Engineering machinery oil consumption monitoring and diagnosing system and method based on multi-source information fusion

Abstract

This invention provides a fuel consumption monitoring and diagnosis system and method for construction machinery based on multi-source information fusion, belonging to the field of intelligent management technology for construction machinery. The system includes an on-board terminal and a cloud platform server. The on-board terminal, through a refueling monitoring module integrated into the fuel tank inlet pipeline, simultaneously collects refueling volume, fuel dielectric constant, and temperature parameters; and through a working condition monitoring module, it collects real-time equipment positioning, engine CAN bus, and inertial measurement data. The cloud platform server executes a rapid fuel quality assessment algorithm and a multi-dimensional fuel consumption anomaly diagnosis algorithm. The latter constructs a personalized theoretical fuel consumption benchmark based on a gradient boosting regression tree model, automatically diagnoses the root causes of fuel consumption anomalies through residual analysis, and generates a visualized diagnostic report including fuel quality judgment, anomaly type, and root cause analysis. This invention achieves closed-loop management of the entire process from source control and accurate metering to intelligent diagnosis.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for construction machinery, specifically to a system and method for monitoring and diagnosing fuel consumption of construction machinery based on multi-source information fusion. Background Technology

[0002] As the construction machinery industry rapidly develops towards intelligence and large-scale operations, large construction companies generally face challenges such as scattered equipment distribution, complex operating environments, and high management difficulty. As a core aspect of equipment operating cost control, fuel consumption management directly impacts the company's economic benefits and core competitiveness in terms of efficiency and accuracy.

[0003] Currently, fuel consumption management for construction machinery mainly relies on manual recording or simple remote monitoring systems, which have the following technical shortcomings: inaccurate fuel consumption measurement; traditional fuel level sensors are greatly affected by vehicle tilt angle, temperature, and fuel tank shape, making them unable to effectively detect fuel theft and minor leaks; uncontrolled fuel quality; construction machinery often operates in the field, with complex refueling sources. Inferior fuel can seriously damage engines, but current technology cannot perform real-time and effective quality checks during refueling, only allowing for retrospective investigation after equipment failure; superficial data analysis; existing remote monitoring platforms mostly only display raw fuel consumption and location data, lacking in-depth analytical capabilities. Managers cannot know the specific reasons for abnormal fuel consumption—whether it's equipment failure, improper operation, fuel quality issues, or fuel theft—making accurate decision-making difficult.

[0004] To address the shortcomings of the existing technologies, this invention provides a system and method for monitoring and diagnosing fuel consumption of engineering machinery based on multi-source information fusion. This system mitigates risks at the source, precisely controls consumption, and quickly identifies problems, providing enterprises with refined and proactive management support. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a system and method for monitoring and diagnosing fuel consumption of construction machinery based on multi-source information fusion. It solves the technical problems of inaccurate fuel consumption measurement, difficulty in monitoring fuel quality, and lack of intelligent anomaly diagnosis in the prior art, and realizes high-precision fuel consumption management and fuel quality control for construction machinery in all scenarios.

[0006] The technical solution adopted in this invention is to provide a fuel consumption monitoring and diagnosis system for engineering machinery based on multi-source information fusion, including an on-board terminal and a cloud platform server: the on-board terminal includes a main controller, and a refueling monitoring module, a working condition monitoring module and a communication module respectively connected to the main controller; the cloud platform server includes a data storage unit and a data analysis and diagnosis unit.

[0007] Preferably, the refueling monitoring module integrates a high-precision flow meter and a fuel quality sensor: the fuel quality multi-parameter sensor is used to collect fuel dielectric constant, viscosity and temperature parameters in real time; the high-precision flow meter accumulates the refueling volume through pulse counting.

[0008] Preferably, the operating condition monitoring module collects GPS positioning data, engine CAN bus data, and IMU acceleration data. The main controller classifies the equipment status into working, idling, or moving states based on the GPS positioning data, engine CAN bus data, and IMU acceleration data.

[0009] Preferably, the cloud platform server includes a data storage unit and a data analysis unit. The data analysis unit is used to parse refueling event data, extract fuel parameter features, and call the fuel quality model for evaluation. The data storage unit aggregates equipment operating data daily, constructs feature vectors, and inputs them into the fuel consumption prediction model to obtain predicted fuel consumption values. The actual fuel consumption is compared with the predicted fuel consumption value, and residual analysis is used to determine whether fuel consumption anomalies have occurred.

