A Digital Twin Model Data Processing Method Applied to the Energy Storage Internet of Things

Abstract

The present invention discloses a digital twin model data processing method applied to the energy storage Internet of Things. A digital twin model based on WASM is constructed. By realizing the synchronization of the digital twin models of WASM between the cloud and the edge, the training of the digital twin model of WASM is completed at the edge, and the model is transmitted to the cloud at the same time. The data of the energy storage device is predicted through the WASM digital twin model, and only the points different from the data prediction are transmitted, so as to effectively compress the data transmitted in the energy storage Internet of Things. When data transmission errors and data loss occur, the digital twin model can reasonably fill in the error data and reduce data discard.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage Internet of Things, and particularly to a method for processing digital twin model data applied to the energy storage Internet of Things. Background Art

[0002] Industrial and commercial energy storage is a typical application of distributed energy storage systems on the user side. Its characteristics are that it is relatively close to both the distributed photovoltaic power source end and the load center. It can not only effectively improve the consumption rate of clean energy, but also effectively reduce the transmission loss of electric energy, contributing to the realization of the "dual carbon" goal. Currently, industrial and commercial energy storage mainly recovers the investment cost through the benefits obtained from peak-valley price arbitrage, capacity electricity charge reduction, demand response, etc. Due to the continuous rise in the prices of upstream raw materials for lithium batteries recently, the economy of industrial and commercial energy storage shows a downward trend.

[0003] In the Internet of Things application scenario of industrial and commercial energy storage equipment, it is not only necessary to collect and process energy storage battery-related signals such as the working voltage, working current, and working temperature of the battery in the energy storage equipment, but also to collect various data related to equipment operation such as the total power consumption of users, the power of each energy-consuming device, and the output power of the user's distributed photovoltaic power station. When all these data are collected in real time, the amount of data is extremely large. In the process of transmitting the data to the cloud, problems such as the network being unable to bear this huge amount of data will inevitably occur, and due to the complex environment of industrial Internet application scenarios and numerous interferences, data will inevitably be retransmitted or even lost during the transmission process.

[0004] In order to handle the scenario of uploading massive data from the Internet of Things to the cloud, a data loading and transformation (ETL) process is usually included in the Internet of Things data processing. This process is used to clean the data, process data alignment and data missing, and compress the data to reduce the required network transmission bandwidth. Methods for errors include methods for processing error data such as deleting entire columns and filling default values for missing values. For data compression, compression algorithms based on data rows and compression algorithms based on data columns can be used. However, these methods are not the optimal data processing strategies. In the current edge-cloud integrated architecture of the Internet of Things, such methods either lose a lot of useful data processing information at the edge end, or place too many data processing tasks on the cloud computing end and may increase the time delay of the data, reducing the value of real-time data sampling.

[0005] Meanwhile, in the industrial and commercial energy storage device Internet of Things application scenario, the charging and discharging of energy storage devices are key tasks for device operation. How to effectively achieve peak shaving and valley filling of electric energy, discharging during peak electricity consumption and charging during valley electricity consumption is an important indicator determining the profitability of industrial and commercial energy storage. The charging and discharging efficiency of the energy storage power station and its generated economic value are determined by a very large number of variables, including the attenuation of energy storage batteries, the temperature change of energy storage batteries during charging and discharging, the consistency between battery packs, the ambient temperature, and even the current electricity consumption behavior of users. The traditional method is to generate rule-based charging and discharging strategies based on manual analysis of user electricity consumption behavior and electricity price periods, which is obviously not intelligent and not the optimal solution. Summary of the Invention

[0006] To solve the above problems, the present invention provides a digital twin model data processing method applied to the energy storage Internet of Things. By using digital twin modeling technology, it realizes the system collaborative calculation and data processing between the edge-side energy storage cabinet and cloud big data, realizes the compression of data during the data transmission process in the entire energy storage Internet of Things scenario and the processing of data errors, and simultaneously realizes the optimization and learning of the intelligent recharging and discharging strategy of energy storage devices.

