Digital twinning-based energy consumption prediction and anomaly detection method and system

Abstract

The invention provides an energy consumption prediction and anomaly detection method and system based on digital twinning, and the method comprises the steps: collecting physical parameters in real time through a sensor, and the physical parameters comprise the cell temperature, the total current of a battery pack, the total voltage of the battery pack, the gradient, the vehicle speed and the mass; collecting battery state parameters by using a battery management system, wherein the battery state parameters comprise real-time energy consumption; the collected physical parameters are input into a Simulink and Cruise combined model for simulation, and model output prediction energy consumption is obtained; wherein the Simulink is used for establishing an equivalent circuit model of a power battery, and the control system, the sensor simulation module and the Cruise simulation model are used for establishing a whole-vehicle simulation model of the pure electric vehicle; comparing the model output energy consumption with the real-time energy consumption, and obtaining an energy consumption deviation value; whether the energy consumption deviation value exceeds a preset range or not is judged, and if yes, energy consumption is judged to be abnormal; and if not, evaluating the energy consumption of the whole vehicle so as to solve the technical problem that the actual energy consumption analysis, energy management and endurance mileage evaluation of the whole vehicle are difficult.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle energy consumption prediction methods, and specifically to an energy consumption prediction and anomaly detection method and system based on digital twins. Background Art

[0002] The cruising range of electric mining trucks is relatively short, especially in the case of continuous heavy-load operation, electric mining trucks need to be charged frequently. In the actual operation of current mining trucks, road conditions, slope, load, speed, motor efficiency, battery status, driver operation, ambient temperature, and the use of other electrical equipment will cause differences in energy consumption assessment. The above factors may cause the economic efficiency of electric mining trucks to decline, resulting in further increases in operating costs. In addition, in cold areas, the accuracy of lithium battery SOC status assessment is difficult to guarantee, which has a great impact on the actual energy consumption analysis, energy management and cruising range assessment of the vehicle. Summary of the invention

[0003] The present application provides an energy consumption prediction and anomaly detection method and system based on digital twins to solve the technical problems of difficult actual energy consumption analysis, energy management and range evaluation of the whole vehicle.

[0004] The present application provides an energy consumption prediction and anomaly detection method based on digital twins, comprising the steps of: Using sensors to collect physical parameters in real time, the physical parameters include battery cell temperature, battery pack total current, battery pack total voltage, slope, vehicle speed and mass; Using the battery management system of the electric vehicle to collect battery status parameters, the battery status parameters include real-time energy consumption; The collected physical parameters are input into the Simulink and Cruise joint model for simulation to obtain the model output predicted energy consumption; Simulink is used to establish the equivalent circuit model of the power battery, the control system and the sensor simulation module, and the Cruise simulation model is used to build the whole vehicle simulation model of the pure electric vehicle; Compare the model output energy consumption with the real-time energy consumption and obtain the energy consumption deviation value; Determine whether the energy consumption deviation value exceeds the preset range. If so, determine that the energy consumption is abnormal; if not, evaluate the energy consumption of the entire vehicle.

[0005] Optionally, in the step of determining whether the energy consumption deviation value exceeds a preset range, the following steps are specifically included: Collect the operating conditions of the electric vehicle, including no-load uphill condition, no-load downhill condition, full-load uphill condition, full-load downhill condition, and waiting condition; When the energy consumption deviation value is within 10%, the energy consumption of the electric vehicle is judged to be in a normal state, and the energy consumption of the whole vehicle is evaluated; When the energy consumption deviation value is between 10% and 20%, the energy consumption is judged to be abnormal and the electric vehicle management platform is fed back to form a record, but the driver is not reminded; When the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the system will report it to the electric vehicle management platform to form a record and remind the driver at the same time.

