Diesel generator fault detection method based on data analysis

Abstract

The invention discloses a diesel generator fault detection method based on data analysis, and particularly relates to the technical field of fault detection. High-order harmonic distribution standard ranges and temperature gradient standard temperature difference change ranges under different load levels are established based on historical data in a normal operation period, high-order harmonic and temperature difference data are collected and analyzed in real time, the accuracy of diesel generator fault detection is comprehensively evaluated, the diesel generator is divided into different grades according to the accuracy result, and the fault detection accuracy is improved. According to the method, the problem of fault detection deviation caused by equipment aging and sensor misalignment in the prior art is solved, and the reliability and timeliness of fault detection are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault detection, and particularly to a fault detection method for diesel generators based on data analysis. Background Art

[0002] With the improvement of the automation level of the power system, diesel generator sets, as standby power supply equipment, are widely used in the power supply guarantee of various important loads. To ensure the safe and stable operation of diesel generator sets, relevant protection and fault detection technologies are crucial. Among many protection technologies, pilot protection and differential protection, as key electrical protection means, have been widely studied and developed in practical applications. Pilot protection refers to the use of relays or protection devices arranged longitudinally between the generator and other equipment to detect the occurrence and location of faults by using the differences in electrical quantities such as current and voltage between two or more sampling points. This protection method can quickly respond to faults inside the generator or between the generator and external equipment, ensuring that faults can be isolated in time to prevent further expansion. Differential protection is mainly based on Kirchhoff's current law to judge faults by comparing whether the currents entering and leaving the protected area are balanced. Specifically for diesel generators, differential protection can effectively detect internal winding short circuits, phase-to-phase short circuits or grounding faults in the unit, and can more accurately reflect the types and locations of faults inside the generator, so as to quickly cut off the faults and ensure the safety of the generator.

[0003] The existing technologies have the following deficiencies:

[0004] During long-term operation, the internal equipment of diesel generators (such as current transformers, voltage transformers, etc.) will gradually age, and the accuracy of sensors will also decrease over time. This aging phenomenon will affect the basic data relied on by pilot protection and differential protection, increasing the deviation of fault detection results and even failing to effectively capture potential faults inside the generator at critical moments. At the same time, if the system overly relies on sensor data, sensor failures may cause serious data distortion, ultimately affecting the reliability of the entire protection system. Summary of the Invention

[0005] The purpose of the present invention is to provide a fault detection method for diesel generators based on data analysis to solve the deficiencies in the background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A fault detection method for diesel generators based on data analysis, including the following steps:

[0007] S1: Based on the historical data during the normal operation period, respectively establish the standard range of high-order harmonic distribution under different load levels in different operating conditions, and determine the standard temperature difference change range under different load levels according to the temperature gradient distribution of different parts of the generator;

[0008] S2: Collect the high - order harmonic components of current and voltage in real - time. By continuously collecting harmonic data within several operating cycles, analyze the abnormal upward trend of the high - order harmonic components; collect the temperature data of different parts of the internal equipment of the diesel generator in real - time through several temperature sensors, and judge the temperature difference fluctuation between different parts.

[0009] S3: Comprehensively analyze the abnormal upward trend of the high - order harmonic components and the temperature difference fluctuation between different parts to evaluate the accuracy of the diesel generator fault detection results.

[0010] S4: Divide the accuracy of the diesel generator fault detection results into different levels, which are divided into accurate detection results, possibly accurate detection results, and inaccurate detection results, and perform corresponding processing.

[0011] Preferably, in S2, after analyzing the abnormal upward trend of the high - order harmonic components, a high - order harmonic abnormal upward index is generated. The method for obtaining the high - order harmonic abnormal upward index is as follows:

[0012] Standardize the real - time collected data of the high - order harmonics. Taking the nth harmonic as an example, at a moment t, the amplitude of the nth harmonic is H n (t), and mark the standardized amplitude as Z n (t);

[0013] Use the moving average method to calculate the average value at M moments. The window size of the moving average is w; the expression is: is the moving average value of the nth harmonic at moment t, and Z n (i) is the standardized value of the nth harmonic at moment i; calculate the high - order harmonic abnormal upward index, and the expression is: In the formula, KM is the high - order harmonic abnormal upward index, and UCL n is the upper control limit of the nth harmonic, and σ n is the standard deviation of the harmonic standardized value.

[0014] Preferably, compare the obtained high - order harmonic abnormal upward index with the reference threshold of the high - order harmonic abnormal upward index set under normal conditions in the historical data. If the high - order harmonic abnormal upward index is greater than or equal to the reference threshold of the high - order harmonic abnormal upward index, it indicates that the abnormal degree of the upward trend of the high - order harmonic components is high. At this time, generate a high - order harmonic component abnormal signal; if the high - order harmonic abnormal upward index is less than the reference threshold of the high - order harmonic abnormal upward index, it indicates that the abnormal degree of the upward trend of the high - order harmonic components is low. At this time, generate a high - order harmonic component normal signal.

