Host power supply operation management method and system based on data processing

Abstract

The invention belongs to the technical field of power supply operation management, and discloses a host power supply operation management method and system based on data processing. The method comprises the following steps: carrying out training data processing based on a host; performing model construction by using an artificial intelligence algorithm; simplifying the artificial intelligence model; acquiring real-time power supply operation monitoring data based on a power supply management device; real-time data processing is carried out based on the host; performing power supply operation analysis by using the simplified power supply operation analysis model; generating a power supply operation management strategy by using the simplified power supply operation management model; and generating a real-time power supply operation management instruction based on a power supply management device according to the real-time power supply operation management strategy, and controlling the host power supply to operate. According to the method, the problems of insufficient data processing capability, high model complexity, poor real-time performance, weak model generalization capability and insufficient interpretability in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power supply operation management, and in particular relates to a host power supply operation management method and system based on data processing. Background Art

[0002] With the rapid development of computer technology, the stable operation and efficient management of host power supply have become increasingly important. Therefore, the significance of host power supply operation management is very important, and it plays a key role in ensuring the stable operation of the host, extending the service life, improving energy efficiency and ensuring data security. However, the existing power management technology has the following defects:

[0003] 1) Insufficient data processing capabilities: Existing host power operation management methods often rely on simple data processing techniques, such as simple statistical analysis and threshold judgment, which are difficult to capture complex patterns and potential anomalies in power operation data;

[0004] 2) High model complexity: Although artificial intelligence algorithms have shown great potential in power operation data analysis, existing host power operation management models based on complex algorithms (such as deep learning) often have high computational complexity, require expensive hardware support, and are not suitable for deployment in resource-constrained environments.

[0005] 3) Poor real-time performance: Existing host power operation management technologies often fail to meet the needs of real-time monitoring and rapid response, especially when faced with sudden power system conditions, and cannot make timely adjustments;

[0006] 4) Weak model generalization ability: The existing host power operation management model has weak generalization ability when facing different working scenarios and load conditions, and it is difficult to adapt to diverse power operation environments;

[0007] 5) Lack of explainability: Although complex AI models can provide accurate analysis and predictions, their internal decision-making processes often lack transparency, making host power operation management strategies difficult to explain and understand. Summary of the invention

[0008] In order to solve the problems of insufficient data processing capability, high model complexity, poor real-time performance, weak model generalization capability and insufficient interpretability in the prior art, the present invention aims to provide a host power operation management method and system based on data processing.

[0009] The technical solution adopted by the present invention is:

[0010] A host power operation management method based on data processing comprises the following steps:

[0011] Based on the host, data processing is performed on a number of historical power operation monitoring data collected by the power management device to obtain a number of processed historical power operation monitoring data;

[0012] Based on the host, according to a number of processed historical power operation monitoring data, artificial intelligence algorithms are used to build models to obtain power operation analysis models and power operation management models;

[0013] Based on the host and according to the host resources, the power operation analysis model and the power operation management model are simplified to obtain a simplified power operation analysis model and a simplified power operation management model;

[0014] Based on the power management device, real-time power operation monitoring data of the host power supply is collected, and the real-time power operation monitoring data is transmitted to the host;

[0015] Based on the host, the real-time power supply operation monitoring data is processed to obtain the processed real-time power supply operation monitoring data;

[0016] Based on the host, according to the processed real-time power operation monitoring data, a simplified power operation analysis model is used to perform power operation analysis to obtain real-time power operation analysis results;

[0017] Based on the host, according to the real-time power operation analysis result, a simplified power operation management model is used to generate a power operation management strategy, obtain the real-time power operation management strategy, and transmit it to the power management device;

[0018] Based on the power management device, according to the real-time power operation management strategy, a real-time power operation management instruction is generated, and according to the real-time power operation management instruction, the host power operation is controlled.

[0019] Further, based on the host, data processing is performed on a number of historical power operation monitoring data collected by the power management device to obtain a number of processed historical power operation monitoring data, including the following steps:

[0020] Based on the power management device, a number of historical power operation monitoring data of the host power supply is collected, and the number of historical power operation monitoring data is transmitted to the host;

[0021] Based on the host, pre-processing is performed on a number of historical power supply operation monitoring data to obtain a number of pre-processed historical power supply operation monitoring data;

[0022] Data dimension reduction is performed on a number of pre-processed historical power supply operation monitoring data to obtain a number of processed historical power supply operation monitoring data.

[0023] Furthermore, the PCA method is used to perform data dimension reduction on a number of pre-processed historical power supply operation monitoring data to obtain a number of processed historical power supply operation monitoring data.

[0024] Furthermore, based on the host, according to a number of processed historical power operation monitoring data, an artificial intelligence algorithm is used to construct a model to obtain a power operation analysis model and a power operation management model, including the following steps:

[0025] Based on the host, a number of processed historical power supply operation monitoring data are divided into a model training set and a model test set in a ratio of 7:3;

[0026] Use a deep learning algorithm to build an initial power supply operation analysis model, and use the model training set to optimize the initial power supply operation analysis model to obtain an optimized power supply operation analysis model;

[0027] Using the model test set, the optimized power supply operation analysis model is tested to obtain the model test accuracy;

[0028] If the model test accuracy is greater than the threshold, the optimal power supply operation analysis model is output, several historical power supply operation analysis results are generated, and the next step is entered; otherwise, the optimization training is continued;

[0029] Taking the power operation management problem as the simulation environment, the reinforcement learning algorithm is used to build the initial power operation management model and the corresponding experience replay pool;

[0030] Using several historical power operation analysis results, the initial power operation management model is optimized and trained to obtain the optimal power operation management model, and several historical power operation management experiences are generated;

[0031] Several historical power operation management experiences are stored in an experience playback pool corresponding to the optimal power operation management model.

