Ship power system situation awareness platform architecture based on data driving

Abstract

The invention relates to a ship power system situation awareness platform architecture based on data driving, which sorts data types, measuring point scales and deployment modes monitored by conventional ship integrated power system monitoring equipment and typical electrical equipment, and combines system state evaluation and prediction requirements. And carrying out the situation awareness overall architecture design of the integrated power system based on data driving. Potential typical data feature analysis is carried out through supplementary data acquisition and edge computing equipment, system modeling, data mining, analysis and other technologies, a comprehensive power system situation prediction model based on data driving is constructed, the operation complexity of an operator can be greatly reduced through the datamation and intelligence degree, and the number of human faults is reduced; as the power system becomes more and more complex, the operability of the system is improved, and the number of operators and the training time are reduced. And a data support is provided for applications such as subsequent energy management, system abnormity identification and early warning, efficient positioning analysis and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of ship power monitoring, and in particular to a data-driven ship power system situation awareness platform architecture. Background Art

[0002] At present, the information of the equipment measurement points monitored by the ship power monitoring system is not comprehensive enough and the collection accuracy is low, which makes it difficult to meet the needs of intelligent, unmanned and data-driven ship power system status assessment, fault diagnosis and trend prediction. It is urgent to establish a new ship power perception system, which can not only ensure the normal operation of the original power monitoring system without interference, but also access the new signal sensing signals required for the health assessment and fault diagnosis of the power system and its important equipment, and process massive data to reduce the complexity of operators analyzing massive data. At the same time, the speed of the processor to process multiple scenarios and big data is much higher than the speed of human brain judgment after operator inspection; at the same time, with the help of technical means such as three-dimensional display, the status display and status assessment results of the power system and its important equipment are intuitively displayed, and the trend analysis and prediction results of the power system are given to reduce the reliance on personnel experience to judge the status of the power system. Summary of the invention

[0003] Aiming at improving the intelligence of ship power monitoring system, a data-driven ship power system situation awareness platform architecture was proposed.

[0004] The technical solution of the present invention is: a data-driven ship power system situation awareness platform architecture, which includes, from bottom to top, a data perception layer, a data acquisition layer, a data layer for data processing and storage, a model layer for data mining, and an interaction layer for analytical applications.

[0005] The data perception layer: uses a new type of sensor to perceive signal data, the signal data is the current power system grid state data and the power generation and distribution equipment working state data;

[0006] The data collection layer collects the real-time data of the data perception layer through the existing local collection equipment, and analyzes the real-time data of each device in combination with the device model locally to obtain the real-time status and status prediction of the device, and uploads them to the integrated network for the interaction layer to call;

[0007] The data layer: receives the real-time status data of the equipment uploaded by the data collection layer, stores the inherent design data of the power system and the power generation and distribution equipment, calls the corresponding model of the model layer, trains the model through the data, obtains the applicable model of the equipment to the equipment configuration management, and calls the data collection layer and the interaction layer;

[0008] The model layer includes a three-dimensional equipment model and a mechanism model;

[0009] The interaction layer: the front-end display page and the human-computer interaction layer integrate a large amount of real-time data, situation data, situation prediction data and fault data of multi-source heterogeneous power generation equipment, power grids and external power-consuming equipment, to achieve multi-dimensional correlation results of all aspects of the unmanned power system, infer the root cause of power grid faults, track fault trajectories, push fault warnings, provide fault analysis reports, establish a fault data archive system, provide auxiliary decision-making suggestions, and equipment maintenance and repair suggestions.

[0010] Preferably, the data acquisition layer includes an edge measurement terminal, which collects special sensor data through IO hardware; the edge measurement terminal collects equipment monitoring data of the machine-side control box and the power station controller through RS485 communication; the edge measurement terminal has built-in data processing, data feature extraction, and data analysis programs, which process, extract data features, and analyze the special sensor data collected by the IO hardware and the equipment monitoring data collected by RS485 communication, and transmit the obtained equipment status data to the integrated network via Ethernet communication.

