Concrete vibrating control method based on neural network and concrete vibrator

Abstract

This invention discloses a concrete vibration control method and a concrete vibrator based on a neural network, comprising: a temperature sensor, a triaxial acceleration sensor, and a rotation speed sensor installed inside the vibrating rod of the concrete vibrator, used to collect temperature signals, triaxial acceleration signals, and rotation speed signals, respectively; constructing multi-channel time-series data based on the temperature signals, triaxial acceleration signals, and rotation speed signals, and labeling the corresponding initial vibration stage, near-completion vibration stage, or completion vibration stage as sample data; training a neural network model based on the sample data to distinguish the concrete vibration stage based on the real-time collected multi-channel time-series data; establishing rotation speed ranges corresponding to different vibration stages; and the processor performing real-time vibration control based on the discrimination results of the neural network model; and forcibly stopping the machine if the real-time collected temperature value exceeds the preset stop temperature.

Description

Technical Field

[0001] This invention relates to the field of concrete construction vibration technology, specifically to a concrete vibration control method and a concrete vibrator based on neural networks. Background Technology

[0002] During concrete pouring and forming, vibration is crucial for removing air bubbles from the mixture, rearranging aggregates and mortar, and increasing density. It's a key process for ensuring structural strength, appearance quality, and durability. However, on construction sites, the insertion depth, spacing, and vibration duration of vibrators are often manually controlled. Influenced by factors such as slump fluctuations, aggregate gradation, reinforcement density, and operator posture, the vibration process is prone to problems like under-vibration, missed vibration, or over-vibration. Under-vibration or missed vibration can lead to voids, slag inclusions, and insufficient density, while over-vibration can cause segregation, bleeding, and other quality issues. Because the internal state of concrete is not visible, operators typically rely on experience and surface observations to judge whether vibration is sufficient, making standardized, quantifiable, and traceable process control difficult, thus leading to rework costs and quality risks.

[0003] To address the aforementioned issues, existing technologies have proposed vibratory compactor solutions with monitoring and alerting functions. For example, publication number CN116446656A discloses an automated concrete vibratory compactor capable of monitoring. This vibratory compactor is equipped with vibration sensors on the vibratory head to detect motion parameters such as amplitude and frequency. A processor receives these motion parameters and, in conjunction with pre-input parameters such as concrete slump and strength, calculates the ideal damping coefficient for achieving compaction. During vibration, the real-time damping coefficient is calculated from the sensor signals and compared with the ideal damping coefficient to determine the vibration status. The operator is alerted via indicator lights or other means. Furthermore, this solution includes a hose reel mechanism to facilitate hose lowering, pulling, and retraction, improving operational convenience.

[0004] However, the above-mentioned solutions still mainly rely on characteristic quantity calculation, comparison and judgment, and prompt output. Their judgment mechanism depends on comparing the ideal damping coefficient with the real-time damping coefficient, and requires input of parameters such as slump and strength before operation. At the execution level, they primarily use indicator lights to remind the operator, who then decides to continue or stop vibration. This makes it difficult to adjust vibration output parameters such as motor speed and vibration velocity in real time, thus hindering the formation of closed-loop control based on the vibration state. In contrast, if the vibration state could be automatically judged using a neural network based on the real-time signal characteristics of the vibration process, and the judgment results directly used for adaptive adjustment of the servo motor speed, a higher speed could be maintained in the early stages of vibration to improve air removal and compaction efficiency. As vibration nears completion, the speed could be automatically reduced to suppress the risk of segregation and bleeding caused by over-vibration. Upon reaching the completion state, the machine could automatically stop and output a completion prompt, thereby reducing reliance on human experience and improving the consistency and controllability of vibration quality under different working conditions. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing concrete vibration operations, such as reliance on manual experience, difficulty in quantitatively judging the vibration state, susceptibility to under-vibration or over-vibration, and difficulty in achieving process control. This invention provides a concrete vibration control method and a concrete vibrator based on a neural network, which can identify and judge the vibration state in real time and automatically adjust the rotation speed of the eccentric mechanism inside the concrete vibrator according to the judgment result, thereby realizing closed-loop control of the vibration process and automatic completion prompts.

