A multi-sensor time-sharing hierarchical acquisition control method and system

Abstract

The application discloses a kind of multi-sensor time-sharing hierarchical acquisition control method and system, it is related to drainage pipeline detection technical field, constructs space, time, function three-dimensional collaborative control framework, space classification is according to the circumferential installation position of three-channel ground penetrating radar Configuration physical isolation acquisition link;Time classification adopts the main time base driven by encoder mileage pulse and the auxiliary time base driven by high-frequency clock, constitutes double time reference;Function classification divides system task into the physical layer of hardware state machine direct drive, the link layer of interrupt driving and the application layer of main cycle execution.The application realizes omnidirectional hierarchical management and control of space, time and function by constructing three-dimensional collaborative control framework, effectively solves the problems of channel crosstalk, low synchronization accuracy, task response delay and other problems in existing methods.Three-channel radar eliminates the time deviation caused by spatial distribution through hardware synchronization locking, improves the acquisition accuracy;Encoder pulse direct drive hardware state machine can complete time-sharing acquisition without software intervention.

Description

Technical Field

[0001] This invention relates to the field of drainage pipeline inspection technology, and in particular to a multi-sensor time-division and hierarchical acquisition and control method and system. Background Technology

[0002] As a crucial component of urban infrastructure, the health of drainage pipes directly impacts urban drainage safety and public health. Currently, drainage pipe inspection often employs robots equipped with multiple sensors working collaboratively. These sensors, such as ground-penetrating radar, video, and gyroscopes, acquire information about the internal and surrounding structures of the pipes, providing data support for pipe defect detection and maintenance.

[0003] Existing multi-sensor acquisition and control methods mostly employ single-dimensional hierarchical or software-driven acquisition modes, which have many shortcomings. Spatially, multi-channel radars lack physical isolation design, making them prone to crosstalk between channels and affecting acquisition accuracy. Temporally, acquisition triggering often relies on software queries or nested interrupts, making it difficult to achieve nanosecond-level synchronization and unable to precisely coordinate with robot movement and displacement. Functionally, the lack of hierarchical management of system tasks leads to delayed responses for high-priority acquisition tasks and excessive resource consumption by low-priority tasks.

[0004] Meanwhile, existing methods do not dynamically adjust the acquisition and transmission strategies according to changes in robot speed, which can easily lead to data loss during high-speed inspections and resource redundancy during low-speed scanning. Furthermore, the lack of a unified time base alignment mechanism for multi-source data results in low data fusion accuracy, making it difficult to meet the high-precision and high-efficiency requirements of pipeline inspection. In addition, the hardware architecture design of traditional acquisition systems is not reasonable, with poor interaction between upper and lower computer systems and poor coordination among hardware modules, which further restricts the overall performance of acquisition and control. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a multi-sensor time-sharing and hierarchical data acquisition and control method and system. The technical solution adopted is as follows:

[0006] A multi-sensor time-division hierarchical acquisition and control method for a drainage pipe detection system including a three-channel ground-penetrating radar, a video acquisition device, a gyroscope, and an encoder, comprising the following steps:

[0007] Step 1: Construct a three-dimensional collaborative control architecture encompassing space, time, and function. The spatial hierarchy is based on the physical isolation acquisition links configured according to the circumferential installation positions of the three-channel ground penetrating radar. The temporal hierarchy uses a primary time base driven by encoder odometer pulses and a secondary time base driven by a high-frequency clock to form a dual time reference. The functional hierarchy divides the system tasks into a physical layer directly driven by a hardware state machine, a link layer driven by interrupts, and an application layer executed by the main loop.

[0008] Step 2: Trigger the three-channel radar by a unified synchronous trigger pulse, analyze the phase difference of the direct wave of each channel to calculate the spatial time delay compensation value, and use a hardware timer to superimpose the compensation value to achieve strict synchronization of the equivalent time axis of the three channels;

[0009] Step 3: Construct a multi-state acquisition state machine in the programmable hardware logic that is driven by the rising edge of the encoder pulse, automatically triggers the three-channel radar time-division acquisition in sequence, latches the pulse number and generates a synchronization event flag after the acquisition is completed.

[0010] Step 4: Real-time monitoring of encoder pulse frequency determines robot speed; sensor acquisition density, data overlay times, and transmission strategy are automatically adjusted according to speed range to match acquisition efficiency with resource utilization.

[0011] Step 5: Based on the acquisition completion signal of the acquisition state machine, control the two row buffers of the dual-port RAM to alternately write data and transmit via Ethernet, so as to realize the overlap of the acquisition and transmission pipelines;

[0012] Step 6: Using the encoder pulse number as the primary key and the hardware latched timestamp as the reference, register and fuse the three-channel radar data, video frame data, and gyroscope attitude data to generate a unified format synchronization data record.

[0013] Optionally, the spatial grading method in step 1 is as follows: the transceiver antennas of the three-channel ground penetrating radar adopt an arc structure and are installed at a uniform angle of 120° around the circumference of the pipe, and are marked as CH1, CH2, and CH3 respectively; during system initialization, each channel is allocated an independent hardware trigger signal line, ADC sampling channel, and DMA data stream channel to form a physically isolated spatial grading acquisition link to ensure that there is no crosstalk between channels.

[0014] Optionally, the time grading method in step 1 is as follows: the main time base is directly driven by the high-precision encoder mileage pulse, and the displacement of the detection robot corresponding to each pulse serves as the only highest priority clock source for triggering the acquisition; the auxiliary time base is composed of a 100MHz high-frequency clock generated by the PLL inside the MCU, which is used to stamp each sampling point with nanosecond-level precision timestamps to provide a reference for subsequent data time base alignment.

