Distributed Control Technologies And Its Application In Smart Distribution Network

With the development of society and the improvement of people’s living standard, the requirements for power supply quality and reliability from users become more and more stringent. As the last part of the power system which interfaces with the end users, distribution network has a direct impact on the quality of power supply. While a large number of distributed energy resources (DER) being required to access to the distribution network, the distribution network become an active network with two-way flow of energy. The dispersion and intermittent characteristics of DER outputs create the volatility of power flow and voltage increased in distribution network. This brings new challenges to the protection and control, as well as operation and management of distribution network. New control methods are urgently needed.Traditional control methods include local control and centralized control. The local control mode only uses the information from the local control device. It is easy to implement with fast action speed, but since the information acquired is very limited, the control performance of this mode is imperfect. The centralized control mode with the master station can use the information from the whole system, so it can optimize the control performance, but the response speed is slow and is strongly dependent on the communication network. The development trend of control mode is towards distributed control technology based on IP peer-to-peer communication. It uses the measurement information from multiple sites to improve the performance of control and protection applications whilst avoiding the long time latency problem associated with the communication and data processing operations of the centralized control mode. Currently, there are already some on-going research works on distributed control of smart distribution network, but no specific concept exists about distributed control is in the smart distribution network. Suitable control strategies are needed and some key technical problems remain unresolved. Distributed control theory system still needs to be established.Distributed control theory and its related key supporting technologies of smart distribution network are studied in this paper. Basic concepts of distributed control applications in smart distribution network are summarized and proposed. The method for the smart terminal units (STUs) to identify real-time topology structure of distribution network autonomously is proposed. The method of transmitting real-time data is explored using GOOSE over UDP transmission mechanism. Therefore, two critical fundamental problems about distributed control implementation are solved. As an application example of distributed control in distribution network protection, a distributed protection algorithm based on wide area blocking signals is proposed, which could overcome the difficulty of the existing overcurrent protection in achieving both rapidity and selectivity simultaneously.The main contributions and innovations of this thesis are as follows:(1) The basic concepts and coordinated control strategies of distributed control applications in smart distribution network are summarized and provided, which lays a theoretical foundation for the study of distributed control. The definitions of control task, control domain and etc. are given. The start mode of control task is classified into 3 kinds:periodic start mode, event trigger mode and remote command mode. The control method is divided into cooperative control and agent control. The characteristics and appropriate applications of different control tasks’starting mode and operation mode are analyzed.(2) A self-identification method for distributed control application topology in smart distribution network based on STUs’successive polling is proposed, thus solving one of the key problems on distributed control realization. The implementation of distributed control applications in a distribution network requires the STUs to store the feeder real-time topology. While the number of sites that a specific distributed control application related is small, the realization of the application does not need the topology information of the whole station. The traditional way is to initially acquire the topology information by the master station, the master station then passes the application related topology information to the STUs. There are several problems existed in this way:the system is highly dependent on the master station, security and reliability risks are high, and the topology identification and update speed is slow. Therefore, several topology obtaining methods and realization mechanisms of topology self-polling by STU are analyzed, and a successive polling method by STU to self-identify and to store the topology is proposed. A STU is first configured with its local topology information and established communications with its adjacent STUs. The static feeder topology information is then obtained by polling other STUs’ local topology information. After that, application topology is structured according to the state of the switches in the control domain. The stored topology of each STU is updated automatically when the feeder topology changes. Good flexibility, small configuration workload and easy to design distributed control application software can then be achieved. Local network of STUs is defined and Common Information Model (CIM) is used to model the feeder topology which can drive the interoperability realization among STUs and between the STUs and the master station.(3) A transmission scheme (GOOSE over UDP) based on UDP(User Datagram Protocol) to transmit GOOSE real-time message is proposed. The real-time message is mapped to the UDP protocol of transport layer, and based on IP peer-to-peer communication network to achieve message transmission in WAN(Wide Area Netork), which can satisfy the requirements of real-time on message transmission. The main impact factors on the communication latency of end to end applications are analyzed, and the communication requirements for distributed control applications of distribution network are given. The existing fast message transmission modes are compared and the performance of transmitting GOOSE messages with the other measurement information in the Ethernet are studied. The idea that GOOSE over UDP transmission scheme should be used to transmit the real-time data in distributed control applications of distribution network is proposed. To ensure reliability and speed of message transmission, several real-time improvement measures such as GOOSE retransmission mechanism, Linux real time operation system, process priority setting and IP layer priority configuration are proposed. The composition of transmission latency is analyzed, the Ping-pong principle and realization method for transmission latency test are given. This method can be realized by programming and does not need the STUs to maintain accurate time synchronization. The principle of separating the packet transmission latency into the internal processing latency of STUs and the network transmission latency between STUs is proposed, the format of application layer message is designed, and the kernel program of STUs’ operating system is modified to test the real-time performance of the proposed transmission scheme. A real-time data transmission test system that could transmit GOOSE over UDP real-time data among LANs(Local Area Networks) was set up. Various factors that impact the real time performance of message transmission latency were tested. The maximum transmission latency between STUs measured was found to be less than 10ms, which is able to meet the requirements of distributed control applications in distribution network.(4) A distributed overcurrent protection algorithm using wide area blocking signals is proposed. STU who detected fault current will start the algorithm, if the of their monitored switches are normal, they will issue blocking signals to the opposite direction of the fault current flow(If the state of the monitored switches are failure, stop issuing the blocking signals immediately). If the switch is not a terminal switch, it will trip till not received a blocking signal in a setting time; if the switch is a terminal switch, it will trip directly as soon as it detected the fault current. Abnormal occasions such as link failure, single point communication failure and breaker failure, etc. are considered. The configuration process is easy. This method can well adapt to the variation of the distribution network topology to achieve fast and selective fault clearance.(5) A test system for distributed overcurrent protection using wide area blocking signals is developed based on the static simulation system of active distribution network. The STUs use the proposed successive polling method to obtain the application topology of distributed overcurrent protection using wide area blocking signals. Then realize protection and control commands such as overcurrent detection results and blocking signals transmission through GOOSE over UDP transmission mechanism. The test results show that the distributed overcurrent protection using wide area blocking signals can be achieved under occasions such as:normal, line structure changes and single point network failure, etc., the outage area can be minimized, and protection action can be completed within 100ms.Since distributed control can achieve good control performance with fast response speed, it will be the development trend of protection and control applications in distribution network. The ability of STUs on topology self-identification and preservation can support the autonomously realization of distributed control, this way has good flexibility, and can not only ensures real-time of the topology information, but also reduces the risks of insecurity and unreliability by checking the topology information with the master station. The GOOSE over UDP transmission mechanism is proposed to realize the transmission of real-time data (such as switch tripping and blocking signals, etc.), it can satisfy the requirements of distributed control applications on the performance of real-time and reliable about fast message transmission in the WAN, the configuration process is simple and the universality is achieved in this mechanism. As an application example of distributed control in distribution network, a distributed overcurrent protection algorithm using wide area blocking signals is proposed. It is able to adapt to the changes in the structure of distribution network and can achieve protection within 100ms. This algorithm can overcome the difficulty of the traditional overcurrent protection in achieving speed and selectivity simultaneously. The research results of this thesis have great significances in promoting the study and application of distributed control, enriching the control measures of distribution network, and promoting the development of smart distribution network.