Energy equipment operation data driven health state evaluation method and system

Abstract

The invention discloses a health state assessment method and system driven by energy equipment operation data, and belongs to the field of data processing.The health state assessment method driven by the energy equipment operation data comprises the following steps that original operation data of energy equipment is collected, data cleaning, feature engineering and data standardization are carried out, and a health state assessment result is obtained; obtaining available data of the model; selecting a regression algorithm to construct a health assessment model, and training model parameters by using health equipment data and fault equipment data; after the original operation data of the to-be-detected energy equipment is preprocessed, the method has the beneficial effects that the operation data of the energy equipment are collected in real time, the health assessment model is constructed, the quantized energy equipment health score is output, and the active sensing of the state of the energy equipment is realized; the detected slight faults, early faults of non-core components and the like are temporarily not maintained, fault evolution data are continuously collected and used for training a more accurate health assessment model, and short-term risk is changed into long-term precision improvement.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and in particular relates to a health status assessment method and system driven by energy equipment operation data. Background Technology

[0002] Currently, the operation and maintenance of energy equipment generally suffers from delayed fault identification. Traditional monitoring methods mainly rely on manual inspections and basic alarm systems, often only detecting faults when obvious abnormalities (such as excessive temperature or abnormal vibration) have already occurred, by which time component damage or performance degradation has often taken place. This reactive maintenance model leads to three major challenges for enterprises: first, sudden shutdowns cause production disruptions (with an average impact time of 8-24 hours); second, emergency repairs are 3-5 times more expensive than preventative maintenance; and third, it disrupts established production schedules and energy efficiency management plans. These systemic defects highlight the limitations of the existing maintenance model and necessitate improvement. Summary of the Invention

[0003] Therefore, it is necessary to provide a data-driven health status assessment method and system for energy equipment to address the above-mentioned problems.

[0004] This invention is implemented as follows: a health status assessment method driven by energy equipment operation data includes the following steps:

[0005] Collect raw operating data of energy equipment (such as temperature, pressure, vibration, current, voltage, etc.), perform data cleaning (handling missing values ​​and outliers), feature engineering (extracting or constructing meaningful features, such as statistics and time-frequency domain features), and data standardization to obtain usable data for the model;

[0006] A health assessment model is constructed using regression algorithms (such as linear regression, gradient boosting tree GBDT, and neural networks), and the model parameters are trained using data from healthy equipment and faulty equipment.

[0007] After preprocessing the raw operating data of the energy equipment to be tested (data cleaning, feature engineering, and data standardization), the data is input into the health assessment model, and the health assessment model outputs a health score.

[0008] Continuously monitor the predictive performance (accuracy, recall, F1 score, etc.) of the health assessment model in real-world environments, and train the health assessment model periodically or as needed based on newly accumulated data, discovered misjudgment cases, and changes in equipment status (aging, etc.).

[0009] In one embodiment, the present invention provides a health status assessment method driven by energy equipment operation data, further comprising:

[0010] When the health score output by the health assessment model (regression model) is higher than the first set threshold (such as 95 points), the equipment is determined to be in normal condition and no maintenance prompt is triggered.

[0011] When the health score is lower than the first set threshold but higher than the second set threshold (e.g., 80 points), the fault classification model (e.g., SVM, random forest) is activated to identify potential fault types (e.g., bearing wear, coil overheating, etc.) based on preprocessed data (data that has undergone data cleaning, feature engineering, and data standardization). If the classification result is an early fault of a non-core component (e.g., slight dust accumulation on the radiator), no repair is prompted, and the system is marked as under observation.

[0012] When the health score falls below the second set threshold, the fault classification model is triggered to output the specific fault type and location; repair is immediately prompted, and a fault diagnosis report is pushed.

[0013] In one embodiment, the present invention provides a health status assessment method driven by energy equipment operation data, further comprising:

[0014] Based on the aging degree of energy equipment (such as years of operation and cumulative load), the first and second set thresholds are dynamically adjusted (for example, the first set threshold for old equipment changes from 95 to 84, and the second set threshold changes from 80 to 67) to avoid false alarms.

[0015] In one embodiment, the present invention provides a health status assessment method driven by energy equipment operation data, further comprising:

[0016] After the fault classification model outputs its results, a causal analysis module (such as DoWhy) is introduced to distinguish the root causes of the fault (e.g., is the voltage fluctuation due to power grid fluctuations or internal capacitor failure?), outputting root cause diagnosis to guide precise maintenance.

