使用python写一个有关水质信息的BP网络（8个输入，1个输出，5个隐藏节点）

以下是使用Python编写的基本BP网络，用于水质信息的预测：

import numpy as np

# 定义sigmoid函数
def sigmoid(x):
    return 1 / (1 + np.exp(-x))

# 定义BP神经网络类
class BPNeuralNetwork:
    def __init__(self, num_inputs, num_hidden, num_outputs):
        self.num_inputs = num_inputs
        self.num_hidden = num_hidden
        self.num_outputs = num_outputs
        # 初始化权重
        self.weights_input_hidden = np.random.randn(self.num_inputs, self.num_hidden)
        self.weights_hidden_output = np.random.randn(self.num_hidden, self.num_outputs)

    # 前向传播
    def forward(self, inputs):
        hidden_inputs = np.dot(inputs, self.weights_input_hidden)
        hidden_outputs = sigmoid(hidden_inputs)
        final_inputs = np.dot(hidden_outputs, self.weights_hidden_output)
        final_outputs = sigmoid(final_inputs)
        return final_outputs

    # 反向传播
    def backward(self, inputs, targets, outputs, learning_rate):
        output_errors = targets - outputs
        hidden_errors = np.dot(output_errors, self.weights_hidden_output.T) * hidden_outputs * (1 - hidden_outputs)
        self.weights_hidden_output += learning_rate * np.dot(hidden_outputs.T, output_errors)
        self.weights_input_hidden += learning_rate * np.dot(inputs.T, hidden_errors)

    # 训练BP神经网络
    def train(self, inputs, targets, num_iterations, learning_rate):
        for i in range(num_iterations):
            for j in range(len(inputs)):
                inputs_j = np.array(inputs[j], ndmin=2)
                targets_j = np.array(targets[j], ndmin=2)
                hidden_inputs = np.dot(inputs_j, self.weights_input_hidden)
                hidden_outputs = sigmoid(hidden_inputs)
                final_inputs = np.dot(hidden_outputs, self.weights_hidden_output)
                final_outputs = sigmoid(final_inputs)
                self.backward(inputs_j, targets_j, final_outputs, learning_rate)

# 测试BP神经网络
if __name__ == '__main__':
    inputs = np.array([[1, 2, 3, 4, 5, 6, 7, 8], [2, 3, 4, 5, 6, 7, 8, 9], [3, 4, 5, 6, 7, 8, 9, 10]])
    targets = np.array([[1], [0], [1]])
    nn = BPNeuralNetwork(8, 5, 1)
    nn.train(inputs, targets, 1000, 0.1)
    print(nn.forward([1, 2, 3, 4, 5, 6, 7, 8]))

这个BP神经网络有8个输入，5个隐藏节点和1个输出。在这个示例中，我们使用了一个简单的数据集进行训练，即三个输入向量和相应的目标向量。