An intelligent side station monitoring system and method for adaptive scheduling of computing power

Abstract

The application relates to an intelligent side station supervision system and method for adaptive scheduling of computing power, wherein a hardware state sensing unit collects power, temperature and charging and discharging state data in real time and inputs the data into a mode control unit, the mode control unit matches a corresponding operation mode according to the hardware state and sets an upper limit of the occupancy rate threshold of global computing power resources in the mode, thereby realizing hierarchical matching of the hardware health state and the computing power demand; a thread pool management unit allocates independent inference thread pools to each algorithm model group for parallel processing of video frame data, a time delay monitoring unit acquires the single frame inference time consumption of each model group in real time and calculates a maximum difference value and inputs the maximum difference value into a scheduling unit, the scheduling unit dynamically adjusts the thread quota of each inference thread pool according to the maximum difference value and the hardware state parameter, under the premise of ensuring the timing synchronization of multi-algorithm concurrent inference, realizes real-time linkage of computing power allocation and the hardware state, and ensures that the dynamic balance of computing power and power consumption can be maintained under different working conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for smart construction sites, specifically to an intelligent on-site monitoring system and method with adaptive computing power scheduling. Background Technology

[0002] With the development of artificial intelligence technology, computer vision recognition algorithms have achieved significant results in target detection, behavior analysis, and hazard warning, and have been gradually applied to fields such as security monitoring, smart construction sites, and traffic management. By collecting video data through cameras and combining it with AI algorithms for analysis, it is possible to automatically identify safety helmet wearing status, intrusion into dangerous areas, and the operating status of mechanical equipment, thereby assisting in on-site management and safety supervision. However, existing construction site monitoring and scheduling solutions suffer from three major unresolved technical shortcomings in terms of portable deployment, concurrent operation of multiple algorithms, and collaborative adaptation for power consumption control: Insufficient hardware integration and adaptability: Existing portable monitoring devices are mostly spliced ​​together from scattered modules, requiring additional external switches, power supplies, AI hosts, speakers, and other peripherals. This results in cumbersome wiring, poor deployment flexibility, and an inability to adapt to the dynamic and temporary monitoring needs of construction site work areas. Furthermore, the hardware is not specifically optimized for edge multi-algorithm computing power scheduling, making it impossible to achieve real-time linkage between hardware status and computing power scheduling. The hardware merely serves as a data acquisition carrier and cannot participate in the collaborative management of computing power and power consumption.

[0003] The computing power scheduling mechanism has inherent defects: the existing edge AI multi-thread scheduling scheme only adjusts the number of threads based on the algorithm inference time, without binding it to core parameters such as hardware power supply mode, battery status, and device temperature; when multiple algorithms run concurrently, it is very easy to have uneven computing power distribution and asynchronous inference time of each algorithm, resulting in system lag, delayed recognition response, and even device overheating, frequency reduction, and crashes. It cannot take into account both the real-time performance of multiple concurrent algorithms and the stability of device operation.

[0004] Power consumption management and computing power scheduling are completely disconnected: The power consumption management of existing portable monitoring devices can only achieve simple coarse operations such as CPU frequency reduction and peripheral shutdown, without deep integration with the fine-grained scheduling of multi-algorithm thread pools; it cannot achieve precise power consumption control through dynamic adjustment of thread quotas while ensuring the stable operation of core monitoring algorithms, and cannot completely solve the core contradiction between high computing power and concurrent operation of multiple algorithms in portable devices and long battery life, which can easily lead to unplanned downtime in construction site scenarios without fixed power supply. Summary of the Invention

[0005] The main objective of this invention is to provide an intelligent on-site monitoring system and method with adaptive computing power scheduling to solve the technical problems of low equipment integration, disconnect between computing power and hardware, lag during concurrent multi-algorithm operation, and difficulty in balancing computing power and battery life in existing construction site monitoring solutions.

[0006] To achieve the above objectives, the present invention provides an intelligent on-site monitoring system with adaptive computing power scheduling, comprising an integrated portable on-site robot. The integrated portable on-site robot includes interconnected hardware status sensing units, thread pool management units, latency monitoring units, scheduling units, and mode control units; wherein; The hardware status sensing unit is used to collect the hardware status parameters of the integrated portable side-by-side robot in real time. The mode control unit is used to match the corresponding operating mode according to the hardware status parameters, so as to execute the upper limit of the global computing power resource occupancy threshold under the preset operating mode. The thread pool management unit has multiple built-in algorithm models. The thread pool management unit classifies the multiple algorithm models into multiple algorithm model groups and allocates an independent inference thread pool to each algorithm model group. The latency monitoring unit is used to acquire the real-time inference time of each algorithm model group in a single frame and calculate the maximum difference in real-time time between each algorithm model group. The scheduling unit is used to dynamically adjust the thread quota of each inference thread pool according to the maximum difference and the hardware status parameters, so that the inference cycle of each algorithm model group is consistent, and the total computing power utilization rate of each algorithm model group is controlled within the upper limit of the global computing power resource utilization rate threshold of the current running mode.