[0006] The objective of this invention is achieved through the following technical solution: a concrete vibration control method based on neural networks, comprising: The vibrator in the concrete vibrator is equipped with a temperature sensor, a triaxial acceleration sensor and a rotation speed sensor, which are used to collect temperature signals, triaxial acceleration signals and rotation speed signals, respectively. Multi-channel time-series data were constructed based on temperature signals, triaxial acceleration signals, and rotational speed signals, and the corresponding initial stage of vibration, near-complete stage of vibration, or complete stage of vibration were labeled as sample data. A neural network model trained based on sample data is used to identify the concrete vibration stage based on real-time acquired multi-channel time-series data. The processor establishes rotation speed ranges corresponding to different vibration stages and performs vibration control in real time based on the judgment results of the neural network model; if the real-time collected temperature value exceeds the preset shutdown temperature, it will be forcibly stopped.

[0007] Furthermore, the neural network model is the PatchMTSC neural network model. The PatchMTSC neural network model first divides the time-series signal into multiple continuous patch segments along the time dimension using the Padding & Patching module, and then uses the built-in Transformer encoder to encode the features of each patch segment. Next, the built-in channel mixing module models the coupling relationship between different sensor channels such as temperature, acceleration, and rotation speed, and the time-series feature extraction module extracts the dynamic evolution features of each patch segment. Finally, the classification layer outputs the determination result of the vibration stage to which the current vibration process belongs.

[0008] Furthermore, the rotation speed ranges corresponding to different vibration stages are established, including: the initial vibration stage corresponds to any value of 7200–12000 r / min for the rotation speed of the eccentric mechanism inside the concrete vibrator; the near completion stage corresponds to any value of 4800–7200 r / min for the rotation speed of the eccentric mechanism inside the concrete vibrator; and the completion stage corresponds to a rotation speed of 0 for the rotation speed of the eccentric mechanism inside the concrete vibrator.

[0009] The present invention also provides a concrete vibrator based on neural network-based intelligent judgment of vibration state, for implementing the above method; the concrete vibrator includes a control module, a drive module and a vibrating rod; a sensing chamber is provided inside the vibrating rod, and a temperature sensor, a triaxial acceleration sensor and a rotation speed sensor are installed inside the sensing chamber. The control module receives data from temperature sensors, triaxial accelerometers, and speed sensors, uses a trained neural network model to determine the current vibration stage, and then combines the processor to generate corresponding drive control commands based on the determination results of the neural network model. The drive module controls the excitation intensity of the vibrator according to the drive control command.

[0010] Furthermore, the drive module includes a servo driver, a servo motor, and a flexible transmission shaft; the vibrating bar is provided with an eccentric mechanism, which is rotatably installed inside the vibrating bar and connected to one end of the flexible transmission shaft, the other end of the flexible transmission shaft is connected to the servo motor, the servo motor is electrically connected to the servo driver, and the servo driver is electrically connected to the main control processor board. The control module includes a main control processor board, a storage module, a signal acquisition and conditioning module, a neural network inference acceleration module, a Bluetooth signal receiving module, and a Bluetooth signal transmitting module. The Bluetooth signal transmitting module is connected to a temperature sensor, a triaxial accelerometer, and a speed sensor via signal transmission lines. It receives sensor data (data acquired by the temperature sensor, triaxial accelerometer, and speed sensor) and sends it to the Bluetooth signal receiving module. The Bluetooth signal receiving module transmits the sensor data to the signal acquisition and conditioning module for synchronous acquisition, filtering, and conditioning of the sensor data, generating multi-channel time-series data. Based on the multi-channel time-series data, the main control processor board calls the neural network inference acceleration module to perform real-time inference calculations. Using the multi-channel time-series data as input to a trained neural network model, it determines the current vibration stage and outputs the vibration stage determination result. The main control processor board generates corresponding servo control commands based on the vibration stage determination result and outputs them to the servo driver. The servo driver adjusts the speed of the servo motor in real time, thereby changing the excitation speed of the eccentric mechanism and achieving vibration control. The storage module is used to store key state parameters, judgment results, servo control commands, and completion records of the vibration process.

