Unattended intelligent inspection method and system for hydropower basin data center

Abstract

The invention discloses an unattended intelligent inspection method and system for a hydropower basin data center, and relates to the technical field of intelligent operation and maintenance scheduling, and the method comprises the steps: collecting and preprocessing the data of each region of the hydropower basin data center, and calculating the abnormal value of each region of the hydropower basin data center based on the preprocessed data; based on the abnormal value of each region of the hydropower basin data center, adopting Poisson random distribution to derive the lowest detection frequency required by each region, and based on the lowest detection frequency, using an NSGA-II algorithm to perform global planning scheduling on inspection; and in the inspection process, the LSTM prediction model is used to predict the fault probability of the inspection equipment. According to the method, the lowest detection frequency is calculated by introducing the Poisson random distribution model, global scheduling planning is carried out in combination with the NSGA-II multi-target optimization algorithm, and the key problems in the aspects of inspection scheduling intelligence and equipment resource optimal configuration in the prior art are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance scheduling technology, and in particular to an unattended intelligent inspection method and system for hydroelectric data centers. Background Technology

[0002] With the widespread application of IoT, big data, and AI technologies, hydropower data centers have become crucial infrastructure for watershed scheduling, ecological monitoring, safety early warning, and comprehensive management. Most existing inspection systems rely on preset time cycles or human experience for inspection path planning, deploying fixed monitoring equipment or unmanned inspection terminals to achieve basic operational status monitoring. However, as data centers expand in scale and equipment diversity increases, new challenges arise in areas such as monitoring frequency scheduling, resource allocation, and equipment maintenance efficiency.

[0003] Current unattended inspection solutions typically determine inspection tasks through statistical features or rule-based anomaly detection models, and then use heuristic or genetic algorithms for task scheduling. While these methods can achieve automated inspection to some extent, they struggle to achieve balanced scheduling at the global level in multi-objective optimization scenarios. Especially in multi-node, heterogeneous environments like hydroelectric data centers, how to achieve adaptive planning of inspection frequency and dynamic optimization of resource allocation while ensuring system stability remains a crucial research direction in the industry. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an unattended intelligent inspection method for hydroelectric data centers, which solves the problems of insufficient adaptive planning of detection frequency and uncoordinated optimization of equipment deployment scale in the inspection of hydroelectric data centers.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an unattended intelligent inspection method for a hydroelectric data center, comprising, Data from each area of ​​the hydroelectric current domain data center is collected and preprocessed, and outliers in each area of ​​the hydroelectric current domain data center are calculated based on the preprocessed data. Based on the outliers in each region of the hydroelectric data center, the minimum detection frequency required for each region is derived using Poisson random distribution. Based on the minimum detection frequency, the NSGA-II algorithm is used to perform global planning and scheduling of the inspection. During the inspection process, an LSTM prediction model is used to predict the failure probability of the inspection equipment. At the same time, a cost-efficiency function is used to calculate the critical point of the deployment scale of the inspection equipment, and the inspection equipment is deployed based on the calculation results.

[0007] As a preferred embodiment of the unattended intelligent inspection method for the water and current domain data center described in this invention, the following steps are taken: The collection and preprocessing of data from various areas of the water and current domain data center refers to setting a collection period; within the collection period, water flow and current data from each area of ​​the water and current domain data center are collected at each moment to obtain a time-series sequence of water flow and current data from each area of ​​the water and current domain data center; the Z-score normalization method is used to normalize the time-series data; and for the normalized time-series data, a sliding time window is defined, and the standard deviation of the data points within the sliding time window is calculated. ; Set the standard deviation threshold ,like Then The data points at each time point are considered normal data. Anomaly detection begins, with M consecutive detection time points. If the standard deviation of the data points is greater than or equal to the set standard deviation threshold within any of the consecutive detection time points, then... Data points within the specified time frame and during continuous detection time are considered valid outliers. If the standard deviation of a data point is less than the set standard deviation threshold during continuous detection time, then... The data point at that time is considered an invalid or outlier data point.

[0008] As a preferred embodiment of the unattended intelligent inspection method for the hydroelectric current domain data center described in this invention, the step of calculating the outlier values ​​of each area of ​​the hydroelectric current domain data center based on preprocessed data refers to placing data points judged as valid outliers into an outlier data point set, and calculating the outlier values ​​of each area of ​​the hydroelectric current domain data center according to the number of outlier data points in the outlier data point set.