[0015] Preferably, after analyzing the temperature difference fluctuation between different parts, a temperature difference fluctuation index of the diesel generator is generated. The method for obtaining the temperature difference fluctuation index of the diesel generator is as follows:

[0016] The temperature differences between different parts are collected in real time through multiple temperature sensors of the diesel generator, including: the temperature difference between the stator winding and the housing: ΔT stator - housing(s); the temperature difference between the rotor winding and the stator winding: ΔT rotor - stator(s); the temperature difference between the inlet and outlet of the cooling system: ΔT cooling(s);

[0017] Determine the smoothing factor α. At the first time point s 0 The temperature difference fluctuation index EWMA 0 Is usually set to the initial temperature difference value, that is, EWMA 0 = ΔT 0 ; where ΔT 0 Is the first temperature difference data collected; starting from the second time point s 1 The temperature difference fluctuation index of the diesel generator is calculated recursively using the formula. The expression is: NG = α·ΔT s +(1 - α)·EWMA s-1 ; In the formula, NG is the temperature difference fluctuation index of the diesel generator, ΔT s Is the temperature difference data actually collected at time s, such as the temperature difference between the stator and the housing ΔT stator - housing(s), and EWMA s-1 Is the temperature difference fluctuation index at time s - 1.

[0018] Preferably, the obtained temperature difference fluctuation index of the diesel generator is compared with the reference threshold of the temperature difference fluctuation index of the diesel generator set under normal conditions. If the temperature difference fluctuation index of the diesel generator is greater than or equal to the reference threshold of the temperature difference fluctuation index of the diesel generator, it indicates that the temperature difference fluctuation between different parts of the diesel generator is large, and at this time, a high temperature difference fluctuation signal is generated; if the temperature difference fluctuation index of the diesel generator is less than the reference threshold of the temperature difference fluctuation index of the diesel generator, it indicates that the temperature difference fluctuation between different parts of the diesel generator is small, and at this time, a low temperature difference fluctuation signal is generated.

[0019] Preferably, in S3, the abnormal upward trend of the high - order harmonic components and the temperature difference fluctuation between different parts are comprehensively analyzed to evaluate the accuracy of the diesel generator fault detection result;

[0020] Convert the high - order harmonic abnormal rise index and the temperature difference fluctuation index of the diesel generator into the first eigenvector. Use the first eigenvector as the input of the machine - learning model. The machine - learning model takes the accuracy value label of predicting the diesel generator fault detection result for each group of the first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy value labels of all diesel generator fault detection results as the training target. Train the machine - learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the accuracy value of the diesel generator fault detection result according to the model output result. Among them, the machine - learning model is a polynomial regression model.

[0021] Preferably, in S4, divide the accuracy of the diesel generator fault detection result into different levels, divide it into accurate detection results, possibly accurate detection results, and inaccurate detection results, and perform corresponding processing. Specifically:

[0022] Compare the obtained accuracy value of the diesel generator fault detection result with the gradient standard thresholds. The gradient standard thresholds include the first standard threshold and the second standard threshold, and the first standard threshold is less than the second standard threshold. Compare the accuracy value of the diesel generator fault detection result with the first standard threshold and the second standard threshold respectively;

[0023] If the accuracy value of the diesel generator fault detection result is greater than the second standard threshold, it indicates that the accuracy of the diesel generator fault detection result is high. At this time, no warning signal is generated, and it is classified as an accurate detection result, indicating that the diesel generator is working normally and the system does not require further intervention;

[0024] If the accuracy value of the diesel generator fault detection result is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it indicates that the accuracy of the diesel generator fault detection result is average. At this time, generate a secondary warning signal to remind relevant personnel to pay attention to the status of the generator, and classify it as a possibly accurate detection result;

[0025] If the accuracy value of the diesel generator fault detection result is less than the first standard threshold, it indicates that the accuracy of the diesel generator fault detection result is low. At this time, generate a primary warning signal and classify it as an inaccurate detection result, indicating that the detection result is seriously inaccurate and repair measures should be taken immediately.

[0026] In the above technical solution, the technical effects and advantages provided by the present invention:

[0027] 1. The present invention collects and analyzes the high - order harmonic components and temperature data during the operation of a diesel generator, and respectively establishes the high - order harmonic distribution standards and temperature difference change standards under different load levels. By continuously monitoring the abnormal upward trends of these data, a high - order harmonic abnormal upward index and a temperature difference fluctuation index are generated, and these indicators are comprehensively analyzed through a machine learning model to evaluate the accuracy of the fault detection results, thereby improving the reliability and timeliness of detection. A polynomial regression model is used to train and predict the accuracy values of the fault detection results, reducing the detection errors caused by sensor aging or data distortion, and ensuring that the system can respond quickly in the event of potential faults.

