Multi-layer guiding method for initialization and recovery of reliable data collector

Abstract

The invention discloses a multi-layer guiding method for initialization and recovery of a reliable data collector, and relates to the technical field of data detection. Obtaining operation data of each power supply terminal, and preprocessing each operation data to obtain multiple pieces of target data; encrypting each piece of target data to obtain a plurality of pieces of encrypted data, and uploading each piece of encrypted data to a central management server, so that the central management server detects each piece of encrypted data to obtain an anomaly detection result; and determining a resource allocation strategy of each power supply terminal according to an abnormal detection result, and sending each resource allocation strategy as an update packet to each power supply terminal. The high-precision sensor collects operation data of the power supply terminal, preprocesses the operation data, encrypts and transmits the operation data, the central management server intelligently analyzes and detects abnormity, a resource allocation strategy is determined accordingly, high-quality data, safe transmission, accurate detection and efficient allocation are achieved, and the reliability and expandability of the system are enhanced.

Description

Technical Field

[0001] The invention belongs to the technical field of data detection, and in particular relates to a multi-layer booting method for initializing and recovering a reliable data collector. Background Art

[0002] Most traditional power terminal systems are designed with a linear initialization process, which is prone to single points of failure. Each component is initialized sequentially, meaning each step depends on the successful completion of the previous one. If one component fails, the terminal cannot proceed to the next step in the sequence, causing the entire process to halt.

[0003] Through a linear initialization process, each component is initialized in sequence; data processing is done through manual analysis or preset threshold judgment; updates are pushed uniformly at fixed time periods; data transmission uses a single communication path, the encryption method is simple, and fault recovery relies on manual operation.

[0004] Traditional power terminal systems use a linear initialization process, which is prone to single point failures and lacks fault tolerance. Manual intervention is required in the event of a failure and recovery options are limited. At the same time, they face scalability challenges as the scale of distributed energy systems expands, making it difficult to meet system integrity and reliability requirements. Summary of the Invention

[0005] The purpose of the present invention is to solve:

[0006] 1. Traditional power terminal systems use a linear initialization process, which is prone to single point failures and lacks fault tolerance. Failure of any component during the initialization phase can cause the entire system to crash.

[0007] 2. Manual intervention is required in the event of a fault, resulting in delayed grid activation, increased costs, and the possibility of human error during the fault resolution process;

[0008] 3. If the terminal fails to initialize, the system may need to completely reset or restart the entire terminal. Failure to recover autonomously will result in low system efficiency and reduced reliability.

[0009] 4. Distributed energy systems face scalability challenges as they scale. The increasing number of power terminals makes management and recovery from single-point failures more difficult, making it difficult to meet system integrity and reliability requirements.

[0010] A multi-layer bootstrapping method for reliable data collector initialization and recovery is proposed.

[0011] In a first aspect of the present invention, a multi-layer boot method for reliable data collector initialization and recovery is first proposed, the method comprising:

[0012] Acquire operation data of each power supply terminal, and pre-process each operation data to obtain multiple target data; the operation data includes: voltage data, current data, temperature and humidity data, and vibration data;

[0013] Encrypting each target data to obtain a plurality of encrypted data, and uploading each encrypted data to a central management server so that the central management server detects each encrypted data to obtain an anomaly detection result;

[0014] The resource allocation strategy of each power supply terminal is determined according to the abnormality detection result, and each resource allocation strategy is sent to each power supply terminal as an update package.

[0015] Optionally, current data sampling is performed according to a preset frequency; the preset frequency is 10 times / second.

[0016] Optionally, data denoising is performed on the voltage data, current data, temperature and humidity data, and vibration data in the operating data to obtain denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data, and the denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data are used as target data.

[0017] Optionally, the target data is encrypted using an AES-256 encryption module.

