Motorcycle instrument fuel quantity display calibration method based on multi-sensor fusion

Abstract

The invention discloses a motorcycle instrument fuel quantity display calibration method based on multi-sensor fusion. The method comprises the following steps: acquiring a data signal of a sensor, fusing and preprocessing data, calculating a fuel consumption rate by adopting a dynamic self-adaptive fuel consumption model, predicting the remaining fuel quantity, and dynamically adjusting the remaining mileage; and self-adaptive calibration: based on a dynamic self-adaptive fuel consumption model, displaying the remaining fuel quantity, the instantaneous fuel consumption, the average fuel consumption and the estimated endurance mileage by an instrument, and displaying a historical fuel consumption trend in a thermodynamic diagram form. The method has the advantages that multi-sensor data and dynamic environment compensation are fused, the liquid level measurement error is smaller than + / -1%, the endurance prediction error is smaller than + / -5%, the overall oil mass calculation precision is improved to 98%, the DAFCM model supports millisecond-level response, the method dynamically adapts to complex driving scenes, resource occupation is low, the method is suitable for a low-end vehicle-mounted controller, and the method is suitable for large-scale popularization and application. And reliable data support is provided for intelligent travel.

Description

Technical Field

[0001] The present invention relates to the technical field of motorcycle fuel gauge calibration, and in particular to a motorcycle instrument oil quantity display calibration method based on multi-sensor fusion. Background Art

[0002] With the development of intelligent motorcycles, the accuracy of fuel display directly affects user experience and vehicle energy efficiency management. Traditional fuel measurement technology relies on a single sensor and static model, which is difficult to cope with complex working conditions and dynamic environments. Although existing technologies have attempted to optimize fuel display logic through sensors and simple algorithms, the following key issues still exist:

[0003] Defects and shortcomings of existing technology:

[0004] 1. Limitations of a single data source: Traditional liquid level sensors are susceptible to interference from factors such as fuel tank shape, vehicle tilt, and oil foam. Existing technologies only compensate for static liquid level height through geometric modeling, ignoring instantaneous volume changes caused by fuel level fluctuations and inertial effects during vehicle movement. Furthermore, they lack support for the integration of multi-dimensional data (such as acceleration and fuel flow rate), resulting in accumulated measurement deviations.

[0005] 2. Poor adaptability to dynamic driving conditions: Existing algorithms are mostly based on linear regression or fixed threshold judgment, and do not fully integrate the nonlinear impact of vehicle dynamic parameters (such as sudden acceleration, sudden deceleration, and continuous cornering) on ​​fuel consumption, resulting in large deviations in instantaneous fuel consumption calculations.

[0006] 3. Weak multi-source data fusion capabilities: Traditional methods rely solely on liquid level sensors and fail to integrate multi-dimensional data such as fuel pressure, flow, and engine operating conditions, resulting in accumulated errors.

[0007] 4. Calibration lag: The calibration cycle relies on fixed time or mileage intervals, and the calibration strategy cannot be dynamically adjusted according to real-time operating conditions (such as frequent starts and stops, long-distance climbing, and sudden load changes), resulting in errors that increase with usage time.

[0008] 5. Lack of edge computing capabilities: Data processing relies on the limited computing power of the on-board ECU. Complex algorithms run inefficiently on low-computing-power on-board controllers, making it difficult to process high-dimensional sensor data in real time, affecting the algorithm's response speed and accuracy. Summary of the Invention

[0009] In view of the above-mentioned deficiencies in the prior art, the present invention provides a motorcycle instrument fuel level display calibration method based on multi-sensor fusion, which realizes high-precision fuel measurement and adaptive calibration.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A motorcycle instrument fuel level display calibration method based on multi-sensor fusion is characterized by comprising the following steps:

[0012] S1. Obtain sensor data signals, including liquid level sensor signals, flow meter signals, IMU signals, and environmental sensor signals;

[0013] S2. Data fusion and preprocessing, including dynamically estimating the true fuel level and eliminating vibration noise based on the fused level array, IMU, and flow meter data; establishing dynamic compensation, including building a database of fuel level fluctuations under different accelerations and attitudes based on the fuel tank CAD model and fluid simulation; dynamically calculating fuel density based on the Clausius-Clapeyron equation and ambient temperature and humidity; and dynamically correcting the level-to-volume conversion coefficient by querying the simulation database based on IMU data.