[0017] In one embodiment, the present invention provides a health status assessment method driven by energy equipment operation data, further comprising:

[0018] By correlating data from multiple energy devices within the same plant area, we can identify hidden related faults (e.g., abnormal vibration of pump A causes overheating of adjacent pump B).

[0019] In one embodiment, the present invention provides a health status assessment system driven by energy equipment operation data, comprising:

[0020] The data collection and processing module is used to collect raw operating data of energy equipment (such as temperature, pressure, vibration, current, voltage, etc.), perform data cleaning (handling missing values ​​and outliers), feature engineering (extracting or constructing meaningful features, such as statistics and time-frequency domain features), and data standardization to obtain usable data for the model.

[0021] The model building module is used to select regression algorithms (such as linear regression, gradient boosting tree GBDT, neural network) to build a health assessment model and train model parameters using healthy equipment data and faulty equipment data.

[0022] The health monitoring module is used to preprocess the raw operating data of the energy equipment to be monitored (data cleaning, feature engineering and data standardization), input it into the health assessment model, and output a health score through the health assessment model;

[0023] The model iteration module is used to continuously monitor the predictive performance (accuracy, recall, F1 score, etc.) of the health assessment model in real-world environments. Based on newly accumulated data, discovered misjudgment cases, and changes in equipment status (aging, etc.), the health assessment model is trained periodically or as needed.

[0024] In one embodiment, the present invention provides a health status assessment system driven by energy equipment operation data, further comprising:

[0025] The minor fault non-prompt module is used to determine that the equipment is in normal condition and does not trigger maintenance prompts when the health score output by the health assessment model (regression model) is higher than the first set threshold (such as 95 points).

[0026] The marking and observation module is used to activate a fault classification model (such as SVM or random forest) when the health score is lower than the first set threshold but higher than the second set threshold (e.g., 80 points). Based on preprocessed data (data that has undergone data cleaning, feature engineering, and data standardization), it identifies potential fault types (such as bearing wear, coil overheating, etc.). If the classification result is an early fault of a non-core component (such as slight dust accumulation on the radiator), no repair is prompted, and it is marked as an observation status.

[0027] The fault alert module is used to trigger the fault classification model to output the specific fault type and location when the health score is lower than the second set threshold; it immediately prompts for repair and pushes a fault diagnosis report.

[0028] In one embodiment, the present invention provides a health status assessment system driven by energy equipment operation data, further comprising:

[0029] The setpoint adjustment module is used to dynamically adjust the first set threshold and the second set threshold based on the aging degree of energy equipment (such as years of operation and cumulative load). For example, the first set threshold for old equipment is changed from 95 to 84, and the second set threshold is changed from 80 to 67, so as to avoid false alarms.

[0030] In one embodiment, the present invention provides a health status assessment system driven by energy equipment operation data, further comprising:

[0031] The fault root cause differentiation module is used to introduce a causal analysis module (such as DoWhy) after the fault classification model outputs results, to differentiate the root cause of the fault (e.g., is the voltage fluctuation due to power grid fluctuation? Or is it due to the failure of internal capacitors in the equipment?), output the root cause diagnosis, and guide precise maintenance.

[0032] In one embodiment, the present invention provides a health status assessment system driven by energy equipment operation data, further comprising:

[0033] The hidden associated fault identification module is used to correlate data from multiple energy devices in the same plant area to identify hidden associated faults (e.g., abnormal vibration of pump A causes overheating of adjacent pump B).

[0034] Compared with the prior art, the beneficial effects of the present invention are: the present invention collects the operating data of energy equipment in real time, constructs a health assessment model, outputs a quantitative health score of energy equipment, and realizes the proactive perception of the status of energy equipment; minor faults detected and early faults of non-core components are not repaired temporarily, and fault evolution data are continuously collected to train a more accurate health assessment model, exchanging short-term risks for long-term accuracy improvement. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the first part of a data-driven health status assessment method for energy equipment provided in an embodiment of the present invention.

[0036] Figure 2 This is a schematic diagram of the second part of a data-driven health status assessment method for energy equipment provided in an embodiment of the present invention.

[0037] Figure 3 This is a schematic diagram of the third part of a data-driven health status assessment method for energy equipment provided in an embodiment of the present invention.

[0038] Figure 4 This is a schematic diagram of the fourth part of a data-driven health status assessment method for energy equipment provided in an embodiment of the present invention.

[0039] Figure 5 This is a schematic diagram of the fifth part of a data-driven health status assessment method for energy equipment provided in an embodiment of the present invention.