[0032] Furthermore, the power operation analysis model is constructed based on the LSTM-Attention-cGAN algorithm, and the power operation analysis model includes an input layer, a feature extraction module constructed based on the LSTM algorithm, an attention weight module constructed based on the Attention mechanism, and an adversarial training module, a classification layer, and an output layer that are sequentially connected;

[0033] The power operation management model is built based on the DQN algorithm, and the power operation management model includes a deep Q network, an intelligent agent, and an experience replay pool.

[0034] Further, based on the host and according to the host resources, the power operation analysis model and the power operation management model are simplified to obtain a simplified power operation analysis model and a simplified power operation management model, including the following steps:

[0035] Based on the host, setting an optimization target according to a first model parameter of the power operation analysis model and a second model parameter of the power operation management model, and setting an objective function according to the host resources and the optimization target;

[0036] According to the objective function, a swarm intelligence optimization algorithm is used to search and optimize to obtain the optimal first model parameters and the optimal second model parameters;

[0037] According to the optimal first model parameter, the power supply operation analysis model is simplified to obtain a simplified power supply operation analysis model;

[0038] The power operation management model is simplified according to the optimal second model parameters to obtain a simplified power operation management model.

[0039] Furthermore, according to the objective function, the ICPO optimization algorithm is used to search and optimize to obtain the optimal first model parameter and the optimal second model parameter.

[0040] Further, based on the host, according to the processed real-time power operation monitoring data, using the simplified power operation analysis model, power operation analysis is performed to obtain the real-time power operation analysis result, including the following steps:

[0041] Based on the host, the processed real-time power operation monitoring data is input into the input layer of the simplified power operation analysis model;

[0042] Using the feature extraction module, extract the real-time data features of the processed real-time power operation monitoring data, and input the real-time data features into the attention weight module respectively;

[0043] Use the attention weight module to perform feature splicing on the real-time data features according to the preset attention weight value to obtain the real-time data splicing features, and input the real-time data splicing features into the adversarial training module;

[0044] Use the adversarial training module to perform adversarial training on the real-time data splicing features to obtain real-time adversarial training features, and input the real-time adversarial training features into the classification layer;

[0045] Use the classification layer to perform power operation analysis and prediction based on the real-time adversarial training features, obtain the real-time power operation analysis and prediction labels, and input the real-time power operation analysis and prediction labels into the output layer;

[0046] Using the output layer, the real-time power operation analysis prediction labels are converted into real-time power operation analysis results, and the real-time power operation analysis results are output.

[0047] Further, based on the host, according to the real-time power operation analysis result, using the simplified power operation management model, generating a power operation management strategy, obtaining a real-time power operation management strategy, and transmitting it to the power management device, including the following steps:

[0048] Based on the host, convert the real-time power operation analysis result into a real-time power operation analysis result encoding vector applicable to the simplified power operation management model;

[0049] According to the encoding vector of the real-time power operation analysis result, a plurality of historical power operation management experiences stored in the experience playback pool are searched to obtain a plurality of matching historical power operation management experiences;

[0050] According to the encoding vector of the real-time power operation analysis result and some matching historical power operation management experience, the model parameters of the simplified deep Q network are updated to obtain the updated model parameters;

[0051] According to the updated model parameters of the simplified deep Q network, the power operation management strategy is generated using the intelligent agent to obtain the real-time power operation management strategy;

[0052] Transmits real-time power operation management policies to the power management device.

[0053] A host power operation management system based on data processing, used to implement a host power operation management method, the system includes a host, a power management device and a host power supply, the host is communicatively connected to the power management device, the power management device is communicatively connected to the host power supply, and the host power supply is electrically connected to the host;

[0054] The host is provided with a data processing unit, a model building unit, a model simplification unit, a power operation analysis unit and a power operation management unit which are connected in sequence.

[0055] The beneficial effects of the present invention are:

[0056] The present invention provides a host power operation management method and system based on data processing. The power operation analysis model constructed by the artificial intelligence algorithm can automatically and efficiently process and analyze data, extract deep information from the host power operation monitoring data, and accurately capture the complex patterns and potential anomalies in the power operation data; the power operation management model constructed by the artificial intelligence algorithm can perform fast and intelligent power operation management strategies according to the power operation analysis results, and adaptively adjust the power operation management measures according to the power operation conditions, thereby improving the flexibility and effectiveness of the host power operation management; the power operation analysis model and the power operation management model are optimized by the group intelligence optimization algorithm. The parameters are simplified, which reduces the computational complexity of the model and enables it to operate in a resource-constrained environment. Real-time monitoring and rapid response based on real-time data collection enable timely adjustments to be made in the face of sudden conditions in the power system, thereby improving the real-time performance of host power operation management. The optimized model structure and parameters improve the generalization ability of the model in different working scenarios and enhance the adaptability of power management. The simplified model structure makes the power management strategy more intuitive and easy to understand, enhancing the interpretability of the entire management system. Data processing, real-time analysis, and management strategy generation are integrated into a complete power operation management system, forming a closed-loop, automated power management logic.

