Instrument module based marine engine power monitoring system and method thereof

Abstract

The application provides an instrument module-based marine engine power monitoring system and method, relates to the technical field of power monitoring, and realizes real-time zero point drift suppression of a strain gauge type torque sensor through an environmental seal and a zero point automatic tracking device, realizes synchronous trigger collection of a coaxially installed auxiliary rotating speed induction probe through a corrected original torque voltage signal, realizes working condition synchronous comparison of a volumetric flowmeter reading in a fuel pipeline through the original torque voltage signal and a rotating speed pulse sequence, realizes cross-tracing determination of a sensor drift trend in a whole machine power monitoring link through a corresponding relationship between instantaneous fuel consumption and mechanical power, and obtains a fault attribution result, so that the application realizes accurate positioning of a sensor fault source, and improves long-term stability of power monitoring and accuracy of maintenance decisions.

Description

Technical Field

[0001] This invention proposes a marine engine power monitoring system and method based on an instrument module, relating to the field of power detection technology, specifically to the field of marine engine power monitoring technology based on an instrument module. Background Technology

[0002] The shaft power of a marine diesel engine is a core parameter for measuring the operating status of the power plant and assessing the ship's energy efficiency (such as EEXI). Its measurement accuracy directly affects the ship's operational economy and emission compliance. Currently, mainstream measurement methods include indirect estimation, strain gauge torque measurement, and non-contact magnetoelastic measurement. Indirect estimation infers power from fuel consumption, but is greatly affected by fuel quality and environmental conditions, resulting in poor real-time performance and significant errors. Strain gauge torque measurement requires attaching strain gauges to the drive shaft surface; its output characteristics are significantly affected by the high temperature, high humidity, and salt spray environment of the engine room. After long-term operation, aging of the strain gauge substrate material and creep of the adhesive cause continuous zero-point drift, leading to deviations from the true power measurement value. While non-contact magnetoelastic sensors avoid the problems of the pasting process, their installation location is limited and their resistance to vibration interference is insufficient. Existing power monitoring systems and fuel consumption measurements are independent. When power and fuel consumption mismatch occurs, it is impossible to distinguish whether the cause is torque sensor drift or flow meter deviation. Traditional internal consistency verification fails due to the lack of an independent reference benchmark. Summary of the Invention

[0003] This invention provides a marine engine power monitoring system and method based on an instrument module to solve the above-mentioned problems: This invention proposes a marine engine power monitoring system and method based on an instrument module, the method comprising: S1. By using an environmental sealing and zero-point automatic tracking device, the strain gauge torque sensor is subjected to real-time zero-point drift suppression to obtain the original torque voltage signal. S2. The coaxially mounted auxiliary speed sensing probe is synchronously triggered and acquired using the calibrated original torque voltage signal to obtain a speed pulse sequence; S3. By comparing the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence, the relationship between instantaneous fuel consumption and mechanical power is obtained. S4. By analyzing the relationship between instantaneous fuel consumption and mechanical power, cross-source analysis is performed on the sensor drift trend in the overall power monitoring link to obtain fault attribution results.

[0004] Furthermore, the system includes: The drift suppression module is used to suppress zero-point drift of the strain gauge torque sensor in real time through environmental sealing and automatic zero-point tracking device, so as to obtain the original torque voltage signal. The trigger acquisition module is used to synchronously trigger and acquire the coaxially mounted auxiliary speed sensing probe using the calibrated original torque voltage signal to obtain a speed pulse sequence; The synchronous comparison module is used to compare the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence to obtain the correspondence between instantaneous fuel consumption and mechanical power. The fault attribution module is used to cross-reference the sensor drift trend in the whole machine power monitoring link by the correspondence between instantaneous fuel consumption and mechanical power, and obtain the fault attribution result.

[0005] The beneficial effects of this invention are: it solves the technical problems of torque measurement and fuel consumption measurement being independent and unable to be cross-verified in traditional power monitoring, as well as the inability to quickly locate whether the problem lies in the torque channel or the fuel consumption channel when a fault occurs. It achieves a fully integrated process of torque signal acquisition, speed synchronization triggering, fuel consumption-power correlation, and cross-fault tracing. This improves the directionality of fault diagnosis, reduces the time cost for on-site maintenance personnel to troubleshoot sensor faults, and avoids incorrect replacement or calibration operations due to misjudgment. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of a marine engine power monitoring method based on an instrument module. Detailed Implementation

[0007] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0008] One embodiment of the present invention provides a method for monitoring the power of a marine engine based on an instrument module, the method comprising: S1. By using an environmental sealing and zero-point automatic tracking device, the strain gauge torque sensor is subjected to real-time zero-point drift suppression to obtain the original torque voltage signal after temperature compensation and zero-point correction. S2. The coaxially mounted auxiliary speed sensing probe is synchronously triggered and acquired using the calibrated original torque voltage signal to obtain the speed pulse sequence corresponding to the torque signal moment. S3. By comparing the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence, the corresponding relationship between instantaneous fuel consumption and mechanical power at the same time section is obtained. S4. By analyzing the correlation between instantaneous fuel consumption and mechanical power, cross-reference the sensor drift trends in the overall machine power monitoring chain to determine the root cause of faults that can distinguish between torque drift and flow meter deviation. Figure 1 As shown.

[0009] In the torque signal acquisition stage, strain gauges are attached to the surface of the drive shaft. As the shaft twists, their resistance changes. Four strain gauges are connected in a full-bridge configuration, converting the resistance change into a differential voltage output. Environmental sealing prevents the convective conduction of high engine compartment temperatures to the strain gauges through a sealed heat insulation layer encapsulated within the sensor housing. Simultaneously, silica gel or molecular sieves in the desiccant compartment adsorb moisture around the bridge circuit, reducing the corrosion rate of the sensitive grid by salt spray. The zero-point automatic tracking device includes a sample-and-hold circuit connected to the bridge circuit output. During engine shutdown, when the drive shaft is not under torque load, the strain gauge bridge circuit should theoretically output a zero-point voltage. This circuit samples the bridge circuit's no-load output value at a fixed frequency and stores this value in non-volatile memory as the current zero-point offset record. Upon the next power-on, the microprocessor reads this record from memory and subtracts it from the initial bridge circuit output value in real time, restoring the zero point to its true position.

[0010] During the power-on compensation process, the zero-point offset recorded at the current shutdown is subtracted from the zero-point value of the previous power-on calibration. Since a complete run-stop cycle occurs between the two calibrations, this difference mainly reflects the additional drift generated by the adhesive under the combined effects of stress release and substrate creep. If there is a difference between the current shutdown duration and the previous shutdown duration, the drift rate is not constant. For example, a longer shutdown time may lead to more complete stress release and a decrease in the rate of drift. Therefore, the ratio of the two shutdown durations is used as a weighting coefficient to correct the compensation amplitude. The pre-compensation zero-point value is obtained using an advanced prediction strategy. The microprocessor first queries the historical drift rate curve of the strain gauge at the same heating rate from the empirical database. This curve is obtained through previous bench test calibration. Then, the difference between the ambient temperature at the time of power-on and the stable operating temperature before shutdown is read. The two are multiplied to obtain the predicted value of the total additional drift during the heating stage, which is subtracted in reverse from the initial power-on no-load voltage. When the first complete rotation cycle arrives after power-on, the bridge output generated by the shaft rotating one revolution is sampled and the average value of each bridge arm is taken. This value is used as the dynamic calibration zero-position confirmation value, which is used to verify and fine-tune the aforementioned pre-compensation value. Finally, the offset is deducted from each sampling point based on the confirmed zero position.