[0040] Figure 6 This is a schematic diagram of the first part of a health status assessment system driven by energy equipment operation data, provided in an embodiment of the present invention.

[0041] Figure 7 This is a schematic diagram of the second part of a health status assessment system driven by energy equipment operation data, provided in an embodiment of the present invention.

[0042] Figure 8 This is a schematic diagram of the third part of a health status assessment system driven by energy equipment operation data, provided as an embodiment of the present invention.

[0043] Figure 9 This is a schematic diagram of the fourth part of a health status assessment system driven by energy equipment operation data, provided in an embodiment of the present invention.

[0044] Figure 10 This is a schematic diagram of the fifth part of a health status assessment system driven by energy equipment operation data, provided in an embodiment of the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0046] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0047] In one embodiment, such as Figure 1 As shown, a data-driven health status assessment method for energy equipment includes the following steps:

[0048] Step S1: Collect raw operating data of energy equipment (such as temperature, pressure, vibration, current, voltage, etc.), perform data cleaning (handling missing values ​​and outliers), feature engineering (extracting or constructing meaningful features, such as statistics and time-frequency domain features), and data standardization to obtain usable data for the model.

[0049] Step S2: Select a regression algorithm (such as linear regression, gradient boosting tree GBDT, neural network) to build a health assessment model, and use healthy equipment data and faulty equipment data to train the model parameters;

[0050] Step S3: After preprocessing the raw operating data of the energy equipment to be tested (data cleaning, feature engineering and data standardization), the data is input into the health assessment model, and the health assessment model outputs a health score.

[0051] Step S4: Continuously monitor the predictive performance (accuracy, recall, F1 score, etc.) of the health assessment model in the real environment. Based on newly accumulated data, discovered misjudgment cases, and changes in equipment status (aging, etc.), train the health assessment model regularly or as needed.

[0052] Step S1: Collect raw operating data of the energy equipment through built-in sensors (temperature / Pt100, pressure / piezoelectric, vibration / accelerometer) and related equipment. Use time series linear interpolation to process missing values, use the isolated forest algorithm to identify outliers, extract the statistical features (mean, variance, kurtosis) of the sliding window (1s window length) and the time-frequency domain features (energy entropy, bandwidth power ratio) of wavelet transform (Daubechies4 basis), and finally output usable data of the model through Z-score normalization (parameters are derived from the training set).

[0053] Step S2: Use 1D-CNN (one-dimensional convolutional neural network) to construct a health assessment model (gradient boosting tree can also be used to construct the model): Design 3 convolutional layers (kernel size 32 / 64 / 128) to extract the temporal features of the original vibration signal, and connect them to a fully connected layer to output the score; The training data includes data of healthy equipment within ±10% of the rated operating conditions, as well as data of faulty equipment (such as equipment data under the 1kHz harmonic component of bearing cracks). The loss function is Huber Loss, and overfitting is prevented by 5-fold time series cross-validation. The target output is a continuous health score of 0-100.

[0054] Step S3: Input the raw data of the device to be tested into the preprocessing pipeline that is completely consistent with S1, perform inference through the health assessment model, and output the health score and 95% confidence interval (calculated based on the residual distribution of the test set).

[0055] Step S4: Continuously calculate the mean absolute error (MAE) and coefficient of determination (R²) between the health assessment model's predicted values ​​and the actual maintenance results. 2 When MAE scores are greater than 8 for 3 consecutive days or the physical state of the equipment changes after a major overhaul, incremental training is triggered: Fine-tuning strategy (a model optimization strategy) is adopted to freeze the bottom feature layer of the pre-trained model, update only the top regression weights, and optimize the parameters using newly accumulated data.

[0056] In one embodiment, such as Figure 2 As shown, a data-driven health status assessment method for energy equipment also includes:

[0057] Step S5: When the health score output by the health assessment model (regression model) is higher than the first set threshold (e.g., 95 points), the equipment is determined to be in normal condition and no maintenance prompt is triggered.

[0058] Step S6: When the health score is lower than the first set threshold but higher than the second set threshold (e.g., 80 points), start the fault classification model (e.g., SVM, random forest) to identify potential fault types (e.g., bearing wear, coil overheating, etc.) based on preprocessed data (data that has undergone data cleaning, feature engineering, and data standardization); if the classification result is an early fault of a non-core component (e.g., slight dust accumulation on the radiator), do not prompt for repair and mark it as an observation state.