[0057] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flowchart of the host power operation management method based on data processing in the present invention.

[0059] Figure 2 It is a structural block diagram of the host power operation management system based on data processing in the present invention. DETAILED DESCRIPTION

[0060] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0061] Embodiment 1:

[0062] like Figure 1 As shown, this embodiment provides a host power operation management method based on data processing, including the following steps:

[0063] S1: Based on the host, a plurality of historical power supply operation monitoring data collected by the power supply management device are processed to obtain a plurality of processed historical power supply operation monitoring data, including the following steps:

[0064] S1-1: Based on the power management device, collect some historical power operation monitoring data of the host power supply, and transmit the some historical power operation monitoring data to the host;

[0065] The power supply operation monitoring data includes a number of power supply operation monitoring index data, and the power supply operation monitoring index data includes the input monitoring index data, output monitoring index data, status monitoring index data, environmental monitoring index data, protection monitoring index data, maintenance and life monitoring index data, and energy efficiency monitoring index data of the host power supply;

[0066] Enter monitoring indicator data:

[0067] Input voltage: monitors the voltage level at the host power input terminal;

[0068] Input frequency: monitor the AC frequency at the host power input terminal;

[0069] Input current: monitors the current drawn by the host power supply from the power grid;

[0070] Input power: Calculate the product of input voltage and input current to get input power;

[0071] Output monitoring indicator data:

[0072] Output voltage: monitors the voltage level at the host power output;

[0073] Output current: monitors the current provided by the host power supply to the load;

[0074] Output power: Calculate the product of output voltage and output current to get the output power;

[0075] Output frequency: For AC output host power supply, monitor the frequency at the output end;

[0076] Condition monitoring indicator data:

[0077] Host power status: whether the host power is on, standby, off or in protection mode; Fault indication: whether the host power detects any faults, such as overvoltage, undervoltage, overheating, etc.; Environmental monitoring indicator data:

[0078] Temperature: monitor the internal temperature of the host power supply to prevent overheating;

[0079] Humidity: For host power supplies with higher environmental requirements, monitor the humidity of the surrounding environment;

[0080] Protection monitoring indicator data:

[0081] Overvoltage protection: monitor whether the host power supply triggers overvoltage protection;

[0082] Undervoltage protection: monitor whether the host power supply triggers undervoltage protection;

[0083] Overload protection: monitor whether the host power supply triggers protection due to excessive output current;

[0084] Short circuit protection: monitor whether the host power supply triggers protection due to short circuit at the output end;

[0085] Maintenance and life monitoring indicator data:

[0086] Running time: the total running time of the host power supply, used to predict the maintenance cycle;

[0087] Switching times: The number of times the host power is turned on and off, especially for switchable host power, which may affect its lifespan;

[0088] Energy efficiency monitoring indicator data:

[0089] Power factor: The ratio of input power to the product of input voltage and current, reflecting the energy efficiency of the host power supply;

[0090] Efficiency: The ratio of output power to input power, usually expressed as a percentage;

[0091] The main functions and effects of power management devices:

[0092] Startup control: When the host is turned on, it is responsible for sending a signal to the host power supply to start supplying power;

[0093] Power status monitoring: monitor the output voltage and current of the host power supply to ensure that the host power supply is within a safe working range;

[0094] Power switch control: control the host power on and off through software instructions;

[0095] Sleep and wake-up: manage the sleep state of the host power supply and wake up the host power supply when necessary;

[0096] Power supply protection: protect the host power supply when the host power supply encounters abnormal conditions such as overvoltage, undervoltage, overload or short circuit;

[0097] Power distribution: According to the needs of the host, reasonably allocate the power supply resources of the host power supply to different hardware components;

[0098] Energy efficiency management: Optimize energy efficiency ratio and reduce energy consumption by adjusting the host power output;

[0099] S1-2: Based on the host, pre-processing is performed on a number of historical power supply operation monitoring data to obtain a number of pre-processed historical power supply operation monitoring data;

[0100] Preprocessing includes format conversion, converting the format of power supply operation monitoring data into a format that can be recognized by the host, data cleaning and screening, removing duplicate, erroneous and noisy data, normalization, unifying the magnitude of different indicator data, eliminating the accuracy impact caused by magnitude differences, and providing data support for subsequent model training;

[0101] S1-3: Using the principal component analysis (PCA) method, data dimension reduction is performed on a number of pre-processed historical power supply operation monitoring data to obtain a number of processed historical power supply operation monitoring data, including the following steps:

[0102] S1-3-1: Perform matrix conversion on some pre-processed historical power supply operation monitoring data to obtain the corresponding historical power supply operation monitoring data matrix. The formula is:

[0103] X=[x1,x2,...x p ,...,x n ] T

[0104] Among them, x p is the row vector of the p-th preprocessed historical power supply operation monitoring data, p is the row vector indicator, and n is the total number of preprocessed historical power supply operation monitoring data;

[0105] The initial row vector of the historical power operation monitoring data matrix is ​​the preprocessed historical power operation monitoring data, and the initial column vector of the historical power operation monitoring data matrix is ​​the historical power operation monitoring index data, each index data corresponds to the data of each column in the historical power operation monitoring data matrix;