[0011] In the speed acquisition and synchronization triggering stage, the original torque voltage signal is amplified and sent to a comparator for comparison with a preset synchronization trigger threshold, which is typically set to about one percent of the peak-to-peak value of the torque signal. When the signal crosses zero, the comparator flips to output a high level, triggering a timestamp counter to latch the current clock value. This timestamp serves as a precise marker of the torque zero-crossing point. Window-locked acquisition uses this timestamp as the center, with a preset duration before and after it as the acquisition window. Pulses from the magnetoelectric speed probe within the window are allowed to enter the subsequent processing link, while pulses outside the window are shielded by a gating circuit. The magnetoelectric speed probe utilizes the principle of electromagnetic induction. When the speed-measuring gear at the end of the drive shaft rotates, the tooth tip and tooth groove alternately pass through the probe's magnetic circuit, causing a change in magnetic reluctance. This induces an alternating electromotive force in the probe coil, the frequency of which is proportional to the gear speed. During the adaptive window adjustment process, several consecutive zero-crossing points are taken forward and backward from the current zero-crossing point as the reference. The time interval between these zero-crossing points is extracted and the standard deviation is calculated. This standard deviation quantifies the instantaneous torsional fluctuation amplitude of the drive shaft. The larger the fluctuation, the more uneven the intervals, and the larger the standard deviation. The microprocessor compares the standard deviation with the normal fluctuation reference value in the window configuration table, calculates the fluctuation amplification factor, and determines the scaling level according to the segment interval to which the factor belongs. The base window width is scaled step by step to obtain the coarse adjustment window width. Then, the number of incomplete pulse cycles truncated at the window boundary in the three most recent complete torsional cycles is counted to calculate the pulse truncation rate. This feedback value is used to perform unidirectional fine adjustment of the coarse adjustment window in the boundary direction. If the truncation rate is too high, the window is expanded unidirectionally; otherwise, it is contracted unidirectionally, until the truncation rate drops to an acceptable range, and finally the adaptive window width is obtained. After the window boundary is determined, the pulses falling within the boundary are sharpened into standard square waves by the shaping circuit, and then the amplitude is unified into a logic level that can be recognized by the digital circuit by the level conversion circuit, forming a standard square wave pulse sequence. The interval between adjacent pulses is latched by a hardware timer. The timer starts counting at the rising edge of the first pulse and stops counting at the rising edge of the second pulse. The latched count value multiplied by the timer clock period gives the actual duration. Finally, the pulses before and after the same torque zero-crossing point are assigned to different periods. Based on the zero-crossing point, it is determined whether each pulse period falls completely within the torsional period to which the zero-crossing point belongs. Pulses that cross the period boundary are discarded, and complete pulses are retained and synthesized into a speed pulse sequence that corresponds one-to-one with the torque signal time.

[0012] In the fuel consumption measurement and data fusion stage, the signal acquisition card contains a programmable clock generator. The microprocessor writes to the configuration register to lock the sampling clocks of the torque simulation channel and the speed frequency channel to the same frequency and phase, ensuring that they are matched to the same sampling time. After the paired data group is formed, the pulses output by the volumetric flow meter in the fuel supply line are accumulated within the same window, using the time window corresponding to the paired data as a reference. The volumetric flow meter uses a pair of meshing rotors that rotate as fuel flows through. Each time it rotates through a fixed volume, it outputs a pulse. The accumulated number of pulses multiplied by the volume corresponding to each pulse gives the volume of fuel flowing through the flow meter. A similar auxiliary flow meter is installed in the return line. Its reading represents the amount of fuel that is not burned by the engine and returns directly to the fuel tank. Subtracting the return amount from the supply amount gives the difference, which is the amount of fuel actually burned in the engine cylinder. The fuel temperature sensor typically uses a platinum resistance temperature detector, whose resistance value changes linearly with temperature. The microprocessor calculates the current fuel temperature by measuring the resistance value. The density correction factor is retrieved by using the oil temperature as an index to look up the density-temperature correction table for that fuel grade, which is obtained through laboratory oil sample analysis. Fuel density decreases as temperature increases. If the volumetric flow rate is directly used for power comparison, the error introduced by temperature fluctuations can reach several percentage points. Therefore, the volumetric flow rate must be multiplied by the correction factor at the corresponding temperature to convert it to the mass flow rate at the standard temperature. Finally, the mass flow rate is merged with the aforementioned paired data sets using timestamps. The acquisition card adds a timestamp label to each sampling point, and fuel consumption data is also added with timestamp labels. The microprocessor uses the timestamps as an index to align torque, speed, and fuel consumption at the same time segment, establishing a one-to-one correspondence between instantaneous fuel consumption and mechanical power at the same moment.

[0013] In the fault cross-source tracing stage, the preset method for the expected reference range of specific fuel consumption is as follows: During the engine's factory bench test or initial sea trial phase, the system collects a large amount of specific fuel consumption data at various typical operating points, i.e., combinations of different speeds and load rates. Specific fuel consumption is defined as hourly fuel consumption divided by output power, and its physical meaning is fuel consumption per unit of work, reflecting the engine's thermal efficiency. The mean and standard deviation range of specific fuel consumption under each operating condition are obtained through statistical processing and stored in the historical database. The actual specific fuel consumption is calculated by dividing instantaneous fuel consumption by mechanical power. The deviation direction and magnitude are obtained by comparing the actual specific fuel consumption value with the upper and lower limits of the expected reference range. If the actual value is higher than the upper limit, the deviation direction is positive and the magnitude is excessive; if it is lower than the lower limit, the deviation direction is negative and the magnitude is insufficient. The process of obtaining the zero-point drift cumulative state value is as follows: The zero-point offset at each shutdown time is retrieved from the data recording module and arranged into an array by time. The difference between two adjacent drift values ​​is calculated and divided by the interval to obtain the rate of change sequence. The three most recent rate values ​​are then used for linear trend extrapolation to obtain the current predicted value. Linear trend extrapolation assumes that the drift rate maintains a uniform change in the short term, and the future value is predicted based on the direction of the line connecting the three most recent points. The microprocessor then determines whether the drift process is in a uniform aging or accelerated aging stage: if the three most recent rate values ​​increase or decrease arithmetically, it is determined to be in a uniform stage, and first-order linear extrapolation is used; if the rate difference itself is continuously increasing, it is determined to be in an accelerated stage, and second-order curve extrapolation is used. Starting with the cumulative drift value at the most recent shutdown time, and using the current running time as the step size, the extrapolation model is substituted to obtain the predicted value of the cumulative drift at the current time. This predicted value is then compared in real-time with the initial value of the bridge circuit no-load voltage actually observed after the current startup. If the deviation between the predicted value and the actual observed value falls within a preset confidence interval, it is retained; otherwise, the predicted value is truncated and corrected. Finally, the corrected confidence interval value is normalized according to the proportion of the current running time within the current shutdown interval. For example, if the current running time accounts for a certain percentage of the entire shutdown interval, the drift amount corresponding to that proportion within the confidence interval is taken as the zero-point drift cumulative state value at the current time. The stability judgment value is obtained by checking the calibration coefficient change record of the fuel flow meter in the previous calibration cycle. The flow meter is periodically calibrated online using a standard measuring cylinder or standard flow source, and the calibration results are recorded in the form of calibration coefficients. If the calibration coefficient changes more than a preset threshold compared to the factory value, it is judged as unstable. In the comprehensive judgment process, the torque drift amplitude and the flow meter stability judgment value are each divided by their respective preset alarm thresholds to obtain two normalized percentage values. The purpose of normalization is to transform the two types of evidence with different physical dimensions and orders of magnitude to a comparable scale, facilitating comprehensive evaluation. The magnitude of the two normalized values ​​is compared to initially determine the dominant factor: if the torque drift normalized value is greater than the flow meter normalized value, the torque drift is initially determined to be the dominant factor; otherwise, the flow meter deviation is initially determined to be the dominant factor.Simultaneously, a deviation label for the specific fuel consumption is attached, recording whether the current actual specific fuel consumption is too high or too low. The microprocessor then reads the current engine load rate setting from the ship's propulsion control system. This setting represents the operator's target command for engine output power. Using this load rate as an index, the standard specific fuel consumption value calibrated by the factory bench test is retrieved. Different load rates result in different engine thermal efficiencies, and consequently, different standard specific fuel consumptions, thus obtaining the theoretically correct direction of the specific fuel consumption deviation at the current load level. For example, as the load rate increases, the combustion chamber pressure increases, and thermal efficiency usually improves; the standard specific fuel consumption should show a decreasing trend. If the actual deviation direction is an increase in specific fuel consumption, it contradicts the theoretical direction, indicating a sensor bias rather than normal load changes. The dominant factor is identified as the corresponding positive or negative direction of the deviation. A positive torque drift will cause the calculated power to be too high, and a negative torque drift will cause the power to be too low. A consistency comparison is performed with the actual specific fuel consumption deviation direction: if the direction is consistent and contradicts the theoretical direction, a penalty is applied to the reliability; if the direction is inconsistent, the interference of this factor is eliminated, and the reliability is preserved. Finally, based on the range of the weighted value, a high-confidence or low-confidence label is added to the initial dominant party identifier. If the weighted value is higher than the confidence limit, the result is high-confidence; if it is lower than the limit, it is low-confidence; if it is between the two, the initial judgment is maintained but an uncertain label is added, and a clear fault attribution result is output, that is, to determine whether the current abnormal power and fuel consumption is caused by torque sensor drift or flow meter deviation.