[0015] Optionally, the functional classification method described in step 1 is as follows: the physical layer is the highest response level, directly driven by hardware interrupts, including encoder pulse capture, three-channel radar time-division trigger pulse generation, and ADC sampling DMA transmission, without software intervention; the link layer is the intermediate response level, triggered by DMA transmission interrupts, responsible for data packaging, dual-port RAM pipeline switching, and Ethernet frame transmission, and is executed in the interrupt service routine; the application layer is the lowest response level, running in the background main loop, including gyroscope data polling acquisition, video frame capture command issuance, and data time base alignment and fusion, and can be preempted by high-priority tasks.

[0016] Optionally, the method for three-channel hardware synchronization locking in step 2 is as follows: the main control MCU generates a unified synchronization trigger pulse, which is sent in parallel to the transmitter trigger input ports of the three radar channels; the three channel transmitters generate high-voltage nanosecond pulses and radiate them through the antennas, and the receivers synchronously collect the echo signals; the system extracts the direct wave signals of each channel, and calculates the relative time delay compensation values ​​ΔT12, ΔT13, and ΔT23 between the channels through cross-correlation calculation or zero-crossing detection; the time delay compensation values ​​are written into the hardware timer compensation register, and the hardware timer automatically adds the compensation values ​​during normal acquisition to eliminate the inherent time deviation caused by the spatial distribution of the antenna.

[0017] Optionally, the acquisition state machine in step 3 includes five hardware states: S_IDLE representing the idle waiting state, S_TRIG_CH1 representing the acquisition state of channel 1, S_TRIG_CH2 representing the acquisition state of channel 2, S_TRIG_CH3 representing the acquisition state of channel 3, and S_SYNC representing the synchronization mark state.

[0018] The trigger condition for S_IDLE is the completion of system initialization, and the action to be performed is to wait for the encoder pulse;

[0019] The trigger condition for S_TRIG_CH1 is the rising edge of the encoder pulse, and the action is to trigger CH1 to transmit and start CH1ADC acquisition.

[0020] The trigger condition for S_TRIG_CH2 is the completion of CH1 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition;

[0021] The trigger condition for S_TRIG_CH3 is the completion of CH2 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition;

[0022] The trigger condition for S_SYNC is the completion of CH3 acquisition. The action is to latch the current pulse number and generate a synchronization event flag. Each state transition is driven by hardware signals, and the CPU can enter a low-power mode during the state machine operation.

[0023] Optionally, the method for dynamic bandwidth adaptive adjustment in step 4 is as follows: Based on the encoder pulse arrival frequency, high-speed mode, medium-speed mode, and low-speed mode are divided. High-speed mode corresponds to a pulse interval of less than 10ms, medium-speed mode corresponds to a pulse interval of 10ms-50ms, and low-speed mode corresponds to a pulse interval greater than 50ms. In high-speed mode, the radar is triggered once per pulse, and video frame acquisition is reduced to one frame per 5 pulses, with Ethernet using large packet aggregation for transmission. In medium-speed mode, all sensors acquire data at the standard frequency and use standard packet transmission. In low-speed mode or when stationary, the radar triggers multiple times per pulse for superimposed acquisition, increasing the video frame rate to one frame per pulse, and improving the signal-to-noise ratio through multiple echo superpositions.

[0024] Optionally, the dual-port RAM pipeline transmission method in step 5 is as follows: two line buffers, LineBuf_0 and LineBuf_1, are pre-allocated, each buffer can store 2048 points × 16 bits of single-channel single-track acquisition data; after the hardware state machine completes the acquisition of a single channel, it generates a channel acquisition completion pulse signal, triggers the DMA write pointer to switch to the idle buffer, and simultaneously triggers the Ethernet DMA to start transmission from the just-filled buffer; by utilizing the time interval of the three-channel time-division acquisition, the data transmission of the previous channel and the acquisition of the current channel are fully pipelined and overlapped.

[0025] Optionally, the specific implementation of multi-source data time base alignment and fusion in step 6 is as follows: the encoder pulse number is used as the main alignment index, and the synchronization timestamp latched by the hardware state machine is used as the reference; the video frame carries a pulse number tag and is directly associated with the radar data with the same number; the gyroscope data carries an independent timestamp, and the nearest neighbor matching combined with a linear interpolation algorithm is used to estimate the attitude angle data corresponding to the sampling time of each radar channel; the finally generated synchronization data record includes the pulse number, timestamp, three-channel radar raw data, video frame data, gyroscope attitude data, current speed mode, and superposition number fields.

[0026] A multi-sensor time-division hierarchical acquisition and control system is provided to implement a multi-sensor time-division hierarchical acquisition and control method, including a radar detection device and an embedded acquisition and control system.

[0027] The radar detection device includes a three-channel ground-penetrating radar, a video acquisition device, a gyroscope module, a high-precision encoder, a radar transmitter, a radar receiver, an arc-shaped transceiver antenna, and a power supply module. The arc-shaped transceiver antenna of the three-channel ground-penetrating radar is installed at a uniform 120° angle around the circumference of the pipeline. The high-precision encoder is used to collect and detect the robot's displacement and output mileage pulses.