[0106] S1-3-2: Standardize the historical power supply operation monitoring data matrix to obtain the corresponding standardized historical power supply operation monitoring data matrix;

[0107] The formula is:

[0108]

[0109] Where X' is the historical power supply operation monitoring data matrix after standardization; μ is the mean of the historical power supply operation monitoring data matrix; σ is the variance of the historical power supply operation monitoring data matrix;

[0110] S1-3-3: Obtain the covariance matrix of the historical power operation monitoring data matrix after standardization, and obtain the corresponding candidate principal component matrix based on the historical power operation monitoring data matrix and the covariance matrix after standardization; the candidate row vectors of the candidate principal component matrix are the historical power operation monitoring data after preprocessing, and the candidate column vectors of the candidate principal component matrix are the historical power operation monitoring indicator data;

[0111] The formula is:

[0112]

[0113] Where D is the covariance matrix of the historical power supply operation monitoring data matrix after standardization; Y is the candidate principal component matrix; P is the conversion matrix; E is the unit eigenvector matrix; n is the total number of historical power supply operation monitoring data after preprocessing;

[0114] Y=PX'

[0115] Where, Y is the candidate principal component matrix; P is the conversion matrix; X' is the historical power supply operation monitoring data matrix after standardization;

[0116] S1-3-4: Set the candidate principal component matrix Y = [y1, y2, ...y l ,...,y L ] Several candidate column vectors y' of the top 90% of the cumulative contribution rate of variance l As the corresponding principal component column vectors, the reduced-dimensional historical power supply operation monitoring data matrix Y'=[y'1,y'2,...y' l ,...,y K ] and the principal component set;

[0117] The key row vector of the historical power operation monitoring data matrix after dimensionality reduction is the historical power operation monitoring data after dimensionality reduction, and the data index corresponding to the key column vector of the historical power operation monitoring data matrix after dimensionality reduction is the principal component, and a principal component set is obtained;

[0118] The formula is:

[0119]

[0120] In the formula, G is the cumulative contribution rate of variance; l The lth candidate principal component y l The variance of ; l is the indicator of the candidate principal component; L is the total number of candidate principal components; K is the total number of principal components;

[0121] S1-3-5: Perform an inverse matrix transformation on the dimension-reduced historical power supply operation monitoring data matrix to obtain a number of dimension-reduced historical power supply operation monitoring data;

[0122] The PCA method transforms the original data into a set of linearly independent representations in each dimension through linear transformation, which is used to remove noise and redundancy in the data, making the data easier to understand and process;

[0123] S2: Based on the host, according to a number of processed historical power operation monitoring data, using artificial intelligence algorithms, model construction is performed to obtain a power operation analysis model and a power operation management model, including the following steps:

[0124] S2-1: Based on the host, a number of processed historical power supply operation monitoring data are divided into a model training set and a model test set in a ratio of 7:3;

[0125] S2-2: Use a deep learning algorithm to build an initial power operation analysis model, and use the model training set to optimize the initial power operation analysis model to obtain an optimized power operation analysis model;

[0126] The power operation analysis model is constructed based on the LSTM-Attention-cGAN algorithm, and the power operation analysis model includes an input layer connected in sequence, a feature extraction module constructed based on the LSTM algorithm, an attention weight module constructed based on the Attention mechanism, and an adversarial training module, a classification layer, and an output layer constructed based on the cGAN algorithm;

[0127] The input layer is used to receive input power operation monitoring data. The feature extraction module realizes feature extraction of data through multiple gating mechanisms and has a memory function. It is very suitable for processing and predicting time series data, improving the accuracy and stability of the model. It is used to extract features from time series data. The LSTM layer converts the processed data into a hidden state sequence. The attention weight module realizes splicing of key features through preset attention weights, integrates scattered features, enhances the influence of key features on label prediction, and increases the flexibility and explanatory power of the model. The adversarial training module generates data based on conditional input, simulates different operation scenarios or generates anomaly detection samples. It is used to input the output of the attention weight module as a condition into cGAN for adversarial training. Adversarial training helps the model learn more complex distributions and improve generalization ability. The cGAN generator tries to generate data similar to the real data distribution, while the discriminator tries to distinguish between the real data and the data generated by the generator. The output of the adversarial training module is sent to the classification layer for feature classification. The classification layer maps the features to different categories for power operation analysis and prediction. The output layer converts the output of the classification layer into the final real-time power operation analysis result.

[0128] S2-3: Use the model test set to perform model testing on the optimized power supply operation analysis model to obtain the model test accuracy;

[0129] S2-4: If the model test accuracy is greater than the threshold, the optimal power supply operation analysis model is output, a number of historical power supply operation analysis results are generated, and the next step is entered; otherwise, the optimization training is continued;

[0130] S2-5: Using the power operation management problem as the simulation environment, the DQN algorithm is used to build the initial power operation management model and the corresponding experience replay pool, including the following steps:

[0131] A-1: Generate the power operation management problem as the simulation environment of the DQN algorithm, and build the intelligent agent and experience replay pool;

[0132] A-2: Based on each power supply operation state type corresponding to the historical power supply operation analysis results, define the state space of the DQN algorithm. The parameters of the state space correspond to each power supply operation state, such as the system load of the host power supply, power supply output prediction, heat prediction, fault prediction, etc.