[0014] A strain gauge is a sensitive element that converts mechanical deformation into a change in electrical resistance. It consists of a metal resistive grid bonded to an insulating substrate. After being bonded to the surface of a drive shaft, the shaft undergoes torsional deformation under torque. The strain gauge is stretched or compressed accordingly, and its resistance value changes accordingly. By measuring the change in resistance, the strain on the shaft surface can be calculated, and then converted into torque. The sensitivity coefficient of a strain gauge is typically around 2.0, representing the rate of change of relative resistance per unit strain.

[0015] A full-bridge circuit is a resistive bridge consisting of four strain gauges connected in a specific manner. One strain gauge is attached to each of the four bridge arms; two are attached along the axis to sense axial strain, and the other two are attached perpendicular to the axis to sense lateral strain. During operation, the resistance changes in adjacent bridge arms in opposite directions, making the differential voltage amplitude of the bridge output several times that of a single strain gauge. The full-bridge configuration has temperature self-compensation characteristics; when changes in ambient temperature cause equal and in-direction changes in the resistance of all four bridge arms, the differential outputs cancel each other out, preventing the generation of false strain signals.

[0016] The zero-point automatic tracking device includes functional units such as a sample-and-hold circuit, a non-volatile memory, and microprocessor control logic. During engine shutdown, the sample-and-hold circuit acquires the bridge circuit's no-load output value at a fixed frequency. The microprocessor determines this value as the current zero-point offset and stores it in the non-volatile memory. Upon the next engine startup, this record is read from the memory and subtracted in real-time from the bridge circuit's initial output value using a subtraction operation.

[0017] The magnetoelectric speed probe is a speed sensor that operates on the principle of electromagnetic induction, consisting of a permanent magnet and an induction coil. When the speed-measuring gear mounted at the end of the drive shaft rotates, the tooth tip and tooth groove alternately pass through the probe's magnetic circuit, causing a periodic change in the magnetic reluctance in the magnetic circuit. This causes a corresponding change in the magnetic flux through the coil, inducing an alternating electromotive force (EMF) with a frequency proportional to the gear's rotational speed. The magnetoelectric probe requires no external power supply, and the output signal amplitude increases with increasing rotational speed, making it suitable for use in ship engine room environments.

[0018] Volumetric flow meter: A flow measurement device that uses a fixed-volume chamber for fluid measurement. It contains a pair of meshing rotors. As fuel flows through, the rotors rotate, and each rotation across a fixed volume outputs a pulse signal. The cumulative number of pulses multiplied by the calibrated volume corresponding to each pulse gives the volume of fuel flowing through the flow meter. This type of flow meter is insensitive to fluid viscosity, and its measurement accuracy is less affected by temperature, making it suitable for online monitoring of marine fuel consumption.

[0019] Platinum resistance temperature detectors utilize the linear relationship between the resistance of platinum metal and temperature. Typically, Pt100 type sensors are used, with a resistance of 100Ω at 0℃, increasing by approximately 0.385Ω for every 1℃ increase in temperature. A microprocessor applies an excitation current to the platinum resistance thermometer via a constant current source, measures the voltage drop across its terminals, calculates the resistance value using Ohm's law, and then uses the resistance-temperature characteristic curve to inversely calculate the current temperature value for temperature correction calculations related to fuel density.

[0020] Specific fuel consumption is defined as the hourly fuel consumption divided by the engine's output power, measured in grams per kilowatt-hour. Its physical meaning is the mass of fuel consumed per unit of power output, and it is a core indicator for measuring engine thermal efficiency. A lower specific fuel consumption value indicates higher engine thermal efficiency. Engine thermal efficiency varies under different load rates, and specific fuel consumption changes accordingly, typically reaching its lowest value in the 75% to 85% load rate range.

[0021] The calibration factor is a correction multiplier obtained at the factory or during periodic calibration of the flow meter. It is used as a correction factor to convert the flow meter pulse count into the actual volumetric flow rate. After a period of use, due to rotor wear or deposits, the actual volume corresponding to each pulse may change. By periodically performing online calibration using a standard measuring cylinder or standard flow source, the current calibration factor can be recalculated. The deviation between this value and the factory value or the previous calibration value is used as the basis for judging the stability of the flow meter's measurement channel.

[0022] Linear trend extrapolation is a mathematical method for predicting future trends based on historical data. It assumes that the variable being predicted changes at a constant rate in the short term, meaning the change remains constant within a unit of time. The slope of the line connecting the three most recent historical data points is calculated, and extending this line forward predicts the variable's value at future times. If the drift process exhibits accelerating characteristics, i.e., the difference between adjacent rates continuously increases, then second-order curve extrapolation is used. A quadratic polynomial is fitted to the three most recent data points, and the prediction is extended forward.

[0023] Normalization is a data processing method that transforms data with different physical dimensions and orders of magnitude to a unified comparable scale. Specifically, it involves dividing the current amplitude of each variable by its respective preset alarm threshold. The resulting ratios are all between 0 and 1, allowing the torque drift amplitude and the flow meter stability judgment value, which were originally not directly comparable, to be compared and weighted under the same benchmark.

[0024] The zero-crossing point is the moment when the voltage value in an AC signal waveform changes from positive to negative or from negative to positive, crossing the zero point. In torque measurement, the zero-crossing point of the AC voltage signal output by the strain gauge bridge circuit corresponds to the moment when the drive shaft rotates to a specific angle. This moment is highly deterministic and can be used as a time reference for synchronously triggering the acquisition of the speed signal.

[0025] The acquisition window is a time interval centered on the torque zero-crossing timestamp, extending forward and backward by a preset duration. Within this interval, the output pulses of the magnetoelectric speed probe are allowed to enter the subsequent processing chain, while pulses outside the interval are shielded by a gating circuit. The window width can be dynamically adjusted according to the degree of torsional fluctuation of the drive shaft to ensure that complete and valid speed pulses are acquired.