[0009] As a preferred embodiment of the unattended intelligent inspection method for the hydroelectric current domain data center described in this invention, wherein: the minimum detection frequency required for each area is derived by using a Poisson random distribution based on the outlier values ​​of each area of ​​the hydroelectric current domain data center refers to the number of times each area needs to be detected per unit time based on the outlier values ​​of each area of ​​the hydroelectric current domain data center; Based on the number of times each area needs to be inspected per unit time, the inspection behavior of the inspection equipment for each area is modeled as a Poisson distribution; The probability of each region being detected at least once per unit time is calculated based on the Poisson distribution function; Set a probability threshold for being detected at least once per unit time. Based on probability threshold Calculate the minimum detection frequency for each region.

[0010] As a preferred embodiment of the unattended intelligent inspection method for hydroelectric data center described in this invention, the step of using the NSGA-II algorithm to perform global planning and scheduling of inspection based on the minimum detection frequency refers to generating a set of inspection scheduling schemes while satisfying the minimum detection frequency of each area, and calculating the optimization objective function value for each scheme in the set of inspection scheduling schemes, including total path length, task balance and operating cost. Perform a non-dominated sort on all inspection plans in the inspection scheduling plan set to obtain a non-dominated inspection scheduling plan set, and calculate the sparsity of the plans in the non-dominated inspection scheduling plan set. ; Set sparsity threshold ,like If so, the solution will be added to the set of preferred inspection and scheduling solutions. If so, the plan will not be adopted; In the set of preferred inspection scheduling schemes, crossover and mutation operations are used to generate new inspection scheduling schemes, and non-dominated sorting and sparsity filtering are performed again. The above operations are repeated. When the number of iterations reaches the set upper limit, the inspection scheduling scheme with the largest non-dominated and sparsity in the set of preferred inspection scheduling schemes is output as the optimal inspection scheduling scheme.

[0011] As a preferred embodiment of the unattended intelligent inspection method for hydroelectric data center described in this invention, in the inspection process, an LSTM prediction model is used to predict the failure probability of the inspection equipment and output the state sequence of the inspection equipment in real time, and the hidden state of the inspection equipment is calculated based on the state sequence of the inspection equipment. Set a prediction time, and calculate the probability of the inspection equipment malfunctioning within the prediction time based on the hidden state of the inspection equipment. ; Set failure probability threshold ,like If the inspection equipment is used normally, then... If the equipment is deemed high-risk, it will not be included in the inspection schedule and will be sent for inspection and repair.

[0012] As a preferred embodiment of the unattended intelligent inspection method for hydroelectric data centers described in this invention, the method of simultaneously using a cost-efficiency function to calculate the critical point of the deployment scale of inspection equipment, and deploying inspection equipment based on the calculation results, refers to calculating a cost function based on operating costs and equipment costs, calculating an efficiency function based on the number of inspection equipment deployed, calculating the unit efficiency cost using the cost function and efficiency function, differentiating the unit efficiency cost, solving for the minimum point where the derivative is zero, and taking the number of inspection equipment deployed at the extreme point as the optimal deployment.

[0013] Secondly, this invention provides an unattended intelligent inspection system for a hydroelectric data center, comprising: The data acquisition module is used to collect data from various areas of the hydroelectric current domain data center and perform normalization processing. The outlier detection module is used to detect abnormal data in the collected data and determine whether the abnormal data is a valid or invalid outlier data point. The detection frequency derivation module is used to calculate the minimum detection frequency for each region based on the outliers in each region using a Poisson random distribution. The inspection scheduling module is used to perform global optimization of the inspection scheduling using the NSGA-II algorithm with the lowest detection frequency as a constraint, and generate the optimal inspection scheduling scheme. The fault prediction module is used to predict faults in the inspection equipment based on its status using an LSTM model. The equipment deployment module is used to calculate the optimal number of inspection equipment to be deployed using a cost-efficiency function.

[0014] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the unattended intelligent inspection method for hydroelectric data centers as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the unattended intelligent inspection method for water current domain data centers as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By deriving the minimum detection frequency based on the outlier values ​​of each region using a Poisson random distribution, and by using the NSGA-II algorithm for global planning and scheduling of inspections, adaptive optimization allocation of inspection tasks is achieved, enabling the inspection frequency to dynamically match the regional operating status, thereby improving the global utilization rate and scheduling accuracy of inspection resources; by introducing an LSTM prediction model to predict the failure probability of inspection equipment during the inspection process, and combining it with a cost-efficiency function to calculate the critical point of equipment deployment scale, a balance between intelligent and economical deployment of inspection equipment is achieved, effectively improving the operational reliability and cost control capabilities of inspection scheduling, and realizing the efficient, accurate, and sustainable operation of unattended inspection processes in hydroelectric data centers. Attached Figure Description

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

[0018] Figure 1 A flowchart for an unattended intelligent inspection method for hydroelectric data centers.