[0028] 2. The present invention divides the accuracy of the detection results into different levels, sets reasonable gradient standard thresholds, which are divided into three levels: accurate, possibly accurate, and inaccurate, and correspondingly generates first - level or second - level warning signals to guide relevant personnel to take different handling measures. This warning mechanism not only improves the accuracy of fault detection, but also can prevent the occurrence of serious faults, ensure the stable operation of the diesel generator, reduce maintenance costs, and extend the service life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0030] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0032] Embodiment, please refer to Figure 1 As shown, the method for fault detection of a diesel generator based on data analysis in this embodiment includes the following steps:

[0033] S1: Based on the historical data during the normal operation period, respectively establish the high - order harmonic distribution standard ranges under different load levels in different operating conditions, and determine the standard temperature difference change ranges under different load levels according to the temperature gradient distribution of different parts of the generator;

[0034] S2: Collect the high - order harmonic components of current and voltage in real - time. By continuously collecting harmonic data within several operating cycles, analyze the abnormal upward trend of the high - order harmonic components. Real - time collect the temperature data of different parts of the internal equipment of the diesel generator through several temperature sensors, and judge the temperature difference fluctuation between different parts.

[0035] S3: Conduct a comprehensive analysis of the abnormal upward trend of the high - order harmonic components and the temperature difference fluctuation between different parts, and evaluate the accuracy of the diesel generator fault detection results.

[0036] S4: Divide the accuracy of the diesel generator fault detection results into different levels, classify them into accurate detection results, possibly accurate detection results, and inaccurate detection results, and conduct corresponding processing.

[0037] Among them, in S1, based on the historical data during the normal operation period, respectively establish the standard range of high - order harmonic distribution under different load levels in different operating conditions, and determine the standard temperature difference change range under different load levels according to the temperature gradient distribution of different parts of the generator.

[0038] High - order harmonics (usually 5th, 7th, 11th, etc.) can reflect the health status of the generator and its electrical system. Under different load levels, the distribution of high - order harmonics will have significant differences, so it is necessary to establish the corresponding standard range according to historical data.

[0039] Data collection: Use high - precision harmonic monitoring equipment to collect the current and voltage waveforms under different load levels for a long time during the normal operation of the generator. Focus on monitoring the amplitudes of the 5th, 7th, 11th and other key harmonics.

[0040] Load classification: Divide the operating conditions of the diesel generator into different intervals according to the load level. For example: light load: 0 - 30% rated load, medium load: 30 - 70% rated load, full load: 70 - 100% rated load, overload: more than 100% rated load;

[0041] Harmonic calculation: Conduct Fourier transform on the collected waveform data, extract the frequency components and amplitudes of each harmonic, and focus on the amplitude changes of the 5th, 7th, 11th harmonics.

[0042] Eliminate abnormal data: Statistically eliminate the collected noise data or abnormal points to ensure the data quality. Abnormal data may be caused by transient interference, power grid fluctuations or occasional equipment failures.

[0043] Statistical range for calculating harmonic amplitudes: For each type of harmonic (5th, 7th, 11th, etc.) under different load levels, based on long-term collected data, calculate its average value, standard deviation, and determine the range of its 95% confidence interval. This range is the standard distribution range of this harmonic under this load level. Harmonic range under light load: For example, under light load conditions, the amplitude of the 5th harmonic may be relatively low, and the standard range is 0.1% - 0.5% of the fundamental wave voltage or current. Harmonic range under medium load: Under medium load conditions, the harmonic content may be moderate, and the standard range of the amplitude of the 5th harmonic is 0.3% - 1.0%. Harmonic range under full load: At full load, the amplitude of the 5th harmonic is usually higher, and it may reach 0.5% - 1.5%. Harmonic range under overload: In the case of overload, the increase in harmonic components is more obvious, and the amplitude of the 5th harmonic may reach 1.0% - 2.5%.

[0044] Distribution standard diagram: Visualize the amplitude ranges of each harmonic under different load conditions as box plots or bar charts to facilitate observing the distribution differences of harmonics under different load levels.

[0045] Archiving and reference: Establish a table of standard distributions of high-order harmonics under each load level to provide reference for subsequent detection or fault analysis. This table can be used as a comparison standard in daily monitoring.

[0046] Temperature gradient, that is, the temperature difference between different parts inside the generator, reflects the health status of the thermodynamic state of the generator. As the load changes, the heat generation of the generator will increase significantly. Therefore, it is necessary to determine the range of temperature difference changes between key parts according to different load levels.

[0047] Temperature sensor arrangement: Install temperature sensors at different parts of the generator, with key positions including the following: Stator winding: The stator winding is one of the core heat-generating parts of the generator.

[0048] Rotor winding: Although the rotor winding is generally in a rotating state, its temperature can also reflect the heat generation inside the generator.

[0049] Generator housing: The temperature of the generator housing is usually relatively low, and monitoring the housing temperature can reflect the heat dissipation situation.

