing process is cumbersome, time-consuming and labor-intensive, especially in large-scale construction sites, and it is difficult to meet the testing needs in a timely manner.

[0005] 2. Inefficiency: Each test takes a long time and cannot achieve real-time monitoring of each batch of concrete, which may lead to delayed discovery of quality problems.

[0006] 3. Limited accuracy: Human factors have a greater impact, and different operators may produce measurement errors, resulting in inconsistent results.

[0007] 4. Unable to conduct full coverage inspection: Due to the limitation of manpower and time, only random inspection can be carried out, and it is impossible to achieve comprehensive monitoring of all concrete output, which poses a quality risk.

[0008] With the rapid development of artificial intelligence and machine vision technology, it is possible to use image processing and deep learning algorithms to perform real-time and automated detection of concrete slump. Some studies have tried to collect images of the concrete flow process through cameras and analyze them in combination with image processing algorithms. However, current research and applications still have the following problems:

[0009] 1. Insufficient algorithm accuracy: Traditional image processing algorithms are difficult to accurately extract concrete flow characteristics. They are affected by factors such as lighting and background, and the reliability of detection results is not high.

[0010] 2. Poor real-time performance: Some algorithms are computationally complex and cannot meet the requirements of real-time detection, affecting on-site applications.

[0011] 3. Lack of system integration: Most of the existing solutions are experimental research, lacking complete system integration and engineering applications, and are difficult to promote in actual production.

[0012] Therefore, there is an urgent need for a concrete quality detection system based on machine vision and deep learning technology that can detect the slump of concrete in real time and accurately, implement one inspection for each vehicle, improve the quality control level, and meet the needs of modern construction. Summary of the invention

[0013] In order to solve the above technical problems, the present invention provides a concrete quality detection system and detection method based on machine vision, which uses advanced deep learning algorithms to perform real-time analysis of the concrete feeding process, accurately calculate the slump value, and realize automated and intelligent quality detection.

[0014] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0015] In a first aspect, the present invention provides a concrete quality detection system based on machine vision, comprising:

[0016] The front-end high-definition camera device is installed at the concrete discharge port to collect the video stream during the concrete discharge process;

[0017] A data transmission module, used for transmitting the video stream to a data center via a dedicated network;

[0018] Data center intelligent analysis system, including:

[0019] a. A video receiving module, used for receiving the video stream;

[0020] b. an algorithm analysis module, using a pre-trained three-dimensional deep convolutional neural network model to perform real-time analysis on the video stream and extract flow characteristics of the concrete mixture;

[0021] c. a slump calculation module, which calculates the slump value of concrete according to the flow characteristics;

[0022] A result display module, used to display the slump value and related video images in real time;

[0023] An alarm module sends an alarm message to a central management platform when the slump value exceeds a preset range;

[0024] The parameter adjustment module allows on-site operators to adjust the material ratio of concrete according to the slump value and alarm information.

[0025] Preferably, the front-end high-definition camera device is a network camera with 8 to 80 million pixels.

[0026] Preferably, the algorithm analysis module adopts an algorithm model based on a three-dimensional deep convolutional neural network, and the model is trained based on a large number of concrete feeding videos and corresponding slump data.

[0027] Preferably, the result display module includes:

[0028] Real-time value display unit, used to display the real-time slump value of the current mixture;

[0029] The historical value display unit is used to display the final slump values ​​of the current batch and the previous batch;

[0030] Project information display unit, used to display the project information, grade and equipment moisture content of the concrete currently being produced;

[0031] The equipment status display unit is used to display the real-time current information and aggregate grading information of the mixer.

[0032] Preferably, the detection response time of the data center intelligent analysis system does not exceed 3 seconds.

[0033] In a second aspect, the present invention provides a concrete quality detection method based on machine vision, comprising the following steps:

[0034] Video acquisition: The video stream of the concrete discharging process is collected through the front-end high-definition camera installed at the concrete discharging port;

[0035] Data transmission: transmitting the video stream to a data center via a dedicated network;

[0036] Intelligent analysis: In the data center intelligent analysis system, a pre-trained three-dimensional deep convolutional neural network model is used to perform real-time analysis on the video stream, extract the flow characteristics of the concrete mixture, and calculate the slump value;

[0037] Result display: the slump value and related video images are displayed in real time in the result display module;

[0038] Abnormal alarm: determine whether the slump value exceeds the preset range. If so, send an alarm message to the central management platform through the alarm module;

[0039] Parameter adjustment: On-site operators adjust the material ratio of concrete according to the slump value and alarm information.