[0028] The embedded data acquisition and control system includes a host computer and a slave main control MCU; the slave main control MCU is communicatively connected to the host computer, and the slave main control MCU has a built-in programmable hardware logic unit and a hardware timer module;

[0029] The programmable hardware logic unit is used to implement the encoder pulse direct drive acquisition state machine, three-channel time-division trigger pulse generation, and synchronization event flag generation;

[0030] The hardware timer module is used to store the hierarchical delay compensation value and generate precise channel acquisition time slot control signals;

[0031] The firmware program in the lower-level main control MCU is used to realize dynamic bandwidth adaptive adjustment, dual-port RAM pipeline transmission control, multi-source data time base alignment and fusion, and communication and interaction with the upper-level computer and various sensor modules.

[0032] In summary, the present invention has at least one of the following beneficial technical effects:

[0033] This invention provides a multi-sensor time-division hierarchical acquisition and control method and system. By constructing a three-dimensional collaborative control architecture, it achieves comprehensive hierarchical control of space, time, and function, effectively solving problems such as channel crosstalk, low synchronization accuracy, and task response delay in existing methods. The three-channel radar eliminates time deviations caused by spatial distribution through hardware synchronization locking, improving acquisition accuracy; the encoder pulse-driven hardware state machine can complete time-division acquisition without software intervention, reducing CPU load and achieving microsecond-level real-time response.

[0034] The dynamic bandwidth adaptive adjustment mechanism can flexibly adjust the acquisition and transmission strategies according to the robot's travel speed, achieving a dynamic match between acquisition efficiency and resource utilization, and avoiding high-speed data loss and low-speed resource redundancy. Dual-port RAM pipeline transmission enables parallel acquisition and transmission, improving data transmission efficiency; multi-source data time base alignment and fusion ensure the consistency of data from various sensors, providing reliable support for accurate pipeline defect location.

[0035] The entire system has a reasonable architecture, efficient collaboration between the upper and lower level machines, and clear division of labor among hardware modules. This not only improves the accuracy and efficiency of multi-sensor acquisition and control but also reduces system power consumption and enhances system stability and practicality, making it suitable for high-efficiency detection scenarios of various drainage pipes. Attached Figure Description

[0036] Figure 1 This is a flowchart illustrating a multi-sensor time-division hierarchical acquisition and control method;

[0037] Figure 2 This is a schematic diagram of the dynamic bandwidth adaptive adjustment process in a specific embodiment of the present invention.

[0038] Explanation of reference numerals in the attached diagram: 1. Multi-channel ground-penetrating radar front end; 2. Clock trigger management module; 3. Dual FIFO ping-pong buffer module; 4. Network status monitoring module; 5. Timeout detection and control unit; 6. Main control MCU; 7. Network interface; 8. Host computer. Detailed Implementation

[0039] The present invention will be further described in detail below with reference to the accompanying drawings.

[0040] This invention discloses a multi-sensor time-division hierarchical acquisition and control method and system.

[0041] Reference Figure 1 and Figure 2 Example 1: A multi-sensor time-division hierarchical acquisition and control method for a drainage pipe detection system including a three-channel ground-penetrating radar, a video acquisition device, a gyroscope, and an encoder, comprising the following steps:

[0042] Step 1: Construct a three-dimensional collaborative control architecture encompassing space, time, and function. The spatial hierarchy is based on the physical isolation acquisition links configured according to the circumferential installation positions of the three-channel ground penetrating radar. The temporal hierarchy uses a primary time base driven by encoder odometer pulses and a secondary time base driven by a high-frequency clock to form a dual time reference. The functional hierarchy divides the system tasks into a physical layer directly driven by a hardware state machine, a link layer driven by interrupts, and an application layer executed by the main loop.

[0043] Step 2: Trigger the three-channel radar by a unified synchronous trigger pulse, analyze the phase difference of the direct wave of each channel to calculate the spatial time delay compensation value, and use a hardware timer to superimpose the compensation value to achieve strict synchronization of the equivalent time axis of the three channels;

[0044] Step 3: Construct a multi-state acquisition state machine in the programmable hardware logic that is driven by the rising edge of the encoder pulse, automatically triggers the three-channel radar time-division acquisition in sequence, latches the pulse number and generates a synchronization event flag after the acquisition is completed.

[0045] Step 4: Real-time monitoring of encoder pulse frequency determines robot speed; sensor acquisition density, data overlay times, and transmission strategy are automatically adjusted according to speed range to match acquisition efficiency with resource utilization.

[0046] Step 5: Based on the acquisition completion signal of the acquisition state machine, control the two row buffers of the dual-port RAM to alternately write data and transmit via Ethernet, so as to realize the overlap of the acquisition and transmission pipelines;

[0047] Step 6: Using the encoder pulse number as the primary key and the hardware latched timestamp as the reference, register and fuse the three-channel radar data, video frame data, and gyroscope attitude data to generate a unified format synchronization data record.

[0048] By adopting the above technical solutions and constructing a three-dimensional collaborative control architecture, a comprehensive hierarchical management foundation is provided for multi-sensor acquisition. Then, through hardware synchronization, direct hardware acquisition, dynamic adjustment, pipeline transmission, and data fusion, the problems of multi-sensor synchronization accuracy, acquisition efficiency, resource matching, transmission speed, and data consistency are solved in turn, forming a complete acquisition and control link. This enables the precision and efficiency of multi-sensor collaborative work and adapts to the actual application needs of drainage pipeline detection systems.

[0049] Example 2, the spatial hierarchical method in step 1 is as follows: the transceiver antenna of the three-channel ground penetrating radar adopts an arc structure and is installed at a uniform angle of 120° around the circumference of the pipe, and is marked as CH1, CH2 and CH3 respectively; during system initialization, each channel is allocated an independent hardware trigger signal line, ADC sampling channel and DMA data stream channel to form a physically isolated spatial hierarchical acquisition link to ensure no crosstalk between channels.

