CN-122008410-A - Vehicle-mounted remote intelligent quality control system and device for concrete mixer

CN122008410ACN 122008410 ACN122008410 ACN 122008410ACN-122008410-A

Abstract

The invention discloses a vehicle-mounted remote intelligent quality control system and device for a concrete mixer truck, and relates to the field of concrete mixer trucks; according to the vehicle-mounted remote intelligent quality control system and device for the concrete mixer truck, the concrete transportation process is changed from the traditional black box period into full-flow visual and interveneable transparentization management and control. The technical staff does not need to visit the site, and can check the high-definition video in each mixer truck tank, the slump value and the vehicle position dynamic of automatic analysis in real time through the mobile phone APP or the PC end. When the concrete state deviates from the preset standard, the system automatically triggers early warning, supports operations such as remote accurate addition of additives, start of lens cleaning and the like, and thoroughly replaces the traditional lag mode of manual climbing observation and experience judgment. The whole flow data is automatically uploaded to the cloud end, a non-tamperable quality file is built, a power assisting technician realizes the whole process from delivery to pouring, quality tracing and responsibility definition are achieved, and stability and reliability of concrete delivery quality are remarkably improved.

Inventors

LIN JINLIANG
LIN YONGJI
YANG CHAOYANG
LIN JIASHENG

Assignees

星迈动力科技(广州)有限公司

Dates

Publication Date: 20260512
Application Date: 20260323

Claims (10)