[0133] A-3: Define the action space of the DQN algorithm according to the actions that need to be output by the power operation management; the action space includes various actions to control the power host, such as power output, changing fan speed, entering sleep mode, etc.;

[0134] A-4: Based on the possible impact of each action in the action space, define the reward function of the DQN algorithm to evaluate the quality or impact of the action;

[0135] A-5: Construct the input layer, several hidden layers, and output layer of the deep Q network, connect the input layer to the state space, and connect the output layer to the action space;

[0136] The power operation management model is built based on the Deep Q Network (DQN) algorithm, and the power operation management model includes a deep Q network, an intelligent agent, and an experience replay pool;

[0137] S2-6: Based on the state space, action space and reward function, use several historical power operation analysis results to optimize the initial power operation management model, obtain the optimal power operation management model, and generate several historical power operation management experiences;

[0138] S2-7: storing a number of historical power supply operation management experiences in an experience playback pool corresponding to the optimal power supply operation management model;

[0139] S3: Based on the host and according to the host resources, simplifying the power operation analysis model and the power operation management model to obtain a simplified power operation analysis model and a simplified power operation management model, including the following steps:

[0140] S3-1: Based on the host, an optimization target is set according to a first model parameter of the power operation analysis model and a second model parameter of the power operation management model, and an objective function is set according to the host resources and the optimization target;

[0141] When the host computing resources are limited and fast response and real-time analysis are required, the power operation analysis model built based on the LSTM-Attention-cGAN algorithm and the power operation management model built based on the DQN algorithm need to be appropriately simplified or adjusted to adapt to specific application scenarios and resource constraints. In this embodiment, minimizing the host resource consumption is used as the optimization goal, and the first model parameters (the number of network layers of the LSTM network, the number of neurons in each network layer, the number of layers of the generator and discriminator of the cGAN network, and the encoding method of the conditional input) and the second model parameters (the number of hidden layers of the deep Q network, the number of neurons in each hidden layer) are simplified to make them more suitable for power operation data analysis while maintaining sufficient performance and low complexity;

[0142] The objective function is:

[0143]

[0144] Where F is the objective function; R i is the demand for host resources i, including host computing resources, host memory resources, host program resources, etc.; C i is the cost of host resource i; n' is the total number of host resources; W R , W c are the host resource demand weight and the host resource cost weight;

[0145] S3-2: According to the objective function, the Improved Crested Porcupine Optimizer (ICPO) optimization algorithm is used to search and optimize to obtain the optimal first model parameters and the optimal second model parameters, including the following steps:

[0146] S3-2-1: Encode the optimization problem of the first model parameter and the second model parameter to obtain a real-time model parameter encoding vector including the first model parameter and the second model parameter;

[0147] S3-2-2: The real-time model parameter encoding vector is used as the ICPO individual of the ICPO optimization algorithm, initialized using the Circle chaotic mapping sequence, generates several initial solutions of the ICPO optimization algorithm, and uses the objective function as the fitness function of the ICPO optimization algorithm;

[0148] The formula is:

[0149]

[0150] In the formula, It is the initial ICPO individual of Circle chaos mapping; is the randomly generated initial ICPO individual; i' is the ICPO individual indicator;

[0151] S3-2-3: Set ICPO population parameters and maximum number of iterations of ICPO optimization algorithm;

[0152] S3-2-4: Introduce a cyclic population reduction mechanism to limit the number of individuals in the ICPO population parameters and obtain updated ICPO population parameters for the next iteration;

[0153] The formula is:

[0154]

[0155] In the formula, S t+1 is the number of individuals in the ICPO population parameter of the t+1th iteration; S t is the number of individuals in the ICPO population parameter of the tth iteration; S min is the minimum number of individuals in the ICPO population parameter; a' is the function evaluation parameter; V is the function evaluation cycle parameter; V max is the maximum function evaluation loop parameter; t is the number of iterations indicator;

[0156] S3-2-5: Calculate the initial fitness value of the initial ICPO individuals in the initial ICPO population according to the fitness function;

[0157] S3-2-6: According to the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population;

[0158] The formula for the first defense strategy is:

[0159]

[0160] In the formula, For the updated ICPO individuals within the first defense range; is the initial ICPO individual within the first defense range; τ1 is a random number based on normal distribution; τ2 is a random value in the interval [0,1]; It is the optimal solution within the first defense range; is the vector generated between the true optimal solution within the first defense range and the optimal solution randomly selected from the ICPO population; i' is the ICPO individual indicator; t is the iteration indicator;

[0161] The formula for the second defense strategy is:

[0162]

[0163] In the formula, For the updated ICPO individuals within the second defense range; The initial ICPO individual within the second defense range; is the search upper limit vector of the second defense range; τ3 is a random value in the interval [0,1]; are the r1th and r2th initial ICPO individuals respectively; r1 and r2 are two random integers between [1, S]; is the vector generated between the true optimal solution within the second defense range and the optimal solution randomly selected from the ICPO population;

[0164] The formula for the third defense strategy is:

[0165]

[0166] In the formula, For the updated ICPO individuals within the third defense range; The initial ICPO individual within the third defense range; is the search upper limit vector of the third defense range; are the r2th and r3th initial ICPO individuals respectively; r3 is a random integer between [1, S]; F i' t The odor diffusion factor defined for the fitness function; λ t It is a defense factor; Control parameters for search direction;