[0006] Optionally, in the step of collecting the operating conditions of the electric vehicle, When the electric vehicle is in an unloaded uphill condition, the battery is in a medium load discharge state; When the electric vehicle is in an unloaded downhill condition, the battery is in a charging state; When an electric vehicle is in a fully loaded uphill condition, the battery is in a high-load discharge state; When an electric vehicle is in a fully loaded downhill condition, the battery is in a high-load charging state; When an electric vehicle is in a waiting state, the battery is in a low energy consumption state.

[0007] Optionally, when the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the electric vehicle management platform is fed back to form a record, and the step of reminding the driver at the same time also includes the steps of: The energy consumption deviation value is input into the classification neural network, and the high energy consumption type is determined, which includes abnormal SOC, low transmission system efficiency, high energy consumption of accessories, high braking intensity, high road rolling resistance, and abnormal BMS management.

[0008] Optionally, in the step of determining whether the energy consumption deviation value exceeds a preset range, if so, determining that the energy consumption is abnormal; if not, evaluating the energy consumption of the entire vehicle, The physical parameters and predicted energy consumption are integrated, and regression training is performed using a deep residual shrinkage network to obtain real-time energy consumption, average energy consumption, and energy consumption under different operating conditions to evaluate the energy consumption of the entire vehicle.

[0009] Optionally, when the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the electric vehicle management platform is fed back to form a record, and the driver is reminded at the same time, the following steps are also included: Use the Redis cache mechanism to achieve real-time model updates.

[0010] Correspondingly, the present application also provides an electronic device, including a memory and a processor, the memory being used to store executable program code; the processor being connected to the memory, and running a computer program corresponding to the executable program code by reading the executable program code, so as to execute the steps in the aforementioned energy consumption prediction and anomaly detection method based on digital twins.

[0011] Correspondingly, the present application also provides an energy consumption prediction and anomaly detection system based on digital twins, which includes an electronic device and a sensor component, the sensor component includes a voltage sensor, a current sensor, a temperature sensor, a slope sensor, a load sensor, and a speed sensor, and the sensor component is electrically connected to the electronic device.

[0012] The present application provides an energy consumption prediction and anomaly detection method and system based on digital twins. By establishing a digital twin model of an electric mining truck, the operating status of the whole vehicle and the battery is simulated in real time, and the twin model is used to analyze eight factors that are strongly related to the energy consumption of the whole vehicle: total current i of the battery pack, total voltage u of the battery pack, cell temperature t, SOC, SOH, vehicle speed v, slope θ and vehicle mass m. Real-time comparison is performed with the data during normal operation of the whole vehicle to predict the energy consumption of the whole vehicle in real time, and then the abnormal energy consumption of the vehicle is clustered and used to predict the remaining cruising range. The energy consumption prediction and anomaly detection method based on digital twins can accurately identify and diagnose the abnormal energy consumption status of the whole electric mining truck. At the same time, it is not limited by the influence of the external environment and temperature on the battery BMS, and the actual energy consumption of the whole vehicle can be monitored in real time to reduce operating costs.

[0013] The comparative analysis method is used to compare the real-time collected data and the predicted data of the digital twin model, and the deep learning-deep residual shrinkage network algorithm and rules are used to accurately determine the causes of abnormal energy consumption and ensure accurate diagnosis of the causes of high energy consumption of the entire vehicle.

[0014] Based on the energy consumption data information of electric mining trucks obtained by the data acquisition system, the energy consumption model is obtained by training big data samples through deep learning. The training method of the deep learning model and the combination of factors related to the energy consumption of electric mining trucks ensure reliable prediction and prompt functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 It is a flowchart of the energy consumption prediction and anomaly detection method based on digital twins. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application. In addition, it should be understood that the specific implementation methods described herein are only used to illustrate and explain the present application and are not used to limit the present application. In the present application, unless otherwise stated, the directional words used, such as "up", "down", "left", and "right", generally refer to the up, down, left, and right of the device in actual use or working state, specifically the drawing direction in the accompanying drawings.