1. On-vehicle remote intelligent quality control device of concrete mixer truck, its characterized in that includes: the controller comprises a data processor, a communication unit and a storage unit and is used for receiving video data, analyzing concrete slump, generating control instructions and uploading the video data to the cloud; The in-tank monitor is arranged on the stirring tank of the stirring truck and is used for acquiring video data in the stirring tank; The additive adder comprises a plurality of additive storage boxes arranged on the mixer truck, wherein a submersible pump, a liquid level meter and a dosage control unit are arranged in the additive storage boxes, and the additive storage boxes are used for adding additives into the mixer truck according to control instructions; The cloud server is deployed with a concrete slump analysis model and is used for analyzing the video data in real time and feeding back the result; The terminal application equipment comprises a mobile phone APP and a PC end and is used for real-time monitoring, remote control and data tracing.
2. The concrete mixer truck-mounted remote intelligent quality control device according to claim 1, wherein the in-tank monitor comprises a mounting shell, a camera, an illuminating lamp and a telescopic driving piece, the mounting shell is mounted on a mixer truck stirring tank, the telescopic driving piece is mounted inside the mounting shell, the camera and the illuminating lamp are fixedly mounted on a base, the base is mounted on a transmission end of the telescopic driving piece and used for enabling the camera and the illuminating lamp to stretch into and out of the lower end of the mounting shell, and the camera and the illuminating lamp face the mixer truck tank after protruding out of the mounting shell.
3. The remote intelligent quality control device for the concrete mixer truck according to claim 2, wherein the mounting shell is further provided with a self-cleaning assembly, the self-cleaning assembly comprises a water conduit arranged in the mounting shell and a flushing hole arranged at the end part of the water conduit, the flushing hole faces the camera, the illuminating lamp is a multicolor LED light source, and the brightness and the color of the illuminating lamp are remotely adjusted through a relay module and a PWM dimmer.
4. The on-vehicle remote intelligent quality control device of concrete mixer truck according to claim 2, wherein at least three additive storage tanks are provided, each tank is provided with an independent liquid outlet pipeline and an electromagnetic valve, and the control module automatically switches tank pump liquid according to control instructions.
5. The vehicle-mounted remote intelligent quality control device of the concrete mixer truck according to claim 1, wherein the controller is further integrated with a Beidou positioning module and a vehicle-mounted video host, and supports 8 paths of video access, voice broadcasting and task list management.
6. A concrete mixer truck-mounted remote intelligent quality control system, characterized in that the system is adapted for a concrete mixer truck-mounted remote intelligent quality control device according to any one of claims 1-5, the system comprising: The in-tank monitoring module is used for acquiring real-time video stream data of the concrete surface in the stirring tank; The motion sensing module is used for collecting the motion parameters of the stirring tank including the rotation speed and the roll angle in real time; The external additive module is used for adding the external additive in the corresponding additive storage box into the external additive with required dosage according to the control instruction; The control module is used for collecting video stream data, motion parameters and liquid level data in real time, uploading the video stream data, the motion parameters and the liquid level data to the cloud server, receiving a control instruction from the cloud, and driving corresponding hardware to execute after analysis; the cloud server module receives the video stream data, the motion parameters and the liquid level data uploaded by the control module, inputs the video stream data, the motion parameters and the liquid level data into a concrete slump analysis model, and calculates a concrete slump value and a confidence coefficient in real time; And the terminal application module is used for carrying out real-time monitoring, remote control, alarm management, data tracing and task management on the system.
7. The system of claim 6, further comprising a self-cleaning module, wherein when the system detects that the sharpness of the video stream data obtained by the in-tank monitoring module is below a predetermined threshold, the system starts the water conduit to direct the water flow to the flushing hole for directional flushing of the camera.
8. The concrete mixer truck-mounted remote intelligent quality control system of claim 6, wherein the system further comprises: The positioning and scheduling module is used for acquiring longitude, latitude, speed and direction of the vehicle in real time, dynamically displaying the position and track of the vehicle on a map, combining with automatic triggering state switching of the electronic fence, linking with a scheduling system and automatically loading current task list information; And the data tracing and auditing module is used for generating a tamper-proof log record for all key operations, uploading the log record to the cloud end, providing a full life cycle quality file, and supporting quality accident responsibility definition and process optimization analysis.
9. The on-board remote intelligent control system for a concrete mixer truck of claim 6, wherein the calculating step of the concrete slump analysis model comprises: Taking video stream data and motion parameters as input data, wherein the motion parameters comprise stirring cylinder rotating speed rpm and stirring cylinder rolling angle theta, aligning all data streams with the same time stamp, extracting a frame every 0.2s of a video frame, and interpolating and synchronizing the motion parameters according to the same time stamp to form a paired time sequence; The method comprises the steps of selecting continuous T=16 frames of images in a sliding window mode as an input of one reasoning, setting the sliding step length of adjacent windows to be 1 frame or 8 frames, scaling each frame of images from original resolution to 224×224 pixels with uniform size, keeping the aspect ratio unchanged, filling the redundant part with black, converting RGB images into floating point tensors, normalizing to [0,1] intervals, carrying out random brightness/contrast adjustment, gaussian noise and random shielding on the images during training, scaling the rotation speed rpm by the maximum rotation speed to [0,1], encoding the roll angle theta into two-dimensional vectors [ sin theta, cos theta ] through sine and cosine, splicing the normalized rotation speed and angle codes to obtain a motion parameter vector m epsilon R3, stacking T frames of images into tensors with the shape of (T, 3,224,224), taking m as independent input, and the shape of (3,) or expanding the T frames into batch dimensions; Sequentially executing a spatial feature extraction module, a time modeling module, a fluid mask prediction module, a motion parameter fusion module and a slump regression module in a model architecture, inputting a video frame tensor X epsilon R T×3×224×224 into the spatial feature extraction module to obtain a spatial feature map sequence { F1, F2, & gt, FT } of a T frame, inputting the spatial feature map sequence into the space-time modeling module to obtain a space-time feature Z, inputting the space-time feature Z into the fluid mask prediction module to obtain a fluid mask sequence { M1, & gt, MT } and inputting the space-time feature Z and a motion parameter vector M into the motion parameter fusion module to obtain a fusion feature h, and inputting the fusion feature h into the slump regression module to obtain a slump predicted value s pred and a confidence coefficient c; smoothing the slump value continuously inferred and output by adopting an Exponential Moving Average (EMA): wherein a=0.3, The initial value is the first predicted value, the prediction confidence coefficient c <0.7 is marked as low confidence coefficient, and the final slump value is the smoothed slump value The confidence level is high/medium/low, or the confidence value is directly output, and the visualized image of the fluid mask is also output.
10. The on-board remote intelligent control system for a concrete mixer truck according to claim 1, wherein: inputting a video frame tensor X epsilon R T×3×224×224 into a spatial feature extraction module, independently processing each frame, dividing an image into 14 multiplied by 14 patches, obtaining patch tokens and dimension 196 multiplied by 96 through linear projection, and finally outputting a spatial feature map Ft epsilon R14 multiplied by 256 corresponding to a T frame through a Swin transducer block of 4 stages to obtain a spatial feature map sequence { F1, F2, the. Inputting the space feature map sequence into a space-time modeling module, adding a learnable space position code and a time index code to form an input token sequence Z (0) ∈R (T×196)×256 , wherein the space-time modeling module comprises L=8 layers, each layer sequentially executes a space self-attention mechanism and a time attention mechanism, carries out self-attention calculation on 196 tokens in each frame, captures a space relationship in the frame, carries out self-attention calculation on tokens in the same space position in different frames, captures time sequence change, and finally outputs enhanced space-time features Z epsilon R (T×196)×256 ; The space-time feature Z is input into a fluid mask prediction module, the feature Z is remodeled into T multiplied by 14 multiplied by 256, each layer is connected with ReLU activation and BatchNorm by a transposed convolution decoder, finally, the fluid mask Mt E [0,1] 56×56 of each frame is output by convolution layer and sigmoid activation, the probability that each pixel belongs to a flowing area is represented, and a fluid mask sequence { M1,.. Inputting a space-time feature Z and a motion parameter vector m into a motion parameter fusion module, carrying out global average pooling on the space-time feature Z along space and time dimensions to obtain a global feature vector Z global ∈R 256 , mapping the motion parameter m to the dimension identical to the global feature through two layers of MLP to obtain m proj ∈R 256 , and splicing Z global and m proj to obtain a fusion feature vector h epsilon R 512 to obtain a fusion feature h; And inputting the fusion characteristic h into a slump regression module, connecting with a ReLU activation and a Dropout through a three-layer full-connection network, outputting a slump predicted value s pred epsilon R at the last layer, branching a single-node sigmoid layer from the front of the last layer full-connection, and outputting a confidence degree c epsilon [0,1] to obtain a slump predicted value s pred and a confidence degree c.