[0050] Cooling system inlet and outlet: Monitoring the temperature of the cooling medium (such as water or oil) helps analyze the effectiveness of the heat dissipation system.

[0051] Load classification: Similar to harmonic analysis, divide the load level into four intervals: light load, medium load, full load, and overload.

[0052] Temperature difference calculation: Real-time collect the temperature data of different parts during the operation of the generator, and focus on calculating the temperature difference between key parts. For example: Winding-housing temperature difference: The temperature difference between the stator winding and the generator housing reflects the internal heat generation and external heat dissipation capacity.

[0053] Cooling system temperature difference: The temperature difference between the inlet and outlet of the cooling medium can evaluate the operating efficiency of the cooling system.

[0054] Data cleaning and smoothing: Similarly, abnormal data in temperature collection needs to be eliminated to ensure accurate temperature difference calculation, and the temperature difference data is smoothed to eliminate errors caused by instantaneous fluctuations.

[0055] For the temperature difference data under different load levels, calculate the average temperature difference, standard deviation, and confidence interval to establish the standard temperature difference range under normal operating conditions. Temperature difference range under light load: Under light load conditions, the internal heat generation of the generator is less, and the temperature difference between the stator winding and the housing may be in the range of 5-10 °C.

[0056] Temperature difference range under medium load: Under medium load conditions, the temperature difference will increase appropriately and may reach 10-20 °C.

[0057] Temperature difference range under full load: Under full load conditions, the heat generated by the generator increases significantly, and the temperature difference between the stator and the housing may reach 20-30 °C.

[0058] Temperature difference range under overload: When the generator is operating under overload, internal overheating may cause the temperature difference to reach 30-40 °C or even higher.

[0059] By analyzing the temperature difference distribution under different load levels, draw a temperature difference change curve or box plot to show the fluctuation range of the temperature difference under different working conditions. According to the statistical results, formulate a standard temperature difference reference table for use in daily operation monitoring and equipment status assessment.

[0060] S2: Real-time collect the high-order harmonic components of current and voltage, analyze the abnormal upward trend of the high-order harmonic components by continuously collecting harmonic data within several operating cycles; real-time collect the temperature data of different parts of the internal equipment of the diesel generator through several temperature sensors to judge the temperature difference fluctuation between different parts.

[0061] Install high-precision current transformers and voltage sensors at key positions of the diesel generator to capture the waveform data of current and voltage in real time. Use a digital signal processor (DSP) or a data acquisition system to perform Fourier transform (FFT) to convert the time-domain signals of current and voltage into frequency-domain signals. Through Fourier analysis, the amplitudes and phases of specific harmonics can be extracted, with particular attention paid to higher harmonics such as the 5th, 7th, 11th, and 13th harmonics. To ensure the representativeness of the data, it is recommended to collect waveform data in real time at a high sampling rate (such as above 10 kHz) to ensure that sufficient detailed harmonic information is captured.

[0062] The collected current and voltage data may contain external electromagnetic interference or instantaneous burst noise. Use a low-pass filter or a band-pass filter to remove non-harmonic signals or high-frequency noise to ensure that the extracted harmonic signals are clean. Eliminate abnormal points during the acquisition process through statistical methods, such as transient abnormal data due to short-term power grid fluctuations or the startup phase of the equipment, as these data may cause analysis errors.

[0063] The operating cycle can be defined according to the operating conditions of the diesel generator and usually includes stages such as startup, stable operation, load change, and shutdown. An operating cycle may last for several hours or even days. To analyze the changing trend of the higher harmonic components, it is necessary to continuously monitor the data of multiple operating cycles. Usually, at least 3 - 5 cycles of harmonic data need to be continuously collected to observe the trend changes rather than accidental fluctuations. For each operating cycle, extract the amplitudes of the main higher harmonics such as the 5th, 7th, 11th harmonics, and time-series these amplitudes. Extract the higher harmonic amplitudes within each cycle to generate the harmonic amplitude curve for that cycle. To facilitate comparison between different cycles, the harmonic data can be normalized to convert the harmonic amplitudes into relative ratios (such as based on the fundamental wave amplitude).

[0064] After analyzing the abnormal rising trend of the higher harmonic components to generate the higher harmonic abnormal rising index, the method for obtaining the higher harmonic abnormal rising index is as follows:

[0065] Perform standardization processing on the real-time acquisition data of the higher harmonics. Taking the nth harmonic as an example, at a moment t, the amplitude of the nth harmonic is H n (t), and mark the standardized amplitude as Z n (t);

[0066] Use the moving average method to calculate the average value at M moments. The window size w of the moving average can be determined according to the specific application scenario and is usually 5 - 10 sampling moments; the expression is: is the moving average value of the nth harmonic at moment t, Z n(i) The normalized value of the nth harmonic at time i; calculate the high - order harmonic abnormal rise index, and the expression is: In the formula, KM is the high - order harmonic abnormal rise index, and σ n is the standard deviation of the harmonic normalized value, and UCL n is the upper control limit (Upper Control Limit) of the nth harmonic, which is usually set based on historical data. Generally set as: UCL n = μ n + 3σ n ; Here, 3σ n means that in 99.7% of the cases, the harmonics during normal operation should fall within three standard deviations of the mean value. If the moving average exceeds the upper control limit UCL n , the abnormal rise index will increase. If it does not exceed, this item is zero.