[0014] S3. A dynamic adaptive fuel consumption model is used to calculate the fuel consumption rate. The sensor data in step S1 is input, including liquid level, flow rate, vehicle speed, acceleration, attitude angle, environmental parameters, engine speed, and load weight. The dynamic adaptive fuel consumption model uses a lightweight LSTM neural network with 12 nodes in the input layer, 8 nodes in the hidden layer, and the output layer is the instantaneous fuel consumption rate.

[0015] S4, remaining fuel prediction, including real-time range calculation, based on the fuel consumption rate output by the dynamic adaptive fuel consumption model, and combined with Tbox map data to obtain the road slope and congestion index ahead, and dynamically adjust the remaining mileage;

[0016] S5, Adaptive Calibration. Triggered by conditions such as detecting a refueling operation, a sharp fuel level fluctuation caused by sudden deceleration, an air pressure change rate greater than 10 hPa / min caused by a sudden change in altitude, or three consecutive deviations of >3% between the model prediction value and the sensor measurement value. The calibration algorithm uses a sliding window least squares method to dynamically adjust the model parameters and synchronize them to the cloud.

[0017] S6. Based on a dynamic adaptive fuel consumption model, the instrument displays the remaining fuel, instantaneous fuel consumption, average fuel consumption, estimated cruising range, and displays historical fuel consumption trends in the form of a heat map.

[0018] Furthermore, in step S2, a timestamp mechanism is used to align the data of multiple sensors, the sampling frequency is the same as 100 Hz, and the timing deviation is eliminated by Kalman filtering.

[0019] Furthermore, in step S3, the dynamic adaptive fuel consumption model adopts an online learning mechanism, injecting the latest data into the model for fine-tuning every 10 minutes, and the weight update adopts quantitative perception training technology to reduce computing power requirements.

[0020] Furthermore, step S4 also includes an abnormality detection to determine whether the difference between the liquid level data and the flow meter data continues to exceed 5%. If so, a self-check program is triggered and the redundant sensor mode is switched.

[0021] Furthermore, the remaining fuel level display status in step S6 includes:

[0022] When a displays 1 grid and the remaining fuel volume is less than 2L, the fuel low alarm indicator icon lights up in yellow and flashes 1 grid at a frequency of 1Hz;

[0023] b. If the fuel sensor is disconnected or short-circuited, 5 grids will flash and the fuel level icon will flash in yellow as an alarm with a flashing frequency of 1Hz. A value greater than 500 ohms is considered disconnected and a value less than 3 ohms is considered short-circuited.

[0024] c. To switch from the normal state to the open state, the open state signal must be input for more than 15 seconds;

[0025] d. To switch from normal state to short circuit state, the short circuit state signal must be input for more than 15 seconds;

[0026] e To switch from open circuit or short circuit state to normal state, the input signal must last for more than 3S.

[0027] Furthermore, the remaining fuel level display status in step S6 includes:

[0028] Without speed:

[0029] a. Fueling status: Displays the corresponding fuel level in real time according to the fuel sensor signal; when the fuel resistance reaches point F, the remaining fuel level on the instrument will display the number of liters of full fuel;

[0030] b. Fuel consumption status: Fuel consumption only decreases but not increases, and is displayed in a cycle of 10 seconds;

[0031] With vehicle speed:

[0032] a. Fuel consumption status: Fuel volume is only decreasing but not increasing. The fuel sensor signal frequency is 1Hz, and the fuel grid value critical range is compared and judged. If the fuel signal value changes and remains stable within the fuel grid value range for more than 15 seconds, the fuel grid display is updated. The instrument updates the fuel grid display every 15 seconds.

[0033] b. When the power is turned off, the current remaining fuel amount and the fuel sensor signal value must be recorded. When the power is turned on again, if the fuel sensor signal value has not changed, the remaining fuel amount recorded last time will be displayed. If the value has changed, the corresponding remaining fuel amount will be updated and displayed.