[0019] Figure 2 This is a schematic diagram of an unattended intelligent inspection system for a hydroelectric data center.

[0020] Figure 3 Flowchart for obtaining the minimum detection frequency for each region. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] Reference Figures 1-3 As one embodiment of the present invention, this embodiment provides an unattended intelligent inspection method for a hydroelectric data center, comprising the following steps: Data from each area of ​​the hydroelectric current domain data center is collected and preprocessed, and outliers in each area of ​​the hydroelectric current domain data center are calculated based on the preprocessed data.

[0025] Specifically, a data acquisition period is set. Within this period, water flow and current data for each area of ​​the water and current domain data center are collected at each time point to obtain a time-series sequence of water flow and current data for each area of ​​the data center. The Z-score normalization method is used to normalize the time-series data. For the normalized time-series data, a sliding time window is defined, and the standard deviation of the data points within the sliding time window is calculated using the following formula: ; in, The center moment of the sliding time window , The first data center for the hydroelectric domain Each region The standard deviation of the sliding time window. The length of the sliding time window. For the first time window within the sliding time window Data points, For the first The data values ​​of each data point This represents the average value of the data points within the sliding time window.

[0026] Set the standard deviation threshold ,like Then The data points at each time point are considered normal data. Anomaly detection begins, with M consecutive detection time points. If the standard deviation of the data points is greater than or equal to the set standard deviation threshold within any of the consecutive detection time points, then... Data points within the specified time frame and during continuous detection time are considered valid outliers. If the standard deviation of a data point is less than the set standard deviation threshold during continuous detection time, then... The data point at that time is considered an invalid or outlier data point.

[0027] By acquiring time-series data of water flow and current in various regions during the acquisition period and normalizing them using the Z-score standardization method, unified quantification of heterogeneous data from different regions was achieved, making subsequent anomaly detection comparable and stable. Subsequently, by setting a sliding time window and calculating the standard deviation of data points within the window, dynamic characterization of data fluctuations was achieved, which can reflect the changing trends of the watershed's operating status in a timely manner. Furthermore, by setting a standard deviation threshold and combining it with continuous detection time to judge the validity of anomalies, the distinction between short-term interference and continuous anomalies was achieved, ensuring the sensitivity of anomaly detection while reducing the false alarm rate, and improving the monitoring accuracy and operational stability of the water and current domain data center.

[0028] Furthermore, data points identified as valid anomalies are added to an anomaly data point set. Based on the number of anomaly data points in the set, the outlier values ​​for each region of the hydroelectric current domain data center are calculated using the following formula: ; in, For the first data center in the hydroelectric domain Outliers in each region For the first The number of outlier data points in the set of outlier data points in each region. The total data collection period is set.

[0029] By incorporating the identified valid abnormal data points into the abnormal data point set, the abnormal information is centrally summarized and structured for management, facilitating subsequent quantitative analysis. Furthermore, based on the number of abnormal data points in the abnormal data point set, the abnormal value of each region is calculated, enabling a quantitative assessment of the degree of abnormality in the operating status of different regions, which can intuitively reflect the stability level of equipment in each region.

[0030] Based on the outliers in each region of the hydroelectric data center, the minimum detection frequency required for each region is derived using the Poisson random distribution. Based on the minimum detection frequency, the NSGA-II algorithm is used for global planning and scheduling of the inspection.

[0031] Specifically, based on the outlier calculation for each region of the hydroelectric data center, the number of times each region needs to be detected per unit time is calculated using the following formula: ; in, For the first The number of times each area needs to be tested. The set global baseline detection frequency, The set outlier weights.

[0032] Based on the number of times each area needs to be inspected per unit time, the inspection behavior of the patrol equipment for each area is modeled as a Poisson distribution, with the following formula: ; in, Let be the Poisson distribution function. For the set unit time, For the first unit of time The number of times each region was detected. To detect the number of times an event occurs.

[0033] The probability of each region being detected at least once per unit time is calculated based on the Poisson distribution function, using the following formula: ; in, For the first unit of time The probability that a region will be detected at least once.

[0034] Set a probability threshold for being detected at least once per unit time. The formula for calculating the minimum detection frequency for each region is: ; in, For the first The minimum detection frequency for each region.