[0067] Compare the obtained high - order harmonic abnormal rise index with the reference threshold of the high - order harmonic abnormal rise index set under normal conditions in historical data. If the high - order harmonic abnormal rise index is greater than or equal to the reference threshold of the high - order harmonic abnormal rise index, it indicates that the abnormal degree of the rising trend of the high - order harmonic component is high, and at this time, a high - order harmonic component abnormal signal is generated; if the high - order harmonic abnormal rise index is less than the reference threshold of the high - order harmonic abnormal rise index, it indicates that the abnormal degree of the rising trend of the high - order harmonic component is low, and at this time, a high - order harmonic component normal signal is generated.

[0068] When the high - order harmonic abnormal rise index increases significantly, it means that the amplitude of the high - order harmonics continuously exceeds its normal control range at multiple consecutive time points. This situation indicates that the abnormal rise of the harmonic component is no longer a short - term fluctuation, but has formed an obvious and gradually deteriorating trend. This may imply the following problems:

[0069] Equipment aging: The aging or damage of electrical equipment (such as generator windings, instrument transformers, etc.) will lead to an enhanced non - linear characteristic of the system, thereby causing a significant increase in the high - order harmonic content.

[0070] Electrical faults: The continuous rise of high - order harmonics may also mean the existence of internal electrical faults, such as winding short - circuits, insulation failures, etc. These faults will exacerbate the distortion of voltage and current waveforms, resulting in a significant deviation of the harmonic components from the normal level.

[0071] Load imbalance or overload: Long - term load imbalance or overload operation will lead to an increase in harmonics. Especially in the case of drastic changes in generator load, a significant increase in harmonic content is a sign of abnormal operating conditions.

[0072] When the high - order harmonic abnormal rise index remains small or even close to zero, it indicates that the fluctuation of the high - order harmonic components in the system is within the normal range and there is no significant abnormal rising trend. At this time, the change of harmonics belongs to the regular fluctuation in daily operation and will not trigger any risk of equipment failure or performance degradation. This situation may occur in the following scenarios:

[0073] Normal operation: The equipment operates under normal load or working conditions, the voltage and current waveforms remain stable, and the high - order harmonic components are maintained within the historical baseline range without significant deviation or abnormal rise.

[0074] Short - term fluctuation correction: Even if there is a slight rise in harmonics in some short periods, as long as this fluctuation quickly returns to the normal range, the high - order harmonic abnormal rise index will still remain small, indicating that there is no long - term trend - like fault in the system.

[0075] Stable load: The generator load remains stable, the harmonic pollution in the power grid is low, and there is no cumulative rise in harmonic components.

[0076] The temperature data of different parts of the internal equipment of the diesel generator is collected in real - time through several temperature sensors to judge the temperature difference fluctuation between different parts. Specifically:

[0077] Install temperature sensors at key parts of the diesel generator to collect the temperature data of these areas in real - time. Usually, the following key parts should be selected:

[0078] Stator winding: This is the place where the most heat is generated inside the diesel generator, directly reflecting the working state and load condition of the generator.

[0079] Rotor winding: Although the rotor is rotating, the change of its temperature will also reflect the fluctuation of electromagnetic performance.

[0080] Generator housing: The temperature of the housing reflects the heat dissipation condition of the generator and has a direct relationship with the internal heat conduction.

[0081] Cooling system inlet and outlet: By measuring the temperature difference between the inlet and outlet of the cooling medium (such as oil or water), the efficiency of the cooling system can be evaluated.

[0082] Other parts that may overheat: Such as generator bearings or terminal blocks, etc. These areas are also prone to heat under long - term high load.

[0083] Data acquisition frequency: The temperature sensors should be set to a high acquisition frequency to ensure capturing the dynamic changes of the temperature of each part. Usually, it is recommended to collect data once a minute, and it can be adjusted specifically according to the operating speed and temperature change of the equipment.

[0084] Data storage: Transmit the temperature data collected by each sensor to the monitoring system to form complete time-series data. The data at each moment s can be expressed as: Ti(s); where Ti(s) is the temperature value recorded by the i-th temperature sensor at moment s.

[0085] By comparing the temperature data of different parts, calculate the temperature difference fluctuation between key parts. Typical temperature difference calculations can include the following combinations:

[0086] The temperature difference between the stator winding and the housing: ΔTstator - housing(s) = Tstator(s) - Thousing(s); This temperature difference reflects the balance between the heat generated inside the generator and the external heat dissipation capacity. If the temperature difference is too large, it may indicate overheating inside the generator or insufficient heat dissipation.

