代码import pandas as pd import numpy as np from scipy.optimize import minimize data = pd.read_csv('E:/lunwenpython/measure

这个错误是由于在objective_function函数中，predicted_failure_time和actual_failure_time都是DataFrame类型，而计算mse时需要将它们转换为numpy数组。可以使用.values将DataFrame转换为numpy数组。具体修改如下： import pandas as pd import numpy as np from scipy.optimize import minimize # 加载数据 data = pd.read_csv('E:/lunwenpython/measures_v21.csv') data= data.iloc[610000:630000, ] data = (data - data.mean()) / data.std() def pearson_fusion(data, weights): # 计算每个特征与目标变量之间的皮尔逊相关系数 corr = data.corrwith(data['pm']) corr.drop('pm', inplace=True) # 根据皮尔逊相关系数和权重计算复合健康指标 composite_index = 0 for i in range(len(corr)): composite_index += weights[i]*corr[i]*data.iloc[:,i] return composite_index, corr # 加载设备实际寿命数据 actual_failure_time = data.iloc[:-1]['time'].values # 假设设备实际寿命为1000秒 # 定义目标函数 def objective_function(x, data): weights = x[:6] T = x[6] data = data.iloc[:-1, :] composite_index, corr = pearson_fusion(data, weights) N = len(composite_index) dt = T/N t = np.linspace(0, T, N+1) dW = np.sqrt(dt)*np.random.randn(N) W = np.cumsum(dW) W = np.insert(W, 0, 0) weights = np.repeat(weights, N+1).reshape((6,N+1)) Z =np.dot(weights, W.flatten()) predicted_failure_time = t[np.argmax(Z > 0.9*np.max(Z))] mse = ((predicted_failure_time - actual_failure_time)**2)/len(data.iloc[:-1]) return mse res = minimize(objective_function, [0.1]*6 + [1000], args=(data,), method='Nelder-Mead') optimized_weights