[0035] By calculating the number of inspections required per unit time based on outliers in each region, a dynamic correlation between inspection frequency and regional operating status is achieved, making the allocation of inspection tasks more in line with actual needs. Subsequently, the inspection behavior of the inspection equipment is modeled as a Poisson distribution, realizing a probabilistic description of random inspection events and objectively reflecting the distribution characteristics of inspection tasks in the time dimension. Then, the probability of each region being inspected at least once per unit time is calculated using the Poisson distribution function, and the minimum inspection frequency is determined in combination with the set probability threshold, realizing the quantification and adaptive control of the inspection frequency.

[0036] Furthermore, while ensuring the minimum detection frequency for each region, a set of inspection scheduling schemes is generated. For each scheme in the set, an optimization objective function value is calculated, including total path length, task balance, and operating cost. The formula is as follows: ; in, The first in the inspection and dispatch scheme set One inspection and scheduling plan For the first The optimization objective function value of each inspection scheduling scheme This is a function of the total path length. The task balance function, This is the running cost function.

[0037] All inspection plans in the inspection scheduling plan set are sorted by non-dominated order to obtain the non-dominated inspection scheduling plan set. The sparsity of the plans in the non-dominated inspection scheduling plan set is calculated using the following formula: ; in, To assess the sparsity of the inspection and dispatch plan, To optimize the total number of objectives, For the first One optimization objective, and The inspection and dispatch plan is as follows: The optimization objective is the objective value of the left and right neighboring solutions in the solution set. and They are the first The maximum and minimum values ​​of each optimization objective.

[0038] Set sparsity threshold ,like If so, the solution will be added to the set of preferred inspection and scheduling solutions. If so, the plan will not be adopted; In the set of preferred inspection scheduling schemes, crossover and mutation operations are used to generate new inspection scheduling schemes, and non-dominated sorting and sparsity filtering are performed again. The above operations are repeated. When the number of iterations reaches the set upper limit, the inspection scheduling scheme with the largest non-dominated and sparsity in the set of preferred inspection scheduling schemes is output as the optimal inspection scheduling scheme.

[0039] By generating a set of inspection scheduling schemes while meeting the minimum detection frequency in each region, and calculating multi-objective optimization functions such as total path length, task balance, and operating cost, a quantitative evaluation of the comprehensive performance of the inspection schemes is achieved, providing an optimization basis for global scheduling. Subsequently, the scheme set is sorted non-dominated and filtered by sparsity to ensure that the diversity of the solution set is maintained while retaining the optimal solution, so that the scheduling result takes into account both efficiency and stability. Then, by performing crossover and mutation operations on the preferred schemes and iteratively filtering them, the adaptive evolution optimization of the inspection schemes is achieved, and finally, an inspection scheduling scheme with both global optimality and balance is obtained.

[0040] During the inspection process, an LSTM prediction model is used to predict the failure probability of the inspection equipment. At the same time, a cost-efficiency function is used to calculate the critical point of the deployment scale of the inspection equipment, and the inspection equipment is deployed based on the calculation results.

[0041] Specifically, the status sequence of the inspection equipment is output in real time, and the hidden state of the inspection equipment is calculated based on the status sequence. The formula is as follows: ; in, For the inspection equipment The hidden state at all times The basic computational unit used to compute the hidden state in the LSTM model. For the inspection equipment The state at any given moment, This refers to the hidden state of the inspection equipment at the previous moment.

[0042] Set a prediction time, and calculate the probability of the inspection equipment malfunctioning within the prediction time based on the hidden state of the inspection equipment. The formula is: ; in, This is the probability of equipment malfunctions during inspection. The function set for the last layer of the LSTM model. The weights are obtained based on the prediction time. This is the bias term obtained based on the prediction time training.

[0043] Set failure probability threshold ,like If the inspection equipment is used normally, then... If the equipment is deemed high-risk, it will not be included in the inspection schedule and will be sent for inspection and repair.

[0044] By outputting the status sequence of the inspection equipment in real time and calculating its hidden state, a deep modeling of the equipment's operating characteristics is achieved, which can identify potential performance degradation trends. Subsequently, based on the hidden state, the probability of equipment failure within a set time is predicted, enabling a forward-looking assessment of the equipment's health status and early identification of high-risk equipment. Furthermore, by setting a failure probability threshold and performing risk classification and scheduling elimination based on the prediction results, intelligent screening and maintenance decision linkage for inspection equipment are realized, ultimately reducing the risk of inspection tasks and improving the reliability and efficiency of equipment operation and maintenance.

