受政策因素和技术环境影响,线上教学已成常态化.为改进已有线上教学质量评价体系的不足,提升线上教学质量,文中以高校线上教学为研究对象,利用人工智能技术提出一种线上教学质量评价方法.首先,基于科学的教学评价理念和原则,充分考虑已有的评价指标和线上教学特点,建立包含多元化评价主体的评价指标体系;然后,利用随机森林方法对评价指标进行重要性评估,达到数据降维和教学反馈的目的;再采用卷积神经网络构建以学生评价、教师自评、同行评价、平台数据为输入,以专家评结果为输出的评价模型;最后,将所提出的线上教学质量评价体系和模型应用于教学实践.实践结果表明,所提方法能够有效评价高校线上教学质量并降低评价成本,具有较高的准确性和应用价值.

随着云计算技术的飞速发展,传统的黑板、粉笔教学向多媒体教学、开放式教学转化,同时课堂空间、教学资源、受众范围也发生转变.利用网络学习已成为学习者的一种学习趋势.通过对云平台的深入了解,研究与设计了一个通过云平台,学习者利用自己碎片化时间获取教育资源的新模式.该平台由构建基于云平台的开放式教育模型、云资源池建设、云空间服务建设三部分组成.通过对该云平台的小范围试用,学生学习能力与教师素质明显提高,取得良好的实践效果.

To ensure the safety of electrical equipment and staffs in wind power plant, it is necessary to design qualified grounding resistance, touch voltage and step voltage. When the grounding resistance meets the requirement, if the grounding resistance meets the requirement, it may also appear high step voltage and touch voltage on the ground and exceed permissible value. There are two main methods to reduce the step voltage and touch voltage: spreading grounding system and using surface layer. But the natural environment of wind turbines tough, it is not easy to expand the grounding grid. Without increasing the total length of earth mat, the unequal spacing arrangement of grounding conductor, or changing the relative position of the foundation and mat can significantly reduce the surface maximum touch voltage and improve the safety level of grounding net. Placing short vertical grounding rod at the corner of grounding grid can also reduce the maximum touch voltage. The relative position of grounding grid and foundation also affects the touch voltage value. In the fact of this, the author simulates and analyses the wind turbine grounding grid with the CDEGS software, and presents a kind of optimization design of wind turbine grounding grid.

This paper researches on the temperature rise of concrete caused by the current divergence when fault current divergence of tower foundation of UHV line happened.Tower foundation model is established based on the thermodynamics finite element analysis theory.The feasibility and effectiveness of the concrete heat simulation is analyzed by comparing the ANSYS simulation with test results.The calculation of maximal temperature rise of concrete on the condition of current divergence of typical vertical four tower bases single pile tower foundation of UHV line shows that the thermal stability of the tower foundation concrete can be guaranteed.

To research the impacts of artificial horizontal grounding body on grounding resistance and current divergence of tower foundation for UHV power transmission,based on CDEGS software a simulation model for typical tower foundation for UHV power transmission and its artificial horizontal grounding body is built,and simulation computation of resistance-reducing rate and current divergence ratio under different soil resistivity,different lengths of horizontal grounding body and horizontal grounding body with different number of electrodes is performed.Simulation results show that the effect of horizontal grounding body on grounding resistance reduction of tower foundation for UHV power transmission is not so evident,so the effect of natural grounding body should be fully utilized to the foundation grounding of towers for UHV power transmission.Current densities in foundation grounding bodies of three typical towers for UHV power transmission are calculated,and calculation results show that the current density in vertical single-pile grounding body is the maximum.When the grounding current is 10A,the maximum current density appears at the bottom segment of the grounding body and the maximum current density at non-bottom segment of the grounding body is 1.156 A/m.

This paper mainly discusses how to inspire students’ enthusiasm for study and improve students' innovative abilities through the reformation on teaching methods,teaching contents,knowledge structure and assessment form in the teaching of computer culture basic course.

According to the principle of boundary element analysis,The influences of horizontal grounding electrode on ultra high voltag(e UHV)tower grounding electrode are analyzed.A algorithm is proposed to calculate the current density distribution of UHV tower grounding electrode.By use of C language,the current density distribution of UHV tower grounding electrode is calculated,and the comparison of the calculated result with that by CDEGS software shows that the proposed method is correct.Calculating the current density distribution of UHV tower natural grounding with two different interpolation methods,the calculation results show that linear interpolation more in line with the actual current density distribution.

In order to calculate the grounding resistance of ultra high voltage(UHV) tower in the two-layer soil,a new method for two-layer soil structure is presented.The equivalent soil resistivity can be calculated by simulating different reflection coefficients,top soil thicknesses and length of grounding electrodes with CDEGS software.A formula is obtained by nonlinear fitting for calculating the equivalent soil resistivity,and accordingly the resistance of UHV tower with single foundation is calculated.In order to calculate the resistance of UHV tower with four foundations,the simplified formula of utilization coefficient is obtained by data fitting.Finally,the grounding resistances of simplified formula and that of the CDEGS simulation have been compared,the results show that the simplified formula is reliable,and it can provide reference for the engineering design.