[0050] By adopting the above technical solution, based on the circumferential installation characteristics of the three-channel ground penetrating radar, an arc-shaped antenna and a 120° uniform layout are used to ensure full circumferential coverage of the pipeline. At the same time, an independent hardware link is allocated to each channel, and signal interference between channels is blocked through physical isolation design to avoid crosstalk affecting the acquisition accuracy, thus laying a spatial foundation for subsequent multi-channel synchronous acquisition.

[0051] Example 3, the time grading method in step 1 is as follows: the main time base is directly driven by the high-precision encoder mileage pulse, and the displacement of the detection robot corresponding to each pulse is used as the only highest priority clock source for acquisition triggering; the auxiliary time base is composed of a 100MHz high-frequency clock generated by the PLL inside the MCU, which is used to stamp each sampling point with nanosecond-level precision timestamps to provide a reference for subsequent data time base alignment.

[0052] By adopting the above technical solution, dual time bases work together. The primary time base is bound to the encoder mileage pulse, which enables precise linkage between data acquisition triggering and robot movement displacement, ensuring that the acquired data corresponds one-to-one with the pipeline position. The secondary time base uses a high-frequency clock to provide a high-precision time mark for each sampling point, solving the problem of inconsistent time bases for multi-source data and providing a time base for subsequent data alignment.

[0053] Example 4: The functional classification method described in step 1 is as follows: The physical layer is the highest response level, directly driven by hardware interrupts, including encoder pulse capture, three-channel radar time-division trigger pulse generation, and ADC sampling DMA transmission, without software intervention; the link layer is the intermediate response level, triggered by DMA transmission interrupts, responsible for data packaging, dual-port RAM pipeline switching, and Ethernet frame transmission, and is executed in the interrupt service routine; the application layer is the lowest response level, running in the background main loop, including gyroscope data polling acquisition, video frame capture command issuance, and data time base alignment and fusion, and can be preempted by high-priority tasks.

[0054] By adopting the above technical solution, layered management is implemented based on the response priority of system tasks. Core data acquisition tasks are allocated to the physical layer, enabling rapid response without software intervention through direct hardware drive, thus avoiding delays. The link layer is responsible for data transmission tasks, ensuring timely data packaging and transmission. The application layer is responsible for assisting data acquisition and data fusion, rationally allocating system resources, and improving overall response efficiency and resource utilization.

[0055] In Example 5, the method for three-channel hardware synchronization locking in step 2 is as follows: the main control MCU generates a unified synchronization trigger pulse, which is sent in parallel to the transmitter trigger input ports of the three radar channels; the three channel transmitters generate high-voltage nanosecond pulses and radiate them through the antennas, and the receivers synchronously collect the echo signals; the system extracts the direct wave signals of each channel, and calculates the relative time delay compensation values ​​ΔT12, ΔT13, and ΔT23 between the channels through cross-correlation calculation or zero-crossing detection; the time delay compensation values ​​are written into the hardware timer compensation register, and the hardware timer automatically adds the compensation values ​​during normal acquisition to eliminate the inherent time deviation caused by the spatial distribution of the antenna.

[0056] By adopting the above technical solution, the three-channel radar is started to acquire data simultaneously by using a unified synchronous trigger pulse. The inherent time delay deviation between channels is calculated by using the phase difference of the direct wave signal. Then, the acquisition time of each channel is calibrated to the same equivalent time axis by superimposing compensation values ​​through a hardware timer, thereby eliminating the synchronization deviation caused by spatial layout and achieving high-precision synchronization of three-channel acquisition.

[0057] Example 6: The acquisition state machine in step 3 includes five hardware states: S_IDLE representing the idle waiting state, S_TRIG_CH1 representing the acquisition state of channel 1, S_TRIG_CH2 representing the acquisition state of channel 2, S_TRIG_CH3 representing the acquisition state of channel 3, and S_SYNC representing the synchronization mark state.

[0058] The trigger condition for S_IDLE is the completion of system initialization, and the action to be performed is to wait for the encoder pulse;

[0059] The trigger condition for S_TRIG_CH1 is the rising edge of the encoder pulse, and the action is to trigger CH1 to transmit and start CH1ADC acquisition.

[0060] The trigger condition for S_TRIG_CH2 is the completion of CH1 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition;

[0061] The trigger condition for S_TRIG_CH3 is the completion of CH2 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition;

[0062] The trigger condition for S_SYNC is the completion of CH3 acquisition. The action is to latch the current pulse number and generate a synchronization event flag. Each state transition is driven by hardware signals, and the CPU can enter a low-power mode during the state machine operation.

[0063] By adopting the above technical solution, a multi-state hardware state machine is constructed using programmable hardware logic. The rising edge of the encoder pulse is used as the only trigger source. Through preset state transition logic, the time-division acquisition of the three-channel radar is automatically completed without CPU intervention, reducing CPU load. At the same time, through state latching and event flags, trigger signals are provided for subsequent data packaging and alignment, ensuring the orderliness and real-time nature of the acquisition process.

[0064] Example 7, the method of dynamic bandwidth adaptive adjustment in step 4 is as follows: according to the encoder pulse arrival frequency, high-speed mode, medium-speed mode and low-speed mode are divided. The pulse interval corresponding to high-speed mode is less than 10ms, the pulse interval corresponding to medium-speed mode is 10ms-50ms, and the pulse interval corresponding to low-speed mode is greater than 50ms. In high-speed mode, the radar is triggered once per pulse, and the video frame acquisition frequency is reduced to one frame per 5 pulses. Ethernet uses large packet aggregation for transmission. In medium-speed mode, all sensors acquire data at the standard frequency and use standard packet transmission. In low-speed mode or when stationary, the radar triggers multiple times per pulse for superimposed acquisition, and the video frame rate is increased to one frame per pulse. The signal-to-noise ratio is improved by averaging multiple echo superpositions.