[0026] The timestamp is the moment value latched by a high-precision clock counter when a specific event occurs, recording the precise time of the event with a resolution of microseconds or nanoseconds. In synchronous acquisition, when the torque zero-crossing triggers the comparator to flip, the system latches the current counter value as the timestamp of that zero-crossing point. Subsequently, this timestamp is used as the anchor point to control the window acquisition of speed pulses and the multi-channel data alignment in data fusion.

[0027] The working principle and technical effect of the above technical solution are as follows: This method constructs a complete closed-loop monitoring link from torque signal acquisition to fault attribution. Real-time suppression of zero-point drift of the torque sensor is achieved through environmental sealing and zero-point tracking devices, ensuring the accuracy of the original torque signal. Using the corrected torque signal as a time reference, the speed probe is synchronously triggered to acquire data, ensuring precise alignment of torque and speed on the time axis and eliminating time misalignment errors caused by independent sampling. Torque, speed, and fuel flow meter readings are compared synchronously under operating conditions, establishing a physical correspondence between fuel consumption and power at a unified time cross-section. This correspondence is used to trace the source in reverse, cross-judging the possible drift trends of the torque sensor and flow meter in the power monitoring link to distinguish whether the fault source is torque drift or flow meter deviation.

[0028] This solution addresses the technical challenges of traditional power monitoring where torque and fuel consumption measurements are independent and cannot be cross-checked, as well as the difficulty in quickly identifying whether a fault originates in the torque or fuel consumption channel. It achieves a fully integrated process encompassing torque signal acquisition, synchronized speed triggering, fuel consumption-power correlation, and cross-fault tracing. This improves the accuracy of fault diagnosis, reduces the time cost for on-site maintenance personnel to troubleshoot sensor faults, and avoids incorrect replacement or calibration operations due to misjudgment.

[0029] In one embodiment of the present invention, S1 includes: By encapsulating a heat-insulating layer and a desiccant chamber inside the sensor housing, the strain gauges attached to the surface of the drive shaft are physically isolated, thus obtaining a strain gauge working environment that is less susceptible to corrosion from high temperature, high humidity, and salt spray in the cabin. By continuously sensing the temperature of the thermistor embedded in the strain gauge bridge circuit through the strain gauge working environment, an analog voltage signal reflecting the current real temperature of the strain gauge is obtained. By dynamically adjusting the resistance of the temperature compensation network connected in series at the output of the bridge circuit using a simulated voltage signal, a preliminary compensation signal is obtained to offset the spurious strain components caused by thermal expansion and contraction. By using the preliminary compensation signal to store the bridge and road no-load output values ​​collected during the downtime as a zero reference, the current zero offset at each downtime is recorded. By recording the current zero-point offset, the initial output value of the bridge circuit at the next startup is subtracted in real time to obtain the original torque voltage signal after temperature compensation and zero-point correction.

[0030] The working principle and technical effects of the above-mentioned technical solution are as follows: This method uses physical isolation to block the direct conduction of high temperature in the engine compartment to the strain gauges through a sealed heat insulation layer. The desiccant compartment absorbs moisture around the bridge circuit, slowing down the corrosion of the sensitive grid by salt spray, thus reducing the rate of drift at the source. A thermistor embedded near the bridge circuit continuously senses the temperature of the strain gauges, converting temperature changes into electrical signals for output, allowing the system to monitor the temperature status in real time. Based on the temperature signal fed back by the thermistor, the resistance of the temperature compensation network connected in series at the bridge circuit output is dynamically adjusted. The compensation voltage generated by this network is equal in amplitude and opposite in polarity to the spurious strain voltage caused by thermal expansion and contraction, thus canceling it out. During engine shutdown, when the shaft system is not under torque load and the bridge circuit should be at its theoretical zero position, the no-load output value of the bridge circuit is collected and stored as the zero-point offset, recording the drift state at each shutdown moment. Upon the next startup, the stored zero-point offset is deducted from the initial output value of the bridge circuit in real time, ensuring that the torque signal starting point returns to the true zero point each time the engine starts.

[0031] This invention solves the problem of continuous zero-point drift in strain gauge torque sensors after long-term operation in the high-temperature, high-humidity, and salt-spray environment of ship engine rooms, caused by temperature changes, adhesive creep, and substrate aging. It organically combines four suppression methods—physical isolation, temperature compensation, zero-point tracking during shutdown, and zero-point deduction during startup—to achieve all-time drift suppression. This improves the long-term stability of the sensor in harsh marine environments, reduces power measurement errors caused by accumulated zero-point drift, extends the effective service life of the sensor in the field, and reduces the maintenance workload of frequent disassembly and calibration.

[0032] In one embodiment of the present invention, the step of real-time subtraction of the bridge initial output value at the next power-on startup based on the current zero-point offset record includes: By comparing the difference between the current zero-point offset record and the previous power-on calibration zero-point value stored in the non-volatile memory, the additional drift increment value caused by strain gauge substrate creep and adhesive stress release during this shutdown is obtained. By assigning time weights to the current downtime and the previous downtime by adding drift increment values, a correction coefficient for drift rate change caused by extended downtime is obtained. By using the drift rate change correction coefficient, the initial no-load voltage of the bridge circuit read after the first power-on after this shutdown is predicted and corrected for drift trend, and a pre-compensation zero value is obtained after deducting the expected additional drift during the future power-on heating phase. By using the pre-compensated zero-point value, the average output value of each bridge arm collected during the first complete rotation cycle after power-on is updated and confirmed in real time to obtain the calibration zero-point confirmation value under the dynamic working condition of power-on. By calibrating the zero-point confirmation value, the instantaneous output value of the bridge circuit at the current operating moment is deducted from the difference at each sampling point to obtain the original torque voltage signal after temperature compensation and zero-point correction.

[0033] Specifically, by using a drift rate change correction coefficient, the initial no-load voltage of the bridge circuit read during the first power-on after this shutdown is corrected for drift trend prediction, resulting in a pre-compensated zero-point value that deducts the expected additional drift during the future power-on heating phase, including: By using the drift rate change correction coefficient, the historical drift rate curves of this type of strain gauge stored in the empirical database under the same heating rate are matched and queried to obtain the expected drift rate benchmark value corresponding to the unit temperature change during the start-up heating stage under the current shutdown duration. By using the expected drift rate benchmark value, the temperature difference between the ambient temperature at the start-up time and the stable operating temperature before shutdown is read to obtain the expected temperature rise required to rise from the current ambient temperature to the operating temperature. By multiplying the expected temperature rise value by the expected drift rate baseline value, the predicted total additional drift of the strain gauge bridge due to temperature rise during the period from power-on to thermal equilibrium establishment is obtained. By adding the total drift prediction value, the initial no-load voltage of the bridge circuit read on the first power-on is subjected to reverse pre-subtraction processing to obtain the pre-compensated zero initial value after subtracting the expected temperature rise drift. By using the pre-compensated zero-position initial value, the validity boundary of the bridge line linear working range in which the initial value is located is checked, and extreme values ​​that exceed the linear range are saturated and truncated to obtain the pre-compensated zero-position value after deducting the expected additional drift amount during the future start-up and heating phase.