[0010] The evaluation process of the oil quality model includes: calculating the average dielectric constant and average temperature during the stable flow phase of a single refueling event; inputting the average values ​​into a pre-trained oil quality model to obtain an oil quality score; and determining that the oil is unqualified and generating an alarm when the relative deviation between the score and the standard value exceeds a preset threshold.

[0011] The fuel consumption prediction model is trained through the following steps: collect historical operating condition data and fuel consumption data of multiple normal equipment of the same model to construct a training dataset; construct feature vectors for daily data, including working duration, idling duration, moving duration, average working speed, average working hydraulic pressure, equipment model and ambient temperature; and use the gradient boosting regression tree algorithm to train the training dataset to obtain the fuel consumption prediction model.

[0012] Preferably, when the data analysis unit performs residual analysis, it calculates the residual between the actual fuel consumption and the predicted fuel consumption for the day. When the absolute value of the residual exceeds the dynamic threshold set based on the standard deviation of historical residuals, it is determined to be an abnormal fuel consumption.

[0013] When the data analysis unit determines that the fuel consumption is abnormally high, it also checks whether there are any records of substandard fuel during the abnormal period. If so, the diagnosis is that the fuel quality is a problem. If not, it analyzes the contribution of each operating condition characteristic to the predicted fuel consumption. If the contribution of the average pressure of the hydraulic system is abnormally high, the diagnosis is that the hydraulic system load is abnormal. If the contribution of the average engine speed is abnormally high, the diagnosis is that the operation is improper.

[0014] When an abnormal fuel loss is detected, the data analysis unit also checks whether there is a rapid drop in fuel level data during non-working periods. If so, it makes a judgment based on the GPS location at the time of the drop: if the location is not in the preset safe zone, the diagnosis is suspected fuel theft; if the location is in the preset safe zone, the diagnosis is suspected leakage.

[0015] This invention also provides a method for monitoring and diagnosing fuel consumption of engineering machinery based on multi-source information fusion, comprising:

[0016] The vehicle-mounted terminal simultaneously collects data on fuel flow, fuel parameters, and equipment operating conditions during refueling. The collected data is uploaded to the cloud platform server; Real-time assessment and alerts for oil quality are provided via cloud platform servers; The theoretical fuel consumption is predicted by a model trained on historical data using a cloud platform server. By comparing the model with the actual fuel consumption, residual analysis is performed to diagnose the root cause of abnormal fuel consumption.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. Real-time fuel quality testing during refueling effectively prevents equipment damage and efficiency loss caused by fuel issues, achieving a shift from passive maintenance to proactive protection.

[0018] 2. Direct metering via refueling pipeline provides a much higher accuracy than fuel level sensors, offering a reliable basis for cost accounting and effectively warning of fuel theft and leakage incidents.

[0019] 3. Through multi-dimensional data fusion algorithms, not only can anomalies be detected, but the root cause of the anomalies can also be accurately located, which greatly improves management efficiency and the level of management refinement.

[0020] 4. A single system solves multiple problems such as oil management, oil consumption measurement, equipment health management and safety monitoring, reducing the user's total cost of ownership. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is the overall system architecture diagram of the present invention; Figure 2 This is a flowchart of the fuel consumption monitoring process for the main controller of this invention; Figure 3 This is a flowchart of the intelligent diagnostic analysis platform of the present invention. Detailed Implementation

[0022] To better understand the purpose, system architecture, and functional implementation of this embodiment, the embodiments and features in the embodiments of this application can be combined with each other without conflict. The exemplary embodiments disclosed in this application will be described below with reference to the accompanying drawings, which include specific technical details disclosed in this embodiment to aid understanding; however, these details should be considered exemplary rather than restrictive.

[0023] Example 1 Figure 1 This is the overall system architecture diagram of the present invention.

[0024] like Figure 1 As shown, a fuel consumption monitoring and diagnostic system for engineering machinery based on multi-source information fusion consists of an on-board terminal and a cloud platform server.