Concerning PID process control of nonlinear dynamic system,a control model and its method of PID parameter adaptive tuning based on Process Neural Network(PNN) identification were proposed in the paper.Using the learning mechanism of PNN for time-varying input-output signals of dynamic system,the PID control with parameter adaptive matching was implemented by tracing change sensitivity information of the output with the controlling input of the controlled object by identifying the controlled object under certain optimal control rule.The PID control system structure and corresponding realization mechanism based on PNN identification were presented in the paper and the experimental results verify the effectiveness of the model and algorithm.

针对一般SVM在机制上难以直接对动态模式进行分类的问题,提出了一种基于函数正交基展开的过程支持向量机.该模型的输入为时变函数,输出为模式类别.在输入函数空间中选择一组适当的正交函数基,将输入函数在该组函数基下进行有限项展开,把展开式系数作为核函数的输入.由于时变函数在基函数映射下与展开式系数一一对应,从而可利用SVM的变换机制实现动态模式分类.给出了基于SMO的求解算法,实验结果验证了模型和算法的有效性.

Well logging lithological identification is a very important basis task in oil-gas exploration and can supply references to choose correct interpretation methods and interpretation parameters for logging interpretation.A comprehensive nonlinear lithological identification model based on process neural network is built in this paper.In order to improve the adaptability to solving practical problems and the efficiency of the algorithm,a learning algorithm of process neural network based on spline function fitting is developed.In the end,actual logging data of Bei 16 region in Hailar Basin is used for identification of lithology.The test results show that the lithological identification method based on spline process neural network avoids the course of establishing complex mathematical or physical model to extract morphologic pattern characteristics of sublayer logs in advance that common BP neural network need,improves the computing speed of the network and the noise immunity for real data effectively,and has better stability and generalization ability.

In view of the fact that pipeline professionals make massive specialized information retrieval,based on systematic analysis and study on implementation technology of topic-specific search engine,adopting multi-thread programming technology,we have designed and developed Pipeline Topic-Specific Search Engine System based on improved PageRank algorithm.This paper designs system structure according to pipeline information work flow and the key technology involved such as hyperlink analysis,searching strategy and page rank sort technology is discussed in detail.This system has been developed using VC++6.0 and the prototype system displays good applicability in practical application.

Aiming at the problem that the inputs and the outputs of real systems are continuous processes related to time, a process neural network model for continuous process approximation is proposed in the paper. The inputs and the outputs of this neural network model are both continuous time functions, and we can use the nonlinear mapping capability of the neural network to realize continuous mapping relationship between system's inputs and outputs. The paper gives a learning algorithm based on function basis expansion integrated with gradient descent and proves effectiveness of the model and the algorithm by an example of exploitation process simulation in oil field.

Aiming at the nonlinear system modeling problem that input and output are time-varying functions,a process neural network model with time-varying inputs and outputs is proposed,and the concrete learning algorithm is given.The inputs and outputs of the proposed process neural network are all time-varying functions.Its spatial-temporal aggregation operators are adopted as space weighted summation and integral depending on time parameter separately.Aggregation operation and activation can reflect the space aggregation function of the time-varying input signals and the stage time additive effect in the input process at the same time.The simulation experiment results show the effectiveness of the proposed model and algorithm.

针对带有奇异值复杂时变信号的模式分类和系统建模问题,提出了一种分式过程神经元网络·该模型是基于有理式函数具有的对复杂过程信号的逼近性质和过程神经元网络对时变信息的非线性变换机制构建的,其基本信息处理单元由两个过程神经元成对偶组成,逻辑上构成一个分式过程神经元,是人工神经网络在结构和信息处理机制上的一种扩展·分析了分式过程神经元网络的连续性和泛函数逼近能力,给出了基于函数正交基展开的学习算法·实验结果表明,分式过程神经元网络对于带有奇异值时变函数样本的学习性质和泛化性质要优于BP网络和一般过程神经元网络,网络隐层数和节点数可较大减少,且算法的学习性质与传统BP算法相同·

Aim at the problems that the inputs and outputs of some practical nonlinear systems are Continuous time signals, we brought forward a Continuous process neuron and process neural networks model. The input and output of the defined process neuron are Continuous time functions, and the space-time aggregation operation can reflect the space aggregation of the input signals and the time cumulative effect in the process of input at the same time, and can also realize the nonlinear real-time mapping between the input and output. A Continuous feedforward process neural networks model is given in this paper, and the corresponding property theorems are also proved, including continuity, function approximation ability and computational capacity.