[0034] Furthermore, the data displayed in step S6 supports active calibration. The refueling amount is automatically recorded during refueling and compared with the predicted value. If the deviation is greater than 3%, the model parameter reset is initiated. Manual calibration by the user is also supported. The user enters the actual refueling amount and mileage through the mobile phone APP, and the system automatically generates a compensation coefficient.

[0035] Furthermore, in step S6, the meter provides a low fuel alarm based on the navigation data, indicating the distance to the nearest gas station and the required fuel amount.

[0036] The beneficial effects of the present invention include: integration of multi-sensor data and dynamic environmental compensation, liquid level measurement error <±1%, cruising range prediction error <±5%, overall fuel quantity calculation accuracy increased to 98%, the DAFCM model supports millisecond-level response, dynamically adapts to complex driving scenarios, has low resource usage, is suitable for low-end vehicle controllers, and provides reliable data support for smart travel. DETAILED DESCRIPTION

[0037] The present invention will be further described in detail below with reference to specific embodiments.

[0038] A high-precision fuel display logic and calibration method for motorcycle instruments based on multi-sensor fusion. 1. Hardware selection

[0039] (1) Multimodal sensor group

[0040] 1. High-precision liquid level sensor: resolution of ±0.5mm, built-in temperature compensation chip (working range: -40℃~125℃), installed in the geometric center of the tank to avoid tilt interference.

[0041] 2. Miniature fuel flow meter: integrated into the fuel line, using the Hall effect principle to measure real-time fuel flow rate (accuracy ±1mL / s) and simultaneously monitor oil pressure fluctuations.

[0042] 3. Six-axis inertial measurement unit (IMU): collects vehicle acceleration, angular velocity, pitch angle, and roll angle (accuracy ±0.1°) to compensate for dynamic tilt and vibration interference.

[0043] 4. Environmental sensor group: including air pressure sensor (monitoring altitude changes) and temperature sensor (synchronously collecting oil temperature and ambient temperature), monitoring atmospheric pressure, temperature, and humidity for dynamic correction of fuel density.

[0044] 5. Seat pressure sensor, used to detect vehicle load.

[0045] 6. Engine speed sensor, used to detect the real-time speed of the engine.

[0046] (2) Edge computing module.

[0047] Equipped with a low-power AI chip (such as an ARM Cortex-M7 + NPU) and a built-in hardware floating-point unit, it supports real-time multi-threaded task scheduling. It supports real-time execution of lightweight deep learning models with a processing frequency of ≥100Hz. It aggregates multi-vehicle data, trains a global model, and distributes it to the edge, optimizing prediction accuracy for special scenarios (such as extreme altitudes and low temperatures).

[0048] 2. Data Fusion and Preprocessing

[0049] 1. Multi-sensor data fusion and noise reduction: Fusion of liquid level array, IMU, and flow meter data dynamically estimates the true fuel level and eliminates vibration noise. A timestamp mechanism is used to align multi-sensor data, with a unified sampling frequency of 100Hz. Kalman filtering is used to eliminate timing deviations.

[0050] 2. Dynamic compensation strategy:

[0051] A. Fluid dynamics simulation: Based on the fuel tank CAD model and FLUENT fluid simulation, a database of fuel level fluctuations under different accelerations and postures is constructed.

[0052] B Fuel density correction: Based on the Clausius-Clapeyron equation, the fuel density is dynamically calculated in combination with the ambient temperature and humidity.

[0053] C Real-time interpolation compensation: query the simulation database based on IMU data and dynamically correct the liquid level-volume conversion coefficient.

[0054] 3. Core Algorithm Design

[0055] (1) Dynamic Adaptive Fuel Consumption Model (DAFCM):

[0056] 1. Input parameters: liquid level, flow rate, vehicle speed, acceleration, engine speed, attitude angle, environmental parameters, and load weight (obtained through seat pressure sensor).