[0007] Furthermore, the integrated portable side-by-side robot includes a computing chip, an integrated circuit board, a power module, a switch module, and a field alarm module. The computing chip, power module, switch module, and field alarm module are all connected to the integrated circuit board. The hardware status perception unit, the thread pool management unit, the latency monitoring unit, the scheduling unit, and the mode control unit are all integrated on the integrated circuit board.

[0008] More preferably, the integrated portable side-by-side robot also includes a housing, in which the computing chip, integrated circuit board, power module, switch module and field alarm module are all disposed. The compartment cover is closable and connected to the housing. The housing is provided with multiple interfaces, and the integrated circuit board is connected to external devices and / or networks through the interfaces.

[0009] More preferably, the integrated circuit board further includes a charging and discharging circuit, which is configured with a charging cutoff voltage and a discharging cutoff voltage to control the power supply and charging of the integrated portable side-by-side robot and external devices.

[0010] More preferably, the power module includes a removable battery, a battery compartment, a compartment cover, and a battery interface. The removable battery is disposed in the battery compartment, the compartment cover is hinged to the battery compartment, and the removable battery is connected to the charging and discharging circuit through the battery interface.

[0011] This invention also provides an intelligent on-site monitoring method with adaptive computing power scheduling, applied to the intelligent on-site monitoring system with adaptive computing power scheduling as described above, comprising the following steps: S1. Real-time acquisition of hardware status parameters of the integrated portable side-by-side robot, and acquisition of locally pre-stored algorithm configuration instructions; the hardware status parameters include at least remaining power data, temperature data, and charging / discharging status data; the algorithm configuration instructions include algorithm scenarios, algorithm priorities, operating mode parameter thresholds, and hardware linkage rules; S2. Obtain the preset operating mode, and match the corresponding operating mode as the current operating mode according to the hardware status parameters and the algorithm configuration instructions; wherein, each of the operating modes is set with a corresponding upper limit of the global computing power resource occupancy rate threshold; the operating modes include real-time mode, energy efficiency balance mode and battery life priority mode; S3. Obtain multiple algorithm models, classify the multiple algorithm models to obtain multiple algorithm model groups, select the algorithm model group to be enabled according to the algorithm configuration instructions, and match an independent inference thread pool for each enabled algorithm model group to process video frame data in parallel. S4. Obtain the real-time time of each inference thread pool to complete a single frame inference, and calculate the maximum difference in real-time time between each algorithm model group; S5. Obtain the preset hardware linkage rules, and perform hardware status compliance verification based on the maximum difference and the hardware status parameters, and determine whether the maximum difference is within the preset synchronization tolerance; if the verification passes and the maximum difference exceeds the preset synchronization tolerance, dynamically adjust the thread quota of each inference thread pool to make the inference cycle of each algorithm model consistent. S6. Determine whether the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode; if yes, readjust the thread quota and return to step S4; if no, maintain the current thread quota of each inference thread pool and complete the adaptive scheduling of computing power.

[0012] More preferably, the dynamic adjustment of thread quotas for each inference thread pool in step S5 specifically includes the following steps: The direction of thread quota adjustment is determined based on the charge / discharge status data in the hardware status parameters. The thread quota adjustment coefficient is calculated based on the remaining power data and the temperature data. The thread quota adjustment coefficient is negatively correlated with the remaining power and the temperature. Based on the thread quota adjustment direction and the thread quota adjustment coefficient, the thread quota adjustment amount for each inference thread pool is generated; Wherein; when the charge / discharge status data is "external power supply", the thread quota adjustment coefficient ; When the charge / discharge status data is "battery powered", the thread quota adjustment coefficient Determined according to the following formula: in, This is the current remaining battery level. Rated power, The current temperature. To ensure the upper limit of safe operating temperature, This is the lower limit of the normal operating temperature.

[0013] More preferably, the dynamic adjustment of thread quotas for each inference thread pool in step S5 specifically includes the following steps: Obtain the average time taken for each inference thread pool to complete a single frame of inference, and determine whether the real-time time is greater than the average time. If yes, mark the algorithm model group in the inference thread pool where the real-time time is greater than the average time as a slow model. If no, mark the algorithm model group in the inference thread pool where the real-time time is less than the average time as a fast model. Reclaim a first number of thread quotas from the inference thread pool of the fast model; Obtain a preset allocation ratio, and allocate the reclaimed thread quota to the inference thread pool of the slow model according to the preset allocation ratio; wherein, the preset allocation ratio is proportional to the deviation between the real-time consumption time and the average consumption time of each slow model.

[0014] More preferably, the readjustment of thread quotas in step S6 includes: When the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode, the thread quota of the inference thread pool corresponding to the algorithm model with lower algorithm priority will be reduced first.