[0065] By adopting the above technical solution, the robot's speed is inferred by monitoring the encoder pulse frequency. Different working modes are divided according to speed, and the acquisition density, data superposition times, and transmission strategies are adjusted for different modes to match the acquisition and transmission strategies with the actual speed. This avoids data loss at high speeds and reduces resource redundancy at low speeds. At the same time, the signal-to-noise ratio of acquisition in low-speed scenarios is improved by superimposing acquisition data multiple times.

[0066] In Example 8, the method for dual-port RAM pipeline transmission in step 5 is as follows: two line buffers, LineBuf_0 and LineBuf_1, are pre-allocated, each buffer can store 2048 points × 16 bits of single-channel single-track acquisition data; after the hardware state machine completes the acquisition of a single channel, it generates a channel acquisition completion pulse signal, triggering the DMA write pointer to switch to an idle buffer, and simultaneously triggering the Ethernet DMA to start transmission from the just-filled buffer; by utilizing the time interval of the three-channel time-division acquisition, the data transmission of the previous channel and the acquisition of the current channel are fully pipelined and overlapped.

[0067] By adopting the above technical solution, the two independent row buffers work alternately to separate the data writing and data sending processes. By using the time interval of the three-channel time-division acquisition, the data transmission of the previous channel is synchronized with the acquisition of the current channel, realizing pipeline overlap of acquisition and transmission, reducing data transmission waiting time, and improving overall data transmission efficiency.

[0068] Example 9, the specific implementation of multi-source data time base alignment and fusion in step 6 is as follows: the encoder pulse number is used as the main alignment index, and the synchronization timestamp latched by the hardware state machine is used as the reference; the video frame carries a pulse number tag and is directly associated with the radar data with the same number; the gyroscope data carries an independent timestamp, and the nearest neighbor matching combined with the linear interpolation algorithm is used to estimate the attitude angle data corresponding to the sampling time of each radar channel; the finally generated synchronization data record includes the pulse number, timestamp, three-channel radar raw data, video frame data, gyroscope attitude data, current speed mode and superposition number fields.

[0069] By adopting the above technical solution, using the encoder pulse number as the core alignment index and combining it with the high-precision timestamp of hardware latch, the data of different sensors are associated with the same time and position reference. The matching and interpolation algorithms are used to make up for the differences in the acquisition time sequence of different sensors, and finally integrate various types of data to generate a unified format record, ensuring the consistency of multi-source data and providing complete data support for pipeline defect location.

[0070] Example 10: A multi-sensor time-division hierarchical acquisition and control system, used to implement a multi-sensor time-division hierarchical acquisition and control method, including a radar detection device and an embedded acquisition and control system;

[0071] The radar detection device includes a three-channel ground-penetrating radar, a video acquisition device, a gyroscope module, a high-precision encoder, a radar transmitter, a radar receiver, an arc-shaped transceiver antenna, and a power supply module. The arc-shaped transceiver antenna of the three-channel ground-penetrating radar is installed at a uniform 120° angle around the circumference of the pipeline. The high-precision encoder is used to collect and detect the robot's displacement and output mileage pulses.

[0072] The embedded data acquisition and control system includes a host computer and a slave main control MCU; the slave main control MCU is communicatively connected to the host computer, and the slave main control MCU has a built-in programmable hardware logic unit and a hardware timer module;

[0073] The programmable hardware logic unit is used to implement the encoder pulse direct drive acquisition state machine, three-channel time-division trigger pulse generation, and synchronization event flag generation;

[0074] The hardware timer module is used to store the hierarchical delay compensation value and generate precise channel acquisition time slot control signals;

[0075] The firmware program in the lower-level main control MCU is used to realize dynamic bandwidth adaptive adjustment, dual-port RAM pipeline transmission control, multi-source data time base alignment and fusion, and communication and interaction with the upper-level computer and various sensor modules.

[0076] By adopting the above technical solution, the radar detection device and the embedded acquisition and control system work together in a division of labor. The radar detection device is responsible for the acquisition and signal conversion of pipeline information. The arc antenna and high-precision encoder ensure the comprehensiveness and position correlation of the acquired data. The embedded acquisition and control system works in collaboration between the upper and lower computer. The hardware module of the lower computer's main control MCU realizes the acquisition trigger and timing control, and the firmware program realizes dynamic adjustment and data fusion. The upper computer is responsible for issuing instructions and data interaction. The overall architecture adapts to the requirements of each step of the acquisition and control method, ensuring the stable implementation of the method.

[0077] The following specific embodiments illustrate the implementation principle of the present invention:

[0078] Applied to the inspection of urban drainage pipes with a diameter of 1.2 meters, a pipe inspection robot equipped with a three-channel ground-penetrating radar, a high-definition video acquisition device, a six-axis gyroscope, and a high-precision encoder is used. The following specific steps are taken to achieve multi-sensor time-sharing and hierarchical acquisition and control, while matching the corresponding system hardware configuration to ensure the feasibility of the method.