[0011] Furthermore, it also includes a backpack chassis, in which the main control processor board, storage module, signal acquisition and conditioning module, neural network inference acceleration module, Bluetooth signal receiving module and servo driver are installed; the servo motor is installed on the backpack chassis.

[0012] Furthermore, it also includes a flexible transmission hose and a power and communication module mounting compartment; the power and communication module mounting compartment is fixed to the flexible transmission hose by a fixing bayonet, and a second power supply and a Bluetooth signal transmitting module are installed inside the power and communication module mounting compartment. The second power supply powers the sensors in the sensing compartment through wires; the flexible transmission hose is used to wrap the wires, flexible transmission shaft and signal transmission line.

[0013] Furthermore, it also includes a display screen, which is connected to the main control processor board via a data transmission line, for displaying vibration frequency, load level, rotation speed information, temperature, and a prompt message indicating whether vibration is complete; wherein, the load level is calculated from the operating parameters of the servo motor or the characteristics of the sensor signal.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described neural network-based concrete vibration control method.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described neural network-based concrete vibration control method.

[0016] Compared with existing technologies, the advantages of this invention are as follows: This invention collects key state signals of the vibration process in real time through sensors and uses a neural network to judge the vibration state online. The judgment result is directly used for adaptive adjustment of the servo motor speed, forming a closed-loop control chain from state perception, intelligent judgment to speed adjustment, transforming the vibration process from experience-driven to data-driven and automatic control. By maintaining a high speed in the early stage of vibration to improve compaction efficiency and automatically reducing the speed near completion to suppress the risk of segregation and bleeding caused by over-vibration, the consistency of vibration quality and construction controllability can be significantly improved, reducing reliance on operator experience. Automatic shutdown and completion notification after judgment can reduce invalid vibration time and human error, improving construction efficiency and safety. At the same time, the integrated layout of the backpack chassis facilitates portability and on-site deployment, and the display screen provides intuitive process status feedback, facilitating standardized operations and quality management. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the overall structure of an adaptive speed-regulating concrete vibrator based on intelligent discrimination.

[0019] Figure 2 Detailed drawing of the power supply and communication module installation compartment.

[0020] Figure 3 This is a schematic diagram of signal transmission between the backpack chassis and the vibrator.

[0021] Figure 4 This is a schematic diagram of the internal structure of the vibrator.

[0022] Figure 5 This is a schematic diagram of the display screen interface.

[0023] In the diagram: 1-Display screen; 2-Data transmission line; 3-Backpack chassis; 4-Servo motor; 5-Fixing bayonet; 6-Power supply and communication module mounting compartment; 7-Flexible transmission hose; 8-Vibration rod; 31-Interface connection board; 32-Main control processor board; 33-Storage module; 34-Signal acquisition and conditioning module; 35-Neural network inference acceleration module; 36-Bluetooth signal receiving module; 37-First power supply; 38-Servo driver; 61-Second power supply; 62-Bluetooth signal transmitting module; 63-Signal transmission line; 81-Flexible drive shaft; 82-Sensor compartment; 83-Eccentric mechanism; 84-Fixing support; 821-Temperature sensor; 822-Triaxial accelerometer; 823-Speed ​​sensor. Detailed Implementation

[0024] The present invention will now be described in detail with reference to the accompanying drawings. Unless otherwise specified, the features of the following embodiments and implementations can be combined with each other.