[0087] The temperature difference between the rotor winding and the stator winding: ΔTrotor - stator(s) = Trotor(s) - Tstator(s); This temperature difference can be used to analyze the changes in the electromagnetic performance of the generator. If the temperature difference fluctuates violently, it may indicate abnormalities in the electromagnetic field.

[0088] The temperature difference between the inlet and outlet of the cooling medium: ΔTcooling(s) = Tcooling outlet(s) - Tcooling inlet(s); This temperature difference reflects the efficiency of the cooling system. If the temperature difference of the cooling medium is small, it indicates good cooling effect; if the temperature difference increases, it may be due to a decrease in the efficiency of the cooling system or a failure in the cooling circuit.

[0089] After analyzing the temperature difference fluctuation between different parts, generate the temperature difference fluctuation index of the diesel generator. The method for obtaining the temperature difference fluctuation index of the diesel generator is as follows:

[0090] Real-time collect the temperature differences between different parts through multiple temperature sensors of the diesel generator, including: the temperature difference between the stator winding and the housing: ΔTstator - housing(s); the temperature difference between the rotor winding and the stator winding: ΔTrotor - stator(s); the temperature difference between the inlet and outlet of the cooling system: ΔTcooling(s);

[0091] Determine the smoothing factor α. α is a parameter that controls the weight between historical data and new data, and its value range is 0 < α < 1. The larger α is, the more sensitive the EWMA is to new data. Usually, α is taken between 0.1 and 0.3 and can be adjusted according to specific situations: α is larger (such as 0.3): suitable for scenarios with fast response, suitable for real-time detection of sensitive temperature difference changes. α is smaller (such as 0.1): suitable for relatively stable scenarios, suitable for tracking long-term trends and reducing the influence of noise.

[0092] At the first time point s 0 the temperature difference fluctuation index EWMA 0 is usually set to the initial temperature difference value, that is, EWMA 0 = ΔT0 ; where ΔT 0 is the first temperature difference data collected; starting from the second time point s 1 the temperature difference fluctuation index of the diesel generator is calculated recursively using the formula, and the expression is: NG = α·ΔT s +(1 - α)·EWMA s-1 ; in the formula, NG is the temperature difference fluctuation index of the diesel generator, ΔT s is the temperature difference data actually collected at time s, such as the temperature difference ΔT stator - housing(s) between the stator and the housing, and EWMA s-1 is the temperature difference fluctuation index at time s - 1, representing the index value of the previous time point.

[0093] Compare the obtained temperature difference fluctuation index of the diesel generator with the reference threshold of the temperature difference fluctuation index of the diesel generator set under normal conditions. If the temperature difference fluctuation index of the diesel generator is greater than or equal to the reference threshold of the temperature difference fluctuation index of the diesel generator, it indicates that the temperature difference fluctuation between different parts of the diesel generator is large, and at this time, a high temperature difference fluctuation signal is generated; if the temperature difference fluctuation index of the diesel generator is less than the reference threshold of the temperature difference fluctuation index of the diesel generator, it indicates that the temperature difference fluctuation between different parts of the diesel generator is small, and at this time, a low temperature difference fluctuation signal is generated.

[0094] S3: Comprehensively analyze the abnormal upward trend of the high - order harmonic components and the temperature difference fluctuation between different parts to evaluate the accuracy of the diesel generator fault detection result.

[0095] Convert the high - order harmonic abnormal rise index and the temperature difference fluctuation index of the diesel generator into a first feature vector, and use the first feature vector as the input of the machine learning model. The machine learning model takes the prediction target as the accuracy value label of the diesel generator fault detection result predicted by each group of first feature vectors, and takes minimizing the sum of the prediction errors of all diesel generator fault detection result accuracy value labels as the training target. Train the machine learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the accuracy value of the diesel generator fault detection result according to the model output result, where the machine learning model is a polynomial regression model.

[0096] The method for obtaining the accuracy value of the diesel generator fault detection result is: from the first feature vector training data of the trained machine learning model, obtain the corresponding function expression: HK = F(KM, NG); in the formula, F is the output function of the model, KM is the high - order harmonic abnormal rise index, NG is the temperature difference fluctuation index of the diesel generator, and HK is the accuracy value of the diesel generator fault detection result.

[0097] S4: Divide the accuracy of the diesel generator fault detection results into different levels, classify them into accurate detection results, possibly accurate detection results, and inaccurate detection results, and perform corresponding processing.