[0011] Preferably, an edge terminal storage device-level perception model is set in the edge measurement terminal. The model is the actual physical model of each device, which is established according to the thermodynamic or electrotechnical characteristics of the device. When the model is used, the input is the special sensor data collected by the edge test terminal IO hardware and the equipment monitoring data collected by RS485 communication, and the output is the monitoring parameters of the device when it is running healthily; the monitoring data actually collected by the edge test terminal is compared with the output data of the model, and the difference between the two groups of data is evaluated through the health assessment algorithm and the fault diagnosis algorithm to obtain the health level of the device, that is, the real-time status of the current device; the status of the device is predicted through the built-in prediction algorithm.

[0012] Preferably, the actual physical model of the device is corrected by bench test data and mooring test data before being built into the edge test terminal to ensure that the model is consistent with the normal operating status of the actual device under various working conditions and is regarded as a healthy model of the device. The healthy model of the device is in the data layer and is obtained through data exchange training between the acquisition layer and the model layer.

[0013] Preferably, the interaction layer includes an industrial computer of an electric power monitoring station, which has a built-in electric power system model, which is an actual physical model of the electric power system. The input of the electric power system model is the real-time status of the equipment calculated and output by the edge test terminal, and the output is the current real-time status of the overall electric power system after integrating the real-time status of each device in the current electric power system.

[0014] Preferably, the edge measurement terminal and the industrial computer of the power monitoring station realize data interaction between different models by managing and calling the application of different mechanism models and algorithm models and managing interfaces between models.

[0015] The beneficial effects of the present invention are as follows: the present invention is based on a data-driven ship power system situation awareness platform architecture. Through the added sensor equipment, the information of the ship power grid and extremely critical equipment that can be perceived is more comprehensive and higher in accuracy. It can also perform state estimation, evaluate the ship power grid energy reserves, optimal energy allocation strategy, equipment health status, fault prediction and other information, and give auxiliary decision-making suggestions, which can effectively reduce the number of operators, reduce the workload of operators, and further reduce the requirements for the operating experience of operators. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is an architecture diagram of the integrated power system situation awareness system based on the digital twin platform of the present invention;

[0017] Figure 2 This is a hardware architecture diagram of the comprehensive power situation awareness system of the present invention. DETAILED DESCRIPTION

[0018] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0019] The data types, measurement point scale and deployment methods of conventional ship integrated power system monitoring equipment and typical electrical equipment monitoring are sorted out, and combined with the system status assessment and prediction needs, the overall architecture design of the integrated power system situation awareness based on data drive is carried out. Through supplementing data collection and edge computing equipment, system modeling, data mining, analysis and other technologies, potential typical data feature analysis is carried out, and the construction of the integrated power system situation prediction model based on data drive is completed. Its dataization and intelligence can greatly reduce the complexity of the operator's operation and reduce the number of man-made failures. As the power system becomes more and more complex, it also improves the operability of the system, reduces the number of operators and the length of training; and provides data support for subsequent energy management, system abnormality identification and warning, and efficient positioning analysis and other applications.

[0020] like Figure 1 The following is a diagram of the architecture of the ship power system situation awareness platform based on data drive. The situation awareness platform architecture includes five layers from bottom to top: data perception layer, data acquisition layer, data layer for data processing and storage, model layer for data mining, and interaction layer for analytical applications. The situation awareness of the integrated power system is mainly aimed at two parts: situation awareness of the integrated power system power grid and situation awareness of important power generation and distribution equipment in the power grid. The system functions mainly include three parts: important equipment status assessment and prediction, power quality perception, and power quality assessment and prediction.

[0021] Data perception layer: New sensors are used to sense the thermal performance of diesel generator sets, instantaneous speed, power station opening and closing status, current, power, insulation and other related signal data. These signal data are the current power system grid status data and the working status data of power generation and distribution equipment.

[0022] Data collection layer: The collection layer uses data transmission protocols such as Modbus RTU, CAN OPEN, Modbus TCP, and OPC to realize data collection and protocol analysis. The real-time data of the data perception layer is collected through the existing local collection equipment, and the real-time data of each device is analyzed locally in combination with the device model to obtain the real-time status and status prediction of the device, which are uniformly uploaded to the integrated network for the interaction layer to call.

[0023] Data layer: including equipment configuration management data, design data, bench test data, real-time ship operation data, etc. Receive the real-time status data of the equipment uploaded by the data acquisition layer, store the inherent design data of the power system and power generation and distribution equipment, call the corresponding model of the model layer, train the model through data, obtain the applicable model of the equipment to the equipment configuration management, and call the data acquisition layer and the interaction layer.