[0079] The system hardware configuration is as follows: In the radar detection device, the transceiver antennas of the three-channel ground-penetrating radar adopt an arc structure and are installed at a uniform angle of 120° around the circumference of the pipeline, corresponding to CH1, CH2, and CH3 respectively. Each channel is equipped with an independent radar transmitter and receiver; the high-precision encoder is an incremental encoder with a resolution of 3000 lines, and each pulse corresponds to a 0.33mm travel displacement of the detection robot, outputting mileage pulses in real time; the video acquisition device uses a high-definition industrial camera with a resolution of 1920×1080; the gyroscope module uses a six-axis attitude sensor with a fixed sampling rate of 100Hz; the power supply module provides a stable DC power supply for the entire detection device.

[0080] In the embedded data acquisition and control system, the lower-level main control MCU is an industrial-grade MCU with a built-in programmable hardware logic unit and a high-precision hardware timer module. The programmable hardware logic unit is an FPGA module, and the hardware timer module supports nanosecond-level timing control. The dual-port RAM uses an 8KB high-speed cache, divided into two 4KB line buffers, LineBuf_0 and LineBuf_1. Each buffer can store 2048 points × 16 bits of single-channel, single-track acquisition data. The upper-level computer is an industrial control computer that communicates with the lower-level main control MCU via Ethernet to realize command issuance and data reception and display.

[0081] The specific data acquisition and control steps are as follows:

[0082] Step 1: Construct a three-dimensional collaborative control architecture encompassing space, time, and function. In terms of spatial hierarchy, based on the circumferential installation position of the three-channel ground-penetrating radar, independent hardware trigger signal lines, ADC sampling channels, and DMA data stream channels are allocated to CH1, CH2, and CH3 respectively, forming physically isolated acquisition links to avoid crosstalk between channels. In terms of temporal hierarchy, the primary time base is directly driven by high-precision encoder odometer pulses, with each pulse corresponding to a 0.33mm travel displacement, serving as the sole highest-priority clock source for acquisition triggering; the secondary time base is generated by a 100MHz high-frequency clock from the MCU's internal PLL, stamping each sampling point with a nanosecond-level timestamp. In terms of functional hierarchy, the physical layer is responsible for encoder pulse acquisition, three-channel radar time-division trigger pulse generation, and ADC sampling DMA transmission, directly driven by hardware interrupts; the link layer is responsible for data packaging, dual-port RAM pipeline switching, and Ethernet frame transmission, with interrupt triggering completed by DMA transmission; the application layer is responsible for gyroscope data polling acquisition, video frame capture command issuance, and data time base alignment and fusion, running in the background main loop.

[0083] Step 2: Achieve hardware synchronization locking for the three channels. The main control MCU generates a unified synchronization trigger pulse, which is sent in parallel to the transmitter trigger input ports of the three radar channels. After receiving the pulse, the transmitters of the three channels synchronously generate high-voltage nanosecond pulses, which are radiated to the pipe wall through the arc antenna. The receivers synchronously collect the echo signals. The system extracts the direct wave signals of each channel and calculates the time delay deviation ΔT12 of CH2 relative to CH1 as 8ns, the time delay deviation ΔT13 of CH3 relative to CH1 as 15ns, and the time delay deviation ΔT23 of CH2 relative to CH3 as 7ns through cross-correlation calculation. The above time delay compensation values ​​are written into the compensation register of the hardware timer. During normal acquisition, the hardware timer automatically superimposes the corresponding compensation values ​​when generating the trigger signals of each channel, achieving strict synchronization of the three channels on the equivalent time axis.

[0084] Step 3: Run the encoder pulse-driven acquisition state machine. Construct an acquisition state machine with five hardware states in the FPGA module: S_IDLE, S_TRIG_CH1, S_TRIG_CH2, S_TRIG_CH3, and S_SYNC. After system initialization, the state machine enters S_IDLE, waiting for encoder pulses. When a rising edge of an encoder pulse is detected, the state machine jumps to S_TRIG_CH1, triggering the CH1 transmitter to emit a pulse and starting CH1's ADC acquisition. After CH1 acquisition is complete, the state machine jumps to S_TRIG_CH2, triggering CH2 to emit and acquire ADC data. After CH2 acquisition is complete, the state machine jumps to S_TRIG_CH3, triggering CH3 to emit and acquire ADC data. After CH3 acquisition is complete, the state machine jumps to S_SYNC, latching the current encoder pulse number, generating a synchronization event flag, and then returning to S_IDLE to wait for the next encoder pulse. The entire state transition process is driven by hardware signals, requiring no CPU intervention. The CPU can perform gyroscope data processing or enter low-power mode during state machine operation.

[0085] Step 4: Perform dynamic bandwidth adaptive adjustment. The system monitors the encoder pulse frequency in real time and divides it into three speed modes based on the pulse interval: high-speed mode with a pulse interval of less than 10ms, corresponding to a robot speed greater than 33mm / s; medium-speed mode with a pulse interval of 10ms to 50ms; and low-speed mode with a pulse interval greater than 50ms, corresponding to a robot speed less than 6.6mm / s. In high-speed mode, the radar triggers one acquisition per pulse, and the video frame acquisition frequency is reduced to one frame every 5 pulses. Ethernet uses large packet aggregation for transmission, packaging the data from 5 radar channels into one network frame for transmission. In medium-speed mode, all sensors acquire data at the standard frequency, and video frames are captured one frame every 2 pulses. Ethernet uses standard packet transmission, and each network frame contains data from one radar channel. In low-speed mode, the radar triggers four superimposed acquisitions per pulse, superimposing and averaging the echo data from the same location. The video frame rate is increased to one frame per pulse, and Ethernet uses standard packet transmission.