[0167] The formula for the fourth defense strategy is:

[0168]

[0169] In the formula, For the updated ICPO individuals within the fourth defense range; The initial ICPO individual within the fourth defense range; is the optimal solution within the fourth defense range; τ4 and τ5 are both random values ​​in the interval [0,1]; λ t It is a defense factor; Control parameters for search direction; is the average force affecting the search direction; a' is the convergence speed factor;

[0170] S3-2-7: Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population;

[0171] The formula is:

[0172]

[0173] In the formula, is the ICPO individual in dynamic reverse direction; γ' is the decreasing inertia coefficient; L max , L min are the maximum and minimum values ​​of the vector space respectively; For the updated ICPO individuals;

[0174] S3-2-3-6: According to the fitness function, calculate the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population, take the ICPO individual with the minimum fitness value as the optimal individual, and retain the optimal individual;

[0175] S3-2-8: If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output;

[0176] S3-2-9: Decode the optimal solution to obtain the optimal first model parameter and the optimal second model parameter;

[0177] S3-3: simplifying the power supply operation analysis model according to the optimal first model parameter to obtain a simplified power supply operation analysis model;

[0178] S3-4: simplifying the power operation management model according to the optimal second model parameters to obtain a simplified power operation management model;

[0179] S4: Based on the power management device, collect real-time power operation monitoring data of the host power supply, and transmit the real-time power operation monitoring data to the host;

[0180] S5: Based on the host, the real-time power supply operation monitoring data is processed to obtain the processed real-time power supply operation monitoring data;

[0181] Data processing includes format conversion and normalization processing, and according to the principal component set, data dimension reduction is performed on the real-time power supply operation monitoring data to obtain processed real-time power supply operation monitoring data;

[0182] S6: Based on the host, according to the processed real-time power operation monitoring data, using the simplified power operation analysis model, the power operation analysis is performed to obtain the real-time power operation analysis results, including the following steps:

[0183] S6-1: Based on the host, the processed real-time power operation monitoring data is input into the input layer of the simplified power operation analysis model;

[0184] S6-2: using the feature extraction module to extract the real-time data features of the processed real-time power supply operation monitoring data, and inputting the real-time data features into the attention weight module respectively;

[0185] S6-3: Use the attention weight module to perform feature splicing on the real-time data features according to the preset attention weight value to obtain the real-time data splicing features, and input the real-time data splicing features into the adversarial training module;

[0186] S6-4: Use the adversarial training module to perform adversarial training on the real-time data splicing features to obtain real-time adversarial training features, and input the real-time adversarial training features into the classification layer;

[0187] S6-5: Use the classification layer to perform power operation analysis and prediction based on the real-time adversarial training features, obtain the real-time power operation analysis and prediction label, and input the real-time power operation analysis and prediction label into the output layer;

[0188] S6-6: using the output layer, converting the real-time power operation analysis prediction label into the real-time power operation analysis result, and outputting the real-time power operation analysis result;

[0189] S7: Based on the host, according to the real-time power operation analysis result, using the simplified power operation management model, generating a power operation management strategy, obtaining the real-time power operation management strategy, and transmitting it to the power management device, including the following steps:

[0190] S7-1: Based on the host, convert the real-time power operation analysis result into a real-time power operation analysis result encoding vector applicable to the simplified power operation management model;

[0191] S7-2: according to the encoding vector of the real-time power operation analysis result, searching among several historical power operation management experiences stored in the experience playback pool, and obtaining several matching historical power operation management experiences;

[0192] S7-3: According to the real-time power operation analysis result encoding vector and some matching historical power operation management experience, the model parameters of the simplified deep Q network are updated to obtain the updated model parameters, including the following steps:

[0193] S7-3-1: Update the simplified state space S'=[s'1,...,s' based on the encoding vector of the real-time power supply operation analysis results i" ,...,s' I' ], where s' i"is the updated i'th state value, i' is the state indicator, and I' is the total number of state space dimensions;

[0194] S7-3-2: Based on some matching historical power supply operation management experience, the action space of the simplified deep Q network is updated, and the updated action space A'=[a'1,...,a' j" ,...,a' I ], where a' j" is the updated j-th "action value", j" is the action indicator, and I is the total number of action space dimensions;

[0195] S7-4: According to the updated model parameters of the simplified deep Q network, using the intelligent agent, a power operation management strategy is generated to obtain a real-time power operation management strategy, including the following steps:

[0196] S7-4-1: Update the state space S' = [s'1, ..., s' i' ,...,s' I ] as the input of the simplified deep Q network, using the agent to generate the updated action space A' = [a'1,...,a' j' ,...,a' I ] is the Q value of each possible action;

[0197] S7-4-2: Use the reward function to obtain the reward value of each possible action in the updated action space, and update the Q value of the possible action according to the reward value to obtain the updated Q value of the possible action;

[0198] The formula is:

[0199] Q(s' p' ,a' p' )=(1-α")·Q(s p' ,a p' )+α"·(R(s p' ,a p' ,s' p' )+γ·Q max (s p' ,a p' ))

[0200] In the formula, Q(s' p' ,a' p' ) is the updated state value s' p' and the updated action value a' p' The corresponding updated Q value; Q(s p' ,a p' ) is the state value s p' and action value a p' The corresponding predicted Q value; α" is the learning rate; Qmax (s p' ,a p' ) is the highest predicted Q value; p' is the comprehensive indicator; γ is the update parameter;

[0201] S7-4-3: Repeat the above steps until the iteration number threshold is reached, use the greedy strategy, take the possible action corresponding to the highest updated Q value in each iteration as the execution action, and integrate all the execution actions to obtain the real-time power operation management strategy;

[0202] S7-5: transmitting the real-time power operation management strategy to the power management device;

[0203] S8: Based on the power management device, generate a real-time power operation management instruction according to the real-time power operation management strategy, and control the host power operation according to the real-time power operation management instruction.