[0024] Model layer: The model layer mainly includes equipment 3D models and mechanism models. Equipment 3D models include equipment models such as diesel engines / generators in diesel generator sets and switches / loads in power stations, which are mainly used for page display. Mechanism models include power grid topology models, data association models, reasoning models, and trend modeling. Various models are combined with data fusion and multiple algorithms to achieve load forecasting, power consumption analysis, fault location, load feature analysis, power quality assessment, equipment detection, short-term load data forecasting, distribution operation reliability analysis, power system transient stability assessment, fault diagnosis and trend forecasting, etc.

[0025] Interaction layer: The interaction layer is the front-end display page and human-computer interaction layer, which can be used to view the health, trend and analysis of each device and component on the power system status monitoring and predictive maintenance power system server, equipment sub-health warning, equipment failure and auxiliary decision-making suggestions, equipment maintenance and repair suggestions, etc. Integrate a large number of real-time data, situation data, situation prediction data and fault data of multi-source heterogeneous power generation equipment, power grid and external power equipment, through the digital twin platform system's rich statistical analysis, multi-dimensional analysis, mining algorithm library, fault situation perception model and data mining tools, and use the platform interface to achieve multi-dimensional correlation results of all links of the unmanned power system, infer the root cause of power grid failure, track fault trajectory, push fault warnings, provide fault analysis reports, establish a fault data archive system, and provide other functions to assist decision-making suggestions, equipment maintenance and repair suggestions, etc.

[0026] like Figure 2As shown in the hardware architecture diagram of the comprehensive electric power situation awareness system, the data layer, model layer, and interaction layer can borrow the existing industrial control computer of the power monitoring station and the ship digital platform. The ship digital platform performs data processing and communicates with the industrial control computer of the power monitoring station in real time. The perception layer and the acquisition layer include edge measurement terminals, machine-side control boxes, power station controllers, and various types of special sensors. The industrial control of the power monitoring station, the energy management system, and the edge measurement terminal are connected to the ship digital platform through Ethernet. Special sensors collect power parameters of the bus and branches in the power system unit equipment and the distribution board, and are connected to the edge measurement terminal.

[0027] The edge measurement terminal is an important part of the platform architecture. It has three functions: first, it collects special sensor data through IO hardware; second, it collects equipment monitoring data from the control box next to the machine and the power station controller through RS485 communication; third, it has built-in data processing, data feature extraction, and data analysis programs to process, extract data features, and analyze the special sensor data collected by IO hardware and the equipment monitoring data collected by RS485 communication, and the new data is transmitted to the industrial computer of the power monitoring station via Ethernet communication.

[0028] The edge measurement terminal is set up with an edge terminal storage device-level perception model, which is the actual physical model of each device, established according to the thermodynamic or electrotechnical characteristics of the device. These models are established in Python, C and other languages. Before the actual physical model is built into the edge test terminal, it is corrected by the bench test data and mooring test data to ensure that the model is the same as the normal operation status of the actual device under various working conditions, and is regarded as the health model of the device (this is in the data layer, obtained by exchanging data training between the acquisition layer and the model layer). When the model is used, the input of the model is the special sensor data collected by the edge test terminal IO hardware and the device monitoring data collected by RS485 communication, and the output is the monitoring parameters of the device when it is running healthily. The monitoring data actually collected by the edge test terminal is compared with the output data of the model, and the difference between the two groups of data is evaluated through the health assessment algorithm and the fault diagnosis algorithm to obtain the health level of the device, that is, the real-time status of the current device; the status of the device can also be predicted through the built-in prediction algorithm.

[0029] The power monitoring station industrial computer has a built-in power system model, which is an actual physical model of the power system. The model is modeled by CLOUDPSS simulation software. The input of the power system model is the real-time status of the equipment calculated and output by the edge test terminal, and the output is the current real-time status of the overall power system after integrating the real-time status of each device in the current power system. According to the degree of deterioration of the equipment status in the system, auxiliary decisions for power system operation are given, and operators are warned to exit equipment in poor health in advance to avoid large-scale power outages in the system. Through the built-in multiple intelligent algorithms, different inter-device topological structures formed by different protective switch actions after the device exits are given to guide operators to make optimal operation decisions.