[0025] The present invention provides a concrete vibration control method based on a neural network, comprising: The vibrator in the concrete vibrator is equipped with a temperature sensor, a triaxial acceleration sensor and a rotation speed sensor, which are used to collect temperature signals, triaxial acceleration signals and rotation speed signals, respectively. Multi-channel time-series data were constructed based on temperature signals, triaxial acceleration signals, and rotational speed signals, and the corresponding initial stage of vibration, near-complete stage of vibration, or complete stage of vibration were labeled as sample data. A neural network model trained based on sample data is used to identify the concrete vibration stage based on real-time acquired multi-channel time-series data. Establish rotation speed ranges corresponding to different vibration stages, and the processor performs vibration control in real time based on the discrimination results of the neural network model.

[0026] This invention also provides a concrete vibrator based on neural network-based intelligent judgment of vibration state for implementing the above method, such as... Figure 1 As shown, the system includes a display screen 1, a data transmission line 2, a backpack housing 3, a servo motor 4, a mounting bracket 5, a power supply and communication module mounting compartment 6, a flexible transmission hose 7, and a vibrator 8. The backpack housing 3 integrates control and computing components. The servo motor 4 is located outside the backpack housing 3 and serves as the power source. The flexible transmission hose 7 connects the backpack housing 3 and the vibrator 8 to achieve synchronous transmission of power and signals. The display screen 1 provides visual prompts for the vibration status and working information. The mounting bracket 5 limits and secures the power supply and communication module mounting compartment 6, which provides installation space for the power supply and communication units.

[0027] like Figure 2As shown, the power supply and communication module installation compartment 6 is equipped with a second power supply 61 and a Bluetooth signal transmitting module 62, and a signal transmission line 63 is provided to transmit the sensor signals to the Bluetooth signal transmitting module 62. The second power supply 61 supplies power to the sensors in the sensor compartment 82 inside the vibrator 8 through wires. The Bluetooth signal transmitting module 62 is used to send the working status data and construction record data of the vibrator 8 to the Bluetooth signal receiving module 36 inside the backpack chassis 3, so as to realize centralized processing and closed-loop control within the backpack chassis 3. The working status data includes at least the temperature data collected by the temperature sensor 821, the triaxial acceleration data collected by the triaxial acceleration sensor 822, and the rotational speed data collected by the rotational speed sensor 823; the construction record data includes, but is not limited to, the vibration start time, the vibration end time, the duration of a single vibration, the judgment results of each stage and their timestamps, servo control commands, the rotational speed range of the corresponding time period, temperature warning, completion prompt information, and the storage record of key process status parameters. The construction record data is automatically generated by the main control processor board 32 during the vibration process based on real-time sensor data, neural network discrimination results and servo drive feedback information, and is written into the storage module 33 for storage.

[0028] like Figure 3 As shown, the backpack chassis 3 is equipped with an interface connection board 31, a main control processor board 32, a storage module 33, a signal acquisition and conditioning module 34, a neural network inference acceleration module 35, a Bluetooth signal receiving module 36, and a servo driver 38. The first power supply 37 is used to supply power to the various control and computing components in the backpack chassis 3.

[0029] The neural network inference acceleration module 35 incorporates a block-multivariate temporal classification neural network model, PatchMTSC, for vibration state discrimination. The PatchMTSC neural network model is trained based on pre-collected vibration process sample data at different vibration stages. The sample data includes at least multi-channel time-series data of temperature signals, triaxial acceleration signals, and rotational speed signals, and is labeled with corresponding initial vibration stage, near-completion vibration stage, and completion vibration stage. The rotational speed in the initial vibration stage is 7200–12000 r / min, the rotational speed in the near-completion vibration stage is 4800–7200 r / min, the root mean square value of the synthetic acceleration (determined based on acceleration signals collected by the triaxial accelerometer) corresponding to the initial vibration stage is preferably 8–25 g, the root mean square value of the synthetic acceleration corresponding to the near-completion vibration stage is preferably 6–8 g, and the root mean square value of the synthetic acceleration corresponding to the completion vibration stage is preferably 2–6 g. For warning temperatures, a range of 70–85℃ is typically used; for forced load reduction or shutdown temperatures, a range of 85–100℃ is typically used. The above ranges for temperature, rotational speed, and vibration acceleration are merely illustrative examples used to characterize the typical variations of multi-channel time-series signals at different vibration stages. These parameter ranges may differ for different equipment models, concrete mix proportions, slumps, aggregate gradations, construction environments, and vibration conditions. The specific ranges can be determined through experimental calibration, empirical correction, or model training.