[0034] The working principle and technical effect of the above technical solution are as follows: This method compares the difference between the zero-point offset recorded during the current shutdown and the zero-point value calibrated during the previous startup, and extracts the pure additional drift increment caused by the creep of the strain gauge substrate and the release of adhesive stress during the current shutdown. Then, the shutdown duration is introduced as a weighting factor. If the current shutdown time is significantly longer than the previous one, the drift rate may decrease due to the continuous release of stress. Based on this, a rate change correction coefficient is calculated to correct the compensation amplitude. On this basis, it is further extended forward: the temperature difference between the ambient temperature before startup and the stable operating temperature before shutdown is read. Combined with the historical drift rate curve of this strain gauge model, the total amount of additional drift caused by thermal expansion during the startup heating stage is estimated, and it is deducted in reverse from the bridge circuit no-load voltage at the first power-on, realizing advanced pre-compensation. When the first complete rotation cycle arrives after startup, the average value of the output of each bridge arm is actually collected to update and confirm the pre-compensation zero point. Finally, the offset is deducted point by point based on the confirmed calibration zero point.

[0035] This solution addresses the technical blind spot in conventional static zero-point subtraction schemes that fail to consider differences in downtime and anticipated drift during the start-up and warm-up phases. It represents an upgrade from simple historical drift compensation to pre-compensation for future trends. This improves the timeliness and foresight of zero-point correction, reduces transient measurement errors caused by thermal shock during the warm-up phase, and enables the acquisition of more accurate power data from the initial engine startup stage, avoiding the waiting time required for effective measurement after thermal equilibrium is established, as is common in traditional solutions.

[0036] In one embodiment of the present invention, S2 includes: The timestamp of the effective zero-crossing point of the torque signal is obtained by comparing the original torque voltage signal with the synchronization trigger threshold in the signal conditioning module. The output pulses of the magnetoelectric speed probe, which is coaxially mounted at the end of the drive shaft, are acquired by locking the window using timestamp markers, so as to obtain a speed pulse group that is only valid within a limited window before and after the timestamp. By shaping and level-converting the rising edge of each pulse using a rotational speed pulse group, a standard square wave pulse sequence with normalized amplitude is obtained. The time interval between adjacent pulses is timed and latched using a standard square wave pulse sequence to obtain the actual duration value corresponding to each pulse period. By determining and merging the pulses before and after the zero-crossing moment of the same torque signal using the actual duration value, a speed pulse sequence corresponding one-to-one with the torque signal moment is obtained.

[0037] The working principle and technical effect of the above technical solution are as follows: This method achieves precise time synchronization between torque and speed signals. The calibrated original torque voltage signal is sent to the signal conditioning module and compared with a preset synchronization trigger threshold. When the voltage waveform crosses zero, a trigger marker is generated, carrying precise arrival time information. Then, using this timestamp marker as an anchor point, a window is opened for acquisition within a limited range before and after the time axis, capturing only the output pulses of the magnetoelectric speed probe within the window and excluding irrelevant pulses outside the window. The rising edges of these pulses are then shaped into standard square waves, with uniform level amplitudes for subsequent counting, and the time interval between adjacent pulses is latched by hardware timing. Finally, the pulses before and after the same torque zero-crossing moment are assigned and merged.

[0038] This solution resolves the time asynchrony issue that existed when torque and speed signals were acquired through separate acquisition channels. By using the torque zero-crossing point as the synchronization anchor point for speed acquisition, precise alignment of the two types of signals on the time axis is achieved. This improves the accuracy of torque-speed pairing in power calculations and avoids instantaneous power fluctuations caused by time misalignment. It also reduces timing ambiguity in subsequent fuel consumption-power comparisons.

[0039] In one embodiment of the present invention, the step of window-locking acquisition of output pulses from a magnetoelectric speed probe coaxially mounted at the end of a drive shaft using timestamp marking includes: The standard deviation of the time interval between several consecutive zero-crossing points before and after the timestamp is calculated to obtain the quantified value of the instantaneous torsional fluctuation amplitude of the current drive shaft. The width of the preset fixed acquisition window is dynamically scaled and adjusted by quantization value to obtain an adaptive window width value that matches the current degree of torsional fluctuation. By symmetrically expanding the adaptive window width value forward and backward with the timestamp as the center, the effective speed pulse acquisition time window boundary on both sides of the torque zero crossing point is obtained. By using the effective rotational speed pulse acquisition time window boundary, all pulses output by the magnetoelectric rotational speed probe within this boundary are numbered and recorded one by one to obtain the original pulse group falling within this window. By eliminating incomplete pulse periods truncated at the window boundary from the original pulse group, a rotational speed pulse group that is valid only within the limited window before and after the timestamp and consists of complete periods is obtained.

[0040] This includes dynamically scaling and adjusting the preset fixed acquisition window width using quantized values ​​to obtain an adaptive window width value that matches the current degree of torsional fluctuation, including: By comparing the quantized value of the instantaneous torsional fluctuation amplitude of the current drive shaft with the normal fluctuation benchmark value stored in the window configuration table, the fluctuation amplification factor of the current fluctuation amplitude relative to the benchmark value is obtained. By using the fluctuation magnification factor, the segmented membership interval of the magnification factor is determined to obtain the window scaling level identifier corresponding to the current fluctuation intensity. By using the window scaling level indicator, the preset base window width value is scaled up and down step by step to obtain the coarse adjustment window width value corresponding to the current fluctuation level. By adjusting the coarse adjustment window width value, the integrity statistics of the pulse cycle at the window boundary within the most recent three complete torsion cycles are retrieved to obtain the pulse truncation rate feedback value used for fine adjustment of the coarse adjustment window boundary. By using the pulse cutoff rate feedback value, the coarse adjustment window width value is expanded or contracted in one direction along the boundary direction to obtain an adaptive window width value that matches the current degree of torsional fluctuation.

[0041] The working principle and technical effect of the above technical solution are as follows: This method adaptively optimizes the window locking acquisition. Centered on the torque zero-crossing timestamp, the time intervals of several consecutive zero-crossing points are extracted and their standard deviations are calculated. This standard deviation quantifies the instantaneous torsional fluctuation of the drive shaft under the current operating condition. The greater the fluctuation, the more uneven the time intervals between zero-crossing points, and the larger the standard deviation. The width of the acquisition window is dynamically adjusted based on this quantified value: when the fluctuation is severe, the window is appropriately widened to cover more effective pulses; when the fluctuation is gentle, the window is narrowed to improve focusing. After determining the window width, the acquisition boundary is symmetrically expanded forward and backward centered on the timestamp. All pulses within the boundary are numbered and recorded one by one. Finally, incomplete pulse periods truncated at the boundary are identified and removed, retaining only complete pulses.

[0042] This invention solves the technical problem that a fixed acquisition window cannot adapt to changes in the torsional fluctuation of the drive shaft under different operating conditions. It achieves adaptive adjustment of the acquisition strategy, improving acquisition accuracy through narrow window focusing under stable operating conditions, and ensuring no valid pulses are lost through wide window coverage under fluctuating operating conditions. This enhances the adaptability of speed pulse acquisition to complex operating conditions, reduces pulse counting errors introduced by window boundary truncation, and ensures the reliability of subsequent speed and frequency calculations.

[0043] In one embodiment of the present invention, S3 includes: By aligning and locking the dual-channel synchronous sampling clock in the signal acquisition card with the original torque voltage signal and speed pulse sequence, a pair of data sets of torque analog quantity and speed frequency quantity at the same sampling time are obtained. By accumulating the output pulses of the volumetric flow meter in the fuel supply line within the same operating window using paired data groups, the instantaneous reading of the fuel volumetric flow rate corresponding to the sampling time can be obtained. By subtracting the difference between the instantaneous reading of fuel volume flow rate and the reading of the auxiliary flow meter in the return oil line, the actual fuel consumption flow rate value after deducting the return oil volume is obtained. The density correction coefficient is queried and multiplied by the current oil temperature measured by the fuel temperature sensor based on the actual fuel consumption flow rate value to obtain the mass flow rate value converted to the standard temperature. By merging and aligning the mass flow rate value with the torque and speed signals in the aforementioned paired data set along the time axis, the correspondence between instantaneous fuel consumption and mechanical power at the same time segment can be obtained.