[0018] Optionally, determining a resource allocation strategy for each power terminal based on the anomaly detection result includes:

[0019] If there is an abnormal detection result, it is determined that the power supply terminal has an abnormality;

[0020] Obtain the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority of the target terminal, and calculate the resource demand parameters based on the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority;

[0021] Optimizing resources of the power supply terminal according to the resource demand parameters to obtain a resource allocation plan;

[0022] Resource requirement parameter calculation formula:

[0023] Among them, RDI i Represents the resource demand parameter, L i Represents the real-time load rate of terminal i, L i，max represents the maximum allowable load rate of terminal i, P i Represents the terminal priority, S i Represents resource utilization, S i，max represents the total amount of resources of terminal i, α, β, and γ are all constant proportional coefficients, and α+β+γ=1.

[0024] Optionally, obtain the historical periodic load of each power terminal, determine the load window period of the power terminal based on the historical periodic load, generate an update package, and push the update package to each power terminal so that each power terminal receives the update package and performs the update.

[0025] Optionally, the operating data is used as training data, and the historical database is updated according to the training data to obtain a target database, so that the machine learning model is trained according to the target database to obtain a fault classification model; the fault classification model is used to classify the operating data.

[0026] Beneficial effects of the present invention:

[0027] This invention proposes a multi-layer bootstrapping method for reliable data collector initialization and recovery. The method obtains operational data from each power supply terminal, preprocesses each operational data to obtain multiple target data sets, encrypts each target data set to obtain multiple encrypted data sets, and uploads each encrypted data set to a central management server, which then detects anomalies. The central management server then determines a resource allocation strategy for each power supply terminal based on the anomaly detection results, and sends each resource allocation strategy as an update package to each power supply terminal. High-precision sensors collect the power supply terminal operational data, preprocess it, and encrypt it for transmission. The central management server intelligently analyzes and detects anomalies, and determines resource allocation strategies accordingly. This method achieves high data quality, secure transmission, accurate detection, and efficient allocation, enhancing system reliability and scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below with reference to the accompanying drawings.

[0029] Figure 1 The present invention provides a flowchart of a multi-layer boot method for reliable data collector initialization and recovery. DETAILED DESCRIPTION

[0030] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments represent only a portion of the embodiments of the present invention, not all of them. The term "and / or" herein simply describes an association relationship between associated objects, indicating that three possible relationships exist. For example, "A" and "B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, references to "first," "second," and so on in the present invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include at least one of these features. Furthermore, the technical solutions of the various embodiments may be combined, but only if they are achievable by a person of ordinary skill in the art. If a combination of technical solutions contradicts or is unachievable, such combination shall be deemed non-existent and outside the scope of protection claimed by the present invention.

[0031] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0032] The embodiment of the present invention provides a multi-layer boot method for reliable data collector initialization and recovery. Figure 1 , Figure 1 A flowchart of a multi-layer boot method for reliable data acquisition initialization and recovery provided by an embodiment of the present invention. The method includes the following steps:

[0033] S101, obtaining operation data of each power supply terminal, and preprocessing each operation data to obtain a plurality of target data;

[0034] S102, encrypting each target data to obtain a plurality of encrypted data, and uploading each encrypted data to a central management server, so that the central management server detects each encrypted data to obtain an anomaly detection result;

[0035] S103: Determine a resource allocation strategy for each power supply terminal according to the abnormality detection result, and send each resource allocation strategy as an update package to each power supply terminal.

[0036] Operation data includes: voltage data, current data, temperature and humidity data, and vibration data;

[0037] Based on a multi-layer boot method for initializing and recovering a reliable data collector provided by an embodiment of the present invention, high-precision sensors are used to collect power terminal operating data, pre-process it, and encrypt it for transmission. A central management server intelligently analyzes and detects anomalies, and determines a resource allocation strategy based on this, thereby achieving high data quality, secure transmission, accurate detection, and efficient allocation, thereby enhancing system reliability and scalability.

[0038] In one implementation, a voltage sensor with ±0.1% accuracy, a current sensor updating every 0.1 seconds, and environmental monitoring sensors (temperature, humidity, and vibration) are deployed to achieve full, real-time awareness of the power supply terminal's operating status. The embedded microcontroller pre-processes the raw data using hardware filtering circuits and multiple verification mechanisms (range verification, logic verification, and redundant sensor cross-validation), effectively eliminating noise and invalid values. This ensures the high reliability, consistency, and availability of the target data transmitted to the central management server, providing a solid data foundation for subsequent anomaly detection and resource allocation.