[0030] During model inference, the main control processor board 32 performs time window truncation on the multi-channel sensor timing signals output by the signal acquisition and conditioning module 34, and inputs the truncated timing signals to the neural network inference acceleration module 35. The PatchMTSC neural network model first divides the timing signals into multiple continuous patch segments along the time dimension using the Padding & Patching module, and then uses the built-in Transformer encoder to perform feature encoding on each patch segment; next, it models the coupling relationship between different sensor channels such as temperature, acceleration, and rotation speed through the built-in channel mixing module, and extracts the dynamic evolution features of each patch segment through the timing feature extraction module; finally, it outputs the vibration stage determination result of the current vibration process through the classification layer.

[0031] The vibration stage determination results include at least the initial vibration stage, the near-completion vibration stage, and the completion vibration stage. The main control processor board 32 generates corresponding servo control commands based on the vibration stage determination results and outputs them to the servo driver 38 to control the rotational speed of the servo motor 4, thereby controlling the rotational speed of the eccentric mechanism 83 and achieving adaptive adjustment of the excitation intensity during vibration. Simultaneously, the vibration stage determination results, servo control commands, and key vibration process status data are written into the storage module 33 for construction record storage and quality traceability. The key vibration process status data includes at least temperature data, triaxial acceleration data, rotational speed data, vibration frequency, load level, and corresponding timestamp information.

[0032] The interface connection board 31 is used to realize power distribution and signal connection between various functional boards.

[0033] like Figure 4 As shown, the vibrating rod 8 is equipped with a flexible drive shaft 81, a sensor chamber 82, an eccentric mechanism 83, and a fixed support 84. The flexible drive shaft 81 is arranged along the flexible transmission hose 7, and one end of it is connected to the output shaft of the servo motor 4 via a coupling, which is used to transmit the rotational power of the servo motor 4 in the backpack housing 3 to the vibrating rod 8. The servo motor 4 is electrically connected to the servo driver 38, which is used to output driving power to the servo motor 4. The servo driver 38 is signal connected to the main control processor board 32, which is used to receive start / stop, target speed, and speed control commands output by the main control processor board 32. At the same time, the servo driver 38 is electrically connected to the first power supply 37, which is used to obtain the working power required for the operation of the servo system, thereby realizing closed-loop drive and speed control of the servo motor 4. The eccentric mechanism 83 is connected to the end of the flexible drive shaft 81, and rotates under the drive of the servo motor 4 to generate eccentric excitation force, causing the vibrating rod 8 to vibrate. The fixed support 84 is used to support and position the rotating shaft and related rotating parts of the eccentric mechanism 83 to improve the transmission stability of the eccentric mechanism 83 during rotation. The sensing chamber 82 is installed inside the vibrating rod 8 and fixedly connected to its structure. The sensing chamber 82 contains a temperature sensor 821, a triaxial accelerometer 822, and a speed sensor 823, used to collect temperature information, vibration response information, and rotational speed information of the vibrating rod 8, respectively. Temperature information characterizes the thermal state of the vibrating rod during operation and is used to identify overload, abnormal heat dissipation, and implement overheat protection. Vibration response information characterizes changes in vibration intensity and concrete compaction state, serving as an important basis for identifying different vibration stages. Speed ​​information characterizes the working state of the excitation source and is used for closed-loop speed regulation and characteristic calculation. These three types of information characterize the vibration process from three dimensions: thermal state, vibration response, and driving state, thus forming a multi-dimensional input signal for intelligent judgment.

[0034] In one embodiment, see Figure 1and Figure 4 The flexible transmission hose 7 serves as both a transmission channel and a signal channel. It houses the flexible transmission shaft 81 and can also accommodate the signal transmission line 63, enabling the sensor signal from the vibrator 8 to be stably transmitted to the Bluetooth signal transmitting module 62, and subsequently successfully sent to the Bluetooth signal receiving module 36 inside the backpack housing 3. The fixing bayonet 5 is used to lock and secure the flexible transmission hose 7 to the power and communication module mounting compartment 6.

