Analysis of the Development of Big Data in Domestic Intelligent Transportation

With the rapid development of the city, traffic jams, traffic pollution have become increasingly serious, and traffic accidents have occurred frequently. These are problems that major cities urgently need to solve. Intelligent transportation is the key to improving urban transport. China's big data industry started late and developed rapidly. With the rapid development of the Internet of Things and the mobile Internet, the speed of data generation and the scale increase have increased. It is urgently necessary to apply big data methods to analyze and process and extract effective information.

Big Data Industry Development Scale Analysis

China's big data industry started late and developed rapidly. With the rapid development of the Internet of Things and the mobile Internet, the speed of data generation and the scale increase have increased. It is urgently necessary to apply big data methods to analyze and process and extract effective information. In 2014, the scale of China's big data market reached 76.7 billion yuan, a year-on-year increase of 27.8%. It is estimated that by 2020, the scale of China's big data industry will reach 822.881 billion yuan. The CAGR of 2015-2017 is 51.5%. In 2014, the scale of China's big data application market was 8.054 billion yuan, a year-on-year increase of 3.2%. In 2015, the market scale increased by approximately 37.3% to 11.056 billion yuan. It is estimated that by 2020, China's big data application market will grow to 501.958 billion yuan. . The compound growth rate for 2015-2017 was 87.8%.

Big Data Applied to Intelligent Transportation Industry

When smart traffic encounters big data, just as manganese dioxide acts as a catalyst in oxygen production experiments, a dramatic chemical reaction aggravates the common development of both parties.

With the rapid development of the city, traffic jams, traffic pollution have become increasingly serious, and traffic accidents have occurred frequently. These are problems that major cities urgently need to solve. Intelligent transportation is the key to improving urban transport. For this reason, timely and accurate acquisition of traffic data and construction of a traffic data processing model are prerequisites for building intelligent traffic, and this problem can be solved by big data technology.

1. Intelligent Transportation Demand and Big Data Integration

The overall framework of intelligent transportation mainly includes physical sensing layer, software application platform, and application of analysis and prediction and optimization management. The physical sensing layer is mainly for the perception of traffic conditions and traffic data collection; the software application platform integrates and transforms the information of each sensing terminal to support the analysis, early warning, and optimization management of application systems; analyzes and predicts and optimizes management applications. It mainly includes application systems such as traffic planning, traffic monitoring, intelligent guidance, and intelligent parking.

The system utilizes advanced video surveillance, intelligent identification and information technology to increase manageable space, time and scope, and continuously improve management breadth, depth and detail. The entire system consists of an information integrated application platform, a signal control system, a video surveillance system, a smart bayonet system, an electronic police system, an information acquisition system, and an information release system. In order to achieve the four goals: to improve traffic capacity, reduce traffic accidents, combat violations, travel information services.

The core of the entire system construction is the collection, storage, and calculation of data, and the most important core idea is "data is value." The question is how to convert data into value. This becomes a technical issue.

From a statistical point of view, any dynamic development in any field, as long as there are enough sample data, will be able to find the dynamic development rules from the sample data. The more data, the higher the accuracy. This "law" is the value of the data. For commercial organizations, users can analyze the laws of user behavior to increase sales; analyze the laws of target markets, and place advertisements to reduce costs; for the public security industry, regional crime trends can be analyzed, and crimes can be prevented and reduced in advance; traffic can also be analyzed. Behavioural laws, advance traffic diversion, and improve the patency rate of traffic, which can truly tap the potential value of the data and improve its social value.

Since the development of the network in the early 20th century, it has entered a highly connected stage. At the same time as networking, data is highly concentrated and the amount of data increases dramatically. According to the IDC report, the data on the Internet now doubles every two years. This growth rate is also effective in the intelligent transportation industry. With the increase in the number of bayonet, electric police, and cameras, and the development of high-definition and intelligent, if we count the various sensors of the Internet of Things, the amount of data in the next few years may increase. It is much higher than this growth rate. This provides the data foundation for the intelligent transportation industry to realize big data.

The ability to quickly obtain valuable information from various types of data is big data technology. From this we look at the 4 Vs (Volume Volume, Variety Variety, Value, Velocity) that IBM sums up: (1) Volume data volume is huge. Jump from TB level to PB level; (2) Variety data types. Including video, pictures, geographic information, sensor data, etc.; (3) Value has low value density and high application value. Take video as an example. In the continuous and continuous monitoring process, the useful data may only have one or two seconds. (4) Velocity processing speed is fast, and the law of 1 second is used.