[0086] Step 5: Implement dual-port RAM pipelined transmission. Two row buffers, LineBuf_0 and LineBuf_1, are pre-allocated. Each buffer can store 2048 points × 16 bits of acquired data per channel. When the hardware state machine completes CH1 acquisition, it generates a channel acquisition completion pulse signal, triggering the DMA write pointer to switch to LineBuf_1. Simultaneously, it triggers the Ethernet DMA to read CH1 data from LineBuf_0 and package it for transmission. During CH2 acquisition, Ethernet completes CH1 data transmission. After CH2 acquisition is completed, the DMA write pointer switches to LineBuf_0, and the Ethernet DMA reads CH2 data from LineBuf_1 and transmits it. This process continues, utilizing the time interval of the three-channel time-division acquisition to achieve complete pipeline overlap between the data transmission of the previous channel and the current channel acquisition, improving data transmission efficiency.

[0087] Step 6: Complete the time-base alignment and fusion of multi-source data. Using the encoder pulse number as the primary alignment index and the nanosecond-level timestamp latched by the hardware state machine as the benchmark, video frames carry corresponding pulse number tags and are directly associated with the three-channel radar data of the same number. Gyroscope data carries independent timestamps. A nearest neighbor matching combined with a linear interpolation algorithm is used to estimate the attitude angle data corresponding to the sampling time of each radar channel. Finally, a unified format of synchronized data records is generated, including pulse number, timestamp, raw radar data from CH1 to CH3, video frame data, gyroscope attitude data, current speed mode, and superposition count fields. This data is synchronously transmitted to the host computer for viewing and analysis by staff, providing data support for pipeline defect location.

[0088] During the inspection, the host computer sends control commands in real time, and the slave computer's main control MCU adjusts the acquisition parameters according to the commands. The entire system operates stably, the acquisition accuracy meets the requirements for pipeline defect detection, the data transmission is lossless and redundant, and the fusion of multi-source data can accurately reflect the internal and surrounding structural conditions of the pipeline, making it suitable for actual drainage pipeline inspection scenarios.

[0089] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-sensor time-division hierarchical acquisition and control method for a drainage pipe detection system comprising a three-channel ground-penetrating radar, a video acquisition device, a gyroscope, and an encoder, characterized in that, Includes the following steps: Step 1: Construct a three-dimensional collaborative control architecture encompassing space, time, and function. The spatial hierarchy is based on the physical isolation acquisition links configured according to the circumferential installation positions of the three-channel ground penetrating radar. The temporal hierarchy uses a primary time base driven by encoder odometer pulses and a secondary time base driven by a high-frequency clock to form a dual time reference. The functional hierarchy divides the system tasks into a physical layer directly driven by a hardware state machine, a link layer driven by interrupts, and an application layer executed by the main loop. Step 2: Trigger the three-channel radar by a unified synchronous trigger pulse, analyze the phase difference of the direct wave of each channel to calculate the spatial time delay compensation value, and use a hardware timer to superimpose the compensation value to achieve strict synchronization of the equivalent time axis of the three channels; Step 3: Construct a multi-state acquisition state machine in the programmable hardware logic that is driven by the rising edge of the encoder pulse, automatically triggers the three-channel radar time-division acquisition in sequence, latches the pulse number and generates a synchronization event flag after the acquisition is completed. Step 4: Real-time monitoring of encoder pulse frequency determines robot speed; sensor acquisition density, data overlay times, and transmission strategy are automatically adjusted according to speed range to match acquisition efficiency with resource utilization. Step 5: Based on the acquisition completion signal of the acquisition state machine, control the two row buffers of the dual-port RAM to alternately write data and transmit via Ethernet, so as to realize the overlap of the acquisition and transmission pipelines; Step 6: Using the encoder pulse number as the primary key and the hardware latched timestamp as the reference, register and fuse the three-channel radar data, video frame data, and gyroscope attitude data to generate a unified format synchronization data record.

2. The multi-sensor time-division hierarchical acquisition and control method according to claim 1, characterized in that, The spatial grading method in step 1 is as follows: The transceiver antennas of the three-channel ground penetrating radar adopt an arc structure and are installed at a uniform angle of 120° around the circumference of the pipe, and are marked as CH1, CH2 and CH3 respectively; during system initialization, each channel is allocated an independent hardware trigger signal line, ADC sampling channel and DMA data stream channel to form a physically isolated spatial grading acquisition link to ensure no crosstalk between channels.

3. The multi-sensor time-division hierarchical acquisition and control method according to claim 2, characterized in that, The time grading method described in step 1 is as follows: the main time base is directly driven by the mileage pulse of the high-precision encoder, and the displacement of the detection robot corresponding to each pulse serves as the only highest priority clock source for triggering the acquisition; the auxiliary time base consists of a 100MHz high-frequency clock generated by the PLL inside the MCU, which is used to stamp each sampling point with nanosecond-level precision timestamps to provide a reference for subsequent data time base alignment.

4. The multi-sensor time-division hierarchical acquisition and control method according to claim 3, characterized in that, The functional grading method described in step 1 is as follows: the physical layer is the highest response level, directly driven by hardware interrupts, including encoder pulse capture, three-channel radar time-division trigger pulse generation, and ADC sampling DMA transmission, without software intervention; the link layer is the intermediate response level, triggered by DMA transmission interrupts, responsible for data packaging, dual-port RAM pipeline switching, and Ethernet frame transmission, and is executed in the interrupt service routine; the application layer is the lowest response level, running in the background main loop, including gyroscope data polling acquisition, video frame capture command issuance, and data time base alignment and fusion, and can be preempted by high-priority tasks.

