火电厂控制系统综合性能评价及优化

发布时间：2018-05-25 03:06

本文选题：性能评价 + 多变量系统　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：控制系统性能监测与评价是工业过程中保证生产运行高效、安全、环保必不可少的一部分。随着生产规模日益扩大,控制回路不断增多,人工监测具有一定难度,必须实现自动的控制系统性能评价及监视。目前,火力发电在我国全社会发电量中所占比例较大,约为75%左右。火电厂设备结构高度复杂,控制回路往往难以实现预期运行效果,需通过一定的评价方法对其性能进行度量,来指导回路中控制器的优化。同时,在经过一段时间的运行后,原来性能良好的回路由于各种内因及外因的影响,性能也会变差,需对其进行及时的监视。因此,要想确保良好的运行状态,进行火电厂控制系统性能评价与优化研究十分必要。针对火电厂控制系统性能评价与优化问题,本论文的主要工作以及取得成果:首先,针对高斯扰动下的单变量系统介绍了经典的最小方差性能评价基准,并将其延伸到多变量控制系统。提出一种基于等效开环传递函数的多变量控制系统性能评价方法,解决了多变量系统最小方差基准求取复杂、限制条件较多等问题,并由仿真实验对其实用性以及准确性进行了验证。其次,针对非高斯噪声扰动下的系统,介绍了最小信息熵性能评价指标,在其基础上引入有理熵,提出了最小有理熵性能评价指标。同时,介绍了一种基于历史数据的Hellinger距离性能评价基准。通过仿真验证了最小信息熵指标、最小有理熵指标以及基于Hellinger距离性能指标这三种指标的有效性。通过对比分析,得出基于Hellinger距离性能指标较适合工业控制系统性能评价,但评价结果不够客观;与最小信息熵指标相比,最小有理熵指标能够更准确的反映系统性能变化的趋势这两点结论。再次,提出一种标量ITAE性能评价指标用以控制系统确定性性能评价,并将该指标通过子目标乘除法与系统随机性性能指相标结合,得到一种综合性能指标。将综合指标作为目标函数,对火电厂中的单元机组协调控制系统控制器进行参数优化,得到了最优自抗扰控制器、广义预测控制器、内模PI控制器参数与对应最优性能指标,并对三种控制器性能进行了对比与监测。实验结论表明广义预测控制综合性能最好,并且在模型参数摄动情况下鲁棒性最好。最后,对于模型参数未知的火电厂控制系统,总结出一种基于数据的综合性能评价方法,并对贵州省盘南电厂3号机组汽包水位控制回路进行了性能评价与监测。
[Abstract]:The performance monitoring and evaluation of control system is an essential part to ensure high efficiency, safety and environmental protection in industrial process. With the increasing scale of production and the increasing number of control loops, manual monitoring is difficult, so it is necessary to realize automatic performance evaluation and monitoring of the control system. At present, the proportion of thermal power generation in the whole society is about 75%. The structure of thermal power plant equipment is highly complex and the control loop is often difficult to achieve the expected operation effect. It is necessary to measure its performance by a certain evaluation method to guide the optimization of the controller in the loop. At the same time, after running for a period of time, the performance of the original circuit with good performance will become poor due to various internal and external factors, so it needs to be monitored in time. Therefore, in order to ensure good operation state, it is necessary to evaluate and optimize the performance of thermal power plant control system. Aiming at the performance evaluation and optimization of control system in thermal power plant, the main work and achievements of this paper are as follows: firstly, the classical minimum variance performance evaluation benchmark is introduced for the single-variable system under Gao Si disturbance. It is extended to multivariable control system. A performance evaluation method for multivariable control systems based on equivalent open loop transfer function is proposed. The practicability and accuracy are verified by simulation experiments. Secondly, the performance evaluation index of minimum information entropy is introduced for the system with non-Gao Si noise disturbance. Based on the introduction of rational entropy, the performance evaluation index of minimum rational entropy is proposed. At the same time, a Hellinger distance performance evaluation benchmark based on historical data is introduced. The effectiveness of the minimum information entropy index, the minimum rational entropy index and the performance index based on Hellinger distance are verified by simulation. Through comparison and analysis, it is concluded that the performance index based on Hellinger distance is more suitable for the performance evaluation of industrial control system, but the evaluation result is not objective enough, and compared with the minimum information entropy index, The minimum rational entropy index can reflect the trend of system performance change more accurately. Thirdly, a scalar ITAE performance evaluation index is proposed to control the deterministic performance evaluation of the system, and a comprehensive performance index is obtained by combining the sub-objective multiplication and division method with the stochastic performance index of the system. Taking the comprehensive index as the objective function, the parameters of the coordinated control system controller of the unit in the thermal power plant are optimized, and the parameters of the optimal ADRC controller, the generalized predictive controller, the internal model Pi controller and the corresponding optimal performance index are obtained. The performance of the three controllers is compared and monitored. The experimental results show that the generalized predictive control has the best comprehensive performance and the best robustness under the perturbation of the model parameters. Finally, for the control system of thermal power plant with unknown model parameters, a comprehensive performance evaluation method based on data is summarized, and the performance evaluation and monitoring of drum water level control loop of No. 3 unit in Pannan Power Plant, Guizhou Province are carried out.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP273;TM621

【参考文献】