[0045] Furthermore, a cost function is calculated based on operating costs and equipment costs, and an efficiency function is calculated based on the number of inspection equipment deployed. The unit efficiency cost is calculated using the cost function and efficiency function. The derivative of the unit efficiency cost is taken, and the minimum point where the derivative is zero is solved. The number of inspection equipment deployed at the extreme point is taken as the optimal deployment.

[0046] By constructing a cost function based on operating costs and equipment costs, and calculating an efficiency function in conjunction with the number of inspection equipment deployed, a quantitative description of the economy and execution capability of inspection scheduling is achieved. Subsequently, the unit efficiency cost is calculated using the cost function and the efficiency function, and the minimum point is obtained by taking their derivatives, enabling a precise analysis of the balance between equipment deployment scale and overall efficiency. Finally, the number of equipment corresponding to the extreme point is taken as the optimal deployment scheme, effectively avoiding resource redundancy and insufficient performance, reducing inspection operating costs, and improving equipment utilization.

[0047] This embodiment also provides an unattended intelligent inspection system for a hydroelectric data center, including: The data acquisition module is used to collect data from various areas of the hydroelectric current domain data center and perform normalization processing. The outlier detection module is used to detect abnormal data in the collected data and determine whether the abnormal data is a valid or invalid outlier data point. The detection frequency derivation module is used to calculate the minimum detection frequency for each region based on the outliers in each region using a Poisson random distribution. The inspection scheduling module is used to perform global optimization of the inspection scheduling using the NSGA-II algorithm with the lowest detection frequency as a constraint, and generate the optimal inspection scheduling scheme. The fault prediction module is used to predict faults in the inspection equipment based on its status using an LSTM model. The equipment deployment module is used to calculate the optimal number of inspection equipment to be deployed using a cost-efficiency function.

[0048] This embodiment also provides a computer device applicable to the unattended intelligent inspection method for hydroelectric data centers, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the unattended intelligent inspection method for hydroelectric data centers proposed in the above embodiment.

[0049] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0050] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the unattended intelligent inspection method for a hydroelectric data center as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0051] In summary, this invention achieves adaptive optimization of inspection task allocation by: deriving the minimum detection frequency based on Poisson random distribution of outliers in each region and using the NSGA-II algorithm for global planning and scheduling of inspections. This allows the inspection frequency to dynamically match the regional operating status, improving the global utilization rate and scheduling accuracy of inspection resources. Furthermore, by introducing an LSTM prediction model to predict the failure probability of inspection equipment during the inspection process and combining it with a cost-efficiency function to calculate the critical point of equipment deployment scale, this invention achieves a balance between intelligent and economical deployment of inspection equipment, effectively improving the operational reliability and cost control capabilities of inspection scheduling. This enables efficient, accurate, and sustainable unattended inspection of hydroelectric data centers.

[0052] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for unattended intelligent inspection of a hydroelectric data center, characterized in that: include, Data from each area of ​​the hydroelectric current domain data center is collected and preprocessed, and outliers in each area of ​​the hydroelectric current domain data center are calculated based on the preprocessed data. Based on the outliers in each region of the hydroelectric data center, the minimum detection frequency required for each region is derived using Poisson random distribution. Based on the minimum detection frequency, the NSGA-II algorithm is used to perform global planning and scheduling of the inspection. During the inspection process, an LSTM prediction model is used to predict the failure probability of the inspection equipment. At the same time, a cost-efficiency function is used to calculate the critical point of the deployment scale of the inspection equipment, and the inspection equipment is deployed based on the calculation results.

2. The unattended intelligent inspection method for hydroelectric data centers as described in claim 1, characterized in that: The process of collecting and preprocessing data from various areas of the water flow and current domain data center involves setting a collection period. Within this period, water flow and current data from each area of ​​the data center are collected at each moment to obtain a time-series sequence of water flow and current data from each area. The Z-score normalization method is then used to normalize the time-series data. For the normalized time-series data, a sliding time window is defined, and the standard deviation of the data points within the sliding time window is calculated. ; Set the standard deviation threshold ,like Then The data points at each time point are considered normal data. Anomaly detection begins, with M consecutive detection time points. If the standard deviation of the data points is greater than or equal to the set standard deviation threshold within any of the consecutive detection time points, then... Data points within the specified time frame and during continuous detection time are considered valid outliers. If the standard deviation of a data point is less than the set standard deviation threshold during continuous detection time, then... The data point at that time is considered an invalid or outlier data point.