[0044] The working principle and technical effect of the above solution are as follows: This method achieves a corresponding correlation between fuel consumption and power at a unified time cross-section. First, the sampling clocks of the two channels, analog torque and speed frequency, are forcibly aligned and locked to ensure that they are paired at the same moment. Then, using the time window where the paired torque-speed data is located as the standard, the output pulses of the volumetric flow meter in the fuel supply line are accumulated within the same window to obtain the instantaneous reading of the fuel volumetric flow rate during that time period. Next, an auxiliary flow meter in the return line is introduced to subtract the return fuel volume from the fuel supply volume to obtain the actual fuel consumption of the engine. Then, based on the current fuel temperature measured by the fuel temperature sensor, the density-temperature correction coefficient of the fuel is looked up and multiplied to convert the volumetric flow rate into mass flow rate, because the fuel volume changes significantly with temperature, and mass flow rate is a stable indicator reflecting the true consumption. Finally, the mass flow rate is merged and aligned with the aforementioned paired data group on the time axis.

[0045] This solution addresses the problem of comparing different physical quantities (torque, speed, fuel consumption) at precisely the same moment due to differences in sampling frequency and clock source. By subtracting return oil and correcting for temperature and density, the interference of return oil circulation and temperature changes on fuel consumption measurement is eliminated. It achieves precise time alignment and unified physical units for fuel consumption and power, ensuring that comparative fuel consumption calculations truly reflect the energy conversion efficiency at the same moment.

[0046] In one embodiment of the present invention, S4 includes: By querying and retrieving the specific fuel consumption characteristic range under preset historical normal operating conditions through the correspondence between instantaneous fuel consumption and mechanical power, the expected reference range of specific fuel consumption under the current operating conditions can be obtained. By comparing the deviation of the actual instantaneous fuel consumption obtained by dividing the actual instantaneous fuel consumption by the mechanical power within the expected reference range, the direction and magnitude of the deviation of the current fuel consumption from the expected range can be obtained. By retrieving the historical trend of torque zero-point offset stored during the downtime period based on the direction and magnitude of the deviation, the cumulative state value of zero-point drift of the torque sensor is obtained. By comparing the cumulative zero-point drift state value with the direction and magnitude of the deviation, the calibration coefficient change records of the fuel flow meter in the previous calibration cycle are compared in parallel to obtain the stability judgment value of the fuel consumption measurement channel. By combining the stability judgment value with the cumulative state value of zero-point drift, the respective deviation contribution weights of the two are comprehensively judged to obtain a fault attribution result that can distinguish whether torque drift is the main cause or flow meter deviation is the main cause.

[0047] The working principle and technical effect of the above technical solution are as follows: This method constructs a dual-evidence cross-tracing judgment framework based on the deviation of fuel consumption. The expected reference range of fuel consumption under the current operating condition is queried from a preset historical normal operating condition database. Then, the actual instantaneous fuel consumption is divided by the mechanical power to obtain the actual fuel consumption value. This value is compared with the expected range to obtain the direction and magnitude of the deviation. A higher fuel consumption means that too much fuel is consumed and less power is produced, while a lower value indicates that more power is produced and less fuel is consumed. Next, the historical trend of torque zero-point offset stored during the shutdown period is retrieved to obtain the drift accumulation state of the torque sensor. Simultaneously, the calibration coefficient change record of the fuel flow meter in the previous calibration cycle is reviewed to determine the stability of the fuel consumption measurement channel. Finally, the torque drift state and the stability of the flow meter are used as two dimensions of evidence, and the deviation contribution weights of the two are comprehensively judged.

[0048] This solution addresses the technical challenge of distinguishing between torque sensor drift and flow meter deviation when power and fuel consumption mismatch occurs. It uses the deviation from fuel consumption as a comprehensive criterion, comparing historical trends in torque zero-point drift and flow meter calibration change records as dual evidence. This allows for clear attribution of the fault source, avoiding blindly replacing sensors or incorrectly adjusting calibration coefficients. It improves the accuracy of maintenance decisions and reduces the costs of over-maintenance and ineffective maintenance.

[0049] In one embodiment of the present invention, retrieving the historical trend of torque zero-point offset stored during the downtime period by means of deviation direction and magnitude includes: By analyzing the direction and magnitude of the deviation of the current specific fuel consumption from the expected range, the zero-point offset sequence of each shutdown time stored in the data recording module is retrieved to obtain an array of zero-point drift history trajectories arranged in chronological order. By calculating the drift difference between two adjacent shutdown records item by item using the zero-point drift history trajectory array, the change rate sequence of drift increment within each running interval is obtained. By extrapolating the rate of change of the three most recent shutdown records using the rate of change sequence, a short-term predicted extension value of the current drift accumulation can be obtained. By comparing and updating the short-term predicted extended value with the initial value of the bridge circuit no-load voltage actually observed after this power-on, the drift accumulation confidence interval reflecting the long-term aging process of the strain gauge and adhesive is obtained. The zero-point drift accumulation state value of the torque sensor at the current moment is obtained by normalizing and converting the drift accumulation confidence interval into the time proportion position of the current running time in this downtime interval.

[0050] Specifically, by extrapolating the rate of change of the three most recent shutdown records using a linear trend from the rate of change sequence, a short-term predicted extension of the current drift accumulation is obtained, including: By calculating the difference between two consecutive rate values ​​using the rate of change values ​​recorded in the three most recent shutdowns, the acceleration sequence value of the drift rate of change is obtained. By using the acceleration sequence values ​​of the drift change rate, the directional consistency of the acceleration sequence is determined, and the process characteristic identifier of whether the current drift process is in the uniform aging stage or the accelerated aging stage is obtained. By identifying process characteristics, a first-order linear extrapolation method is used when the process is determined to be in a uniform speed stage, and a second-order curve extrapolation method is used when the process is determined to be in an acceleration stage, so as to obtain an extrapolation mathematical model that matches the current aging process. Using an extrapolation mathematical model, starting with the cumulative drift value at the zero point of the most recent shutdown, and using the running time since the current startup as the extrapolation step size, the model is substituted to obtain the predicted value of the drift accumulation based on trend extension. By comparing the predicted value of the drift accumulation with the actual maximum drift change observed during the current runtime, and truncating and correcting the predicted value that exceeds the preset confidence interval, a short-term predicted extension value of the drift accumulation at the current time is obtained.

[0051] The working principle and technical effect of the above technical solution are as follows: This method performs trend analysis processing on the retrieval of the historical trend of torque zero-point drift. First, the zero-point offset stored at each shutdown time is retrieved from the data recording module and arranged in chronological order to form a drift history trajectory array. Then, the drift difference between two adjacent shutdown records is calculated item by item, divided by the corresponding running time interval to obtain the drift change rate sequence within each running interval. Based on this, the rate values ​​of the three most recent shutdown records are extracted, and the acceleration sequence is calculated from the adjacent rate differences. Based on this, it is determined whether the current drift is uniform aging or accelerated aging. For uniform aging, first-order linear extrapolation is used, and for accelerated aging, second-order curve extrapolation is used. Taking the cumulative drift value at the most recent shutdown time as the starting point, and using the current running time as the step size, it is substituted into the extrapolation model to obtain the predicted value of the cumulative drift at the current time. This predicted value is then compared and corrected with the actual observed value to obtain a confidence interval and normalize it to the current time.