[0098] Compare the accuracy value of the obtained diesel generator fault detection results with the gradient standard thresholds. The gradient standard thresholds include the first standard threshold and the second standard threshold, and the first standard threshold is less than the second standard threshold. Compare the accuracy value of the diesel generator fault detection results with the first standard threshold and the second standard threshold respectively;

[0099] If the accuracy value of the diesel generator fault detection results is greater than the second standard threshold, it indicates that the accuracy of the diesel generator fault detection results is high. At this time, no warning signal is generated, and it is classified as an accurate detection result, indicating that the diesel generator is working normally and the system does not require further intervention;

[0100] If the accuracy value of the diesel generator fault detection results is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it indicates that the accuracy of the diesel generator fault detection results is average. At this time, a secondary warning signal is generated to remind relevant personnel to pay attention to the status of the generator, and it is classified as a possibly accurate detection result;

[0101] If the accuracy value of the diesel generator fault detection results is less than the first standard threshold, it indicates that the accuracy of the diesel generator fault detection results is low. At this time, a primary warning signal is generated, and it is classified as an inaccurate detection result, indicating that the detection result is seriously inaccurate and there may be a fault in the system or equipment, and immediate measures should be taken.

[0102] It should be noted here that the importance level of the primary warning signal is greater than that of the secondary warning signal, and relevant personnel can take corresponding processing measures according to different warning signal levels.

[0103] Processing measures for the primary warning signal:

[0104] Triggering condition: When the accuracy value of the detection results is less than the first standard threshold.

[0105] Importance: The primary warning signal indicates that there may be a serious fault and immediate emergency measures should be taken.

[0106] Processing measures: The operation and maintenance personnel should quickly conduct a comprehensive inspection of the diesel generator to find possible fault causes, such as electrical problems, mechanical damage, or cooling system failures. Check the working status of the sensors and data acquisition system to ensure the accuracy of the data. If necessary, shut down the generator set to prevent further damage.

[0107] Processing measures for the secondary warning signal:

[0108] Trigger condition: When the accuracy value of the detection result is between the first standard threshold and the second standard threshold.

[0109] Importance: The secondary warning signal indicates that the system operation status is in a warning state, and there may be minor anomalies or potential risks.

[0110] Handling measures: The operation and maintenance personnel should closely monitor the operation status of the diesel generator and conduct regular inspections. Record the changes in the temperature difference fluctuation index and the abnormal rise index of high-order harmonics. Perform equipment maintenance and debugging when necessary. Observe whether there are further signs of deterioration in the system to prevent it from developing into a primary warning situation.

[0111] It should be noted here that by comparing the accuracy value of the diesel generator fault detection result with the gradient standard threshold, the detection result level can be effectively divided, and corresponding warning signals can be generated according to the detection result. The primary warning signal indicates a serious problem that needs to be dealt with immediately, while the secondary warning signal indicates that there may be minor faults that need to be closely monitored. Precise threshold setting and effective warning mechanisms can significantly improve the reliability of fault detection and the safety of the system.

[0112] In this embodiment, based on the historical data during the normal operation period of the diesel generator, first establish the standard range of high-order harmonic distribution and the standard temperature difference change range of the temperature gradient at each part under different load levels. Subsequently, collect the high-order harmonic components of current and voltage and the temperature data at different parts in real time, and analyze the abnormal rise trend of high-order harmonics and the temperature difference fluctuation through multi-cycle continuous collection. Then, comprehensively analyze the abnormal trend of high-order harmonics and the temperature difference fluctuation to evaluate the accuracy of the fault detection result. Finally, according to the accuracy result, divide the fault detection into three categories: accurate, possibly accurate, and inaccurate, and generate corresponding warning signals according to the preset threshold standard to guide the adoption of further handling measures.

[0113] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0114] Those of ordinary skill in the art can realize that, combining the units and algorithm steps of each example described in the embodiments disclosed herein, can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0115] The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. A diesel generator fault detection method based on data analysis, characterized in that: The steps include: S1: Based on the historical data of the normal operation period, the standard range of high-order harmonic distribution under different operating conditions and different load levels is established, and the standard temperature difference range under different load levels is determined according to the temperature gradient distribution of different parts of the generator; S2: Real-time collection of high-order harmonic components of current and voltage, and analysis of abnormal rising trends of high-order harmonic components by continuously collecting harmonic data within several operating cycles; real-time collection of temperature data of different parts of the internal equipment of the diesel generator through several temperature sensors to determine the temperature difference fluctuations between different parts; S3: Comprehensively analyze the abnormal rising trend of high-order harmonic components and the temperature difference fluctuations between different parts to evaluate the accuracy of diesel generator fault detection results; S4: Divide the accuracy of the diesel generator fault detection results into different levels, such as accurate detection results, possibly accurate detection results and inaccurate detection results, and perform corresponding processing.