[0030] The industrial computer of the power monitoring station has built-in storage card and computing power card, which can store data uploaded by edge test terminals, machine-side control boxes, power station controllers, etc. for 180 natural days; the computing power card ensures that the industrial computer can operate normally during model calling and three-dimensional display.

[0031] The operating environment of the edge measurement terminal and the industrial computer of the power monitoring station includes the operating system, language, and model algorithm dependency framework. The operating system uses Kylin as the system foundation. The software protocol stipulates that the algorithm service provides services of the http protocol for business service calls, and requests and responses are transmitted in the json data format. Docking process: Business service calls algorithm service: The business service calls the algorithm service, and the algorithm service returns the result for the business service to use. Coordination between different algorithm services: The algorithm service provides a single algorithm capability, and the business service calls the algorithm service and calls other algorithm services based on the results returned by the algorithm service.

[0032] The edge measurement terminal and the power monitoring station industrial computer realize data interaction between different models by managing and calling the application of different mechanism models and algorithm models and managing the interfaces between models. The main functions of the model and advanced application layer are to store device digital models and algorithm models, and provide the operating environment and data interface for the model to ensure the real-time operation of the model and data interaction.

[0033] The data processing and actual process of the edge test terminal in the perception platform is divided into five parts: data acquisition, data processing, data feature extraction, data analysis and data display. There are two main ways of data acquisition. One is to convert the IO signal generated by the sensor into data information through the acquisition module. The conventional power system acquisition types are mainly temperature, pressure, flow, current and voltage signals. This perception system also includes vibration, lubricating oil quality, and video signals collected from the intelligent edge terminal to assist in the health evaluation and fault diagnosis of the equipment; the other is communication data, including CAN, RS485, Ethernet and other signals from the unit, power distribution and other systems.

[0034] Data processing includes two processes: data screening and data filtering. Due to the accuracy of sensor acquisition and other reasons, the collected data has a large repeatability, so it is necessary to screen the data to ensure that the data input into the model can fully reflect the operating status of the diesel generator set, improve the accuracy of model evaluation, reduce the model scale, and speed up the model operation; after screening, the original data still contains a lot of high-frequency disturbance signals, which seriously affects the accuracy of the state evaluation model, so it is necessary to filter the data. According to the data sample form, after algorithm comparison, an efficient algorithm is used for smoothing filtering to improve the accuracy of the data and reduce sample noise without changing the signal trend.

[0035] Data feature extraction is to explore the potential fault feature information hidden in the time series of equipment operating parameters. The mode feature value is calculated for a specific fault mode to find out the sensitive factors of this fault mode. For these hidden information, statistical indicators based on time domain signal analysis are needed as fault feature information. In power station status monitoring and fault diagnosis, dimensional feature values ​​can reflect the fluctuations of the abnormality of the periodic stationary signal. Commonly used dimensional feature values ​​include peak value, peak-to-peak value, mean value, absolute mean value, mean square value, root mean square value, standard deviation, variance, root mean square amplitude, etc. The following characteristic values ​​of the main analysis parameters are: mean value, maximum value, minimum value, fluctuation range, dispersion, linear slope, etc.; the interval of parameter characteristic values ​​needs to be divided according to the degree of fluctuation, change trend, and margin.

[0036] For data analysis algorithms, there are Bayesian algorithms and SVM algorithms based on the edge measurement terminals of different devices. Bayesian classification algorithm is a statistical classification method that can be applied to large databases. It is simple, has high classification accuracy, and is fast. SVM algorithm, namely support vector machines, is a binary classification model that is suitable for small and medium-sized data samples, nonlinear, and high-dimensional classification problems.

[0037] The database used for data storage of the industrial computer of the power monitoring station is determined by the data management layer according to the different data requirements of the business. It mainly involves the use of relational databases, message queues and time series databases. The relational database is used to store data with relational models such as equipment attributes and alarm information; the message queue is used to cache real-time data to meet the business's demand for real-time data; the time series data is used to store the sensor data uploaded by the device in real time to improve the efficiency of the business's use of data. The relational database uses the domestic database Dameng; the message queue uses the domestic RcoketMQ; the time series database is planned to use the domestic Tdengine or CeresDB.