[0052] This invention addresses the technical challenge of distinguishing between steady-state gradual change and accelerated deterioration of drift based solely on current single-point drift observations. By constructing historical trajectory arrays and calculating rate and acceleration sequences, it enables the identification of drift process stage characteristics. Upgrading static queries to dynamic trend analysis enhances the foresight of drift state assessment, allowing the system to provide early warning signals before drift accelerates and deteriorates, thus reducing the risk of sudden measurement failures.

[0053] In one embodiment of the present invention, the step of comprehensively judging the deviation contribution weights of the stability determination value and the zero-point drift cumulative state value includes: By normalizing the percentage of the current amplitude of each of the stability judgment value and the cumulative state value of torque zero-point drift relative to their respective preset alarm thresholds, the normalized contribution intensity values ​​of the two deviation sources are obtained. The relative ratios of the two are determined by normalizing the contribution intensity values ​​to obtain the initial deviation dominance identifier, and the identifier is then labeled with the current specific fuel consumption deviation direction. By using the initial deviation dominant party identifier and deviation direction label, the current engine load rate setting value is retrieved to obtain a load state reference for determining whether the deviation direction is consistent with the load change trend. By combining the physical correlation between the deviation direction and the positive and negative directions of torque drift with the load state reference, the confidence weighted value of the deviation contribution after excluding load fluctuation interference is obtained. By weighting the confidence level of the deviation contribution, the initial dominant party identifier is reconciled and finally confirmed, resulting in a fault attribution result that can distinguish between torque drift as the main cause and flowmeter deviation as the main cause.

[0054] The verification process involves combining load status references with the physical correlation between the deviation direction and the positive and negative directions of torque drift to obtain a weighted value for the reliability of the deviation contribution after excluding load fluctuation interference. This includes: By comparing the positive and negative directions of the deviation corresponding to the initial deviation dominant party identifier and the deviation direction label, the positive and negative directions of the deviation corresponding to the dominant party identifier are compared with the positive and negative directions of the deviation direction to obtain the direction consistency mark of whether the two directions are consistent or contradictory. By referencing the load condition, the standard specific fuel consumption value calibrated by the engine's factory bench test under this load rate is read, and the theoretically correct direction of the specific fuel consumption deviation under the current load level is obtained. By correctly pointing out the theoretical direction, and matching the current actual fuel consumption deviation direction label, the ratio of the current deviation that can be attributed to normal load changes to the non-attributable part is obtained. By combining the proportional relationship value with the aforementioned directional consistency marker, a joint weighted calculation is performed. When the directional consistency marker is consistent and the actual deviation is inconsistent with the theoretical direction, the credibility is penalized and deducted to obtain the preliminary credibility weighted value. By using the initial confidence weighted value, the confidence range of the weighted value is classified and evaluated. Based on the classification results, a confidence label with high or low confidence is added to the initial deviation leading party, thus obtaining the confidence weighted value of deviation contribution after excluding load fluctuation interference.

[0055] The working principle and technical effect of the above technical solution are as follows: This method performs refined weighted processing on the comprehensive discrimination of dual evidence. The current amplitude of the accumulated torque drift value and the flow meter stability judgment value are each normalized to a percentage with a preset alarm threshold, unifying the two types of evidence with different dimensions onto a comparable scale. Then, the normalized ratios of the two are compared to initially determine the dominant party of the deviation and attach a label indicating the direction of fuel consumption deviation. Next, the current engine load rate setting value is introduced as a third-dimensional reference. The standard fuel consumption calibrated on the factory bench under this load is queried to obtain the theoretically correct direction of the deviation. For example, fuel consumption usually decreases when the load increases. If the actual deviation direction does not match this, it indicates a sensor deviation rather than normal load change. The positive and negative directions of the deviation corresponding to the dominant party label are compared with the actual deviation direction. If the directions are consistent but do not match the theoretical direction, a penalty is applied to the credibility. Finally, based on the interval of the weighted value, a high-confidence or low-confidence label is attached to the initial dominant party label to complete the final judgment and confirmation.

[0056] This solution addresses the engineering challenge of distinguishing between sensor drift and normal load fluctuations in fuel consumption deviations during ship navigation under conditions of frequent load changes. By introducing load status references and physical correlation consistency checks, the normal impact of load changes on fuel consumption is separated from sensor deviations, avoiding false alarms caused by misinterpreting normal operating condition fluctuations as sensor malfunctions. This achieves more accurate and reliable fault attribution, reduces the interference of false alarms on crew operations, and improves the availability and reliability of the monitoring system.

[0057] According to one embodiment of the present invention, the system includes: The drift suppression module is used to suppress zero-point drift of the strain gauge torque sensor in real time through environmental sealing and automatic zero-point tracking device, and obtain the original torque voltage signal after temperature compensation and zero-point correction. The trigger acquisition module is used to synchronously trigger the coaxially mounted auxiliary speed sensing probe to acquire the speed pulse sequence corresponding to the moment of the torque signal using the calibrated original torque voltage signal. The synchronous comparison module is used to compare the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence to obtain the correspondence between instantaneous fuel consumption and mechanical power at the same time section. The source tracing and determination module is used to cross-trace and determine the sensor drift trend in the whole machine power monitoring link by the correspondence between instantaneous fuel consumption and mechanical power, and obtain fault attribution results that can distinguish between torque drift and flow meter deviation.

[0058] The working principle and technical effects of the above technical solution are as follows: This system constructs a complete closed-loop monitoring link from torque signal acquisition to fault attribution. Real-time suppression of zero-point drift of the torque sensor is achieved through environmental sealing and zero-point tracking devices, ensuring the accuracy of the original torque signal. Using the corrected torque signal as a time reference, the speed probe is synchronously triggered to acquire data, ensuring precise alignment of torque and speed on the time axis and eliminating time misalignment errors caused by independent sampling. Torque, speed, and fuel flow meter readings are compared synchronously under operating conditions, establishing a physical correspondence between fuel consumption and power at a unified time cross-section. This correspondence is used to trace the source in reverse, cross-judging the possible drift trends of the torque sensor and flow meter in the power monitoring link to distinguish whether the fault source is torque drift or flow meter deviation.

[0059] This system solves the technical problems of traditional power monitoring, where torque measurement and fuel consumption measurement are independent and cannot be cross-verified, as well as the inability to quickly pinpoint whether a fault originates in the torque or fuel consumption channel. It achieves a fully integrated process for torque signal acquisition, synchronized speed triggering, fuel consumption-power correlation, and cross-fault tracing. This improves the accuracy of fault diagnosis, reduces the time cost for on-site maintenance personnel to troubleshoot sensor faults, and avoids incorrect replacement or calibration operations due to misjudgment.

[0060] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring the power of a marine engine based on an instrument module, characterized in that, The method includes: S1. By using an environmental sealing and zero-point automatic tracking device, the strain gauge torque sensor is subjected to real-time zero-point drift suppression to obtain the original torque voltage signal. S2. The coaxially mounted auxiliary speed sensing probe is synchronously triggered and acquired using the calibrated original torque voltage signal to obtain a speed pulse sequence; S3. By comparing the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence, the relationship between instantaneous fuel consumption and mechanical power is obtained. S4. By analyzing the relationship between instantaneous fuel consumption and mechanical power, cross-source analysis is performed on the sensor drift trend in the overall power monitoring link to obtain fault attribution results.

2. The method for monitoring marine engine power based on an instrument module according to claim 1, characterized in that, S1 includes: By encapsulating a heat-insulating layer and a desiccant chamber inside the sensor housing, the strain gauges attached to the surface of the drive shaft are physically isolated, thus obtaining the working environment of the strain gauges. By continuously sensing the temperature of the thermistor embedded in the strain gauge bridge circuit through the strain gauge's working environment, an analog voltage signal is obtained. A preliminary compensation signal is obtained by dynamically adjusting the resistance of the temperature compensation network connected in series at the output of the bridge circuit using a simulated voltage signal. The zero-point offset record is obtained by storing the bridge and road no-load output values ​​collected during the downtime using the preliminary compensation signal and storing the zero-point reference. The original torque voltage signal is obtained by subtracting the initial output value of the bridge circuit in real time from the current zero-point offset record during the next power-on startup.

