On December 23, 2020, China’s steam research successfully held the “2020 3rd New Energy Automobile Test Evaluation Technology International Forum”. China Steam Research will continue to push a wonderful speech record, this article, this paper, this paper, this article, “Data-based electric vehicle charging system security assessment” brought by the Minister of Data Minister Pan Bo Mo.
1 Safety protection system and transformation of ideas
Special electricity is mainly due to charging safety, mainly in data and system boundaries, when charging is less, the charging safety boundary characteristics are not obvious, with the increase of charging, such as 1 million times, the security boundaries of the system will gradually appear When charging reaches 10 million times, based on data to obtain boundary and characteristics, the expert library is established. When charging reaches 100 million times, the industrial large data can establish charging large data, based on charging large data, real-time Protection, detect the safety status of the vehicle in advance, and inform the user to repair to ensure the safety of the vehicle.
Two-layer security protection technology system has been established on charging network. The picture below shows the vehicle charging process, which is equivalent to a running charge safety protection process, and the entire charging network is approximately 400,000 orders. Each time the vehicle is charged, the CMS device protection model and the large data protection model will perform real-time detection on the information flow and energy streams between BMS, charging piles and vehicles. When the abnormality is detected, the order will be terminated. Enter the abnormal order library.
Two-layer safety protection architecture, first based on charging net real-time data, such as BMS data, charging pile data, user data, environmental data, and combined with charging network history data, including vehicle archives (safety score, battery health score, use Customized score) and model file (safety score, battery health score, usage habits) Establish two-layer safety protection structure: establish CMS active protective layer on the device side, including monomer overvoltage protection model, battery oversight protection model, temperature difference Big protective model, temperature rise exception protection model, etc .; establish large data protection layers on large data sides, including power battery imbalance model, vehicle battery life prediction model, charging behavior model and battery safety score model. Safety decisions will be made for vehicles after two floors, such as blocking charging, alarm notification, monitoring assessment. After the safety decision is complete, the operation system continues to track the order and feeds the entire data to the model, timely corrects the model, and guarantee the accuracy of the model.
Charging safety full lifecycle management process is: When charging, the two-story protection model is protected. When there is an abnormal order, on the one hand, through the APP alarm, or notify C-terminal users, B-terminal users, special electricity operations, main machine, The battery factory carries the maintenance of the vehicle, the maintenance feedback results are also fed back to the protective model; on the other hand, the abnormal order will be updated to the vehicle, the model file, and the high-risk vehicle is evaluated, and the high-risk vehicle is identified, and the work order is used. The start of the hand continues to track the vehicle, send a safety protection report to the client, the maintenance is complete, the work order is closed, and the work order information will also feed back to the protection model, which is built into a forward charging safety protection and benign loop mechanism.
The charging safety protection ideas are: Identify high-risk vehicles through the safety protection model, inform the vehicle maintenance. Protection is summarized as both sides of the biaxial axis, the timeline shaft, the device side platform side, the time axis utilizes the power of the large data cloud, pay attention to the trend; the spatial shaft is different; the device side receives charging parameters, real-time protection; platform side Upload charging curve, long-term protection. Classification of protection, such as from passenger cars, commercial passenger cars, commercial trucks, different charging rates, lithium iron phosphate, three yuan, lithium manganate, lithium titanate, old car and new cars, different regions these dimensions.
2 Safety protection framework, model and data
The overall architecture of safety protection is shown below. The theoretical basis is divided into large-scale electric vehicle safety charging key technologies and equipment research, intelligent charging network and distributed energy integration technology research, cloud technology-based charging network interconnection data platform Research. The implementation of the system is divided into charging equipment, charging management, collaborative scheduling, platform service. Output based on charging security, there are industry standards for special electricity or participation.
Based on charging process data on charging security risk analysis, the scope of data includes longitudinal time dimensions, transverse to models extended data range; data changes include gradient changes of process data, process data deterioration; data is associated with data between data relationship.
Based on normal distribution of charging index abnormal detection method: in normal distribution, interval (μ-3σ, μ + 3σ) is the actual possible range of values of random variables X, and X falls outside the interval less than 3%. Third, in practice, this event is generally considered to happen, and if the variable belongs to this, it is an abnormal point. In the normal distribution of charging abnormal detection, “variable” can be the highest temperature, maximum temperature difference, temperature rise rate, SOC rate, monomer maximum pressure difference, etc., the sample data is from the same area information, the same vehicle model information, the same Charging record of time information.
Example: Variable of the big data analysis indicator, can be the highest temperature, maximum temperature difference, maximum temperature rise rate, battery pressure difference, maximum SOC rate and other indicators, sample data from the same area information, the same vehicle model information, the same time information ( A charging record per day or monthly). The normal distribution n (μ, σ2) is uniquely determined by mean μ and standard deviation σ. The sample data is estimated to estimate the mean U and the standard deviation σ, the number of samples is n, and the formula is as follows. 3 safety protection application
Safety evaluation, currently implemented the various battery index scores of each car under the platform, forming a vehicle scivile file, provides a basis for the qualitative and quantitative evaluation of the vehicle: the platform can be identified according to various indicators to identify high-risk vehicles; according to vehicle scores, customers can assist customers Vehicles are classified and managed, and the batch updates; the vehicle score can make users simply intuitively feel the vehicle security status.
Safety protection products, such as bus exclusive security report; charging firewall, customers can limit different vehicles in different ranges according to the dimensions of urban, charging stations, brands, models, license plate numbers, accounts, VIN, etc., restricting SOC values freely. Using the “Safety Firewall” function to limit / prohibit high-risk vehicles, limit / prohibit a vehicle charging, subsidiaries set up this city restriction strategy, restrict / prohibit all vehicles with a high-hazard station. Currently, requirements include unified five high-risk models and 3 brand all models of the country’s regular burning vehicles; 50 m power stations around the refueling gas station are charged.
The platform can be a pop-up window and lighting warning of the C-terminal user. Similarly, it can be displayed for C-terminal users, and the vehicle’s recent key indicators can be used in real time to view the vehicle security status analysis.