[0040] Preferably, in step 3, the detection response time of the data center intelligent analysis system does not exceed 3 seconds.

[0041] Preferably, the detection rate of the slump test is higher than 90%, and the detection accuracy deviation does not exceed 10%.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. Strong real-time performance: The system can complete slump detection within a response time of no more than 3 seconds and realize real-time monitoring.

[0044] 2. High accuracy: Through advanced deep learning algorithms, the detection rate of slump detection is higher than 90%, and the detection accuracy deviation does not exceed 10%.

[0045] 3. High degree of automation: No human intervention is required, and detection and alarm are automatically completed, which reduces labor costs and improves work efficiency.

[0046] 4. High system integration: It provides a complete software and hardware solution that is easy to deploy and apply in actual production.

[0047] 5. Strong scalability: The system architecture is flexible, and the number of cameras and detection items can be increased according to demand to adapt to production environments of different scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Schematic diagram of the topological structure of the system of the present invention. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] Example 1

[0051] System deployment

[0052] In a concrete production plant, the concrete quality detection system of the present invention is installed. The specific steps are as follows:

[0053] Front-end equipment installation: Install an 8-megapixel high-definition network camera above the concrete discharge port. The camera is firmly fixed and the viewing angle covers the entire discharge process.

[0054] Network connection: The camera is connected to the data center via a dedicated network line to ensure stable transmission of the video stream.

[0055] Data center configuration: deploy an intelligent analysis system in the data center, including a video receiving module, an algorithm analysis module, and a slump calculation module. The algorithm model is pre-trained with a large number of concrete feeding videos and corresponding slump data.

[0056] Result display: Install a display terminal in the on-site operation room, run the result display module, and display the test results and video images in real time.

[0057] System operation: When concrete begins to be discharged, the camera collects video streams in real time and transmits them to the data center through the network. The intelligent analysis system processes the video streams in real time, extracts the flow characteristics of concrete, and calculates the current slump value. The test results are updated in real time on the display terminal, including the current real-time value, the final value of this time, and the last value. If the slump value exceeds the preset range (for example, deviates from the target value by ±10%), the system will issue an alarm through the alarm module, highlight the prompt on the display screen, and send an alarm message to the central management platform.

[0058] On-site operators adjust material ratios in a timely manner based on test results and alarm information to ensure that concrete quality meets requirements.

[0059] Implementation effect:

[0060] 1. Real-time detection of each batch of concrete is realized to avoid the accumulation of quality problems.

[0061] 2. Improved production efficiency and reduced the workload and errors of manual inspection.

[0062] 3. Improved the stability of product quality.

[0063] Example 2

[0064] Based on the above embodiment 1, the monitoring of multiple production lines is added. Specific steps:

[0065] Add cameras: Install high-definition cameras at the discharge ports of the other two concrete production lines.

[0066] Network configuration: Connect the new cameras to a dedicated network to ensure bandwidth and stability of the video stream.

[0067] System adjustment: In the intelligent analysis system of the data center, increase the video stream processing of newly added cameras to ensure parallel analysis of multiple videos.

[0068] Interface optimization: In the result display module, the display of multiple production lines is added, and the information of each production line is displayed independently, including slump value, video screen, project information, etc.

[0069] System operation: The system can simultaneously detect and analyze the concrete feeding process of the three production lines in real time. When any production line has abnormal slump, the system can accurately locate and issue an alarm, and the operator can make targeted adjustments.

[0070] Implementation Effect

[0071] 1. Expanded the monitoring scope of the system and improved the overall production quality management level.

[0072] 2. Real-time analysis of multiple video channels demonstrates the high performance and scalability of the system.

[0073] Example 3

[0074] The algorithm model is optimized for special situations that occur during the production process:

[0075] Data collection: Collect concrete unloading videos under different weather and lighting conditions to enrich the training data set.

