设P=[-1:0.1:1]；T=[-0.96,-0.577,-0.0729,0.377,0.641,0.66,0.461,0.1336,-0.201,-0.434,-0.5,-0.393,-0.1647,0.0988,0.3072,0.396,0.3449,0.1816,-0.0312,-0.2183,-0.3201].试用Matlab编程同时设计和实现分别包含6~12个隐单元的三层前馈BP神经网络，并通过计算误差，确定选择哪种隐单元个数的网络可得到相对最好的BP网络。

好的，我会根据您的要求进行回答。针对您的问题，可以参考以下的Matlab代码实现：

% 数据准备
P = [-1:0.1:1];
T = [-0.96,-0.577,-0.0729,0.377,0.641,0.66,0.461,0.1336,-0.201,-0.434,-0.5,-0.393,-0.1647,0.0988,0.3072,0.396,0.3449,0.1816,-0.0312,-0.2183,-0.3201];

% 网络训练和测试
hiddenLayerSizes = 6:12;
errors = zeros(size(hiddenLayerSizes));
for i = 1:length(hiddenLayerSizes)
    hiddenLayerSize = hiddenLayerSizes(i);
    net = fitnet(hiddenLayerSize);
    net = train(net, P, T);
    Y = net(P);
    errors(i) = perform(net, T, Y);
end

% 可视化误差结果
plot(hiddenLayerSizes, errors, '-o');
xlabel('Number of hidden units');
ylabel('Error');
title('Performance of BP network for different numbers of hidden units');

代码中使用了`fitnet`函数创建了一个三层前馈BP神经网络，`hiddenLayerSizes`定义了隐层单元数的范围，并在循环中依次训练和测试了不同隐层单元数的网络，并计算了每个网络的误差。最后使用`plot`函数将不同隐层单元数的误差进行可视化。

您可以根据可视化结果，选择误差最小的隐层单元数作为最终的BP网络模型。