[0057] x t =[h t ,q t ,v t ,a t ,ω t ,θ t ,φ t ,T t ,P t ,ρ t ,W t ]∈R 12

[0058] h t : Liquid level height (mm);

[0059] qt : fuel flow rate (mL / s);

[0060] v t : vehicle speed (km / h);

[0061] a t :Acceleration(m / s 2 );

[0062] ω t : engine speed (rpm);

[0063] θ t ,φ t : vehicle pitch angle and roll angle (°);

[0064] T t ,P t ,ρ t :Ambient temperature (℃), air pressure (kPa), air density (kg / m 3 );

[0065] W t : Load weight (kg, obtained through seat pressure sensor);

[0066] x t is the input vector at time step tt, containing 12 dimensions of sensor data (such as liquid level, vehicle speed, acceleration, etc.);

[0067] 2. Model structure: A lightweight LSTM neural network is used, with 12 nodes in the input layer, 8 nodes in the hidden layer, and the output layer is the instantaneous fuel consumption rate (L / 100km).

[0068] 2.1 Gating Mechanism Formula

[0069] Input gate: controls the input of new information

[0070] i t =σ(W xi x t +W hi h t-1 +b i )

[0071] i t : The activation value of the input gate (value range [0,1]), which controls the proportion of new information entering the memory unit.

[0072] σ: Sigmoid activation function, which compresses the input to the range [0,1].

[0073] W xi : Input data x t Weight matrix to the input gate (dimensions: hidden layer size × input dimension).

[0074] x t : Input vector for the current time step (dimension: input dimension × 1).

[0075] W hi : Previous hidden state h t-1 Weight matrix to the input gate (dimensions: hidden layer size × hidden layer size). h t-1 : The hidden state at the previous time step (dimension: hidden layer size × 1).

[0076] b i : Bias term for the input gate (dimension: hidden layer size × 1).

[0077] Forget Gate: Control the forgetting of historical information

[0078] f t =σ(W xf x t +W hf h t-1 +b f )

[0079] f t : The activation value of the forget gate (value range [0,1]), which controls the previous memory unit C t-1 The forgetting ratio.

[0080] W xf : Input data x t To the weight matrix of the forget gate.

[0081] W hf : Previous hidden state h t-1 To the weight matrix of the forget gate.

[0082] b f : Bias term of forget gate.

[0083] Output gate: controls the output of the hidden state

[0084] o t =σ(W xo x t +W ho h t-1 +b o )

[0085] o t : The activation value of the output gate (value range [0,1]), which controls the current memory unit C t For the hidden state h t contribution ratio.

[0086] W xo : Input data x tWeight matrix to the output gate.

[0087] W ho : Previous hidden state h t-1 Weight matrix to the output gate.

[0088] b o : Bias term for the output gate.

[0089] Candidate memory units: generating new memory content

[0090] Q t =tanh(W xc x t +W hc h t-1 +b c )

[0091] Q t : The value of the candidate memory cell (range [-1, 1]), which represents the new memory content that may be generated by the current input. tanh: Hyperbolic tangent activation function, which compresses the input to the interval [-1, 1].

[0092] W xc : Input data x t to the weight matrix of the candidate memory unit.

[0093] W hc : Previous hidden state h t-1 to the weight matrix of the candidate memory unit.

[0094] b c : Bias term of candidate memory unit

[0095] 2.2 Memory Unit and Hidden State Update

[0096] Memory unit:

[0097] C t =f t ⊙C t-1 +i t ⊙Q t

[0098] ⊙: Element-wise multiplication (Hadamard product).

[0099] C t : The memory cell value at the current time step, combined with the results of the forget gate and the input gate

[0100] Hidden state:

[0101] h t =o t ⊙tanh(C t )

[0102] h t : The hidden state of the current time step, used to pass to the next time step or output prediction

[0103] 2.3 Output Layer Mapping

[0104] Output instantaneous fuel consumption rate yt (unit: L / 100km):

[0105] y t =W y h t +b y

[0106] W y ∈R 1×8 , b y ∈R is the output layer weight and bias

[0107] 3. Online learning mechanism: The latest data is injected into the model for fine-tuning every 10 minutes, and the weight update adopts Quantization-Aware Training (QAT) technology to reduce computing power requirements.