2. The diesel generator fault detection method based on data analysis according to claim 1 is characterized in that: In S2, the abnormal rising trend of the high-order harmonic components is analyzed to generate the abnormal rising index of the high-order harmonics. The method for obtaining the abnormal rising index of the high-order harmonics is: Standardize the real-time data of high-order harmonics. Take the nth harmonic as an example. At a time t, the amplitude of the nth harmonic is H n (t), the amplitude is normalized and marked as Z n (t); Use the sliding average method to calculate the average value of M moments, and the sliding average window size is w; the expression is: is the sliding average value of the nth harmonic at time t, Z n (i) Normalized value of the nth harmonic at time i; Calculate the abnormal rise index of high-order harmonics, the expression is: Where KM is the abnormal rise index of high-order harmonics, UCL is n is the upper control limit of the nth harmonic, σ n is the standard deviation of the harmonic normalized values.

3. The diesel generator fault detection method based on data analysis according to claim 2 is characterized in that: The acquired abnormal rising index of high-order harmonics is compared with the reference threshold of abnormal rising index of high-order harmonics set under normal conditions in historical data. If the abnormal rising index of high-order harmonics is greater than or equal to the reference threshold of abnormal rising index of high-order harmonics, it indicates that the abnormal degree of the rising trend of high-order harmonic components is high, and a high-order harmonic component abnormal signal is generated at this time; if the abnormal rising index of high-order harmonics is less than the reference threshold of abnormal rising index of high-order harmonics, it indicates that the abnormal degree of the rising trend of high-order harmonic components is low, and a normal signal of high-order harmonic components is generated at this time.

4. The diesel generator fault detection method based on data analysis according to claim 3 is characterized in that: After analyzing the temperature difference fluctuations between different parts, the diesel generator temperature difference fluctuation index is generated. The method for obtaining the diesel generator temperature difference fluctuation index is: The temperature difference between different parts is collected in real time through multiple temperature sensors of the diesel generator, including: the temperature difference between the stator winding and the casing: ΔT stator-casing (s); the temperature difference between the rotor winding and the stator winding: ΔT rotor-stator (s); the temperature difference between the inlet and outlet of the cooling system: ΔT cooling (s); Determine the smoothing factor α. At the first time point s0, the temperature difference fluctuation index EWMA0 is usually set to the initial temperature difference value, that is, EWMA0 = ΔT0; where ΔT0 is the first temperature difference data collected; starting from the second time point s1, the diesel generator temperature difference fluctuation index is calculated recursively using the formula, and the expression is: NG = α · ΔT s +(1-α)·EWMA s-1 ; In the formula, NG is the temperature fluctuation index of the diesel generator, ΔT s is the actual temperature difference data collected at time s, such as the temperature difference between the stator and the casing ΔT stator-casing (s), EWMA s-1 is the temperature difference fluctuation index at time s-1.

5. The diesel generator fault detection method based on data analysis according to claim 4 is characterized in that: The obtained diesel generator temperature difference fluctuation index is compared with the diesel generator temperature difference fluctuation index reference threshold set under the normal state of the diesel generator. If the diesel generator temperature difference fluctuation index is greater than or equal to the diesel generator temperature difference fluctuation index reference threshold, it means that the temperature difference fluctuation between different parts of the diesel generator is large, and a high temperature difference fluctuation signal is generated at this time; if the diesel generator temperature difference fluctuation index is less than the diesel generator temperature difference fluctuation index reference threshold, it means that the temperature difference fluctuation between different parts of the diesel generator is small, and a low temperature difference fluctuation signal is generated at this time.

6. The diesel generator fault detection method based on data analysis according to claim 5 is characterized in that: In S3, the abnormal rising trend of high-order harmonic components and the temperature difference fluctuation between different parts are comprehensively analyzed to evaluate the accuracy of the diesel generator fault detection results; The abnormal high-order harmonic rise index and the diesel generator temperature difference fluctuation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the accuracy value label of the diesel generator fault detection result predicted by each group of first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy value labels of all diesel generator fault detection results as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The accuracy value of the diesel generator fault detection result is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

7. The method for detecting diesel generator faults based on data analysis according to claim 6 is characterized in that: In S4, the accuracy of the diesel generator fault detection result is divided into different levels, namely, accurate detection result, possibly accurate detection result and inaccurate detection result, and corresponding processing is performed, specifically: Compare the accuracy value of the acquired diesel generator fault detection result with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and compare the accuracy value of the diesel generator fault detection result with the first standard threshold and the second standard threshold respectively; If the accuracy value of the diesel generator fault detection result is greater than the second standard threshold, it means that the accuracy of the diesel generator fault detection result is high, and no warning signal is generated at this time, and it is classified as an accurate detection result, indicating that the diesel generator is working normally and the system does not need further intervention; If the accuracy value of the diesel generator fault detection result is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of the diesel generator fault detection result is general, and a secondary warning signal is generated to remind relevant personnel to pay attention to the status of the generator and classify it as a possible accurate detection result; If the accuracy value of the diesel generator fault detection result is less than the first standard threshold, it means that the accuracy of the diesel generator fault detection result is low. At this time, a first-level warning signal is generated and it is classified as an inaccurate detection result, indicating that the detection result is seriously inaccurate and repair measures should be taken immediately.

Images