3. The unattended intelligent inspection method for hydroelectric data centers as described in claim 2, characterized in that: The calculation of outliers in each region of the hydroelectric data center based on preprocessed data refers to placing data points judged as valid outliers into an outlier data point set, and calculating the outliers in each region of the hydroelectric data center based on the number of outlier data points in the outlier data point set.

4. The unattended intelligent inspection method for hydroelectric data centers as described in claim 3, characterized in that: The minimum detection frequency required for each region is derived from the Poisson random distribution of the outlier values ​​in each region of the hydroelectric current domain data center. This means that the number of times each region needs to be detected per unit time is calculated based on the outlier values ​​in each region of the hydroelectric current domain data center. Based on the number of times each area needs to be inspected per unit time, the inspection behavior of the inspection equipment for each area is modeled as a Poisson distribution; The probability of each region being detected at least once per unit time is calculated based on the Poisson distribution function; Set a probability threshold for being detected at least once per unit time. Based on probability threshold Calculate the minimum detection frequency for each region.

5. The unattended intelligent inspection method for hydroelectric data centers as described in claim 4, characterized in that: The global planning and scheduling of inspection based on the minimum detection frequency using the NSGA-II algorithm refers to generating a set of inspection scheduling schemes while satisfying the minimum detection frequency of each region, and calculating the optimization objective function value for each scheme in the set of inspection scheduling schemes, including total path length, task balance and operating cost. Perform a non-dominated sort on all inspection plans in the inspection scheduling plan set to obtain a non-dominated inspection scheduling plan set, and calculate the sparsity of the plans in the non-dominated inspection scheduling plan set. ; Set sparsity threshold ,like If so, the solution will be added to the set of preferred inspection and scheduling solutions. If so, the plan will not be adopted; In the set of preferred inspection scheduling schemes, crossover and mutation operations are used to generate new inspection scheduling schemes, and non-dominated sorting and sparsity filtering are performed again. The above operations are repeated. When the number of iterations reaches the set upper limit, the inspection scheduling scheme with the largest non-dominated and sparsity in the set of preferred inspection scheduling schemes is output as the optimal inspection scheduling scheme.

6. The unattended intelligent inspection method for hydroelectric data centers as described in claim 5, characterized in that: During the inspection process, the LSTM prediction model is used to predict the failure probability of the inspection equipment and output the state sequence of the inspection equipment in real time. The hidden state of the inspection equipment is calculated based on the state sequence of the inspection equipment. Set a prediction time, and calculate the probability of the inspection equipment malfunctioning within the prediction time based on the hidden state of the inspection equipment. ; Set failure probability threshold ,like If the inspection equipment is used normally, then... If the equipment is deemed high-risk, it will not be included in the inspection schedule and will be sent for inspection and repair.

7. The unattended intelligent inspection method for hydroelectric data centers as described in claim 6, characterized in that: The simultaneous use of cost-efficiency functions to calculate the critical point of the deployment scale of inspection equipment, and the deployment of inspection equipment based on the calculation results, refers to calculating the cost function based on operating costs and equipment costs, calculating the efficiency function based on the number of inspection equipment deployed, using the cost function and efficiency function to calculate the unit efficiency cost, taking the derivative of the unit efficiency cost, solving for the minimum point where the derivative is zero, and taking the number of inspection equipment deployed at the extreme point as the optimal deployment.

8. An unattended intelligent inspection system for a hydroelectric data center, based on the unattended intelligent inspection method for a hydroelectric data center according to any one of claims 1 to 7, characterized in that: include, The data acquisition module is used to collect data from various areas of the hydroelectric current domain data center and perform normalization processing. The outlier detection module is used to detect abnormal data in the collected data and determine whether the abnormal data is a valid or invalid outlier data point. The detection frequency derivation module is used to calculate the minimum detection frequency for each region based on the outliers in each region using a Poisson random distribution. The inspection scheduling module is used to perform global optimization of the inspection scheduling using the NSGA-II algorithm with the lowest detection frequency as a constraint, and generate the optimal inspection scheduling scheme. The fault prediction module is used to predict faults in the inspection equipment based on its status using an LSTM model. The equipment deployment module is used to calculate the optimal number of inspection equipment to be deployed using a cost-efficiency function.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the unattended intelligent inspection method for hydroelectric data centers as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the unattended intelligent inspection method for hydroelectric data centers as described in any one of claims 1 to 7.

Images