[0015] More preferably, after step S6, the following steps are also included: Execute the corresponding computing power management strategy according to the current operating mode; where; If the current operating mode is real-time mode, then the thread rate limiting is cancelled, and the inference thread pool is allowed to expand as needed until the computing power utilization rate reaches the upper limit of the utilization rate threshold of real-time mode; If the current operating mode is the energy efficiency balance mode, then the total number of concurrent threads will be limited while maintaining the consistency of the inference cycle of each algorithm model, and the total computing power utilization rate will be controlled within the upper limit of the utilization rate threshold of the energy efficiency balance mode. If the current operating mode is the battery life priority mode, then the frame skipping sampling strategy is executed to obtain the non-core algorithm model, reduce the thread quota of the inference thread pool of the non-core algorithm model, and control the total computing power utilization rate within the upper limit of the utilization rate threshold of the battery life priority mode.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention integrates a hardware status sensing unit, a thread pool management unit, a latency monitoring unit, a scheduling unit, and a mode control unit into a single device using an integrated portable side-monitoring robot. This allows hardware status parameters to participate in computing power scheduling decisions in real time. The hardware status sensing unit collects power, temperature, and charging / discharging status data in real time and inputs them to the mode control unit. The mode control unit matches the corresponding operating mode based on the hardware status and sets an upper limit for the global computing power resource occupancy rate under that mode, achieving hierarchical matching between hardware health status and computing power demand. The thread pool management unit allocates an independent inference thread pool to each algorithm model group for parallel processing. The latency monitoring unit acquires the inference time of each model group in real time and calculates the maximum difference, which is then input to the scheduling unit. The scheduling unit dynamically adjusts the thread quota of each inference thread pool based on the maximum difference and hardware status parameters. This ensures that the inference cycle of each algorithm model group is consistent while keeping the total computing power utilization rate within the upper limit of the current operating mode threshold. Thus, while ensuring the synchronization of the concurrent inference sequence of multiple algorithms, it achieves real-time linkage between computing power resource allocation and hardware power supply mode, battery power, and device temperature. This ensures that the device can maintain a dynamic balance between computing power and power consumption under different operating conditions, significantly extending the continuous working time in battery power mode. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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 the structures shown in these drawings without creative effort.

[0018] Figure 1 This is a three-dimensional schematic diagram of the overall structure of an integrated portable side-monitoring robot according to an embodiment of the present invention; Figure 2 This is a simplified three-dimensional schematic diagram of the internal structure of an integrated portable side-by-side robot according to an embodiment of the present invention; Figure 3This is a side view of the internal structure of an integrated portable side-monitoring robot according to an embodiment of the present invention; Figure 4 This is a side cross-sectional view of the internal structure of an integrated portable side-monitoring robot according to an embodiment of the present invention; Figure 5 This is a top view of the integrated circuit board location of an integrated portable side-by-side robot according to an embodiment of the present invention; Figure 6 This is a bottom-view schematic diagram of an integrated portable side-view robot according to an embodiment of the present invention; Figure 7 This is a flowchart illustrating the intelligent on-site monitoring method for adaptive computing power scheduling according to the present invention.

[0019] In the picture: 1. Integrated portable robot; 2. Shell; 3. On-site alarm module; 4. Power module; 41. Removable battery; 42. Battery compartment; 43. Compartment cover; 44. Battery interface; 5. Integrated circuit board; 6. Computing chip; 7. Interface; 8. Switch module.

[0020] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0024] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0025] Please see Figures 1 to 6 This embodiment provides an intelligent on-site monitoring system with adaptive computing power scheduling, including an integrated portable on-site robot 1. The integrated portable on-site robot 1 includes interconnected hardware status sensing unit, thread pool management unit, latency monitoring unit, scheduling unit, and mode control unit; wherein; The hardware status sensing unit is used to collect the hardware status parameters of the integrated portable side-by-side robot 1 in real time; in this embodiment, the hardware status parameters collected in real time include at least the remaining power data, temperature data, and charging / discharging status data. The mode control unit is used to match the corresponding operating mode according to the hardware status parameters, so as to execute the upper limit of the global computing power resource occupancy threshold under the preset operating mode. The operating modes in this embodiment include real-time mode, energy efficiency balance mode, and battery life priority mode; each of the operating modes has a corresponding upper limit for the global computing power resource utilization threshold; wherein; In this embodiment, the upper limit of the global computing power resource occupancy threshold corresponding to the real-time mode is set to >90%; The energy efficiency balance mode, also known as the high-speed mode, has a global computing power resource utilization threshold upper limit set at 40%-60%. The aforementioned battery life priority mode, also known as low power mode, has a global computing resource utilization threshold set at 5%-20%. The thread pool management unit has multiple built-in algorithm models. The thread pool management unit classifies the multiple algorithm models into multiple algorithm model groups and allocates an independent inference thread pool to each algorithm model group. For example, the first inference thread pool is allocated an algorithm model group consisting of multiple general safety identification models, the second inference thread pool is allocated an algorithm model group consisting of multiple closed safety inspection models, the third inference thread pool is allocated an algorithm model group consisting of multiple high-altitude operation risk identification models, and so on, forming several algorithm model groups.