[0039] In one implementation, an AES-256 hardware encryption module is used to perform end-to-end encryption on preprocessed data. This is combined with primary / secondary dual communication paths (e.g., wired Ethernet + wireless LTE) and a 10-millisecond failover mechanism to create a secure and reliable data transmission link. Hardware-level TLS encryption ensures confidentiality and integrity during data transmission, protecting against cyberattacks. The dual-path redundancy design eliminates the risk of interruption associated with traditional single-link communications. Even if the primary communication path fails, the system can seamlessly switch to the backup path, ensuring uninterrupted data flow. Field measurements have shown that data transmission latency is consistently controlled within 10 milliseconds, improving the real-time performance and reliability of scenarios such as smart grids.

[0040] In one implementation, the Central Management Server (CMS) conducts in-depth analysis of encrypted and transmitted target data by integrating statistical cleaning algorithms, machine learning models (such as unsupervised learning), and historical baseline comparison mechanisms. The system can automatically identify abnormal conditions that deviate from the normal operating range (such as voltage overlimit, load mutation), and issue graded warnings based on the severity of the abnormality (critical / serious / medium / low). Compared with the traditional manual inspection mode, this mechanism shortens the anomaly detection time from hours to seconds, and can predict potential failures in advance (such as analyzing the probability of load mutation through historical trends), achieving a leap from passive response to active prevention, significantly reducing the risk of system downtime and operation and maintenance costs.

[0041] In one implementation, based on the anomaly detection results, the central management server generates a differentiated resource allocation strategy through a cost-optimized linear programming algorithm (simple scenario) or a heuristic algorithm (complex multi-terminal scenario). This strategy combines terminal priority (such as priority for key equipment such as distribution stations), real-time load rate, and remaining resources to achieve economical and efficient allocation of key resources such as computing power and bandwidth. After receiving the update package, the terminal autonomously adjusts the local configuration (such as voltage set point, task scheduling) through the hardware-controlled allocation module, and feeds back performance data to the CMS in real time to form a closed-loop optimization. Actual measurements have shown that this mechanism can increase resource utilization by more than 30%, reduce the failure rate of key equipment by 45%, and enhance the system's resilience in high-load or emergency scenarios.

[0042] In one implementation, a multi-level redundancy and automatic recovery system is built through a closed-loop management process encompassing "data collection - encrypted transmission - intelligent analysis - policy execution - real-time feedback." A central management server pushes update packages during low-load periods, while terminals utilize isolated installation environments (e.g., independent boot partitions) and hardware-level rollback mechanisms to improve the success rate of software updates and restore to stable versions in the event of failure. A distributed storage system and high-throughput message broker support the simultaneous access of thousands of terminals, with data processing capacity scaling linearly with terminal size. This thoroughly eliminates the single points of failure and performance bottlenecks that plague traditional systems in large-scale deployments, providing a highly reliable and scalable underlying architecture for distributed energy systems such as smart grids.

[0043] In one implementation, the multi-layer bootstrapping method divides initialization into three layers: the first layer (corresponding to step S101) activates the primary and backup power supplies and performs self-tests, automatically switching to the backup power supply in the event of a primary power failure; the second layer (corresponding to step S102) establishes a secure connection via Ethernet / LTE dual communication channels and hardware TLS encryption, automatically switching channels in the event of a failure; and the third layer (corresponding to step S103) deploys multiple sensors to cross-validate monitoring parameters, with redundant sensors automatically taking over in the event of a failure. This layered redundant architecture and automatic switching mechanism eliminates the single point of failure risk inherent in traditional linear processes, ensuring the continuity and reliability of power supply, data transmission, and status monitoring, and improving the system's fault tolerance and scalability during the initialization phase.