[0025] The main controller of the vehicle terminal adopts an automotive-grade ARM Cortex-A53 processor with an operating temperature range of -40℃ to 85℃. It integrates a CAN bus controller, a 4G / 5G communication module, and an IMU interface. The vehicle terminal is fixedly installed in the cab or engine compartment of the construction machinery to ensure that the main controller and communication module are not affected by vibration or dust.

[0026] The refueling monitoring module is installed in series in the fuel tank inlet pipeline via a DN50 stainless steel flange, no more than 0.5m away from the refueling port to reduce pipeline oil capacity error. The module has a built-in volumetric flow meter, a wide-band dielectric constant sensor and a PT1000 temperature sensor to ensure that fuel flows only through the module during refueling. The sensor probe is fully immersed in the fuel flow channel to ensure detection accuracy.

[0027] In the operational condition monitoring module, the GPS positioning module uses an UBLOX F9P chip and is installed on the top of the equipment in an unobstructed position. The CAN bus interface reads engine speed, fuel injection quantity, and hydraulic main pressure via the SAE J1939 protocol. The IMU uses a six-axis sensor MPU-6050 with a sampling frequency of 100Hz. The communication module is an industrial-grade 4G / 5G module that supports the MQTT protocol, with a data upload cycle set to 10Hz. The IMU is installed at the center of gravity of the equipment body for accurate acquisition of equipment attitude and vibration data.

[0028] The cloud platform server is deployed on Alibaba Cloud ECS instances. The data storage unit uses the time-series database InfluxDB to store high-frequency sensor data, and the relational database to store device files and diagnostic reports. The data analysis and diagnostic unit constructs a personalized theoretical fuel consumption benchmark by intelligently comparing actual fuel consumption with predicted fuel consumption, and realizes fuel consumption anomaly detection and root cause tracing through residual analysis.

[0029] Figure 2 This is a flowchart of the working condition monitoring module of the present invention.

[0030] like Figure 2 As shown, the operating condition monitoring module continuously collects equipment operation data. The GPS module acquires location information such as the equipment's position, speed, and time to determine whether the equipment is moving and the operating area. The engine CAN bus interface reads data such as engine speed, fuel injection quantity, and hydraulic system pressure. The IMU monitors the equipment's attitude, vibration, and acceleration data. Data collected by sensors is transmitted to the main controller. After receiving the data, the controller performs data synchronization, filtering, and preliminary calculations. Based on data such as positioning speed, engine speed, and IMU acceleration, the equipment status is classified and judged. When both high engine load and a specific speed / attitude are met, the equipment is determined to be in working condition. When the engine speed is non-zero and the position / acceleration is close to stationary, the equipment is determined to be in idling condition. When the speed displayed by the positioning / IMU is non-zero and the engine load is present, the equipment is determined to be in moving condition. These three state branches eventually converge to generate a structured operating condition data packet with status tags, which is transmitted via the 5G communication module.

[0031] Figure 3 This is a flowchart of the intelligent diagnostic analysis platform of the present invention.

[0032] When the external refueling nozzle begins refueling, fuel flows through the refueling monitoring module. The flow meter starts pulse counting to accumulate the refueling volume, and the fuel quality sensor initiates high-frequency sampling to collect dielectric constant and temperature data. The main controller packages the refueling volume, dielectric constant sequence, temperature sequence, GPS location, and timestamp into a refueling data packet and transmits it via the 5G communication module, uploading it to the cloud platform server at a preset 10Hz interval.

[0033] After receiving refueling data packets and equipment operating condition data packets, the system performs data parsing and cleaning, processing data format, missing values, and outliers to ensure data quality. It then performs parallel processing in two branches: oil quality monitoring and oil consumption prediction analysis. The left branch extracts word-based refueling event data, calls a pre-trained model, and uses a rapid oil quality assessment algorithm to detect oil consumption. It removes air interference data from the initial and final stages, calculates the dielectric constant and temperature, inputs these values ​​into the oil quality model to obtain the oil quality score, and determines whether the oil quality is up to standard based on the relative deviation. The calculation method is shown in the following formula.