[0035] In one embodiment, see Figure 3 The signal acquisition and conditioning module 34 is used to synchronously acquire, filter, and condition the signals collected by the temperature sensor 821, the triaxial accelerometer 822, and the speed sensor 823, forming multi-channel time-series signal data for vibration state discrimination, and inputting the multi-channel time-series signal data to the main control processor board 32. Based on the multi-channel time-series signal data, the main control processor board 32 calls the neural network inference acceleration module 35 to perform real-time inference calculations, using the multi-channel time-series signal data as input to the neural network model, to discriminate the vibration stage to which the current vibration process belongs, and outputs the vibration stage discrimination result. The vibration stage includes at least the initial vibration stage, the near completion vibration stage, and the completion vibration stage. The main control processor board 32 generates corresponding servo control commands based on the vibration stage discrimination result and outputs them to the servo driver 38. The servo driver 38 adjusts the speed of the servo motor 4 in real time, thereby changing the excitation speed of the eccentric mechanism 83, realizing adaptive speed regulation closed-loop control of the vibration process.

[0036] In one embodiment, see Figure 5The display screen 1 is electrically connected to the main control processor board 32 via data transmission line 2, and is used to display vibration frequency, load level, rotation speed information, temperature, and a prompt indicating whether vibration is complete. The vibration frequency displayed on the display screen 1 can be calculated from the rotation speed data collected by the rotation speed sensor 823. In the embodiment where the eccentric mechanism 83 generates one main vibration excitation per revolution, the vibration frequency f satisfies f=n / 60, where n is the rotation speed measured by the rotation speed sensor 823 in r / min, and f is in Hz; in the embodiment where the eccentric mechanism 83 generates z excitations per revolution, the vibration frequency f satisfies f=z·n / 60. The load level display is used to intuitively represent the current working load status of the vibrator 8, making it easy for operators to quickly understand the degree of resistance of the equipment in the concrete, the changes in the vibration stage, and whether the operation is abnormal. Through the load level display, it can help identify working conditions such as local aggregate density, steel bar interference, excessive insertion resistance, or abnormal equipment load, thereby improving the intuitiveness and safety of on-site operation. The load level is calculated by the main control processor board 32 based on the operating parameters of the servo motor 4 or the characteristics of the sensor signals. The load level is calculated by the main control processor board 32 based on the operating parameters of the servo motor 4 and the characteristics of the sensor signals. Specifically, the main control processor board 32 obtains the current value I, actual speed n, and speed fluctuation Δn fed back by the servo driver 38, and acquires the temperature value T collected by the temperature sensor 821, the triaxial acceleration signal collected by the triaxial acceleration sensor 822, and the speed signal collected by the speed sensor 823. The main control processor board 32 extracts the composite root mean square value aRMS of acceleration based on the triaxial acceleration signal. The main control processor board 32 constructs an operating load index Lm based on the current value I, actual speed n, and speed fluctuation Δn, and constructs a sensor load index Ls based on the temperature value T, the composite root mean square value aRMS of acceleration, and the speed signal. The operating load index Lm and the sensor load index Ls are then weighted and fused to obtain the comprehensive load index L. Then, the main control processor board 32 compares the comprehensive load index L with a preset threshold range to classify the current working condition into one or more levels of low load, medium load, and high load, and outputs the corresponding load level to the display screen 1 for display. In one embodiment, the operating load index Lm can be obtained by a weighted combination of the current value I, the actual speed n, and the speed fluctuation Δn; the sensor load index Ls can be obtained by a weighted combination of the temperature value T, the root mean square value of the composite acceleration aRMS, and the speed signal; and the comprehensive load index L can be obtained by a weighted fusion of the operating load index Lm and the sensor load index Ls. When the neural network determines that the vibration is in the initial stage, the main control processor board 32 controls the servo motor 4 to maintain a high speed (preferably about 7200–12000 r / min) to improve the exhaust and compaction efficiency.When the neural network determines that vibration is nearing completion, the main control processor board 32 controls the servo motor 4 to automatically reduce its speed to (preferably approximately 4800–7200 r / min) to reduce the risk of over-vibration and complete the final stage of vibration. When the neural network determines that the vibration has reached the completion stage, the main control processor board 32 controls the servo motor 4 to stop outputting and displays the completion information on the display screen 1, thereby reducing human judgment errors and ineffective vibration time. The above-mentioned speed range can be adjusted through experimental calibration according to the specifications of the vibrator, the concrete slump, and the density of the reinforcing bars to reduce the risk of over-vibration and complete the final stage of vibration. Furthermore, based on the vibration status determination, the main control processor board can also combine temperature data collected by a temperature sensor for safety protection control. When the detected temperature reaches a preset warning threshold, the display screen outputs a warning message; when the detected temperature reaches a preset protection threshold, the main control processor board controls the servo driver to stop the servo motor. The preset warning threshold can be around 75°C, and the preset protection threshold can be 90-100°C. The specific values ​​can be determined by calibration according to different working conditions.