[0026] In practical applications, the integrated portable monitoring robot 1, acting as an edge terminal, doesn't try to handle everything at once. Instead, technicians select application scenarios based on the actual needs of the construction site, flexibly activating the necessary algorithm model groups for computation. This enables site monitoring tasks such as "personnel entry prohibited, not wearing safety helmets, and fire detection." The integrated portable monitoring robot 1 distributes tasks to different inference thread pools. For example, an algorithm model group composed of general safety identification models specifically monitors violations, while an algorithm model group composed of closed-loop safety inspection models specifically manages environmental safety. This decoupled design ensures that the intelligent monitoring system maintains high efficiency even when multiple scenarios are activated simultaneously.

[0027] The latency monitoring unit is used to acquire the real-time inference time of each algorithm model group in a single frame and calculate the maximum difference in real-time time between each algorithm model group. The scheduling unit is used to dynamically adjust the thread quota of each inference thread pool according to the maximum difference and the hardware status parameters, so that the inference cycle of each algorithm model group is consistent, and the total computing power utilization rate of each algorithm model group is controlled within the upper limit of the global computing power resource utilization rate threshold of the current running mode.

[0028] In this embodiment, thread quotas are added to the algorithm model group with long real-time inference time for single-frame inference, and redundant thread quotas are reclaimed for the algorithm model group with long real-time inference time for single-frame inference, so that the inference cycle of each algorithm model group tends to be consistent, and the time synchronization waiting delay caused by uneven distribution of computing power is eliminated.

[0029] This embodiment integrates a hardware status sensing unit, a thread pool management unit, a latency monitoring unit, a scheduling unit, and a mode control unit into a single device using an integrated portable side-monitoring robot 1. This allows hardware status parameters to participate in computing power scheduling decisions in real time. The hardware status sensing unit collects power, temperature, and charging / discharging status data in real time and inputs them to the mode control unit. The mode control unit matches the corresponding operating mode based on the hardware status and sets an upper limit for the global computing power resource occupancy rate under that mode, achieving hierarchical matching between hardware health status and computing power demand. The thread pool management unit allocates an independent inference thread pool to each algorithm model group for parallel processing. The latency monitoring unit obtains the inference time of each model group in single frame in real time and calculates the maximum difference, which is then input to the scheduling unit. The scheduling unit dynamically adjusts the thread quota of each inference thread pool based on the maximum difference and hardware status parameters. This ensures that the inference cycle of each algorithm model group is consistent while keeping the total computing power utilization rate within the upper limit of the current operating mode. Thus, while ensuring the synchronization of the concurrent inference sequence of multiple algorithms, it achieves real-time linkage between computing power resource allocation and hardware power supply mode, battery power, and device temperature. This ensures that the device can maintain a dynamic balance between computing power and power consumption under different operating conditions, significantly extending the continuous working time in battery power mode.

[0030] In this embodiment, the integrated portable on-site robot 1 further includes a computing chip 6, an integrated circuit board 5, a power module 4, a switch module 8, and a field alarm module 3. The computing chip 6, power module 4, switch module 8, and field alarm module 3 are all connected to the integrated circuit board 5. The hardware status perception unit, thread pool management unit, latency monitoring unit, scheduling unit, and mode control unit are all integrated into the integrated circuit board 5. In this embodiment, a high-power speaker is configured as the field alarm module 3 to directly broadcast monitoring and identification results at the construction site, enabling real-time on-site supervision.

[0031] As a further preferred embodiment, the integrated portable side-by-side robot 1 also includes a housing 2, in which the computing chip 6, integrated circuit board 5, power module 4, switch module 8 and field alarm module 3 are all disposed. The compartment cover 43 is closable and connected to the housing 2. The housing 2 is provided with multiple interfaces 7, and the integrated circuit board 5 is connected to external devices and / or networks through the interfaces 7.

[0032] In this embodiment, the integrated portable monitoring robot 1 connects to the camera via interface 7 located at the bottom of the housing 2, enabling power supply and data transmission to the camera. Interface 7 has a 12V-2A input, compatible with most 4-inch PTZ cameras on the market. Interface 7 also includes two network ports; either port can be connected to the camera to form a local area network, allowing the integrated portable monitoring robot 1 to retrieve the camera's video stream. The other port can connect to the construction site's local area network, enabling the camera and the integrated portable monitoring robot 1 to complete site networking. Compared to ordinary AI cameras, this embodiment possesses higher AI computing power, supporting multi-task recognition and analysis under complex working conditions.

[0033] In one embodiment, a handle is also provided on one side of the housing 2 to facilitate the carrying and installation of the integrated portable side-by-side robot 1 by technicians during deployment.

[0034] In one embodiment, the integrated circuit board 5 further includes a charging and discharging circuit, which sets a charging cutoff voltage and a discharging cutoff voltage to control the power supply and charging of the integrated portable side-by-side robot 1 and external devices.

[0035] In this embodiment, the integrated circuit board 5 is compatible with a 15V-24V power adapter as a charger. The charger prioritizes powering the device and simultaneously charges the removable battery 41, which is 12V-20000mAh. In this embodiment, the charging cutoff voltage is set to 12.6V, meaning that charging automatically stops when the charging voltage reaches 12.6V. When no charger is connected, power is drawn from the battery, and the discharge cutoff voltage is set to 11.1V.