[0044] In one implementation, the implementation of a multi-layered bootstrap approach yielded significant results and benefits, collectively enhancing the performance, reliability, and scalability of the power terminal system. By distributing initialization responsibilities across multiple tiers and incorporating redundant hardware paths, the system eliminates single points of failure. This ensures that the distribution network remains operational even if individual components or communication paths fail. Automatic failover mechanisms and rapid recovery protocols minimize system downtime. In the event of a failure, the system automatically switches to an alternate path and restores functionality without requiring human intervention, maintaining continuous power distribution. The tiered redundant architecture enables scalable deployment and can accommodate growth in the number of power terminals without compromising system performance. A high-throughput message broker and distributed storage solution enable the system to efficiently handle increasing data volumes. Hardware-based encryption modules and secure communication protocols ensure data integrity and confidentiality, protecting the system from potential cyber threats and unauthorized access. Autonomous configuration adjustments and seamless application updates simplify maintenance processes and reduce the need for human intervention. This improves resource utilization and reduces operating costs.

[0045] In one embodiment, current data sampling is performed according to a preset frequency; the preset frequency is 10 times / second.

[0046] In one implementation, current data sampling is performed at a preset frequency (10 times / second), which can achieve high-frequency real-time perception of the current status of the power supply terminal, provide high-density data support for the system to dynamically monitor load changes and quickly respond to anomalies, and improve the timeliness and continuity of operating status monitoring.

[0047] In one implementation, high-frequency sampling gives current data higher temporal resolution, making it easier to analyze load trends and identify transient anomalies through historical data sequences. This provides a basis for the central management server to perform precise resource allocation and fault prediction, thereby enhancing the stability and reliability of system operation.

[0048] In one embodiment, the voltage data, current data, temperature and humidity data, and vibration data in the operating data are subjected to data denoising to obtain denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data, and the denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data are used as target data.

[0049] In one implementation, at system startup, each power terminal is equipped with a comprehensive suite of high-precision embedded sensors, including a voltage sensor with ±0.1% accuracy, a current sensor that updates every 0.1 seconds, a load condition monitor, and environmental sensors such as temperature, humidity, and vibration detectors. These sensors interface with a powerful embedded microcontroller, which performs initial data preprocessing. This preprocessing includes hardware-based filtering to reduce noise and validation checks to discard any anomalous or invalid readings, ensuring data integrity before transmission.

[0050] In one implementation, the voltage, current, temperature, humidity, and vibration data are subjected to noise reduction processing separately to effectively eliminate noise interference in the signal (such as electromagnetic interference and environmental fluctuation errors), making the target data closer to the actual operating status, providing a reliable data basis for the load trend prediction, anomaly detection and other analyses of the central management server, and reducing the risk of misjudgment due to data distortion.

[0051] In one implementation, noise reduction processing is used to reduce the invalid components in data fluctuations, making key features such as voltage overlimit, current mutation, temperature and humidity abnormalities easier to identify, thereby improving the system's detection sensitivity and accuracy for potential faults in power terminals (such as poor contact and equipment overheating), ensuring the effectiveness of system operation status monitoring, and providing an accurate basis for subsequent resource allocation and fault response.

[0052] In one implementation, data processing operations include: 1. hardware-level filtering circuits (such as low-pass filtering) to eliminate high-frequency noise (such as switching power supply interference) in current / voltage signals, smooth data curves, improve signal integrity, and facilitate analysis of load trends; 2. range verification and logic verification to eliminate invalid data such as voltage values ​​exceeding the rated range and negative current values ​​to ensure that the data conforms to physical laws and avoid abnormal values ​​caused by sensor failures from contaminating the analysis results; 3. redundant sensor cross-validation to perform consistency comparison on multi-sensor data of the same parameter (such as voltage) to eliminate mismeasurements caused by single sensor failure or drift and enhance data credibility; 4. environmental data decoupling processing to separate background interference (such as ventilation system vibration) in temperature, humidity and vibration data, accurately locate characteristic quantities related to the operating status of the equipment, and improve the contribution rate of environmental factors to fault warning.

[0053] In one embodiment, the target data is encrypted using an AES-256 encryption module.

[0054] In one implementation, the target data is encrypted using the AES-256 encryption module, and its high-strength encryption algorithm with a 256-bit key length is used to effectively resist brute force cracking and conventional network attacks, ensuring that sensitive operating data such as voltage and current are not illegally stolen or monitored during transmission between the terminal and the central management server, meeting the security requirements for data confidentiality in scenarios such as smart grids.