[0059] Step S7: When the health score is lower than the second set threshold, the fault classification model is triggered to output the specific fault type and location; repair is immediately prompted, and a fault diagnosis report is pushed.

[0060] In step S5, when the health score is higher than the first set threshold α (initial value 95), the device is determined to be in the safe operating zone. The fault diagnosis process is skipped through the edge rule engine (Apache Drools), and only the score is recorded to the device file.

[0061] In step S6, when the score ∈ (β, α) (the second set threshold β is initially 80), the fault classification model constructed by random forest (500 trees, Gini splitting criterion) is started, the feature vector after preprocessing in step S1 is input, and the fault type probability distribution is output; if the highest probability fault belongs to the predefined list of non-core components (such as dust accumulation on the radiator, rust on the casing) and the confidence level is >90%, it is marked as an observation state and the monitoring frequency is increased.

[0062] In step S7, when the score is less than β, the fault classification model outputs the specific fault type and location (such as "bearing inner ring wear"), and combines the Grad-CAM (gradient weighted class activation mapping) heat map to locate the abnormal sensor channel, automatically pushes the maintenance strategy of the associated knowledge base (such as "replace SKF-6321 bearing, torque needs to reach 25Nm"), and provides maintenance methods.

[0063] In one embodiment, such as Figure 3 As shown, a data-driven health status assessment method for energy equipment also includes:

[0064] Step S8: Based on the aging degree of energy equipment (such as years of operation and cumulative load), dynamically adjust the first set threshold and the second set threshold (for example, the first set threshold for old equipment changes from 95 to 84, and the second set threshold changes from 80 to 67) to avoid false alarms.

[0065] Step S8: Based on the equipment aging coefficient formula γ = 1 - e raised to the power of (-0.15 * years of operation), dynamically lower the threshold: α = 95 - 20γ, β = 80 - 25γ (e.g., for 5-year-old equipment, γ = 0.53 → α = 84, β = 67) to avoid aging equipment from falsely triggering alarms due to performance degradation.

[0066] In one embodiment, such as Figure 4 As shown, a data-driven health status assessment method for energy equipment also includes:

[0067] Step S9: After the fault classification model outputs the results, introduce a causal analysis module (such as DoWhy) to distinguish the root causes of the fault (e.g., is the voltage fluctuation due to power grid fluctuation? Or is it due to the failure of the internal capacitor of the equipment?), output the root cause diagnosis, and guide accurate maintenance.

[0068] Step S9: After the fault classification model outputs the fault type, the DoWhy causal inference framework is used: a causal graph is constructed based on the equipment schematic diagram, the causal effect is calculated through instrumental variables (such as power grid load data), and a root cause probability report is output (such as "the probability of voltage fluctuation caused by capacitor aging is 82%").

[0069] In one embodiment, such as Figure 5 As shown, a data-driven health status assessment method for energy equipment also includes:

[0070] Step S10: Associate data from multiple energy devices in the same plant area to identify hidden associated faults (e.g., abnormal vibration of pump A causes overheating of adjacent pump B).

[0071] Step S10: Use a time-series alignment algorithm to match data streams from multiple devices in the same factory area. Input a graph neural network (GNN) to model the physical connection relationship of the devices. When a synchronization anomaly mode (vibration or current fluctuation with a correlation coefficient > 0.7) is detected in the device group, locate the root cause device and trigger the root cause analysis in step S9.

[0072] In one embodiment, such as Figure 6 As shown, a health status assessment system driven by energy equipment operation data includes:

[0073] Data collection and processing module 1 is used to collect raw operating data of energy equipment (such as temperature, pressure, vibration, current, voltage, etc.), perform data cleaning (handling missing values ​​and outliers), feature engineering (extracting or constructing meaningful features, such as statistics and time-frequency domain features), and data standardization to obtain usable data for the model.

[0074] Model building module 2 is used to select regression algorithms (such as linear regression, gradient boosting tree GBDT, neural network) to build a health assessment model and train model parameters using healthy equipment data and faulty equipment data;

[0075] The health detection module 3 is used to preprocess the raw operating data of the energy equipment to be detected (data cleaning, feature engineering and data standardization), input it into the health assessment model, and output a health score through the health assessment model;

[0076] Model iteration module 4 is used to continuously monitor the predictive performance (accuracy, recall, F1 score, etc.) of the health assessment model in the real environment. Based on newly accumulated data, discovered misjudgment cases, and changes in equipment status (aging, etc.), the health assessment model is trained periodically or as needed.