[0007] For this purpose, the technical solution of the present invention is: a digital twin model data processing method applied to the energy storage Internet of Things, including the following steps:

[0008] 1) Construct and train a digital twin model: Construct a digital twin model and run it on the cloud, and use the historical data of all devices stored on the cloud to train the digital twin model;

[0009] 2) Divide the trained digital twin model into an immutable part and a mutable part. The immutable part generates the WASM model part, and the mutable part combines the configuration information of a specific device on the cloud to generate the model data part, and forms multiple digital twin models corresponding to the edge-side energy storage devices;

[0010] 3) Package and send multiple digital twin models to the corresponding edge-side energy storage devices, and the devices run the corresponding digital twin models to predict the data of the system;

[0011] 4) When the prediction accuracy reaches the expected value, the digital twin model continuously performs data prediction on the edge-side energy storage device; meanwhile, the same digital twin model runs on the data bus ETL module on the cloud to generate prediction data;

[0012] 5) The edge-side energy storage device compresses the predicted correct data, and this part of the data is filled with the prediction data obtained by the same digital twin model on the cloud. Only the predicted incorrect data is transmitted between the edge-side energy storage device and the cloud.

[0013] Further, in step 4), when the prediction accuracy does not reach the expected value, model training is performed on the energy storage device at the edge side, and the model data part in the digital twin model is adjusted through online training to fine-tune and correct the model. After the training is completed, the trained digital twin model and the corresponding training data are synchronized to the cloud to realize the closed-loop of the digital twin model cloud-edge collaboration process.

[0014] Further, in step 5), when data loss occurs in the cloud, the predicted data of the digital twin model is used to directly fill the missing data part.

[0015] Further, the WASM model part is a program structure, which is a description of the behavior of the energy storage device, or a mathematical description form of the physical decay model of the energy storage battery, or a user energy consumption behavior prediction model, or a charging and discharging strategy of the device.

[0016] Further, the energy storage Internet of Things system is divided into an edge side and a cloud side. The edge-side energy storage device includes an EMS module, a PCS module, a BMS module, and a battery module. The cloud side includes a data bus ETL module and a policy model and data processing platform. Digital twin models are respectively deployed in the EMS module, the BMS module, the data bus ETL module, and the policy model and data processing platform.

[0017] Further, the trained digital twin model in step 1) can predict the working states of the battery module, the EMS module, the BMS module, and the PCS module in the energy storage Internet of Things system, ensuring that the accuracy rate of data prediction reaches the preset range.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0019] 1. A digital twin model based on WASM is constructed. By realizing the synchronization of the WASM digital twin models between the cloud and the edge side, the training of the WASM digital twin model is completed at the edge side, and the model is transmitted to the cloud at the same time. The data of the energy storage device is predicted by the WASM digital twin model, and only the points different from the data prediction are transmitted, so as to effectively compress the transmitted data in the energy storage Internet of Things. When data transmission errors and data loss occur, the digital twin model can reasonably fill the error data and reduce data discard.

[0020] 2. In the energy storage Internet of Things edge-cloud collaboration scenario, the digital twin technology is used to realize the efficient collaboration of effective data processing work between the edge side and the cloud computing side of the Internet of Things, and the problems of data loss and data compression occurring in the Internet of Things transmission process are effectively processed by using the global information and the digital twin model. At the same time, the digital twin model is used to realize the intelligence of the energy storage charging and discharging strategy, improving the efficiency and income of the energy storage. Description of the Drawings

[0021] The following will further elaborate in conjunction with the accompanying drawings and the embodiments of the present invention.

[0022] Figure 1 It is a system diagram of the energy storage Internet of Things of the present invention;

[0023] Figure 2 It is a cloud-edge collaboration flowchart of the present invention;

[0024] Figure 3 It is a data compression flowchart of the present invention; Specific embodiments

[0025] In this embodiment, in the energy storage Internet of Things cloud-edge collaboration scenario, digital twin technology is used to achieve efficient collaboration of effective data processing work between the Internet of Things edge side and the cloud computing side, and use global information and digital twin models to effectively handle data loss and data compression problems that occur during the Internet of Things transmission process. At the same time, the digital twin model is used to realize the intelligence of the energy storage charge and discharge strategy, improving the efficiency and benefits of energy storage.