[0204] Embodiment 2:

[0205] like Figure 2 As shown, this embodiment provides a host power operation management system based on data processing, which is used to implement a host power operation management method. The system includes a host, a power management device and a host power supply. The host is communicatively connected to the power management device, the power management device is communicatively connected to the host power supply, and the host power supply is electrically connected to the host.

[0206] A power management device, used to collect real-time power operation monitoring data of the host power supply and transmit the real-time power operation monitoring data to the host; generate real-time power operation management instructions according to the real-time power operation management strategy, and control the host power operation according to the real-time power operation management instructions;

[0207] Host power supply, used to provide power to the host;

[0208] The host is provided with a data processing unit, a model building unit, a model simplification unit, a power operation analysis unit and a power operation management unit connected in sequence;

[0209] The data processing unit is used to process a number of historical power operation monitoring data collected by the power management device to obtain a number of processed historical power operation monitoring data; and to process the real-time power operation monitoring data to obtain processed real-time power operation monitoring data;

[0210] A model building unit, used to build a model based on a number of processed historical power supply operation monitoring data using an artificial intelligence algorithm to obtain a power supply operation analysis model and a power supply operation management model;

[0211] A model simplification unit, used to simplify the power operation analysis model and the power operation management model according to the host resources to obtain a simplified power operation analysis model and a simplified power operation management model;

[0212] A power operation analysis unit, used to perform power operation analysis based on the processed real-time power operation monitoring data and using a simplified power operation analysis model to obtain a real-time power operation analysis result;

[0213] The power operation management unit is used to generate a power operation management strategy based on the real-time power operation analysis result and using a simplified power operation management model to obtain the real-time power operation management strategy and transmit it to the power management device.

[0214] The present invention provides a host power operation management method and system based on data processing. The power operation analysis model constructed by the artificial intelligence algorithm can automatically and efficiently process and analyze data, extract deep information from the host power operation monitoring data, and accurately capture the complex patterns and potential anomalies in the power operation data; the power operation management model constructed by the artificial intelligence algorithm can perform fast and intelligent power operation management strategies according to the power operation analysis results, and adaptively adjust the power operation management measures according to the power operation conditions, thereby improving the flexibility and effectiveness of the host power operation management; the power operation analysis model and the power operation management model are optimized by the group intelligence optimization algorithm. The parameters are simplified, which reduces the computational complexity of the model and enables it to operate in a resource-constrained environment. Real-time monitoring and rapid response based on real-time data collection enable timely adjustments to be made in the face of sudden conditions in the power system, thereby improving the real-time performance of host power operation management. The optimized model structure and parameters improve the generalization ability of the model in different working scenarios and enhance the adaptability of power management. The simplified model structure makes the power management strategy more intuitive and easy to understand, enhancing the interpretability of the entire management system. Data processing, real-time analysis, and management strategy generation are integrated into a complete power operation management system, forming a closed-loop, automated power management logic.

[0215] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A host power operation management method based on data processing, characterized in that: The steps include: Based on the host, data processing is performed on a number of historical power operation monitoring data collected by the power management device to obtain a number of processed historical power operation monitoring data; Based on the host, according to a number of processed historical power operation monitoring data, artificial intelligence algorithms are used to build models to obtain power operation analysis models and power operation management models; Based on the host and according to the host resources, the power operation analysis model and the power operation management model are simplified to obtain a simplified power operation analysis model and a simplified power operation management model; Based on the power management device, real-time power operation monitoring data of the host power supply is collected, and the real-time power operation monitoring data is transmitted to the host; Based on the host, the real-time power supply operation monitoring data is processed to obtain the processed real-time power supply operation monitoring data; Based on the host, according to the processed real-time power operation monitoring data, a simplified power operation analysis model is used to perform power operation analysis to obtain real-time power operation analysis results; Based on the host, according to the real-time power operation analysis result, a simplified power operation management model is used to generate a power operation management strategy, obtain the real-time power operation management strategy, and transmit it to the power management device; Based on the power management device, according to the real-time power operation management strategy, a real-time power operation management instruction is generated, and according to the real-time power operation management instruction, the host power operation is controlled.

2. The host power operation management method based on data processing according to claim 1, characterized in that: Based on the host, a plurality of historical power supply operation monitoring data collected by the power supply management device are processed to obtain a plurality of processed historical power supply operation monitoring data, including the following steps: Based on the power management device, a number of historical power operation monitoring data of the host power supply is collected, and the number of historical power operation monitoring data is transmitted to the host; Based on the host, pre-processing is performed on a number of historical power supply operation monitoring data to obtain a number of pre-processed historical power supply operation monitoring data; Data dimension reduction is performed on a number of pre-processed historical power supply operation monitoring data to obtain a number of processed historical power supply operation monitoring data.