[0108] (2) Remaining fuel prediction algorithm:

[0109] 1. Real-time range calculation: Based on DAFCM output and combined with Tbox map data, the slope and congestion index of the road ahead are obtained, and the remaining mileage is dynamically adjusted.

[0110] 2. Abnormal detection: If the difference between the liquid level and the flow meter data exceeds 5% continuously, the self-test program is triggered and the redundant sensor mode is switched. (III) Adaptive calibration strategy:

[0111] 1. Event-driven: Refueling operation, sudden deceleration (violent fluctuation of fuel level), or sudden altitude change (pressure change rate > 10hPa / min) is detected.

[0112] 2. Error accumulation threshold: Calibration is triggered when the deviation between the model prediction value and the sensor measured value is greater than 3% for three consecutive times.

[0113] 3. Calibration algorithm: Use sliding window least squares method to dynamically adjust model parameters and synchronize them to the cloud.

[0114] (IV) Instrument fuel display strategy:

[0115] 1. Instrument fuel display judgment conditions:

[0116] When a displays 1 grid and the remaining fuel volume is less than 2L, the fuel low level alarm indicator icon lights up in yellow and flashes 1 grid at a frequency of 1Hz.

[0117] b. If the fuel sensor is disconnected or short-circuited, 5 bars will flash and the fuel level icon will flash in a yellow alarm at a frequency of 1Hz. A value greater than 500 ohms is considered disconnected, and less than 3 ohms is considered short-circuited.

[0118] c To switch from normal state to open state, the open state signal must be input for more than 15S.

[0119] d To switch from normal state to short-circuit state, the short-circuit state signal must be input for more than 15 seconds.

[0120] e To switch from open circuit or short circuit state to normal state, the input signal must last for more than 3S.

[0121] 2. Algorithm logic strategy:

[0122] No speed:

[0123] a. Fueling status: The fuel level is displayed in real time based on the fuel sensor signal. When the fuel resistance reaches point F, the remaining fuel level on the instrument displays the number of liters of full fuel. (For example, if the full fuel level is 15L, but there is a 3L fuel blind spot at the top dead center, no matter how many liters of fuel are in the blind spot, as long as the resistance reaches point F, the remaining fuel level will display 15L.)

[0124] b Fuel consumption status: Fuel volume only decreases and does not increase, and is displayed in a cycle of 10 seconds.

[0125] With vehicle speed:

[0126] a. Fuel consumption status: The fuel volume is only decreasing but not increasing. The fuel sensor signal (frequency 1Hz) is collected in real time, and the fuel grid value critical range points are compared and judged. After the fuel signal value changes and remains stable in this fuel grid value range for more than 15 seconds, the fuel grid display is updated. The instrument updates the fuel grid display in a cycle of 15 seconds.

[0127] b When the power is turned off for 15 seconds, the current remaining fuel amount and the fuel sensor signal value must be recorded. When the power is turned on again, if the fuel sensor signal value has not changed, the remaining fuel amount recorded last time will be displayed. If the value has changed, the corresponding remaining fuel amount will be updated and displayed.

[0128] 4. User Interaction Design

[0129] 1. Multi-level feedback system:

[0130] A Active calibration: Automatically records the amount of fuel added during refueling and compares it with the predicted value. If the deviation is greater than 3%, the model parameters are reset.

[0131] B User manual calibration: supports inputting actual refueling amount and mileage through the mobile phone APP, and the system automatically generates a compensation coefficient.

[0132] 2. Visual interface:

[0133] The instrument panel A displays the remaining fuel, instantaneous fuel consumption, average fuel consumption, estimated range, and displays historical fuel consumption trends in the form of a heat map (differentiating between flat road / climbing / downhill scenarios).

[0134] B provides low fuel warning (combined with navigation data, prompts the distance to the nearest gas station and the required fuel amount).

[0135] This embodiment of the multi-sensor fusion fuel display system includes a MEMS liquid level sensor, an IMU attitude compensation module, and a lightweight LSTM prediction model. Its dynamic environmental compensation algorithm is implemented based on the Clausius-Clapeyron equation. Through multi-sensor data fusion, a dynamic adaptive algorithm, and edge computing optimization, this system achieves millimeter-level accuracy and second-level response for motorcycle fuel display, providing reliable data support for smart travel.