[0036] In one embodiment, the power module 4 includes a removable battery 41, a battery compartment 42, a compartment cover 43, and a battery interface 44. The removable battery 41 is disposed in the battery compartment 42, the compartment cover 43 is hinged to the battery compartment 42, and the removable battery 41 is connected to the charging and discharging circuit through the battery interface 44. It is understood that this embodiment also includes a push-pull latch that makes the cover 43 a self-locking flip-top mechanism, allowing technicians at the construction site to quickly open the cover 43 to replace the battery without tools when the battery is depleted, significantly improving the equipment's range and maintainability in scenarios without a fixed power supply at the construction site; the removable battery 41 is connected to the charging and discharging circuit through the battery interface 44, and when the battery is low, simply replacing it with a fully charged battery can quickly restore the equipment to full-load operation, allowing it to continue executing high-performance multi-algorithm concurrent inference tasks without waiting for the built-in battery to charge; the hinged design of the flip-top mechanism ensures the reliability of the connection between the cover 43 and the battery compartment 42, effectively protecting the battery and battery interface 44 from dust and water stains in harsh construction site environments, and maintaining the electrical connection stability of the battery interface 44.

[0037] Please see Figure 7 The present invention also provides an intelligent on-site monitoring method for adaptive computing power scheduling, applied to the intelligent on-site monitoring system for adaptive computing power scheduling as described above, comprising the following steps: S1. Real-time acquisition of hardware status parameters of the integrated portable side-by-side robot 1, and acquisition of locally pre-stored algorithm configuration instructions; the hardware status parameters include at least remaining power data, temperature data, and charging / discharging status data; the algorithm configuration instructions include algorithm scenarios, algorithm priorities, operating mode parameter thresholds, and hardware linkage rules; the operating mode parameter thresholds refer to the pre-set critical values ​​for different operating modes used to limit the utilization rate of global computing resources (CPU / GPU / NPU), i.e., the upper bound of the utilization rate threshold; S2. Obtain the preset operating mode, and match the corresponding operating mode as the current operating mode according to the hardware status parameters and the algorithm configuration instructions; wherein, each of the operating modes is set with a corresponding upper limit of the global computing power resource occupancy rate threshold; the operating modes include real-time mode, energy efficiency balance mode and battery life priority mode; S3. Obtain multiple algorithm models, classify the multiple algorithm models to obtain multiple algorithm model groups, select the algorithm model group to be enabled according to the algorithm configuration instructions, and match an independent inference thread pool for each enabled algorithm model group to process video frame data in parallel. S4. Obtain the real-time time of each inference thread pool to complete a single frame inference, and calculate the maximum difference in real-time time between each algorithm model group; S5. Obtain the preset hardware linkage rules, and perform hardware status compliance verification based on the maximum difference and the hardware status parameters, and determine whether the maximum difference is within the preset synchronization tolerance; if the verification passes and the maximum difference exceeds the preset synchronization tolerance, dynamically adjust the thread quota of each inference thread pool to make the inference cycle of each algorithm model consistent.

[0038] The hardware linkage rule in this embodiment is a pre-defined criterion for judging the health of the hardware itself. For example: Temperature rule: If the chip temperature is >85℃, thread quota increases are prohibited, and frequency reduction protection is activated.

[0039] Battery rule: If the battery level is less than 20% and the system is powered by battery, new algorithm models are prohibited from being started, and the thread quota of non-core inference thread pools will be reduced.

[0040] Power supply rules: If the power supply is battery-powered and the battery level is below the preset threshold, the thread quota adjustment range must be halved.

[0041] In this embodiment, if the hardware status compliance check passes, it indicates that the hardware status is good, and dynamic adjustment of the thread quota can be performed. If the check fails, it indicates that the hardware status is critical. In this case, it is not advisable to push the hardware to its limits in order to pursue algorithm timing synchronization, so dynamic adjustment of the thread quota of each inference thread pool is prohibited. Instead, the thread quota of each inference thread pool should be reduced synchronously.

[0042] S6. Determine whether the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode; if yes, readjust the thread quota and return to step S4; if no, maintain the current thread quota of each inference thread pool and complete the adaptive scheduling of computing power.

[0043] In this embodiment, hardware status parameters and algorithm configuration instructions provide real-time hardware health data and preset control basis for subsequent scheduling decisions; step S2 matches the corresponding real-time mode, energy efficiency balance mode, or battery life priority mode as the current operating mode according to the hardware status parameters, so that the intelligent side-by-side monitoring system's operating strategy is dynamically aligned with the hardware's current power supply status and health level; step S3 allocates an independent inference thread pool to each activated algorithm model group to process video frame data in parallel; step S4 monitors the real-time inference time of each thread pool for a single frame and calculates the maximum difference, thereby quantifying the timing synchronization deviation of multiple concurrent algorithms; step S5 combines hardware linkage rules to perform hardware status compliance verification. When the verification passes and the maximum difference exceeds the preset synchronization tolerance, the thread quota of each inference thread pool is dynamically adjusted to make the inference cycle of each algorithm model consistent, thus achieving multi-algorithm timing synchronization within the hardware safety boundary. Step S6 determines whether the adjusted total computing power utilization rate exceeds the upper limit of the utilization rate threshold of the current operating mode. If it does, the thread quota is readjusted until the constraint is met, thereby ensuring that the computing power resource utilization is always controlled within the preset threshold range of the current operating mode. Finally, real-time linkage scheduling of hardware status and computing power demand is realized, enabling the device to maintain the balance between computing power and power consumption through mode switching and dynamic adjustment of thread quota in battery-powered mode, effectively extending the continuous working time.