[0076] Model training: Use a deeper three-dimensional convolutional neural network model to improve adaptability to complex scenarios.

[0077] Parameter adjustment: Adjust the model’s hyperparameters, optimize the learning rate, batch size, etc., to improve the model’s generalization ability.

[0078] System operation: The optimized algorithm model is deployed in the intelligent analysis system to process the video stream in real time. Under complex lighting conditions (such as cloudy days and insufficient lighting at night), the system can still accurately detect the slump value.

[0079] Implementation effect:

[0080] 1. The detection accuracy of the system is further improved and its adaptability is enhanced.

[0081] 2. Reduce the impact of the external environment on the test results and improve the stability of the system.

[0082] like Figure 1 As shown:

[0083] An 8-megapixel high-definition network camera is installed at the front end, and a dedicated network is used to transmit data to the data center. The central intelligent analysis system obtains the front-end video stream, performs real-time slump analysis, and displays the detection values ​​to the on-site operators through the application page. When an abnormal slump occurs, the data is transmitted to the central management platform and an alarm is issued. On-site personnel can make corresponding adjustments to the concrete material ratio in a timely manner as needed to prevent quality problems. The system provided by the present invention can greatly improve the detection efficiency, reduce the use of manpower and material resources, and is very suitable for actual industrial production processes.

[0084] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered within the protection scope of the present invention.

Claims

1. A concrete quality detection system based on machine vision, characterized in that: include: The front-end high-definition camera device is installed at the concrete discharge port to collect the video stream during the concrete discharge process; A data transmission module, used for transmitting the video stream to a data center via a dedicated network; Data center intelligent analysis system, including: a. A video receiving module, used for receiving the video stream; b. an algorithm analysis module, using a pre-trained three-dimensional deep convolutional neural network model to perform real-time analysis on the video stream and extract flow characteristics of the concrete mixture; c. a slump calculation module, which calculates the slump value of concrete according to the flow characteristics; A result display module, used to display the slump value and related video images in real time; An alarm module sends an alarm message to a central management platform when the slump value exceeds a preset range; The parameter adjustment module allows on-site operators to adjust the material ratio of concrete according to the slump value and alarm information.

2. The detection system according to claim 1, characterized in that: The front-end high-definition camera device is a network camera with 8 to 80 million pixels.

3. The detection system according to claim 1, characterized in that: The algorithm analysis module adopts an algorithm model based on a three-dimensional deep convolutional neural network, and the model is trained based on a large number of concrete feeding videos and corresponding slump data.

4. The detection system according to claim 1, characterized in that: The result display module includes: Real-time value display unit, used to display the real-time slump value of the current mixture; The historical value display unit is used to display the final slump values ​​of the current batch and the previous batch; Project information display unit, used to display the project information, grade and equipment moisture content of the concrete currently being produced; The equipment status display unit is used to display the real-time current information and aggregate grading information of the mixer.

5. The detection system according to claim 1, characterized in that: The detection response time of the data center intelligent analysis system does not exceed 3 seconds.

6. A concrete quality detection method based on machine vision, characterized in that: The detection system according to any one of claims 1 to 5 is used; specifically comprising the following steps: Step 1: Collect the video stream of the concrete discharging process through the front-end high-definition camera device installed at the concrete discharging port; Step 2: Transmit the video stream to a data center via a dedicated network; Step 3: In the data center intelligent analysis system, a pre-trained three-dimensional deep convolutional neural network model is used to perform real-time analysis on the video stream, extract the flow characteristics of the concrete mixture, and calculate the slump value; Step 4: Displaying the slump value and related video images in real time in the result display module; Step 5: Determine whether the slump value exceeds a preset range. If so, send an alarm message to a central management platform through an alarm module; Step 6: On-site operators adjust the material ratio of concrete according to the slump value and alarm information.

7. The detection method according to claim 6, characterized in that: In step 3, the detection response time of the data center intelligent analysis system does not exceed 3 seconds.

8. The detection method according to claim 6, characterized in that: The detection rate of the slump test is higher than 90%, and the detection accuracy deviation does not exceed 10%.

Images