别等服务器炸了才慌！深度学习让运维更“未卜先知”

import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.layers import LSTM, Dense

# 生成模拟服务器数据
time_steps = 50
features = 3  # CPU、内存、磁盘IO
X_train = np.random.rand(1000, time_steps, features)
y_train = np.random.rand(1000, features)

# 构建LSTM模型
model = keras.Sequential([
    LSTM(64, return_sequences=False, input_shape=(time_steps, features)),
    Dense(features)
])

model.compile(optimizer='adam', loss='mse')
model.fit(X_train, y_train, epochs=10)

# 预测下一时刻
predicted_value = model.predict(X_train[:1])
print(predicted_value)

from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.models import Model

# 构建自编码器
input_dim = 100  # 假设日志向量化后有100维
input_layer = Input(shape=(input_dim,))
encoded = Dense(64, activation='relu')(input_layer)
decoded = Dense(input_dim, activation='sigmoid')(encoded)

autoencoder = Model(input_layer, decoded)
autoencoder.compile(optimizer='adam', loss='mse')

# 训练模型
X_logs = np.random.rand(5000, input_dim)  # 模拟日志数据
autoencoder.fit(X_logs, X_logs, epochs=10)

# 识别异常日志
reconstructed_logs = autoencoder.predict(X_logs)
anomaly_score = np.abs(X_logs - reconstructed_logs).mean(axis=1)
print("异常分数:", anomaly_score)