[0026] As Figure 1 shown, in the energy storage cloud-edge system processing system, the energy storage system includes an EMS module, a PCS module, a BMS module, and a battery module, etc. EMS is an energy management system, mainly monitoring the energy usage of the user side and controlling the energy storage system to perform charge and discharge operations based on this data. BMS is a battery perfusion system, mainly monitoring data such as the voltage, current, and temperature of the battery to ensure the normal operation of the battery and prevent battery damage caused by overvoltage or overcurrent. At the same time, it is necessary to model the state of charge (SOC) of the battery to estimate the total energy of the current battery. PCS is an energy storage edge converter, mainly realizing the conversion of the direct current of the battery into alternating current to discharge the battery and converting the alternating current into direct current to charge the battery.

[0027] And in the energy storage Internet of Things cloud service background, it includes a data bus ETL module, a policy model, and a data processing platform. The data bus ETL module realizes data compression and decompression and the processing of error data. The policy model and the data processing platform realize the training and evaluation of the overall intelligent charge and discharge strategy model.

[0028] Add a digital twin module to the EMS module and BMS module of the edge - side energy storage device, and the data bus ETL module, policy model, and data processing platform in the cloud. Since the programming languages and operating systems used in the development of each module of EMS, BMS, and the cloud platform are different, in order for the digital twin model to run in different programming languages and different operating system environments, the WASM technology is used as a unified digital twin model representation method and extended to support the cloud scenarios of the energy storage Internet of Things. Various different programming languages can be uniformly compiled into the WASM format and can cover the deployment and execution of different environments at the edge - side, pipeline - side, and cloud - service sides.

[0029] In the digital twin model, it includes a WASM model part described by WASM and a model data part. Among them, WASM is a new portable, small - sized, fast - loading, and Web - compatible format. The WASM model part is a program. This program is a description of the behavior of the energy storage device, which can be a mathematical description form of the physical decay model of the energy storage battery, or a user energy - consumption behavior prediction model, or the charge - discharge strategy of the device. This WASM program will have some adjustable parameters, and these parameters are stored together with the WASM model as the model data part to form a digital twin model.

[0030] As Figure 2 shown, the cloud - edge collaboration process of the digital twin model includes the following steps:

[0031] 1) Run the digital twin model in the cloud, that is, use the device on the policy model and data processing platform. Train this digital twin model by using the historical data of all devices stored in the cloud so that it can predict the working states of the battery module, EMS module, BMS module, and PCS module in the energy storage system, ensuring that the accuracy rate of data prediction reaches the given requirements.

[0032] 2) Divide the trained model into a unified immutable part and a variable part that can be updated according to different devices. The immutable part generates the WASM model part in the digital twin model, and the variable part combines the configuration information of the specific device in the cloud to generate the model data part of the digital twin model. For example, both the EMS module and the BMS module are added to the digital twin model. The WASM model parts of the two digital twin models are the same, but the model data parts are different, and some adjustable parameters need to be changed according to the device configuration information.

[0033] 3) Package and send the digital twin model to the corresponding edge - side energy storage device. The device runs the corresponding digital twin model to predict the data of the system. If the prediction accuracy meets the requirements, then continuously perform data prediction. At the same time, the same model also runs on the data bus ETL module in the cloud to generate prediction data.

[0034] 4) Since the prediction is accurate, the edge-side energy storage device can compress this part of the correctly predicted data and only transmit a small part of the incorrectly predicted data, while the remaining correctly predicted data is filled with the predicted data of the same digital twin model in the cloud. In this way, the compression of the transmitted data is completed using the digital twin model, as Figure 3 shown.

[0035] 5) If data is missing in the cloud, the missing values can be directly filled using the predicted data of the digital twin model.

[0036] No matter how well the digital twin model is designed or trained, it is inevitable that its accuracy will decline. When the model prediction fails to meet the requirements, the model training process is started at the edge side, and the model data part of the digital twin model is adjusted through online training to fine-tune and correct the model. After the training is completed, the trained digital twin model and the corresponding training data are synchronized to the cloud to realize the closed-loop of the cloud-edge collaboration process of the digital twin model.