Description

Vehicle-mounted remote intelligent quality control system and device for concrete mixer Technical Field The invention relates to a concrete mixer truck technology, in particular to a vehicle-mounted remote intelligent quality control system and device of a concrete mixer truck. Background In recent years, with the increasing demand of the construction industry for concrete quality, a concrete mixer truck is used as a transportation core device, and quality control in the transportation process of the concrete mixer truck is the focus of industry. At present, the existing mixer truck mainly relies on manual experience or simple camera monitoring to evaluate the concrete state in the tank during transportation, but the traditional camera is extremely easy to be shielded by splashed concrete slurry and is exposed in a complex and severe construction site environment for a long time, so that the fault rate is high, the maintenance is difficult, and key quality parameters such as concrete slump and the like cannot be accurately obtained in real time. When the concrete is segregated or dried due to overlong transportation time or environmental factors, a driver often has difficulty in finding out in time, lacks a precise online regulation and control means, and can only wait for the concrete to reach a construction site and then be processed by field technicians, so that the 'post-remediation' mode is easy to cause the concrete to be scrapped, and the situation that each construction site is provided with a full-time technician residence is directly caused, and the labor cost is greatly increased. Meanwhile, a field technician needs to repeatedly climb the mixer truck in the seasoning process, and serious personal safety risks exist on sites with dense vehicles and complex environments. In addition, the whole transportation process lacks an effective data recording and tracing mechanism, information of each link is isolated, once quality problems occur, retrospective analysis is difficult, and difficulty is brought to engineering quality management and responsibility definition. From the above, no matter the field technician or the driver can effectively control the concrete state in the transportation process, a set of intelligent equipment system capable of meeting the accurate control requirement of complex operation scenes is required to be designed aiming at the problem, and the power assisting technician can effectively control the quality of the transported concrete in the whole process. Disclosure of Invention The invention aims to provide a vehicle-mounted remote intelligent quality control system and device for a concrete mixer truck, which are used for solving the problem that technicians cannot effectively control the quality of concrete loaded and transported in a mixer truck tank in the transportation process. In order to achieve the aim, the invention provides the technical scheme that the vehicle-mounted remote intelligent quality control device of the concrete mixer comprises the following components: the controller comprises a data processor, a communication unit and a storage unit and is used for receiving video data, analyzing concrete slump, generating control instructions and uploading the video data to the cloud; The in-tank monitor is arranged on the stirring tank of the stirring truck and is used for acquiring video data in the stirring tank; The additive adder comprises a plurality of additive storage boxes arranged on the mixer truck, wherein a submersible pump, a liquid level meter and a dosage control unit are arranged in the additive storage boxes, and the additive storage boxes are used for adding additives into the mixer truck according to control instructions; The cloud server is deployed with a concrete slump analysis model and is used for analyzing the video data in real time and feeding back the result; The terminal application equipment comprises a mobile phone APP and a PC end and is used for real-time monitoring, remote control and data tracing. Preferably, the in-tank monitor comprises a mounting shell, a camera, an illuminating lamp and a telescopic driving piece, wherein the mounting shell is arranged on a stirring tank of the stirring truck, the telescopic driving piece is arranged inside the mounting shell, the camera and the illuminating lamp are fixedly arranged on a base, the base is arranged on a transmission end of the telescopic driving piece and used for realizing telescopic access of the camera and the illuminating lamp to the lower end of the mounting shell, and the camera and the illuminating lamp extend out of the mounting shell and then face the stirring truck tank. Preferably, the mounting shell is further provided with a self-cleaning assembly, the self-cleaning assembly comprises a water conduit arranged in the mounting shell and a flushing hole arranged at the end part of the water conduit, the flushing hole faces the camera, the illuminating lamp is a m