[0018] The present application provides an energy consumption prediction and anomaly detection method and system based on digital twins, which are described in detail below. It should be noted that the description order of the following embodiments is not intended to limit the preferred order of the embodiments of the present application. In the following embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0019] See also Figure 1 This application provides a method for energy consumption prediction and anomaly detection based on digital twins. By establishing a digital twin model of an electric mining truck, the operating status of the vehicle and the battery is simulated in real time. The twin model is used to analyze eight factors that are strongly related to the energy consumption of the vehicle: total battery current i, total battery voltage u, battery cell temperature t, SOC, SOH, vehicle speed v, slope θ and vehicle mass m. The data is compared with the data during normal operation of the vehicle in real time to predict the energy consumption of the vehicle in real time, and then the abnormal energy consumption of the vehicle is clustered and used to predict the remaining cruising range.

[0020] See also Figure 1 , an energy consumption prediction and anomaly detection method based on digital twins, specifically comprising the following steps: S100, using sensors to collect physical parameters in real time, the physical parameters including battery cell temperature, battery pack total current, battery pack total voltage, slope, vehicle speed, and mass; In this application, a certain electric mining truck is used as a physical entity. A sensor assembly formed by multiple sensors installed in the battery system collects physical parameters in real time and uses them as real-time data.

[0021] The sensor components include voltage sensor, current sensor, temperature sensor, slope sensor, load sensor and speed sensor. The voltage sensor is used to obtain the total voltage of the battery pack in real time, the current sensor is used to obtain the total current of the battery pack in real time, the temperature sensor is used to obtain the battery cell temperature in real time, the slope sensor is used to obtain the external slope of the vehicle in real time, the load sensor is used to obtain the vehicle mass in real time, and the speed sensor is used to obtain the vehicle speed in real time.

[0022] S200, collecting battery status parameters using a battery management system of the electric vehicle, wherein the battery status parameters include real-time energy consumption; The battery management system collects battery status parameters as the basis for energy consumption changes and uses them as virtual layer data.

[0023] S300, input the collected physical parameters into the Simulink and Cruise joint model for simulation, and obtain the model output predicted energy consumption; wherein, Simulink is used to establish the equivalent circuit model, control system and sensor simulation module of the power battery, and the Cruise simulation model is used to build a pure electric vehicle whole vehicle simulation model; ‌‌Simulink simulation principle is based on dynamic system modeling and simulation, which is realized through interactive graphical environment and customizable module library. ‌ Simulink is a toolbox of MATLAB for building, simulating and analyzing dynamic system models.

[0024] The core principle of the Cruise simulation model is based on the Cruise software model and Simulink joint simulation, which predicts and optimizes the performance of hybrid vehicles through high-precision simulation. The model is highly flexible and scalable, and can be adjusted accordingly according to different vehicle control strategies to provide more accurate simulation results.

[0025] The real-time data collected by current, temperature, voltage, vehicle speed, slope, mass and speed sensors are used as the input of the digital virtual model, and the real-time energy consumption and other parameters output by the joint simulation virtual model are used as the output. Through comparative analysis, the real-time collected data are compared with the predicted data of the digital twin model to find the difference between the actual energy consumption of the whole vehicle and the digital twin model.

[0026] S400, comparing the model output energy consumption with the real-time energy consumption, and obtaining an energy consumption deviation value; The deep learning-deep residual shrinkage network model is a key prediction model. It is crucial for the identification and diagnosis of abnormal energy consumption. Based on the comparative analysis method, it can accurately determine the type and cause of high energy consumption.

[0027] The deep residual shrinkage network is composed of residual shrinkage units, which embodies the idea of ​​an attention mechanism: highlighting the importance of strongly correlated data by reducing the proportion of weakly correlated data. Although this method may remove features that are helpful for transfer learning, these features will still be retained in the network through the identity mapping in the residual module. It’s just that after passing through the module, the proportion of these features will be reduced.