[0044] As a further embodiment, the dynamic adjustment of the thread quota of each inference thread pool in step S5 specifically includes the following steps: The direction of thread quota adjustment is determined based on the charge / discharge status data in the hardware status parameters. The thread quota adjustment coefficient is calculated based on the remaining power data and the temperature data. The thread quota adjustment coefficient is negatively correlated with the remaining power and the temperature. Based on the thread quota adjustment direction and the thread quota adjustment coefficient, the thread quota adjustment amount for each inference thread pool is generated; Wherein; when the charge / discharge status data is "external power supply", the thread quota adjustment coefficient ; When the charge / discharge status data is "battery powered", the thread quota adjustment coefficient Determined according to the following formula: in, This is the current remaining battery level. Rated power, The current temperature. To ensure the upper limit of safe operating temperature, This is the lower limit of the normal operating temperature.

[0045] In this embodiment, the direction of thread quota adjustment is determined based on charge / discharge status data. This allows the intelligent side-station monitoring system to prioritize performance when powered by an external power source and prioritize battery life when powered by a battery, achieving dynamic adaptation between power supply mode and computing power scheduling intent. A thread quota adjustment coefficient negatively correlated with remaining power and temperature is calculated based on remaining power and temperature data. This coefficient automatically decreases when powered by a battery and the remaining power decreases or the temperature increases, thus constraining the increment of thread quota. The thread quota adjustment amount for each inference thread pool is generated based on the adjustment direction and adjustment coefficient, ensuring that the expansion of thread quota in battery-powered mode is doubly suppressed by remaining power and temperature levels. The thread quota adjustment coefficient is adjusted when the charge / discharge status data indicates external power supply. This allows for sufficient adjustment of thread quotas to achieve consistency in algorithm inference cycles; the thread quota adjustment factor is adjusted when the charge / discharge state data is battery powered. Based on the quantitative formula of power and temperature, the adjustment coefficient is smaller when the power is lower and the temperature is higher, and the adjustment range of thread quota is more limited. Thus, when the battery is powered and the power is insufficient or the temperature is too high, the expansion of computing power is actively suppressed, and the total computing power utilization rate is controlled within the threshold of the current operating mode, maintaining the device's continuous operation capability under limited power and preventing overheating.

[0046] In another embodiment, the dynamic adjustment of thread quotas for each inference thread pool described in step S5 specifically includes the following steps: Obtain the average time taken for each inference thread pool to complete a single frame of inference, and determine whether the real-time time is greater than the average time. If yes, mark the algorithm model group in the inference thread pool where the real-time time is greater than the average time as a slow model. If no, mark the algorithm model group in the inference thread pool where the real-time time is less than the average time as a fast model. Reclaim a first number of thread quotas from the inference thread pool of the fast model; Obtain a preset allocation ratio, and allocate the reclaimed thread quota to the inference thread pool of the slow model according to the preset allocation ratio; wherein, the preset allocation ratio is proportional to the deviation between the real-time consumption time and the average consumption time of each slow model.

[0047] This embodiment achieves accurate identification of the differences in computing power requirements during concurrent operation of multiple algorithms by marking slow and fast models. A first number of thread quotas are reclaimed from the inference thread pool of the fast model, releasing idle or redundant computing resources from the faster algorithms. A preset allocation ratio is obtained, and the reclaimed thread quotas are allocated to the inference thread pool of the slow model according to this ratio. The preset allocation ratio is proportional to the deviation between the real-time latency and the average latency of each slow model, so that the slower the processing speed and the larger the latency deviation, the more additional thread quotas the algorithm model receives. This ensures that limited computing resources are precisely allocated to the most scarce algorithm stages. Through this differentiated allocation mechanism based on latency deviation, the inference speed of each slow model is specifically improved, ultimately achieving consistency in the inference cycle of all algorithm models. This achieves temporal synchronization of concurrent inference of multiple algorithms without increasing the total computing power occupancy rate, avoiding overall waiting delays caused by the lag in processing individual algorithms, and improving the parallel processing efficiency of the intelligent monitoring system.

[0048] As a further preferred embodiment, the readjustment of thread quotas in step S6 includes: When the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode, the thread quota of the inference thread pool corresponding to the algorithm model with lower algorithm priority will be reduced first.