[0037] Taking the most important charge and discharge strategy scenario of the energy storage Internet of Things as an example:

[0038] Place the model prediction of the user's energy consumption load in the WASM model, place the intelligent charge and discharge strategy of the device based on the load prediction in the WASM model, and then place some configurations and data related to a single user in the model data part. For example, the user's peak evaluation electricity price can be placed in the model data, and the user's production scheduling data can also be placed in the data model, so that this digital twin model can output the user's charge and discharge strategy and the user's predicted income.

[0039] In this way, the same set of digital twin models runs at the edge side and in the cloud, and only a small amount of data transmission and synchronization cost is incurred. In this way, the user's load prediction results, the charge and discharge behavior data of the device, and the user's relevant income curve can be obtained. At the same time, using the training ability of the local digital twin model, the charge and discharge strategy model can be further optimized to optimize the user's income.

[0040] Thus, through the cloud-edge collaboration method based on the digital twin model in this embodiment, the compression of the transmitted data of the energy storage Internet of Things is realized, the correction of the error data is achieved, and at the same time, the intelligence of the charge and discharge strategy is realized.

[0041] The above is only the preferred implementation mode of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art in this technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for processing digital twin model data applied to the energy storage Internet of Things, characterized in that: Including the following steps: 1) Construct and train a digital twin model: Construct a digital twin model and run it in the cloud, and use the historical data of all devices stored in the cloud to train the digital twin model; 2) Divide the trained digital twin model into an immutable part and a mutable part. The immutable part generates a WASM model part described by WASM, and the mutable part combines the configuration information of the device in the cloud to generate a model data part, and forms multiple digital twin models corresponding to the edge-side energy storage devices; The energy storage Internet of Things system is divided into an edge side and a cloud side. The edge-side energy storage device includes an EMS module, a PCS module, a BMS module, and a battery module. The cloud side includes a data bus ETL module and a policy model and data processing platform; The digital twin models are respectively deployed in the EMS module, the BMS module, the data bus ETL module, and the policy model and data processing platform; 3) Package and send multiple digital twin models to the corresponding edge-side energy storage devices, and the devices run the corresponding digital twin models to predict the data of the system; 4) When the prediction accuracy reaches the expected value, the digital twin model continuously performs data prediction on the edge-side energy storage device; at the same time, the same digital twin model runs on the data bus ETL module in the cloud to generate prediction data; 5) The edge-side energy storage device compresses the correctly predicted data, and this part of the data is filled with the prediction data obtained by the same digital twin model in the cloud. Only the incorrectly predicted data is transmitted between the edge-side energy storage device and the cloud.

2. The method for processing digital twin model data applied to the energy storage Internet of Things according to claim 1, characterized in that: In step 4), when the prediction accuracy does not reach the expected value, model training is performed on the edge-side energy storage device, and the model data part in the digital twin model is adjusted by means of online training to fine-tune and correct the model; After training is completed, the trained digital twin model and the corresponding training data are synchronized to the cloud to realize the closed-loop of the digital twin model cloud-edge collaboration process.

3. The method for processing digital twin model data applied to the energy storage Internet of Things according to claim 1, characterized in that: In step 5), when there is a data missing situation in the cloud, the prediction data of the digital twin model is used to directly fill the missing data part.

4. The method for processing digital twin model data applied to the energy storage Internet of Things according to claim 1, characterized in that: The WASM model part is a program structure. The WASM model part is a description of the behavior of the energy storage device, or a mathematical description form of the physical decay model of the energy storage battery, or a user energy consumption behavior prediction model, or a charge and discharge strategy of the device.

5. The method for processing digital twin model data applied to the energy storage Internet of Things according to claim 1, characterized in that: The digital twin model trained in step 1) can predict the working states of the battery module, EMS module, BMS module, and PCS module in the energy storage Internet of Things system, and ensure that the accuracy rate of data prediction reaches the preset range.

Images