[0136] The following are the preparatory steps for implementing the present invention:

[0137] Step 1: System initialization and sensor calibration

[0138] After power is applied, the level sensor performs a self-test and then injects a standard amount of oil (e.g., 1L) to calibrate the zero and full-scale values. The IMU sensor undergoes static calibration to eliminate installation errors and ensure that the pitch and roll angles return to zero.

[0139] Step 2: Data acquisition and dynamic compensation

[0140] The liquid level data is converted into fuel volume through cubic spline interpolation and corrected in real time based on the IMU angle.

[0141] Fuel density calculation:

[0142] ρ=ρ 20℃ ×[1-α(T-20)+β(P-P0)]

[0143] in:

[0144] ρ 20℃ It is the fuel density under standard reference conditions (ambient temperature 20°C).

[0145] α is the temperature expansion coefficient of the fuel, which indicates the relative rate of change of density when the temperature rises by 1°C.

[0146] T is the actual temperature of the current fuel, in °C.

[0147] β is the pressure compressibility coefficient of the fuel, which indicates the relative rate of change of density when the pressure increases by 1 Pa.

[0148] P is the actual pressure of the current fuel, in Pa.

[0149] P0 is the reference pressure (usually standard atmospheric pressure or the reference pressure for fuel density calibration).

[0150] Step 3: Model inference and oil volume prediction

[0151] The DAFCM model receives an input vector every 0.1 seconds and outputs the instantaneous fuel consumption rate.

[0152] The remaining oil amount calculation formula is:

[0153] Q 剩余 =Q 初始 -∫0 t DAFCM(t)dt

[0154] Dynamic update of driving range:

[0155]

[0156] where Q 剩余 Indicates the remaining oil volume, Q 初始 represents the initial fuel volume, DAFM(t) represents the instantaneous fuel consumption rate calculated by the model, and D 续航 Indicates the remaining cruising range, DAFCM(t) 滑动平均 represents the average fuel consumption rate calibrated by the sliding window least squares method.

[0157] Step 4: Real-time calibration and troubleshooting

[0158] 1. Refueling event trigger: When the liquid level rises continuously for 5 seconds and the flow meter has no output, it is determined to be a refueling operation and the refueling amount is recorded as Q 加油 .

[0159] 2. Calibration logic: If |Q 预测 -Q 加油 ∣>0.3LL, start the gradient descent algorithm to optimize the LSTM weights, where Q 预测 is the refueling amount predicted by the model.

[0160] 3. Fault handling: When the pressure sensor fails, switch to flow meter + IMU fusion mode and alarm through the CAN bus.

[0161] Step 5: Data cloud synchronization and OTA upgrade

[0162] 1. Data encryption upload: Upload compressed driving data (JSON format) to the cloud platform every day to generate a fuel consumption analysis report.

[0163] 2. Global model training: The cloud uses a federated learning framework to aggregate data from tens of thousands of vehicles and generate optimized model weights. The cloud model version is tested monthly, and any updates are pushed to the vehicle terminal via Delta Update.