[0049] In one embodiment, after step S6, the following steps are further included: Execute the corresponding computing power management strategy according to the current operating mode; where; If the current operating mode is real-time mode, then the thread rate limiting is cancelled, and the inference thread pool is allowed to expand as needed until the computing power utilization rate reaches the upper limit of the utilization rate threshold of real-time mode; If the current operating mode is the energy efficiency balance mode, then the total number of concurrent threads will be limited while maintaining the consistency of the inference cycle of each algorithm model, and the total computing power utilization rate will be controlled within the upper limit of the utilization rate threshold of the energy efficiency balance mode. If the current operating mode is the battery life priority mode, then the frame skipping sampling strategy is executed to obtain the non-core algorithm model, reduce the thread quota of the inference thread pool of the non-core algorithm model, and control the total computing power utilization rate within the upper limit of the utilization rate threshold of the battery life priority mode.

[0050] This embodiment achieves deep binding between the operating mode and the underlying computing power scheduling logic by executing corresponding computing power management strategies according to the current operating mode. If the current operating mode is real-time mode, thread rate limiting is lifted and the inference thread pool is allowed to expand as needed until the computing power utilization rate reaches the upper limit of the real-time mode's utilization rate threshold. This allows the intelligent side-by-side monitoring system to operate at full load in scenarios with external power supply or extremely high response speed requirements, minimizing recognition latency. If the current operating mode is energy efficiency balance mode, the total number of concurrent threads is limited while maintaining the consistency of the inference cycle of each algorithm model. This controls the total computing power utilization rate within the upper limit of the energy efficiency balance mode's utilization rate threshold, ensuring that the intelligent side-by-side monitoring system can guarantee multi-processing... While synchronizing the timing of the algorithm, the system avoids long-term full-load operation of the hardware, effectively suppressing device heat generation and maintaining a stable energy efficiency ratio. If the current operating mode is the battery life priority mode, the system executes a frame skipping sampling strategy and obtains non-core algorithm models. It reduces the thread quota of the inference thread pool of the non-core algorithm models, keeping the total computing power utilization rate within the upper limit of the utilization rate threshold of the battery life priority mode. This allows the intelligent side-station monitoring system to significantly reduce computing power consumption by reducing the sampling frequency and shutting down non-core algorithms when the battery is powered and the power is limited. It prioritizes the continuous operation of the core monitoring algorithm with limited power, thereby achieving a dynamic balance between hardware status and business needs. This ensures that the device can maintain stable operation with the optimal computing power-power ratio under different operating conditions.

[0051] In one embodiment, according to the current operating mode, the corresponding computing power management strategy is executed, and a computing power-power balance operating state matching the operating mode is output; if the computing power-power balance operating state is detected to deviate from the preset range, the dynamic adjustment logic of steps S4 to S6 is re-triggered to make the intelligent side-by-side monitoring system maintain a stable operating state that is strongly consistent with the hardware state and the preset operating mode.

[0052] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An intelligent on-site monitoring system with adaptive computing power scheduling, characterized in that, The system includes an integrated portable side-monitoring robot, which internally comprises interconnected hardware status sensing units, thread pool management units, latency monitoring units, scheduling units, and mode control units. in; The hardware status sensing unit is used to collect the hardware status parameters of the integrated portable side-by-side robot in real time. The mode control unit is used to match the corresponding operating mode according to the hardware status parameters, so as to execute the upper limit of the global computing power resource occupancy threshold under the preset operating mode. The thread pool management unit has multiple built-in algorithm models. The thread pool management unit classifies the multiple algorithm models into multiple algorithm model groups and allocates an independent inference thread pool to each algorithm model group. The latency monitoring unit is used to acquire the real-time inference time of each algorithm model group in a single frame and calculate the maximum difference in real-time time between each algorithm model group. The scheduling unit is used to dynamically adjust the thread quota of each inference thread pool according to the maximum difference and the hardware status parameters, so that the inference cycle of each algorithm model group is consistent, and the total computing power utilization rate of each algorithm model group is controlled within the upper limit of the global computing power resource utilization rate threshold of the current running mode.

2. The intelligent on-site monitoring system with adaptive computing power scheduling according to claim 1, characterized in that, The integrated portable side-by-side robot includes a computing chip, an integrated circuit board, a power module, a switch module, and a field alarm module. The computing chip, power module, switch module, and field alarm module are all connected to the integrated circuit board. The hardware status perception unit, the thread pool management unit, the latency monitoring unit, the scheduling unit, and the mode control unit are all integrated on the integrated circuit board.

3. The intelligent on-site monitoring system with adaptive computing power scheduling according to claim 2, characterized in that, The integrated portable on-site robot also includes a housing, in which the computing chip, integrated circuit board, power module, switch module and on-site alarm module are all housed. The housing has multiple interfaces, and the integrated circuit board is connected to external devices and / or networks through these interfaces.

4. The intelligent on-site monitoring system with adaptive computing power scheduling according to claim 3, characterized in that, The integrated circuit board also includes a charging and discharging circuit, which sets a charging cutoff voltage and a discharging cutoff voltage to control the power supply and charging of the integrated portable side-by-side robot and external devices.

5. The intelligent on-site monitoring system with adaptive computing power scheduling according to claim 4, characterized in that, The power module includes a removable battery, a battery compartment, a compartment cover, and a battery interface. The removable battery is disposed in the battery compartment, the compartment cover is hinged to the battery compartment, and the removable battery is connected to the charging and discharging circuit through the battery interface.