[0037] In one embodiment, the vibrator of the present invention can implement its closed-loop control process as follows: After construction begins, the servo motor 4 drives the flexible transmission shaft 81 to rotate the eccentric mechanism 83 and generate vibration. The sensor chamber 82 collects signals such as temperature, acceleration and rotation speed in real time and transmits them to the backpack housing 3. After the signal acquisition and conditioning module 34 completes the signal conditioning, it is input to the main control processor board 32. The main control processor board 32 completes the vibration state judgment through the neural network inference acceleration module 35, and controls the servo driver 38 to control the servo motor 4 through the judgment result, forming a closed-loop chain of state perception, intelligent judgment and speed adjustment. At the same time, the main control processor board 32 writes the process data into the storage module 33 and outputs the real-time status and completion prompts to the operator through the display screen 1. Meanwhile, the main control processor board 32 writes the process data into the storage module 33 and outputs the real-time status and completion prompts to the operator through the display screen 1; wherein, the process data includes temperature data, triaxial acceleration data and rotational speed data collected by the sensor cabin 82, vibration stage discrimination results, servo control commands, vibration frequency and load level output by the main control processor board 32, as well as corresponding timestamps, abnormal warning records and vibration completion records.

[0038] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described neural network-based concrete vibration control method.

[0039] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described neural network-based concrete vibration control method.

[0040] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0041] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0042] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0043] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0044] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method for controlling concrete vibration based on neural networks, characterized in that, include: The vibrator in the concrete vibrator is equipped with a temperature sensor, a triaxial acceleration sensor and a rotation speed sensor, which are used to collect temperature signals, triaxial acceleration signals and rotation speed signals, respectively. Multi-channel time-series data were constructed based on temperature signals, triaxial acceleration signals, and rotational speed signals, and the corresponding initial stage of vibration, near-complete stage of vibration, or complete stage of vibration were labeled as sample data. A neural network model trained based on sample data is used to identify the concrete vibration stage based on real-time acquired multi-channel time-series data. Establish rotation speed ranges corresponding to different vibration stages, and the processor performs vibration control in real time based on the discrimination results of the neural network model; If the real-time temperature value exceeds the preset shutdown temperature, the system will be forcibly stopped.

2. The method according to claim 1, characterized in that, The neural network model is the PatchMTSC neural network model. The PatchMTSC neural network model first divides the time-series signal into multiple continuous patch segments along the time dimension through the Padding & Patching module, and then uses the built-in Transformer encoder to encode the features of each patch segment. Next, the coupling relationship between the temperature, acceleration and speed sensor channels is modeled through the built-in channel mixing module, and the dynamic evolution features of each patch segment are extracted through the time-series feature extraction module. Finally, the classification layer outputs the determination result of the vibration stage to which the current vibration process belongs.