[0034] (1) in, Output oil quality to the model. For standard oil quality, when If the value is less than the preset threshold of 0.05, the model outputs an oil quality assessment indicating that the oil is substandard, triggering an alarm to alert management personnel. If the model outputs that the oil quality is qualified, it will be recorded as normal data. The cloud platform will record parameters such as the oil quality, refueling volume, and location of this refueling, and bind them to the equipment number for subsequent analysis or model optimization.

[0035] The right branch performs multi-dimensional fuel consumption anomaly diagnosis. It uses machine learning models to digitize and model the relationship between complex multi-dimensional operating conditions and fuel consumption, thereby generating a theoretical fuel consumption benchmark. It captures anomalies through residual analysis and performs intelligent inference to achieve accurate anomaly diagnosis and root cause analysis.

[0036] Model training is performed offline, on the cloud platform backend, to establish an accurate fuel consumption prediction model. Then, historical data is collected and cleaned from a large number of normal engineering machines of the same model to construct a training dataset. A data sample is constructed for each day's equipment operation data. Data cleaning is used to remove incomplete and abnormal data, ensuring that all training data comes from normal equipment. Feature engineering and feature vectors are constructed. Feature vectors are calculated for the time sample data of each operating state. The average engine speed is shown in equation (2).

[0037] (2) in, This represents the average engine speed under operating conditions. Let be the rotational speed of the i-th time segment. The duration of this event segment. This represents the total working hours for the day.

[0038] The average value of the main pressure of the hydraulic system is shown in the following formula.

[0039] (3) in, This represents the average value of the main pressure of the hydraulic system under operating conditions. For the principal pressure of the i-th time segment, The duration of this event segment. This represents the total working hours for the day.

[0040] The equipment model is processed using one-hot encoding, which collects the average ambient temperature of the day. The aforementioned feature vectors are then input into a Gradient Boosting Regression Tree (GBRT) algorithm for model training. The goal is to minimize the loss function between the predicted and actual values. After training, the model outputs a function; inputting the feature vectors of the day yields the predicted fuel consumption value. Finally, the prediction residuals on normal data are calculated using the validation set. As shown in the following formula.

[0041] (4) in, This represents the actual fuel consumption value. To predict fuel consumption, the standard deviation σ of the residual sequence is calculated, and an anomaly threshold of 3σ is set. Based on the predicted residual and the anomaly threshold, it is determined whether the fuel consumption is in an abnormal state. If the absolute value of the residual value on a given day is greater than the anomaly threshold, the fuel consumption on that day is determined to be abnormal. If the residual value on a given day is positive, it is determined to be excessive fuel consumption. If the residual value on a given day is negative, it is determined to be abnormal fuel loss.

[0042] For anomalies of excessively high fuel consumption, the refueling records for the current period are checked. If any fuel quality is assessed as substandard, the diagnosis is that poor fuel quality leads to low combustion efficiency. If the fuel is qualified, the model interpretation tool is used to analyze the contribution of each feature in the feature vector of the day to the high predicted value. If the analysis finds... The abnormally high feature contribution indicates that the predicted value is overestimated due to the continuous high load on the hydraulic system. If the specific value is significantly higher than the average value of similar equipment, the diagnostic conclusion is that the hydraulic system load is abnormally high, suspected to be due to internal leakage in the main pump or main valve, or improper operation. If the contribution is the highest and the value is relatively high, the conclusion is that improper operation caused the equipment to operate at high speed for a long time.

[0043] For abnormal fuel loss, check the fuel level sensor data for the day to see if there was a rapid drop in fuel level during the downtime. If the rapid drop in fuel level occurs outside of working hours and the GPS location is not in the regular parking area, the diagnosis is abnormal fuel reduction in an unsafe area, highly suspected to be fuel theft. If the rapid drop in fuel level occurs outside of working hours but the GPS location is in the regular parking area, the diagnosis is abnormal fuel reduction in a safe area, suspected to be fuel leakage, and management is advised to investigate immediately. The cloud platform automatically generates a diagnostic report containing the equipment number, date, and outputs a comparison curve of actual fuel consumption and predicted fuel consumption, residual value, anomaly type, and diagnostic conclusion, along with suggested measures. Finally, all the above quality alarms, diagnostic reports, and normal status updates are displayed in a visual interface, allowing maintenance personnel to intuitively monitor fuel quality and equipment fuel consumption status.