[0077] In data collection and processing module 1, direct modeling of raw sensor data containing noise or missing information can lead to misjudgments. Through data cleaning, feature engineering, and data standardization, key signals of the equipment status (such as abnormal vibration frequency bands) are extracted to provide high-quality input for the model.

[0078] In model building module 2, equipment health is a continuously changing process (not a simple binary classification of good and bad). The health assessment model built by the regression algorithm outputs a health score of 0-100, which reflects the decline trend (e.g., a slow drop from 95 to 60) with more granularity than the classification algorithm (the subsequent fault classification model is built using a classification algorithm), supporting predictive maintenance.

[0079] In health monitoring module 3, rapid response is required in industrial settings, so the data processing step S1 preprocessing step is reused.

[0080] In model iteration module 4, equipment aging or changes in operating conditions can cause the health assessment model to fail (such as the offset of newly installed sensors). Incremental training enables the health assessment model to dynamically adapt to real-world changes and maintain prediction accuracy.

[0081] In one embodiment, such as Figure 7 As shown, a health status assessment system driven by energy equipment operation data also includes:

[0082] The minor fault non-prompt module 5 is used to determine that the equipment is in normal condition and not trigger maintenance prompts when the health score output by the health assessment model (regression model) is higher than the first set threshold (such as 95 points).

[0083] The marking and observation module 6 is used to activate a fault classification model (such as SVM or random forest) when the health score is lower than the first set threshold but higher than the second set threshold (e.g., 80 points). Based on preprocessed data (data that has undergone data cleaning, feature engineering, and data standardization), it identifies potential fault types (such as bearing wear, coil overheating, etc.). If the classification result is an early fault of a non-core component (such as slight dust accumulation on the radiator), no repair is prompted, and it is marked as an observation state.

[0084] The fault indication module 7 is used to trigger the fault classification model to output the specific fault type and location when the health score is lower than the second set threshold; immediately prompt for repair and push the fault diagnosis report.

[0085] In Module 5, which does not alert users to minor faults, high-scoring devices have an extremely low probability of failure. By skipping fault diagnosis, the computational load is reduced (especially for edge devices), meaningless analysis is reduced, and resources are focused on risky devices.

[0086] In the marking and observation module 6, some faults develop slowly (such as dust accumulation), so they are not repaired for the time being, but the monitoring frequency is increased to avoid excessive maintenance costs. At the same time, fault evolution data is accumulated to optimize the health assessment model (in step S4).

[0087] In fault indication module 7, low-scoring devices are at risk of immediate failure. By combining fault location and maintenance strategy recommendations, the diagnosis and repair cycle can be shortened, reducing unexpected downtime losses.

[0088] Minor faults (scoring above 95) and early-stage faults in non-core components will not be repaired for the time being. Instead, fault evolution data will be continuously collected to train a more accurate health assessment model, trading short-term low risk for long-term accuracy improvement.

[0089] In one embodiment, such as Figure 8 As shown, a health status assessment system driven by energy equipment operation data also includes:

[0090] The setpoint adjustment module 8 is used to dynamically adjust the first set threshold and the second set threshold based on the aging degree of energy equipment (such as years of operation and cumulative load). For example, the first set threshold for old equipment changes from 95 to 84, and the second set threshold changes from 80 to 67, to avoid false alarms.

[0091] In the setpoint adjustment module 8, the performance of older equipment naturally degrades, and a fixed threshold will lead to false alarms (such as the higher vibration baseline of old fans). The formulaic lowering of the threshold adapts to the aging process and reduces false alarms.

[0092] In one embodiment, such as Figure 9 As shown, a health status assessment system driven by energy equipment operation data also includes:

[0093] The fault root cause differentiation module 9 is used to introduce a causal analysis module (such as DoWhy) after the fault classification model outputs results, to differentiate the root cause of the fault (e.g., is the voltage fluctuation due to power grid fluctuation? Or is it due to the failure of internal capacitors in the equipment?), output the root cause diagnosis, and guide precise maintenance.

[0094] In the fault root cause differentiation module 9, the same fault phenomenon may be caused by different reasons (such as overheating may be caused by heat dissipation failure or voltage abnormality). Causal analysis distinguishes the root cause and guides targeted repair (replacing capacitors instead of cleaning dust) to avoid repeated faults.

[0095] In one embodiment, such as Figure 10 As shown, a health status assessment system driven by energy equipment operation data also includes:

[0096] The hidden correlation fault identification module 10 is used to correlate data from multiple energy devices in the same plant area to identify hidden correlation faults.