[0164] The technical solutions provided by the embodiments of the present invention are introduced in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the embodiments of the present invention. The description of the above embodiments is only applicable to help understand the principles of the embodiments of the present invention. At the same time, for those skilled in the art, according to the embodiments of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A motorcycle instrument fuel level display calibration method based on multi-sensor fusion, characterized by: The following steps are included: S1. Obtain sensor data signals, including liquid level sensor signals, flow meter signals, IMU signals, and environmental sensor signals; S2, data fusion and preprocessing, including dynamically estimating the true height of the fuel level based on the fused liquid level array, IMU and flow meter data, and eliminating vibration noise; Establish dynamic compensation, including building a database of fuel level fluctuations under different accelerations and attitudes based on the fuel tank CAD model and fluid simulation; dynamically calculating fuel density based on the Clausius-Clapeyron equation and ambient temperature and humidity; and dynamically correcting the level-to-volume conversion coefficient by querying the simulation database based on IMU data. S3. A dynamic adaptive fuel consumption model is used to calculate the fuel consumption rate. The sensor data in step S1 is input, including liquid level, flow rate, vehicle speed, acceleration, attitude angle, environmental parameters, engine speed, and load weight. The dynamic adaptive fuel consumption model uses a lightweight LSTM neural network with 12 nodes in the input layer, 8 nodes in the hidden layer, and the output layer is the instantaneous fuel consumption rate. S4, remaining fuel prediction, including real-time range calculation, based on the fuel consumption rate output by the dynamic adaptive fuel consumption model, and combined with Tbox map data to obtain the road slope and congestion index ahead, and dynamically adjust the remaining mileage; S5, Adaptive Calibration. Triggered by conditions such as detecting a refueling operation, a sharp fuel level fluctuation caused by sudden deceleration, an air pressure change rate greater than 10 hPa / min caused by a sudden change in altitude, or three consecutive deviations of >3% between the model prediction value and the sensor measurement value. The calibration algorithm uses a sliding window least squares method to dynamically adjust the model parameters and synchronize them to the cloud. S6. Based on a dynamic adaptive fuel consumption model, the instrument displays the remaining fuel, instantaneous fuel consumption, average fuel consumption, estimated cruising range, and displays historical fuel consumption trends in the form of a heat map.

2. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: In step S2, a timestamp mechanism is used to align the data of multiple sensors, with the same sampling frequency of 100 Hz, and the timing deviation is eliminated through Kalman filtering.

3. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: In step S3, the dynamic adaptive fuel consumption model adopts an online learning mechanism, injecting the latest data into the model for fine-tuning every 10 minutes, and using quantitative perception training technology to update the weights to reduce computing power requirements.

4. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: Step S4 also includes anomaly detection to determine whether the difference between the liquid level data and the flow meter data continues to exceed 5%. If so, a self-check program is triggered and the redundant sensor mode is switched.

5. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: The remaining fuel level display status in step S6 includes: When a displays 1 grid and the remaining fuel volume is less than 2L, the fuel low alarm indicator icon lights up in yellow and flashes 1 grid at a frequency of 1Hz; b. If the fuel sensor is disconnected or short-circuited, 5 grids will flash and the fuel level icon will flash in yellow as an alarm with a flashing frequency of 1Hz. A value greater than 500 ohms is considered disconnected and a value less than 3 ohms is considered short-circuited. c. To switch from the normal state to the open state, the open state signal must be input for more than 15 seconds; d. To switch from normal state to short circuit state, the short circuit state signal must be input for more than 15 seconds; e To switch from open circuit or short circuit state to normal state, the input signal must last for more than 3S.

6. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 5, characterized in that: The remaining fuel level display status in step S6 includes: Without speed: a. Fueling status: Displays the corresponding fuel level in real time according to the fuel sensor signal; when the fuel resistance reaches point F, the remaining fuel level on the instrument will display the number of liters of full fuel; b. Fuel consumption status: Fuel consumption only decreases but not increases, and is displayed in a cycle of 10 seconds; With vehicle speed: a. Fuel consumption status: Fuel volume is only decreasing but not increasing. The fuel sensor signal frequency is 1Hz, and the fuel grid value critical range is compared and judged. If the fuel signal value changes and remains stable within the fuel grid value range for more than 15 seconds, the fuel grid display is updated. The instrument updates the fuel grid display every 15 seconds. b. When the power is turned off, the current remaining fuel amount and the fuel sensor signal value must be recorded. When the power is turned on again, if the fuel sensor signal value has not changed, the remaining fuel amount recorded last time will be displayed. If the value has changed, the corresponding remaining fuel amount will be updated and displayed.

7. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: The data displayed in step S6 supports active calibration. The refueling amount is automatically recorded during refueling and compared with the predicted value. If the deviation is greater than 3%, the model parameter reset is initiated. Manual calibration is also supported. The user enters the actual refueling amount and mileage through the mobile phone APP, and the system automatically generates a compensation coefficient.

8. The motorcycle instrument fuel level display calibration method based on multi-sensor fusion according to claim 1, characterized in that: In step S6, the instrument panel provides a low fuel alarm based on the navigation data, indicating the distance to the nearest gas station and the required fuel amount.

Images