6. A method for intelligent on-site monitoring with adaptive computing power scheduling, applied to the intelligent on-site monitoring system with adaptive computing power scheduling as described in any one of claims 1-5, characterized in that, Includes the following steps: S1. Real-time acquisition of hardware status parameters of the integrated portable side-monitoring robot, and acquisition of locally pre-stored algorithm configuration instructions; among which; The hardware status parameters include at least remaining power data, temperature data, and charge / discharge status data; the algorithm configuration instructions include algorithm scenarios, algorithm priorities, operating mode parameter thresholds, and hardware linkage rules. S2. Obtain the preset operating mode, and match the corresponding operating mode as the current operating mode according to the hardware status parameters and the algorithm configuration instructions; wherein, each of the operating modes is set with a corresponding upper limit of the global computing power resource occupancy rate threshold; the operating modes include real-time mode, energy efficiency balance mode and battery life priority mode; S3. Acquire multiple algorithm models and video frame data, classify the multiple algorithm models to obtain multiple algorithm model groups, select the algorithm model group to be enabled according to the algorithm configuration instructions, and match an independent inference thread pool for each enabled algorithm model group to process video frame data in parallel. S4. Obtain the real-time time of each inference thread pool to complete a single frame inference, and calculate the maximum difference in real-time time between each algorithm model group; S5. Obtain the preset hardware linkage rules, and perform hardware status compliance verification based on the maximum difference and the hardware status parameters, and determine whether the maximum difference is within the preset synchronization tolerance; if the verification passes and the maximum difference exceeds the preset synchronization tolerance, dynamically adjust the thread quota of each inference thread pool to make the inference cycle of each algorithm model consistent. S6. Determine whether the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode; if yes, readjust the thread quota and return to step S4; if no, maintain the current thread quota of each inference thread pool and complete the adaptive scheduling of computing power.

7. The intelligent on-site monitoring method for adaptive computing power scheduling according to claim 6, characterized in that, The dynamic adjustment of thread quotas for each inference thread pool described in step S5 specifically includes the following steps: The direction of thread quota adjustment is determined based on the charge / discharge status data in the hardware status parameters. The thread quota adjustment coefficient is calculated based on the remaining power data and the temperature data. The thread quota adjustment coefficient is negatively correlated with the remaining power and the temperature. Based on the thread quota adjustment direction and the thread quota adjustment coefficient, the thread quota adjustment amount for each inference thread pool is generated; in ； When the charge / discharge status data is "external power supply", the thread quota adjustment coefficient ; When the charge / discharge status data is "battery powered", the thread quota adjustment coefficient Determined according to the following formula: in, This is the current remaining battery level. Rated power, The current temperature. To ensure the upper limit of safe operating temperature, This is the lower limit of the normal operating temperature.

8. The intelligent on-site monitoring method for adaptive computing power scheduling according to claim 6, characterized in that, The dynamic adjustment of thread quotas for each inference thread pool described in step S5 specifically includes the following steps: Obtain the average time taken for each inference thread pool to complete a single frame of inference, and determine whether the real-time time is greater than the average time. If yes, mark the algorithm model group in the inference thread pool where the real-time time is greater than the average time as a slow model. If no, mark the algorithm model group in the inference thread pool where the real-time time is less than the average time as a fast model. Reclaim a first number of thread quotas from the inference thread pool of the fast model; Obtain a preset allocation ratio, and allocate the reclaimed thread quota to the inference thread pool of the slow model according to the preset allocation ratio; wherein, the preset allocation ratio is proportional to the deviation between the real-time consumption time and the average consumption time of each slow model.

9. The intelligent on-site monitoring method for adaptive computing power scheduling according to claim 6, characterized in that, Step S6, which involves readjusting thread quotas, includes: When the total computing power utilization rate of each inference thread pool after adjustment exceeds the upper limit of the utilization rate threshold of the current running mode, the thread quota of the inference thread pool corresponding to the algorithm model with lower algorithm priority will be reduced first.

10. The intelligent on-site monitoring method for adaptive computing power scheduling according to claim 6, characterized in that, Step S6 is followed by the following steps: Execute the corresponding computing power management strategy according to the current operating mode; where; If the current operating mode is real-time mode, then the thread rate limiting is cancelled, and the inference thread pool is allowed to expand as needed until the computing power utilization rate reaches the upper limit of the utilization rate threshold of real-time mode; If the current operating mode is the energy efficiency balance mode, then the total number of concurrent threads will be limited while maintaining the consistency of the inference cycle of each algorithm model, and the total computing power utilization rate will be controlled within the upper limit of the utilization rate threshold of the energy efficiency balance mode. If the current operating mode is the battery life priority mode, then the frame skipping sampling strategy is executed to obtain the non-core algorithm model, reduce the thread quota of the inference thread pool of the non-core algorithm model, and control the total computing power utilization rate within the upper limit of the utilization rate threshold of the battery life priority mode.

Images