[0097] In the hidden associated fault identification module 10, a single device failure may originate from a systemic problem (such as power grid harmonic pollution of multiple devices). By identifying chain reactions through group analysis, the problem can be solved at its source (by addressing the power grid rather than replacing the equipment).

[0098] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0099] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0100] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0101] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0103] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A health status assessment method driven by energy equipment operation data, characterized in that, The energy equipment operation data-driven health status assessment method includes the following steps: Collect raw operating data from energy equipment, perform data cleaning, feature engineering, and data standardization to obtain usable data for the model; A regression algorithm was selected to construct a health assessment model, and the model parameters were trained using data from healthy equipment and faulty equipment. After preprocessing the raw operating data of the energy equipment to be tested, the data is input into the health assessment model, and the health assessment model outputs a health score. Continuously monitor the predictive performance of the health assessment model in real-world environments, and train the health assessment model periodically or as needed based on newly accumulated data, discovered misjudgment cases, and changes in equipment status.

2. The energy equipment operation data-driven health status assessment method according to claim 1, characterized in that, The energy equipment operation data-driven health status assessment method also includes: When the health score output by the health assessment model is higher than the first set threshold, the equipment is determined to be in normal condition and no maintenance prompt is triggered. When the health score is lower than the first set threshold but higher than the second set threshold, the fault classification model is activated to identify potential fault types based on preprocessed data; if the classification result is an early fault of a non-core component, no repair is prompted and the system is marked as under observation. When the health score falls below the second set threshold, the fault classification model is triggered to output the specific fault type and location; repair is immediately prompted, and a fault diagnosis report is pushed.

3. The energy equipment operation data-driven health status assessment method according to claim 2, characterized in that, The energy equipment operation data-driven health status assessment method also includes: Based on the aging level of energy equipment, the first and second set thresholds are dynamically adjusted to avoid false alarms.

4. The energy equipment operation data-driven health status assessment method according to claim 2, characterized in that, The energy equipment operation data-driven health status assessment method also includes: After the fault classification model outputs its results, a causal analysis module is introduced to distinguish the root causes of the faults, output root cause diagnoses, and guide precise maintenance.

5. The energy equipment operation data-driven health status assessment method according to claim 4, characterized in that, The energy equipment operation data-driven health status assessment method also includes: By correlating data from multiple energy devices within the same plant area, hidden associated faults can be identified.

6. A health status assessment system driven by energy equipment operation data, characterized in that, include: The data collection and processing module is used to collect raw operating data from energy equipment, perform data cleaning, feature engineering, and data standardization to obtain usable data for the model. The model building module is used to select a regression algorithm to build a health assessment model and to train the model parameters using data from healthy equipment and faulty equipment. The health monitoring module is used to preprocess the raw operating data of the energy equipment to be monitored and input it into the health assessment model, and output a health score through the health assessment model. The model iteration module is used to continuously monitor the predictive performance of the health assessment model in real-world environments. Based on newly accumulated data, discovered misjudgment cases, and changes in equipment status, the health assessment model is trained periodically or as needed.

7. The energy equipment operation data-driven health status assessment system according to claim 6, characterized in that, Also includes: The minor fault non-prompt module is used to determine that the equipment is in normal condition and not trigger maintenance prompts when the health score output by the health assessment model is higher than the first set threshold. The labeling and observation module is used to activate the fault classification model when the health score is lower than the first set threshold but higher than the second set threshold, and to identify potential fault types based on preprocessed data. If the classification result is an early failure of a non-core component, no repair is prompted at this time, and it is marked as an observation status; The fault alert module is used to trigger the fault classification model to output the specific fault type and location when the health score is lower than the second set threshold; it immediately prompts for repair and pushes a fault diagnosis report.

8. The energy equipment operation data-driven health status assessment system according to claim 7, characterized in that, Also includes: The setpoint adjustment module is used to dynamically adjust the first set threshold and the second set threshold based on the aging degree of the energy equipment to avoid false alarms.

9. The energy equipment operation data-driven health status assessment system according to claim 7, characterized in that, Also includes: The fault root cause differentiation module is used to introduce the causal analysis module after the fault classification model outputs the results, differentiate the fault root causes, output root cause diagnosis, and guide precise maintenance.

10. The energy equipment operation data-driven health status assessment system according to claim 9, characterized in that, Also includes: The hidden correlation fault identification module is used to correlate data from multiple energy devices in the same plant area to identify hidden correlation faults.

Images