[0028] S500, determining whether the energy consumption deviation value exceeds a preset range, if so, determining that the energy consumption is abnormal; if not, evaluating the energy consumption of the entire vehicle; The system is based on comparative analysis and energy consumption anomaly diagnosis. Once the energy consumption is monitored to exceed the set range, the energy consumption anomaly diagnosis module will evaluate the severity of the anomaly.

[0029] The step of determining whether the energy consumption deviation value exceeds a preset range specifically includes the following steps: Collect the operating conditions of the electric vehicle, including no-load uphill condition, no-load downhill condition, full-load uphill condition, full-load downhill condition, and waiting condition; When an electric vehicle is in an unloaded uphill condition, it needs a large amount of power to overcome its own gravity and rolling resistance when driving uphill. At this time, the battery is in a medium load discharge state; When an electric vehicle is in an unloaded downhill condition, a portion of the kinetic energy and gravitational energy can be converted into electrical energy through the brake energy recovery system and fed back to the battery. At this time, the battery is in a charging state; When an electric vehicle is fully loaded and running uphill, the load weight is usually much higher than its own weight, and greater power is required to overcome the gravity and rolling resistance of the vehicle and the load. At this time, the battery is in a high-load discharge state; When an electric vehicle is in a fully loaded downhill condition, the gravity of the load and its own weight does work, and the battery is in a high-load charging state; When electric vehicles are in waiting conditions, waiting to load and unload, electric mining trucks may only consume a small amount of electricity for some basic electronic equipment (such as lighting and communication equipment), so the battery discharge rate may be very low, and the battery is in a low energy consumption state.

[0030] When the energy consumption deviation value is within 10%, the energy consumption of the electric vehicle is judged to be in a normal state, and the energy consumption of the whole vehicle is evaluated; When the energy consumption deviation value is between 10% and 20%, the energy consumption is judged to be abnormal and the electric vehicle management platform is fed back to form a record, but the driver is not reminded; When the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the electric vehicle management platform is fed back to form a record, and the driver is reminded at the same time. At the same time, the energy consumption deviation value is input into the classification neural network, and the high energy consumption type is judged, which includes abnormal SOC, low transmission system efficiency, high energy consumption of accessories, high braking intensity, high road rolling resistance, and abnormal BMS management. The prompt system of the electric vehicle will give voice prompts according to the cause of high energy consumption. Otherwise, the prompt system will not be triggered.

[0031] The physical parameters and predicted energy consumption are integrated, and regression training is performed using a deep residual shrinkage network to obtain real-time energy consumption, average energy consumption, and energy consumption under different operating conditions to evaluate the energy consumption of the entire vehicle.

[0032] Based on the energy consumption data information of electric mining trucks obtained by the data acquisition system, the energy consumption model of electric mining trucks is obtained through deep learning training of big data samples. The training method of the deep learning model and the combination of factors related to the energy consumption of electric mining trucks ensure reliable prediction and prompt functions.

[0033] S600, using the Redis cache mechanism to achieve real-time model updates.

[0034] The digital twin model is established by using the physical and simulation models of the electric mining truck and the battery pack, which is integrated with the real-time sensor data, and a model update strategy is proposed. The Redis cache mechanism is introduced to achieve real-time model updates. The modeling method, model and real-time data fusion strategy, and model update cache mechanism ensure that the model is synchronized with the actual vehicle and battery status.

[0035] This embodiment also provides an energy consumption prediction and anomaly detection system based on digital twins, including an electronic device and a sensor component, wherein the sensor component includes a voltage sensor, a current sensor, a temperature sensor, a slope sensor, a load sensor, and a speed sensor, and the sensor component is electrically connected to the electronic device.

[0036] The electronic device includes a memory and a processor, wherein the memory is used to store executable program code; the processor is connected to the memory, and runs a computer program corresponding to the executable program code by reading the executable program code to execute the steps in the aforementioned energy consumption prediction and anomaly detection method based on digital twins.