3. The method for monitoring marine engine power based on an instrument module according to claim 2, characterized in that, The step of real-time subtraction of the bridge's initial output value at the next power-on startup based on the current zero-point offset record includes: By comparing the difference between the current zero-point offset record and the previous power-on calibration zero-point value stored in the non-volatile memory, the additional drift increment value is obtained. By assigning time weights to the current downtime and the previous downtime by adding drift increment values, a drift rate change correction coefficient is obtained. By using the drift rate change correction coefficient, the initial no-load voltage of the bridge circuit read after the first power-on after this shutdown is predicted and corrected for drift trend, and the pre-compensated zero value is obtained. By using the pre-compensated zero-point value, the average output value of each bridge arm collected during the first complete rotation cycle after power-on is updated and confirmed in real time to obtain the calibration zero-point confirmation value. By calibrating the zero-point confirmation value, the instantaneous output value of the bridge circuit at the current operating moment is subtracted from the difference at each sampling point to obtain the original torque voltage signal.

4. The method for monitoring marine engine power based on an instrument module according to claim 1, characterized in that, S2 includes: The timestamp is obtained by comparing the original torque voltage signal with the synchronization trigger threshold in the signal conditioning module. The output pulses of the magnetoelectric speed probe, which is coaxially mounted at the end of the drive shaft, are acquired by locking the window using timestamp marking to obtain a speed pulse group. A standard square wave pulse sequence is obtained by shaping and level-converting the rising edge of each pulse using a rotation speed pulse group. The actual duration value is obtained by hardware timing and latching the time interval between adjacent pulses using a standard square wave pulse sequence. By determining and merging the pulses before and after the zero-crossing point of the same torque signal using the actual duration value, a speed pulse sequence is obtained.

5. The method for monitoring marine engine power based on an instrument module according to claim 4, characterized in that, The step of window-locking acquisition of output pulses from a magnetoelectric speed probe coaxially mounted at the end of a drive shaft using timestamp marking includes: The standard deviation of the time interval between several consecutive zero-crossing points before and after the timestamp is calculated to obtain the quantified value of the instantaneous torsional fluctuation amplitude of the current drive shaft. By dynamically scaling and adjusting the preset fixed acquisition window width using quantization values, an adaptive window width value can be obtained. By symmetrically expanding the adaptive window width value forward and backward with the timestamp as the center, the boundary of the effective rotational speed pulse acquisition time window is obtained. By using the effective rotational speed pulse acquisition time window boundary, all pulses output by the magnetoelectric rotational speed probe within this boundary are numbered and recorded one by one to obtain the original pulse group; The rotational speed pulse group is obtained by eliminating incomplete pulse periods truncated at the window boundary from the original pulse group.

6. The method for monitoring marine engine power based on an instrument module according to claim 1, characterized in that, S3 includes: By aligning and locking the dual-channel synchronous sampling clock in the signal acquisition card with the original torque voltage signal and speed pulse sequence, a paired data group is obtained; The output pulses of the volumetric flow meter in the fuel supply line are accumulated within the same operating window by matching data groups to obtain the instantaneous reading of the fuel volumetric flow rate. By subtracting the difference between the instantaneous reading of the fuel volume flow rate and the reading of the auxiliary flow meter in the return line, the actual fuel consumption flow rate is obtained. The mass flow rate is obtained by querying and multiplying the density correction coefficient of the current oil temperature measured by the fuel temperature sensor based on the actual fuel consumption flow rate; By merging and aligning the mass flow rate value with the torque and speed signals in the aforementioned paired data set along the time axis, the correspondence between instantaneous fuel consumption and mechanical power can be obtained.

7. The method for monitoring marine engine power based on an instrument module according to claim 1, characterized in that, S4 includes: By querying and retrieving the specific fuel consumption characteristic range under preset historical normal operating conditions through the correspondence between instantaneous fuel consumption and mechanical power, the expected reference range of specific fuel consumption under the current operating conditions can be obtained. By comparing the deviation of the actual instantaneous fuel consumption obtained by dividing the actual instantaneous fuel consumption by the mechanical power within the expected reference range, the direction and magnitude of the deviation are obtained. By retrieving the historical trend of torque zero-point offset stored during the downtime period based on the direction and magnitude of the deviation, the cumulative state value of zero-point drift can be obtained. By comparing the cumulative state value of zero-point drift with the direction and magnitude of deviation, the calibration coefficient change records of the fuel flow meter in the previous calibration cycle are compared in parallel to obtain the stability judgment value. By combining the stability judgment value with the zero-point drift cumulative state value, the deviation contribution weights of the two are comprehensively judged to obtain the fault attribution result.

8. The method for monitoring marine engine power based on an instrument module according to claim 7, characterized in that, The process of retrieving the historical trend of torque zero-point offset stored during the downtime period by analyzing the direction and magnitude of the deviation includes: By analyzing the direction and magnitude of the deviation of the current specific fuel consumption from the expected range, the zero-point offset sequence at each shutdown moment stored in the data recording module is retrieved to obtain the zero-point drift history trajectory array. The rate of change sequence is obtained by calculating the drift difference between two adjacent shutdown records item by item using the zero-point drift history trajectory array; By extrapolating the rate of change of the three most recent shutdown records using the rate of change sequence, a short-term forecast extension value is obtained. The drift accumulation confidence interval is obtained by comparing and updating the short-term predicted extended value with the initial value of the bridge no-load voltage actually observed after the start-up. The zero-point drift accumulation state value is obtained by normalizing and converting the drift accumulation confidence interval into the time proportion position of the current runtime in the downtime interval.

9. The method for monitoring marine engine power based on an instrument module according to claim 7, characterized in that, The method of comprehensively judging the deviation contribution weights of the stability determination value and the zero-point drift cumulative state value includes: By normalizing the percentage of the current amplitude of each of the stability judgment value and the cumulative state value of torque zero-point drift relative to their respective preset alarm thresholds, a normalized contribution intensity value is obtained. The relative ratios of the two are determined by normalizing the contribution intensity values ​​to obtain the initial deviation dominance identifier, and the identifier is then labeled with the current specific fuel consumption deviation direction. By using the initial deviation dominant party identifier and deviation direction label, the current engine load rate setting value is retrieved to obtain a load status reference. The deviation contribution credibility weighted value is obtained by verifying the consistency of the physical correlation between the deviation direction and the positive and negative directions of torque drift by combining the load state reference. The initial bias leader identifier is corrected and finally confirmed by weighting the bias contribution confidence value to obtain the fault attribution result.

10. A marine engine power monitoring system based on an instrument module, characterized in that, The system includes: The drift suppression module is used to suppress zero-point drift of the strain gauge torque sensor in real time through environmental sealing and automatic zero-point tracking device, so as to obtain the original torque voltage signal. The trigger acquisition module is used to synchronously trigger and acquire the coaxially mounted auxiliary speed sensing probe using the calibrated original torque voltage signal to obtain a speed pulse sequence; The synchronous comparison module is used to compare the readings of the volumetric flow meter in the fuel line with the original torque voltage signal and the speed pulse sequence to obtain the correspondence between instantaneous fuel consumption and mechanical power. The fault attribution module is used to cross-reference the sensor drift trend in the whole machine power monitoring link by the correspondence between instantaneous fuel consumption and mechanical power, and obtain the fault attribution result.

Images