目前普遍使用的B/S计算方案难以胜任大运算负荷、非稳定、交互型的计算任务。原因在于 ,目前基于Web的计算方案的主要计算中心建立在服务器端 ,从而在客户与服务器之间的通讯、多用户并发请求、计算稳定性等方面都存在很多问题。本文在保留目前B/S网络计算模式的优点的前提下 ,提出以客户端为计算中心 ,服务器为数据交换中心的网络计算模型 ,避免了上述问题的出现。文章将提出的网络计算模型应用于人工神经网络 ,结合化工应用中的典型计算任务 ,例示了以客户为中心的B/S网络计算模型的要素。结果表明 ,该模型对耗时、非稳定、交互的科学计算任务具有明显的优越性 ,且比以服务器为中心的计算模型更具有发展潜力。同时 ,这种模型也应该是未来软件出租服务的模式之一 Currently widely used B / S computing solution is not competent for large computing load, non-stable, interactive computing tasks. The reason is that at present, the main computing center of the web-based computing solution is set up on the server side, and thus there are many problems in terms of client-server communication, multi-user concurrent requests, computational stability and the like. Based on the advantages of current B / S network computing mode, this paper proposes a network computing model that uses the client as the computing center and the server as the data exchange center, and avoids the above problems. The article presents the network computing model applied to artificial neural network, and illustrates the key elements of client-centric B / S network computing model in combination with typical computing tasks in chemical applications. The results show that this model has obvious advantages for time-consuming, unstable and interactive scientific computing tasks and has more potential than server-centered computing models. At the same time, this model should also be one of the models for future software rental services