3. The method according to claim 1, characterized in that, Establish the rotation speed ranges corresponding to different vibration stages, including: the initial vibration stage corresponds to any value of 7200–12000 r / min for the rotation speed of the eccentric mechanism inside the concrete vibrator; the near completion stage corresponds to any value of 4800–7200 r / min for the rotation speed of the eccentric mechanism inside the concrete vibrator; and the completion stage corresponds to 0 r / min for the rotation speed of the eccentric mechanism inside the concrete vibrator.

4. A concrete vibrator based on neural network-based intelligent judgment of vibration state, characterized in that, The concrete vibrator is used to implement the method according to any one of claims 1-3; the concrete vibrator includes a control module, a drive module and a vibrating rod; a sensing chamber is provided inside the vibrating rod, and a temperature sensor, a triaxial acceleration sensor and a rotation speed sensor are installed inside the sensing chamber; The control module receives data from temperature sensors, triaxial accelerometers, and speed sensors, uses a trained neural network model to determine the current vibration stage, and then combines the processor to generate corresponding drive control commands based on the determination results of the neural network model. The drive module controls the excitation intensity of the vibrator according to the drive control command.

5. The concrete vibrator according to claim 4, characterized in that, The drive module includes a servo driver, a servo motor, and a flexible transmission shaft; the vibrating bar is equipped with an eccentric mechanism, which is rotatably installed inside the vibrating bar and connected to one end of the flexible transmission shaft. The other end of the flexible transmission shaft is connected to the servo motor, the servo motor is electrically connected to the servo driver, and the servo driver is electrically connected to the main control processor board. The control module includes a main control processor board, a storage module, a signal acquisition and conditioning module, a neural network inference acceleration module, a Bluetooth signal receiving module, and a Bluetooth signal transmitting module. The Bluetooth signal transmitting module is connected to a temperature sensor, a triaxial accelerometer, and a speed sensor via signal transmission lines, respectively, to receive sensor data and send it to the Bluetooth signal receiving module. The Bluetooth signal receiving module transmits the sensor data to the signal acquisition and conditioning module for synchronous acquisition, filtering, and conditioning of the sensor data to generate multi-channel time-series data. Based on the multi-channel time-series data, the main control processor board calls the neural network inference acceleration module to perform real-time inference operations, using the multi-channel time-series data as input to a trained neural network model to determine the vibration stage of the current vibration process and output the vibration stage determination result. The main control processor board generates corresponding servo control commands based on the vibration stage discrimination results and outputs them to the servo driver. The servo driver adjusts the speed of the servo motor in real time, thereby changing the excitation speed of the eccentric mechanism and realizing vibration control. The storage module is used to store key state parameters, judgment results, servo control commands, and completion records of the vibration process.

6. The concrete vibrator according to claim 5, characterized in that, It also includes a backpack chassis, in which the main control processor board, storage module, signal acquisition and conditioning module, neural network inference acceleration module, Bluetooth signal receiving module and servo driver are installed; the servo motor is installed on the backpack chassis.

7. The concrete vibrator according to claim 5, characterized in that, It also includes a flexible transmission hose and a power and communication module mounting compartment; the power and communication module mounting compartment is fixed to the flexible transmission hose by a fixing bayonet, and a second power supply and a Bluetooth signal transmitting module are installed in the power and communication module mounting compartment. The second power supply powers the sensors in the sensing compartment through wires; the flexible transmission hose is used to wrap the wires, flexible transmission shaft and signal transmission line.

8. The concrete vibrator according to claim 5, characterized in that, It also includes a display screen, which is connected to the main control processor board via a data transmission line, to display vibration frequency, load level, rotation speed information, temperature, and a prompt message indicating whether vibration is complete; wherein, the load level is calculated from the operating parameters of the servo motor or the characteristics of the sensor signal.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements a neural network-based concrete vibration control method as described in any one of claims 1-3.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements a neural network-based concrete vibration control method as described in any one of claims 1-3.

Images