This last point is also fundamentally different from traditional data mining techniques. In the field of transportation, massive data mainly includes four types of data: sensor data (position, temperature, pressure, image, speed, RFID, etc.); system data (logs, equipment records, MIBs, etc.); service data (charge information , Internet services and other information); application data (generating information on manufacturers, energy, transportation, performance, compatibility, etc.). The types of traffic data are numerous and huge. The volume Volume and variety Variety is due to the complexity of data types and the dramatic increase in data volume, which determines that the application mode of the original simple cause and effect relationship is extremely low in data usage and can not fully play the role of data; Velocity is a process, huge The amount of computation determines that the speed must be fast; the value of Value is the final result.

2. Big data collection

In the process of building smart transportation in each city, more and more data such as video surveillance, bayonet traffic police, road condition information, management and control information, operation information, GPS positioning information, and RFID identification information will be generated. The amount of data generated each day can be Reached the PB level and showed exponential growth.

3, big data value-added applications

In-depth mining of data values, the introduction of big data models such as vehicle trajectories, road traffic, case clustering, etc. in the fields of intelligent transportation and public security operations. Based on the big data model, data value-added applications such as smart decking, smart follow-up analysis, trajectory collision, face comparison, and public opinion analysis are introduced, and the deep-seated problems of the industry are gradually solved.

4, massive data calculation

Through cloud computing clusters, distributed high-speed computing of massive data is realized, and efficient analysis and mining of massive data are supported. The cloud computing cluster is a distributed computing system with an M/S architecture. As the scheduling management server, the master is responsible for the task decomposition and scheduling, and the unified management of computing resources. Slave is composed of a large number of computing servers and is responsible for completing the tasks assigned by the Master.

5, massive data retrieval

Based on the characteristics of industry data query, the search engine is optimized and customized to support the second-level high-speed query of billions of records. Through the clustering mechanism, high reliability, high fault tolerance, and high scalability of search services are realized.

6, massive data storage

For massive data storage, HBase distributed storage systems are used. Compared with the traditional relational database, there are four characteristics: data format flexibility, high availability, horizontal scalability and access efficiency.

At the same time, it can seamlessly integrate and quickly import existing historical data from a relational database. Provides high reliability, high fault-tolerance, high-performance massive data storage solutions, and supports seamless capacity expansion.

7, Big Data Analysis and Application

Efficient cloud computing capabilities, which can bring in second-second retrieval capabilities of hundreds of billions of data, provide rapid protection for big data analytics applications. Intelligent analysis algorithms based on deep learning provide powerful tools for big data analysis applications. The analysis of traffic big data brings more effective support for traffic management, decision-making, planning, service, and active safety precautions.

Using big data technology, combined with high-definition surveillance video, bayonet data, coil micro-acquisition wave data, etc., supplemented by intelligent research and judgment, can basically realize intersection self-adaption and signal timing optimization. Through the analysis of big data, the comprehensive traffic capacity of multiple intersections in the area is obtained, which is used to optimize the time-sharing of traffic lights at multiple intersections in the area, and to improve the traffic efficiency in a single intersection or area. For example, according to weekdays/holidays, morning and evening peaks/other hours, key road junctions/sub-critical junctions/normal junctions, day/night, etc., different timings can be automatically set manually or by the system to greatly increase traffic in the area. ability.

The big data analysis and judgment function can also support the second recognition of the bayonet data and video surveillance data, improve the accuracy of vehicle information, and then use big data to achieve trajectory analysis, foothold analysis, concealed vehicle analysis and other functions. Carry out in-depth digging of vehicle big data to realize the needs of different scenes for comprehensive prior monitoring, timely tracking, and accurate backtracking afterwards. The vehicle big data platform built in Changzhou assisted relevant departments to automatically discover more than 10 deck vehicles each day. Then, according to the vehicle's trajectory analysis and end-of-sale analysis, it quickly found decked vehicles for punishment management.

Combining intelligent algorithms, secondary recognition and other functions, can more accurately identify the characteristics of the license plate, body color, vehicle model, vehicle standard, and annual models, as well as the sun visor detection, seat belt detection, call detection, driver face recognition, etc. Conduct the analysis.