[0055] In one implementation, the encryption process is combined with hardware-level error checking mechanisms (such as checksum verification) to detect in real time whether data has been tampered with or damaged during transmission, ensuring that the encrypted data received by the central management server is completely consistent with the original target data sent by the terminal, providing a reliable data foundation for subsequent anomaly detection, resource allocation and other analyses.

[0056] In one implementation, a hardware-implemented AES-256 encryption module within each power endpoint encrypts pre-processed data. This encryption ensures confidentiality and integrity during data transmission. To ensure uninterrupted data flow, each endpoint is configured with dual communication paths—typically a primary wired Ethernet interface and a secondary wireless LTE module. This redundancy ensures that if the primary communication path fails, the system can seamlessly failover to the secondary path without interrupting data transmission. The hardware configuration is optimized to maintain transmission latency below 10 milliseconds, supporting real-time data applications. These redundant paths facilitate data transmission from each endpoint to the Central Management Server (CMS), ensuring continuous connectivity and data flow. The CMS, deployed on a high-availability server cluster with redundant power supplies and network interfaces, aggregates incoming data streams from all connected power endpoints. It employs a high-throughput message broker managed by a load balancer such as HAProxy or Nginx to efficiently handle simultaneous communications from thousands of endpoints. This setup ensures scalable and fault-tolerant data ingestion, preventing bottlenecks and maintaining system performance under high load conditions.

[0057] In one embodiment, determining a resource allocation strategy for each power terminal based on anomaly detection results includes:

[0058] If there is an abnormal detection result, it is determined that the power supply terminal has an abnormality;

[0059] Obtain the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority of the target terminal, and calculate the resource demand parameters based on the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority;

[0060] Optimize the resources of the power terminal according to the resource demand parameters to obtain a resource allocation plan;

[0061] Resource requirement parameter calculation formula:

[0062] Among them, RDI i Represents the resource demand parameter, L i Represents the real-time load rate of terminal i, L i，max represents the maximum allowable load rate of terminal i, P i Represents the terminal priority, S i Represents resource utilization, S i，max represents the total amount of resources of terminal i, α, β, and γ are all constant proportional coefficients, and α+β+γ=1.

[0063] In one implementation, abnormality of the power supply terminal is determined by the abnormality detection results, and operational hazards such as voltage overlimit and load mutation can be identified in a timely manner, providing clear goals for subsequent resource optimization, improving the system's early warning capability and response efficiency to potential faults, and ensuring the stability of the power supply terminal operation.

[0064] In one implementation method, resource demand parameters are calculated by comprehensively considering parameters such as real-time load rate, maximum load rate, resource utilization rate, total resource amount and terminal priority. A quantitative evaluation model is established from dimensions such as power intensity, safety threshold, resource efficiency and equipment criticality to ensure that the resource allocation strategy fully reflects the actual needs of the terminal and avoids allocation deviation caused by a single indicator.

[0065] In one implementation, based on the resource demand parameter calculation formula, the weights are flexibly configured through a constant proportional coefficient, enabling the system to dynamically adjust the evaluation focus according to business needs (such as reliability priority or efficiency priority, 1 for key terminals such as distribution stations, and 0.5 for ordinary terminals), generate scientific and adaptive resource allocation plans for power terminals, and improve the utilization efficiency and allocation rationality of key resources such as computing power and bandwidth. Resource optimization is implemented through, for example, meta-heuristic algorithms such as particle swarm optimization algorithm, genetic algorithm simulated annealing algorithm, etc.

[0066] In one embodiment, the historical periodic load of each power terminal is obtained, the load window period of the power terminal is determined based on the historical periodic load, an update package is generated, and the update package is pushed to each power terminal so that each power terminal receives the update package and performs the update.

[0067] In one implementation method, the load window period is determined by obtaining the historical cycle load of the power terminal, and the update package is pushed during the period when the terminal business demand is low, so as to avoid the update process running in parallel with high-load business, minimize the risk of service interruption, and ensure the continuous and stable operation of the terminal.