[0038] The data display of the perception platform mainly includes two ways: WEB end and large screen. The WEB end is mainly used to realize the management and control of all businesses in the digital platform; the large screen is mainly used to display the final display status of the business, so the two use different technical frameworks. The expected display effect is shown in the figure below. The display end mainly shows the device used to display the front-end interface. Since the platform adopts B / S architecture, the platform can be displayed differently on devices such as PCs and mobile terminals.

[0039] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

Claims

1. A data-driven ship power system situation awareness platform architecture, characterized in that: From bottom to top, it includes the data perception layer, data collection layer, data layer for data processing and storage, model layer for data mining, and interaction layer for analytical applications. The data perception layer: uses a new type of sensor to perceive signal data, the signal data is the current power system grid state data and the power generation and distribution equipment working state data; The data collection layer collects the real-time data of the data perception layer through the existing local collection equipment, and analyzes the real-time data of each device in combination with the device model locally to obtain the real-time status and status prediction of the device, and uploads them to the integrated network for the interaction layer to call; The data layer: receives the real-time status data of the equipment uploaded by the data collection layer, stores the inherent design data of the power system and the power generation and distribution equipment, calls the corresponding model of the model layer, trains the model through the data, obtains the applicable model of the equipment to the equipment configuration management, and calls the data collection layer and the interaction layer; The model layer includes a three-dimensional equipment model and a mechanism model; The interaction layer: the front-end display page and the human-computer interaction layer integrate a large amount of real-time data, situation data, situation prediction data and fault data of multi-source heterogeneous power generation equipment, power grids and external power-consuming equipment, to achieve multi-dimensional correlation results of all aspects of the unmanned power system, infer the root cause of power grid faults, track fault trajectories, push fault warnings, provide fault analysis reports, establish a fault data archive system, provide auxiliary decision-making suggestions, and equipment maintenance and repair suggestions.

2. According to the data-driven ship power system situation awareness platform architecture of claim 1, it is characterized in that: The data acquisition layer includes an edge measurement terminal, which collects special sensor data through IO hardware; the edge measurement terminal collects equipment monitoring data of the machine-side control box and the power station controller through RS485 communication; the edge measurement terminal has built-in data processing, data feature extraction, and data analysis programs, which process, extract data features, and analyze the special sensor data collected by IO hardware and the equipment monitoring data collected by RS485 communication, and transmit the obtained equipment status data to the integrated network via Ethernet communication.

3. According to the data-driven ship power system situation awareness platform architecture of claim 2, it is characterized in that: The edge measurement terminal is provided with an edge terminal storage device-level perception model, which is an actual physical model of each device, established according to the thermodynamic or electrotechnical characteristics of the device. When the model is used, the input is the special sensor data collected by the edge test terminal IO hardware and the device monitoring data collected by RS485 communication, and the output is the monitoring parameters of the device when it is running healthily; The monitoring data actually collected by the edge test terminal is compared with the output data of the model. The difference between the two sets of data is evaluated through the health assessment algorithm and the fault diagnosis algorithm to obtain the health level of the device, which is the real-time status of the current device; the status of the device is predicted through the built-in prediction algorithm.

4. The data-driven ship power system situation awareness platform architecture according to claim 3 is characterized in that: The actual physical model of the device is corrected by bench test data and mooring test data before being built into the edge test terminal to ensure that the model is consistent with the normal operating status of the actual device under various working conditions. It is regarded as a healthy model of the device. The healthy model of the device is in the data layer and is obtained through data exchange training between the acquisition layer and the model layer.

5. The data-driven ship power system situation awareness platform architecture according to claim 4 is characterized in that: The interaction layer includes an industrial computer for power monitoring, which has a built-in power system model, which is an actual physical model of the power system. The input of the power system model is the real-time status of the equipment calculated and output by the edge test terminal, and the output is the real-time status of the current overall power system after integrating the real-time status of each device in the current power system.

6. The data-driven ship power system situation awareness platform architecture according to claim 5 is characterized in that: The edge measurement terminal and the power monitoring station industrial computer realize data interaction between different models by managing and calling the application of different mechanism models and algorithm models and managing the interfaces between models.

Images