5. The multi-sensor time-division hierarchical acquisition and control method according to claim 4, characterized in that, The method for three-channel hardware synchronization locking in step 2 is as follows: the main control MCU generates a unified synchronization trigger pulse, which is sent in parallel to the transmitter trigger input ports of the three radar channels; the three channel transmitters generate high-voltage nanosecond pulses and are radiated through the antennas, and the receivers synchronously collect the echo signals; the system extracts the direct wave signals of each channel, and calculates the relative time delay compensation values ​​ΔT12, ΔT13, and ΔT23 between the channels through cross-correlation calculation or zero-crossing detection; the time delay compensation values ​​are written into the hardware timer compensation register, and the hardware timer automatically adds the compensation values ​​during normal acquisition to eliminate the inherent time deviation caused by the spatial distribution of the antenna.

6. The multi-sensor time-division hierarchical acquisition and control method according to claim 5, characterized in that, The acquisition state machine described in step 3 includes five hardware states: S_IDLE, representing the idle waiting state; S_TRIG_CH1, representing the acquisition state of channel 1; S_TRIG_CH2, representing the acquisition state of channel 2; S_TRIG_CH3, representing the acquisition state of channel 3; and S_SYNC, representing the synchronization flag state. The trigger condition for S_IDLE is the completion of system initialization, and the action to be performed is to wait for the encoder pulse; The trigger condition for S_TRIG_CH1 is the rising edge of the encoder pulse, and the action is to trigger CH1 to transmit and start CH1ADC acquisition. The trigger condition for S_TRIG_CH2 is the completion of CH1 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition; The trigger condition for S_TRIG_CH3 is the completion of CH2 acquisition, and the action is to trigger CH2 transmission and start CH2ADC acquisition; The trigger condition for S_SYNC is the completion of CH3 acquisition. The action is to latch the current pulse number and generate a synchronization event flag. Each state transition is driven by hardware signals, and the CPU can enter a low-power mode during the state machine operation.

7. The multi-sensor time-division hierarchical acquisition and control method according to claim 6, characterized in that, The method for dynamic bandwidth adaptive adjustment in step 4 is as follows: Based on the encoder pulse arrival frequency, high-speed mode, medium-speed mode, and low-speed mode are divided. High-speed mode corresponds to a pulse interval of less than 10ms, medium-speed mode corresponds to a pulse interval of 10ms-50ms, and low-speed mode corresponds to a pulse interval of more than 50ms. In high-speed mode, the radar is triggered once per pulse, and the video frame acquisition frequency is reduced to one frame per 5 pulses. Ethernet uses large packet aggregation for transmission. In medium-speed mode, all sensors acquire data at the standard frequency and use standard packet transmission. In low-speed mode or when stationary, the radar triggers multiple times per pulse for superimposed acquisition, and the video frame rate is increased to one frame per pulse. The signal-to-noise ratio is improved by averaging multiple echo superpositions.

8. The multi-sensor time-division hierarchical acquisition and control method according to claim 7, characterized in that, The method for dual-port RAM pipeline transmission in step 5 is as follows: two line buffers, LineBuf_0 and LineBuf_1, are pre-allocated, and each buffer can store 2048 points × 16 bits of single-channel single-track acquisition data; After the hardware state machine completes the acquisition of a single channel, it generates a channel acquisition completion pulse signal, triggering the DMA write pointer to switch to the idle buffer, and at the same time triggering the Ethernet DMA to start sending from the just-filled buffer; by utilizing the time interval of the three-channel time-division acquisition, the data transmission of the previous channel and the acquisition of the current channel are fully pipelined and overlapped.

9. The multi-sensor time-division hierarchical acquisition and control method according to claim 8, characterized in that, The specific implementation of multi-source data time base alignment and fusion in step 6 is as follows: the encoder pulse number is used as the main alignment index, and the synchronization timestamp latched by the hardware state machine is used as the reference; the video frame carries a pulse number tag and is directly associated with the radar data with the same number; the gyroscope data carries an independent timestamp, and the nearest neighbor matching combined with the linear interpolation algorithm is used to estimate the attitude angle data corresponding to the sampling time of each radar channel; the finally generated synchronization data record includes the pulse number, timestamp, three-channel radar raw data, video frame data, gyroscope attitude data, current speed mode and superposition number fields.

10. A multi-sensor time-sharing and hierarchical data acquisition and control system, characterized in that: The method for implementing the multi-sensor time-division hierarchical acquisition and control method of claim 9 includes a radar detection device and an embedded acquisition and control system. The radar detection device includes a three-channel ground-penetrating radar, a video acquisition device, a gyroscope module, a high-precision encoder, a radar transmitter, a radar receiver, an arc-shaped transceiver antenna, and a power supply module. The arc-shaped transceiver antenna of the three-channel ground-penetrating radar is installed at a uniform 120° angle around the circumference of the pipeline. The high-precision encoder is used to collect and detect the robot's displacement and output mileage pulses. The embedded data acquisition and control system includes a host computer and a slave main control MCU; the slave main control MCU is communicatively connected to the host computer, and the slave main control MCU has a built-in programmable hardware logic unit and a hardware timer module; The programmable hardware logic unit is used to implement the encoder pulse direct drive acquisition state machine, three-channel time-division trigger pulse generation, and synchronization event flag generation; The hardware timer module is used to store the hierarchical delay compensation value and generate precise channel acquisition time slot control signals; The firmware program in the lower-level main control MCU is used to realize dynamic bandwidth adaptive adjustment, dual-port RAM pipeline transmission control, multi-source data time base alignment and fusion, and communication and interaction with the upper-level computer and various sensor modules.

Images