[0037] The above is a detailed introduction to the energy consumption prediction and anomaly detection method and system based on digital twins provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for energy consumption prediction and anomaly detection based on digital twins, characterized in that: Includes steps: Using sensors to collect physical parameters in real time, the physical parameters include battery cell temperature, battery pack total current, battery pack total voltage, slope, vehicle speed and mass; Using the battery management system of the electric vehicle to collect battery status parameters, the battery status parameters include real-time energy consumption; The collected physical parameters are input into the Simulink and Cruise joint model for simulation to obtain the model output predicted energy consumption; Simulink is used to establish the equivalent circuit model of the power battery, the control system and the sensor simulation module, and the Cruise simulation model is used to build the whole vehicle simulation model of the pure electric vehicle; Compare the model output energy consumption with the real-time energy consumption and obtain the energy consumption deviation value; Determine whether the energy consumption deviation value exceeds the preset range. If so, determine that the energy consumption is abnormal; if not, evaluate the energy consumption of the entire vehicle, where: The step of determining whether the energy consumption deviation value exceeds a preset range specifically includes the following steps: Collect the operating conditions of the electric vehicle, including no-load uphill condition, no-load downhill condition, full-load uphill condition, full-load downhill condition, and waiting condition; When the energy consumption deviation value is within 10%, the energy consumption of the electric vehicle is judged to be in a normal state, and the energy consumption of the whole vehicle is evaluated; When the energy consumption deviation value is between 10% and 20%, the energy consumption is judged to be abnormal and the electric vehicle management platform is fed back to form a record, but the driver is not reminded; When the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the system will report it to the electric vehicle management platform to form a record and remind the driver at the same time.

2. The energy consumption prediction and anomaly detection method based on digital twin according to claim 1 is characterized in that: In the step of collecting the operating conditions of the electric vehicle, When the electric vehicle is in an unloaded uphill condition, the battery is in a medium load discharge state; When the electric vehicle is in an unloaded downhill condition, the battery is in a charging state; When an electric vehicle is in a fully loaded uphill condition, the battery is in a high-load discharge state; When an electric vehicle is in a fully loaded downhill condition, the battery is in a high-load charging state; When an electric vehicle is in a waiting state, the battery is in a low energy consumption state.

3. The energy consumption prediction and anomaly detection method based on digital twin according to claim 1 is characterized in that: When the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the electric vehicle management platform is fed back to form a record, and the driver is reminded at the same time, the steps also include: The energy consumption deviation value is input into the classification neural network, and the high energy consumption type is determined, which includes abnormal SOC, low transmission system efficiency, high energy consumption of accessories, high braking intensity, high road rolling resistance, and abnormal BMS management.

4. The energy consumption prediction and anomaly detection method based on digital twin according to claim 1 is characterized in that: In the step of determining whether the energy consumption deviation value exceeds the preset range, if so, determining that the energy consumption is abnormal; if not, evaluating the energy consumption of the whole vehicle, The physical parameters and predicted energy consumption are integrated, and regression training is performed using a deep residual shrinkage network to obtain real-time energy consumption, average energy consumption, and energy consumption under different operating conditions to evaluate the energy consumption of the entire vehicle.

5. The energy consumption prediction and anomaly detection method based on digital twin according to claim 1 is characterized in that: When the energy consumption deviation value is above 20%, the energy consumption is judged to be abnormal, and the electric vehicle management platform is fed back to form a record, and the driver is reminded at the same time, the following steps are also included: Use the Redis cache mechanism to achieve real-time model updates.

6. An electronic device, characterized in that: include: A memory for storing executable program codes; as well as A processor, connected to the memory, runs a computer program corresponding to the executable program code by reading the executable program code, so as to execute the steps in the energy consumption prediction and anomaly detection method based on digital twins as described in any one of claims 1 to 6.

7. An energy consumption prediction and anomaly detection system based on digital twins, characterized in that: include: The electronic device according to claim 6; as well as The sensor assembly includes a voltage sensor, a current sensor, a temperature sensor, a slope sensor, a load sensor, and a speed sensor, and the sensor assembly is electrically connected to the electronic device.

Images