Automatic generation method of power grid maintenance operation ticket based on GA-ANN optimization algorithm

The automatic generation method of power grid maintenance operation ticket based on the GA-ANN optimization algorithm is designed. An automatic generation model of power grid maintenance operation ticket, including data communication, power grid information topology, task planning, path search, path selection, operation ticket generation and simulation of rehearsal agent, is built. This paper takes the operation sequence generated by different equipment performing operation tasks as the training sample of the GA-ANN model, obtains the change rule of operation sequence of each equipment agent, and realizes the automatic generation of maintenance operation tickets for different equipment. It is verified by experiments that the time of automatic generation of operation tickets is fast and can complete the maintenance and invoicing of various equipment.


Introduction
Automatic generation of operation tickets can provide a more comprehensive guarantee for the maintenance of power grid equipment.The GA-ANN optimization algorithm is a learning algorithm that integrates a genetic algorithm and an artificial neural network model.The algorithm can obtain the best data result through continuous training.Therefore, this paper studies the automatic generation method of power grid maintenance operation order based on the GA-ANN optimization algorithm, constructs the automatic generation model of power grid maintenance operation order, and designs the individual coding, fitness function, genetic operator, and other parameters of genetic algorithm (GA) to optimize the artificial neural network (ANN).This paper also constructs the GA-ANN model, takes the operation sequence generated by different equipment performing the operation task as the training sample of the GA-ANN model, obtains the change rule of the operation sequence of each equipment agent, and finally makes the operation order automatically generated.

Automatic generation model of power grid maintenance operation order
Because of the communication relationship between multiple devices in the process of maintenance, the tasks of each device are not the same, so the number of agents is different.The flexible architecture is used to design the model [1][2][3].If the task handled by the device is relatively complex, it is decomposed into multiple independent tasks, which are performed by multiple departments.Figure 1 shows the model structure.

Operation data processing of power grid maintenance equipment based on GA-ANN
The bus operation Agent is trained.The bus operation order information is used as the input vector [4][5] of the GA-ANN network, and the operation sequence information of equipment components is used as the output vector of the network.There are 12 input vectors and 8 output vectors in the GA-ANN model.The input vectors are [ ] , where 1

X and 2
X are used to describe the bus device when 1 1 X = and 2 0 X = are used to describe the bus device, 1 0 X = and 2 1 X = are used to describe the bus switch, and 1 1 X = and 2 1 X = are used to describe the combination operation of the bus and the bus switch.
3 X , 4 X , and 5 X all represent operation types.When the busbar is in standby condition, there is no distinction between cold standby and hot standby states, which only include operation, standby and maintenance states [6][7], so different operating states can be represented by binary coding.
The bus connection is represented by 6 X , 7 X , 8 X and 9 X respectively.There are eight types of bus connection in the power grid, and there are internal bridge connection and external bridge connection, which are respectively represented by 1000 and 1001.
The voltage levels are represented by 10 X and 11 X .They take the following forms and are represented by binary numbers: 00 for 220 kV, 10 for 35 kV, 01 for 110 kV, and 11 for 10 kV.
The bus type, represented by 12 X , has a main bus and a by-pass bus, which can be described by 0 and 1.

The output vector is
[ ] . The specific definition of each parameter is shown in Table 1.The connection weights matrix between the two layers is obtained through the ANN learning process, and the following steps are adopted to demonstrate the automatic generation process of bus operation.Agent maintenance operation orders: Generation task: 110 kV bus, double bus section with bypass, running into maintenance.
( Because the learning process of other power grid equipment and bus operation is basically the same, and so on, we can analyze other equipment.

Experimental analysis
Based on the OMS mode, the structure of automatic generation of power grid maintenance operation tickets is constructed, that is, the database server and the application server use the existing OMS system hardware.Only by configuring desktop computers for relevant business personnel, users can visually browse the status of power grid maintenance operation ticket generation.
This paper compares the methods in [3] and [4] with this method, and analyzes the time needed for different methods to generate operation order.The analysis results are shown in Figure 2. According to Figure 2, the invoicing time of the proposed method is not affected by the number of devices, which only increases slightly, and the maximum is less than 10 seconds.
Analysis of the application of this method can automatically issue a reasonable operation ticket to the grid equipment that needs to be repaired, and the results are shown in Table 2.In Table 2, the completion of bus operation Agent maintenance order automatic generation process is the realization of automatic generation of grid maintenance order.

Conclusion
Through the analysis of many key and difficult problems, the optimization and management of the maintenance plan are realized, and the safety check of the maintenance plan is realized.

Figure 2 .
Figure 2. Time required to generate an action order

Table 2 .
Analysis of invoicing transactions