[0044] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A fuel consumption monitoring and diagnostic system for engineering machinery based on multi-source information fusion, characterized in that, Including in-vehicle terminals and cloud platform servers: The vehicle-mounted terminal includes a main controller, as well as a refueling monitoring module, a working condition monitoring module, and a communication module, which are respectively connected to the main controller; Cloud platform servers include data storage units and data analysis and diagnostic units.

2. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion as described in claim 1, characterized in that, The refueling monitoring module integrates a high-precision flow meter and a fuel quality sensor. The multi-parameter oil quality sensor is used to collect the dielectric constant, viscosity and temperature parameters of fuel in real time; The high-precision flow meter accumulates the amount of fuel dispensed by counting pulses.

3. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 1, characterized in that, The operating condition monitoring module collects GPS positioning data, engine CAN bus data, and IMU acceleration data; The main controller classifies the equipment status into working, idling, or moving states based on GPS positioning data, engine CAN bus data, and IMU acceleration data.

4. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 1, characterized in that, Cloud platform servers include data storage units and data analysis units: The data analysis unit is used to parse refueling event data, extract oil parameter characteristics, and call the oil quality model for evaluation; The data storage unit aggregates equipment operating data daily, constructs feature vectors, and inputs them into the fuel consumption prediction model to obtain the predicted fuel consumption value; By comparing the actual fuel consumption with the predicted fuel consumption, residual analysis is used to determine whether any abnormal fuel consumption has occurred.

5. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 4, characterized in that, The evaluation process for the oil quality model includes: Calculate the average dielectric constant and average temperature during the steady flow phase of a single refueling event; The average value is input into the pre-trained oil quality model to obtain the oil quality score; When the relative deviation between the score and the standard value exceeds a preset threshold, the oil is deemed unqualified and an alarm is generated.

6. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 4, characterized in that, The fuel consumption prediction model is trained through the following steps: Collect historical operating condition data and fuel consumption data from multiple normal devices of the same model to construct a training dataset; A feature vector is constructed from the daily data, and the features include working duration, idling duration, moving duration, average working speed, average working hydraulic pressure, equipment model and ambient temperature; The gradient boosting regression tree algorithm is used to train the training dataset to obtain the fuel consumption prediction model.

7. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 4, characterized in that, When the data analysis unit performs residual analysis, it calculates the residual between the actual fuel consumption and the predicted fuel consumption for the day. When the absolute value of the residual exceeds the dynamic threshold set based on the standard deviation of historical residuals, it is judged as abnormal fuel consumption. When the residual is positive, it is judged as an abnormally high fuel consumption. When the residual is negative, it is judged as an abnormal fuel loss.

8. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 7, characterized in that, When the data analysis unit is determined to be abnormally high fuel consumption, it is also used for: Check if there are any records of substandard oil products during the abnormal period. If so, the diagnosis is an oil quality problem. If not, analyze the contribution of each operating condition characteristic to the predicted fuel consumption. If the average pressure contribution of the hydraulic system is abnormally high, the diagnosis is that the hydraulic system is under abnormal load. If the contribution of the engine's average speed is abnormally high, the diagnosis is improper operation.

9. The engineering machinery fuel consumption monitoring and diagnosis system based on multi-source information fusion according to claim 7, characterized in that, When an abnormal fuel loss is detected, the data analysis unit is also used to check whether there is a rapid drop in fuel level data during non-working periods. If it exists, then determine the location based on the GPS position at the time of descent: If the location is not within the preset safe zone, the diagnosis is suspected oil theft; If the location is within the preset safe zone, the diagnosis is suspected leakage.

10. A method for monitoring and diagnosing fuel consumption of engineering machinery based on multi-source information fusion, characterized in that, include: The vehicle-mounted terminal simultaneously collects data on fuel flow, fuel parameters, and equipment operating conditions during refueling. The collected data is uploaded to the cloud platform server; Real-time assessment and alerts for oil quality are provided via cloud platform servers; The theoretical fuel consumption is predicted by a model trained on historical data using a cloud platform server. By comparing the model with the actual fuel consumption, residual analysis is performed to diagnose the root cause of abnormal fuel consumption.

Images