News as of October 8, 2021: Recently, news about power outages and electricity restrictions has emerged in an endless stream. High electricity load and energy conservation and emission reduction have both led to an energy shortage. Faced with such challenges, when formulating power restriction policies, it is naturally necessary to make an accurate assessment of energy consumption. This is one of the advantages of smart meters: they liberate human resources and realize remote meter reading. However, there are differences in the ways to realize smart meters. Some tend to directly switch to IoT meters, while others try to introduce AI technology.Replacing with IoT MetersIn the eyes of many, replacing with more intelligent IoT meters is the key to building an intelligent meter reading network. At present, more and more cities have begun to build the Internet of Things in the field of public services. For example, they use NB-IoT and cloud service providers to create intelligent meter reading businesses. Water, gas, and power grid enterprises use smart meters with "cloud + pipe + terminal" to transmit usage information, which not only facilitates users to query in real-time but also provides better management and operation means for resource supply enterprises. The traditional meter reading method in the past required manual meter reading, which was extremely inefficient. After integrating with the NB-IoT network and cloud platform, it can not only achieve low power consumption and massive connections but also use big data for refined operations, such as fault investigation and monitoring, and regional pressure regulation, so as to avoid the shortage of electricity and water. In addition to NB-IoT, it is also quite common to make smart meters based on protocols such as WI-SUN. Take Silicon Labs' EFR32FG12 as an example. This wireless SoC uses an Arm Cortex M4 core, which not only supports common protocols such as WI-SUN and M-BUS but also supports proprietary protocols. In addition to the essential low-power feature, the 1024 kB FLASH and 256 kB RAM of EFR32FG12 also ensure support for subsequent OTA meter upgrades. For applications such as intelligent meter reading, the number of connected devices is quite large. Therefore, the traditional IPv4 pool is insufficient to support its large-scale expansion. However, WI-SUN can provide a large number of address resources by supporting IPv6.AI "Modification" for Meter ReadingAlthough the method of connecting IoT meters is relatively simple to deploy, it puts forward high requirements for equipment transformation, especially for meters such as water meters. Moreover, the current IoT meters are not cheap. To convert the past mechanical meters into electronic ones, problems such as power supply and wireless transmission must be considered, especially the automatic meter reading realized by PLC technology. With the development of deep learning, cloud computing, and chips, artificial intelligence has become a hot technology pursued by various semiconductor industries. However, since many AI applications need to use high-performance chips, the utilization rate of artificial intelligence in some edge or local deployment is not high. Now many manufacturers have begun to study how to integrate AI into low-power local terminals and introduce artificial intelligence technology into embedded devices or devices with low processing capacity such as MCU. By adding or modifying modules, using image sensors to take pictures and convert them into numbers, and then sending data through wireless transmission methods such as NB-IoT, there is no need to replace existing equipment. These visual recognition and image recognition technologies are AI algorithms deployed on the local terminal, which turn traditional mechanical meters into smart meters. This modification method greatly reduces costs and improves efficiency. Although the chips used in this scheme have low computing power, they must support simple visual algorithms. Take Jianan Technology's K210 as an example. The chip uses a 64-bit dual-core RISC-V CPU, with a computing power of 1TOPS but a power consumption of only 300mW. The chip is equipped with Jianan's self-developed core neural network accelerator KPU, which facilitates processing machine vision tasks on the local end. Jianan Technology said that the design of this "non-invasive" transformation costs only a quarter of that of traditional smart meters. However, most manufacturers do not have the ability to develop artificial intelligence algorithms. In addition to directly adopting the recognition algorithm schemes of chip manufacturers, they often directly use the schemes of other AI manufacturers. Take Baidu's PaddleX development tool as an example. PaddleX first uses a target detection model to detect the marks in the image, then uses a semantic segmentation model to segment the pointers and scales of the meter to obtain the approximate digital image of the meter. Then, the image is corroded, the annular dial is unfolded into a rectangular image, which is then converted into a one-dimensional array, and the mean value of the scale is binarized. The actual reading of the mark is calculated according to the relative position of the pointer and the predicted range.Intelligent Meter Reading: Let Data "Run Errands"Carrying a ladder in one hand, a meter reading book in the other, and a tool bag on his shoulder, leaving home at five or six in the morning and returning at eight or nine in the evening—this was Lao Li's original daily work. Lao Li is a