3. The host power operation management method based on data processing according to claim 2 is characterized in that: The PCA method is used to perform data dimension reduction on a number of pre-processed historical power supply operation monitoring data to obtain a number of processed historical power supply operation monitoring data.

4. The host power operation management method based on data processing according to claim 1, characterized in that: Based on the host, according to a number of processed historical power operation monitoring data, using artificial intelligence algorithms, model construction is carried out to obtain a power operation analysis model and a power operation management model, including the following steps: Based on the host, a number of processed historical power supply operation monitoring data are divided into a model training set and a model test set in a ratio of 7:3; Use a deep learning algorithm to build an initial power supply operation analysis model, and use the model training set to optimize the initial power supply operation analysis model to obtain an optimized power supply operation analysis model; Using the model test set, the optimized power supply operation analysis model is tested to obtain the model test accuracy; If the model test accuracy is greater than the threshold, the optimal power supply operation analysis model is output, several historical power supply operation analysis results are generated, and the next step is entered; otherwise, the optimization training is continued; Taking the power operation management problem as the simulation environment, the reinforcement learning algorithm is used to build the initial power operation management model and the corresponding experience replay pool; Using several historical power operation analysis results, the initial power operation management model is optimized and trained to obtain the optimal power operation management model, and several historical power operation management experiences are generated; Several historical power operation management experiences are stored in an experience playback pool corresponding to the optimal power operation management model.

5. The host power operation management method based on data processing according to claim 4 is characterized in that: The power operation analysis model is constructed based on the LSTM-Attention-cGAN algorithm, and the power operation analysis model includes an input layer connected in sequence, a feature extraction module constructed based on the LSTM algorithm, an attention weight module constructed based on the Attention mechanism, and an adversarial training module, a classification layer, and an output layer constructed based on the cGAN algorithm; The power operation management model is constructed based on the DQN algorithm, and the power operation management model includes a deep Q network, an intelligent agent, and an experience replay pool.

6. The host power operation management method based on data processing according to claim 1, characterized in that: Based on the host and according to the host resources, the power operation analysis model and the power operation management model are simplified to obtain a simplified power operation analysis model and a simplified power operation management model, including the following steps: Based on the host, setting an optimization target according to a first model parameter of the power operation analysis model and a second model parameter of the power operation management model, and setting an objective function according to the host resources and the optimization target; According to the objective function, a swarm intelligence optimization algorithm is used to search and optimize to obtain the optimal first model parameters and the optimal second model parameters; According to the optimal first model parameter, the power supply operation analysis model is simplified to obtain a simplified power supply operation analysis model; The power operation management model is simplified according to the optimal second model parameters to obtain a simplified power operation management model.

7. The host power operation management method based on data processing according to claim 6 is characterized in that: According to the objective function, the ICPO optimization algorithm is used to search and optimize to obtain the optimal first model parameters and the optimal second model parameters.

8. The host power operation management method based on data processing according to claim 5 is characterized in that: Based on the host, according to the processed real-time power operation monitoring data, using the simplified power operation analysis model, power operation analysis is performed to obtain the real-time power operation analysis results, including the following steps: Based on the host, the processed real-time power operation monitoring data is input into the input layer of the simplified power operation analysis model; Using the feature extraction module, extract the real-time data features of the processed real-time power operation monitoring data, and input the real-time data features into the attention weight module respectively; Use the attention weight module to perform feature splicing on the real-time data features according to the preset attention weight value to obtain the real-time data splicing features, and input the real-time data splicing features into the adversarial training module; Use the adversarial training module to perform adversarial training on the real-time data splicing features to obtain real-time adversarial training features, and input the real-time adversarial training features into the classification layer; Use the classification layer to perform power operation analysis and prediction based on the real-time adversarial training features, obtain the real-time power operation analysis and prediction labels, and input the real-time power operation analysis and prediction labels into the output layer; Using the output layer, the real-time power operation analysis prediction labels are converted into real-time power operation analysis results, and the real-time power operation analysis results are output.

9. The host power operation management method based on data processing according to claim 5, characterized in that: Based on the host, according to the real-time power operation analysis result, using the simplified power operation management model, generating a power operation management strategy, obtaining the real-time power operation management strategy, and transmitting it to the power management device, including the following steps: Based on the host, convert the real-time power operation analysis result into a real-time power operation analysis result encoding vector applicable to the simplified power operation management model; According to the encoding vector of the real-time power operation analysis result, a plurality of historical power operation management experiences stored in the experience playback pool are searched to obtain a plurality of matching historical power operation management experiences; According to the encoding vector of the real-time power operation analysis result and some matching historical power operation management experience, the model parameters of the simplified deep Q network are updated to obtain the updated model parameters; According to the updated model parameters of the simplified deep Q network, the power operation management strategy is generated using the intelligent agent to obtain the real-time power operation management strategy; Transmits real-time power operation management policies to the power management device.

10. A host power operation management system based on data processing, used to implement the host power operation management method according to any one of claims 1 to 9, characterized in that: The system comprises a host, a power management device and a host power supply, wherein the host is communicatively connected to the power management device, the power management device is communicatively connected to the host power supply, and the host power supply is electrically connected to the host; The host is provided with a data processing unit, a model building unit, a model simplification unit, a power operation analysis unit and a power operation management unit which are connected in sequence.

Images