[0068] In one implementation, an update strategy is generated based on historical load data to make the update window match the actual operation rules of the terminal, change the blindness of traditional random or fixed period updates, improve the fit between the update operation and the terminal load characteristics, and reduce the probability of update failure.

[0069] In one implementation, the CMS (Central Management Server) enables seamless application updates through a hardware-centric deployment mechanism. The CMS schedules update deployments based on real-time workload assessments, determining the optimal window during periods of low demand to minimize service disruptions. Update packages are securely transmitted over established communication channels, leveraging hardware-based error checking mechanisms such as checksum verification to ensure data integrity. Each endpoint installs the update in an isolated hardware environment, such as a separate boot partition or containerized module, to prevent interference with active services. If the installation fails, a hardware-controlled rollback system automatically reverts to the previous stable software version and logs detailed error information for management review, ensuring uninterrupted service.

[0070] In one embodiment, the operating data is used as training data, and the historical database is updated according to the training data to obtain a target database, so that the machine learning model is trained according to the target database to obtain a fault classification model; the fault classification model is used to classify the operating data.

[0071] In one implementation method, operating data is used as training data to continuously update the historical database, so that the target database reflects the latest characteristics of the operating status of the power supply terminal in real time, providing dynamic and fresh training samples for the machine learning model, ensuring that the model can adapt to the drift of operating rules caused by factors such as equipment aging and environmental changes, and improving the timeliness and adaptability of data-driven analysis.

[0072] In one implementation, the fault classification model obtained through target database training can automatically learn fault characteristic patterns based on massive historical operation data, efficiently classify real-time operation data, and accurately identify fault types such as voltage anomalies and load mutations. Compared with traditional manual analysis methods, it greatly improves the efficiency and accuracy of fault identification, providing technical support for the system's proactive early warning and rapid response.

[0073] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A multi-layer boot method for reliable data collector initialization and recovery, characterized in that: The method comprises: Acquire operation data of each power supply terminal, and pre-process each operation data to obtain multiple target data; the operation data includes: voltage data, current data, temperature and humidity data, and vibration data; Encrypting each target data to obtain a plurality of encrypted data, and uploading each encrypted data to a central management server so that the central management server detects each encrypted data to obtain an anomaly detection result; The resource allocation strategy of each power supply terminal is determined according to the abnormality detection result, and each resource allocation strategy is sent to each power supply terminal as an update package.

2. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: The current data is sampled according to a preset frequency; the preset frequency is 10 times / second.

3. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: The voltage data, current data, temperature and humidity data, and vibration data in the operating data are subjected to data denoising to obtain denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data, and the denoised voltage data, denoised current data, denoised temperature and humidity data, and denoised vibration data are used as target data.

4. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: The target data is encrypted using the AES-256 encryption module.

5. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: Determine the resource allocation strategy for each power terminal based on the anomaly detection results, including: If there is an abnormal detection result, it is determined that the power supply terminal has an abnormality; Obtain the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority of the target terminal, and calculate the resource demand parameters based on the real-time load rate, maximum load rate, resource utilization rate, total resources, and terminal priority; Optimizing resources of the power supply terminal according to the resource demand parameters to obtain a resource allocation plan; Resource requirement parameter calculation formula: Among them, RDI i Represents the resource demand parameter, L i Represents the real-time load rate of terminal i, L i，max represents the maximum allowable load rate of terminal i, P i Represents the terminal priority, S i Represents resource utilization, S i，max represents the total amount of resources of terminal i, α, β, and γ are all constant proportional coefficients, and α+β+γ=1.

6. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: Obtain the historical periodic load of each power terminal, determine the load window period of the power terminal based on the historical periodic load, generate an update package, and push the update package to each power terminal so that each power terminal receives the update package and performs the update.

7. A multi-layer boot method for reliable data collector initialization and recovery according to claim 1, characterized in that: The operating data is used as training data, and the historical database is updated according to the training data to obtain a target database, so that the machine learning model is trained according to the target database to obtain a fault classification model; the fault classification model is used to classify the operating data.

Images