distribution operator in a power supply station in Bijie. For many years, he had to take tools to read meters door-to-door on every meter reading day. Recalling the details of his past work, he said: "At that time, I had to walk dozens of kilometers a day and read two or three hundred meters. It was common to be exposed to wind and sun, chased by dogs, and bitten by insects." But in 2017, Guizhou Power Grid launched the construction of a "measurement automation system", and all this changed. "Since we have smart electricity meters, I can realize 'second reading' with just a click of my finger and finish the work while blowing the air conditioner," Lao Li's eyes lit up again. With digital empowerment, now he and his colleagues can receive all the electricity meter data in the jurisdiction without going out. When talking about the convenience brought by technology, Lao Li was quite emotional: "The measurement system has 'liberated' our hands and feet. Now, I can get the electricity data of each household in a few seconds by opening the computer and clicking the mouse." In 2018, supported by the measurement automation system, Guizhou Power Grid launched the "double coverage" battle of "smart electricity meters and low-voltage centralized reading". It was also in that year that Longshine Technology Group (300682.SZ) formally cooperated with Guizhou Power Grid to develop and build this "smart brain" of the power grid for Guizhou Province. With the support of Longshine Technology's "energy digitalization" capability, the new "copy, check, and collect" mode has the functions of "centralized automatic meter reading + centralized automatic accounting + centralized automatic issuance", eliminating the disadvantages of time-consuming, laborious, and inaccurate manual meter reading, and making electricity fee reading and checking management more lean, intensive, and intelligent. At the same time, the measurement automation system can also remotely analyze electricity meter data to help Guizhou Power Grid master customers' electricity consumption information. Once an abnormality is found, it can carry out timely maintenance to fully ensure the electricity consumption of residents. At present, the system supports more than 560,000 measurement terminals and about 16.9 million users in Guizhou, with an average automatic meter reading rate of 99.5%, fulfilling Guizhou Power Grid's service commitment to let residents "use electricity" to "use electricity well".Digitalization Makes Power Technicians All "Sharpshooters"The Guizhou Power Grid measurement automation system developed by Longshine Technology is an automation system mainly focusing on the collection and application of electrical energy data. In addition to the collection and analysis of smart electricity meter data, it also has functions such as measurement device monitoring, measurement operation and maintenance management, and centralized management of electrical energy data. It is an "all-rounder" that fully supports business and serves the grass-roots level. "Three-phase load imbalance will lead to increased line loss, and in serious cases, it will cause overload and open-phase faults." Through the online monitoring platform of electrical energy data, Song Jingjing, deputy director of the urban power supply station of Tongren Yuping Power Supply Bureau in Guizhou, checked the three-phase load of Hujiagou台区 on the 10kV Xinqi Line and found that the load rate of phase B was 15%, slightly lower than that of phase A and phase C. She immediately organized on-site staff to carry out work and installed the newly added 5 electricity meters on phase B to keep the three-phase load in a balanced state. Song Jingjing said that with the automation system and online monitoring platform, it is like having "piercing eyes", which can quickly find and solve large and small faults, effectively improving the power supply service capacity and line loss control quality. The measurement automation system helps to realize lean management of line loss. Through the analysis and management of line loss abnormalities by region, voltage level, line, and台区, it promotes Guizhou Power Grid to continuously reduce losses. By the end of 2020, the comprehensive line loss rate of Guizhou Power Grid had dropped to 4.79%, gradually approaching that of advanced developed regions.ConclusionThese two intelligent meter reading methods have their own advantages, and it is more appropriate to determine the scheme after investigating different scenarios. For example, in some cases where there is a lack of external hanging environment and the meters are located in public areas, it is more suitable to replace them with new meters, such as most electrical meters now. In some cases that require low-cost and high-efficiency installation or are difficult to modify and enter the household, such as indoor water meters, the AI addition scheme is more suitable. In addition to the meters themselves, suppliers such as water utilities and power grids often cooperate with cloud platforms to manage and store meter reading data. Cloud platforms will also work together to create intelligent applications based on meter reading data, such as developing Apps and supporting online payment and query. For these suppliers who have invested a lot of money in modification, whether these data will have more value-added space in the future remains to be further explored in the business model, but intelligent meter reading is definitely a necessary step. Article source: Electronic Enthusiast, Longshine Technology