An oil-free digital energy air compressor station intelligent heat dissipation system

Abstract

This invention provides an oil-free digital energy air compressor station intelligent heat dissipation system, relating to the technical field of air compressor station heat dissipation. It includes a sensor module with communication connectivity, a correlation analysis module, and a heat dissipation execution module. By analyzing the statistical, causal, and temporal correlations between various operating parameters and temperature, a physical mechanism is introduced to construct a hybrid correlation model. Combined with predicted temperature change trends, the system analyzes the specific operating parameters affecting the current temperature change, locates the specific equipment based on these parameters, and executes a heat dissipation strategy. Through a closed loop of model prediction – parameter attribution – equipment location – precise heat dissipation, it combines data-driven correlation analysis with causal reasoning based on physical mechanisms, achieving rapid location from temperature anomalies to specific heat sources, avoiding the inefficiency of traditional equipment-by-equipment troubleshooting.

Description

Technical Field

[0001] This invention relates to the field of air compressor station heat dissipation, and more particularly to an oil-free digital energy air compressor station intelligent heat dissipation system. Background Technology

[0002] In oil-based digital energy air compressor stations, oil participates in air compression during the compression process, absorbing the heat generated. It is then cooled by equipment such as an oil cooler, and the cooled oil is circulated back into the compressor for further heat dissipation. The heat source is relatively concentrated. In addition, the oil also lubricates the moving parts inside the compressor. However, even after oil-gas separation, the compressed air still contains a small amount of oil, which can pollute the compressed air. Furthermore, the disposal of waste oil is also an environmental issue.

[0003] To address this issue, oil-free digital energy-based air compressor stations have emerged. These stations do not require oil during compression and rely primarily on mediums such as air and water for heat dissipation. Heat is carried away by airflow or water circulation to overcome the high temperatures generated during compression. While oil-free operation reduces environmental pollution, it requires the design of additional heat dissipation structures, and the heat sources are relatively dispersed.

[0004] Currently, there are many drawbacks in the heat dissipation of oil-free air compressor stations. Common passive cooling methods, such as relying solely on fans with fixed speeds or cooling water with constant flow rates, lack awareness and response to the actual operating status of the equipment, resulting in low heat dissipation efficiency. Although some air compressor stations employ simple temperature control systems, most are based on a single temperature threshold to trigger heat dissipation, failing to accurately identify the root cause of abnormal temperatures. This easily leads to "overheating" or "underheating," wasting energy and potentially shortening equipment lifespan due to sustained high temperatures in critical components. Furthermore, traditional methods struggle to cope with the coupled effects of multiple parameters under complex operating conditions. For example, when equipment load fluctuates or ambient temperature changes, it is impossible to accurately determine the specific impact of each operating parameter on temperature, leading to difficulties in fault location and high maintenance costs.

[0005] Therefore, how to accurately locate the heat-generating parts and implement precise heat dissipation strategies by sensing and responding to the actual operating status of the equipment is a technical problem that urgently needs to be solved. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides an oil-free digital energy air compressor station intelligent heat dissipation system. This system can analyze the statistical, causal, and temporal correlations between various operating parameters and temperature, introduce physical mechanisms, construct a hybrid correlation model, and combine the predicted temperature change trend to analyze the specific operating parameters affecting the temperature change at the current moment. Based on these operating parameters, it can locate specific equipment and execute corresponding heat dissipation strategies.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] This invention provides an oil-free digital energy air compressor station intelligent heat dissipation system, comprising a sensor module, a correlation analysis module, and a heat dissipation execution module that are connected in sequence via communication.

[0009] The sensor module is used to collect various operating parameters of the air compressor station equipment through multiple types of sensors and to preprocess the collected data.

[0010] The correlation analysis module is used to analyze the statistical correlation, causal correlation and temporal correlation between various operating parameters and temperature, and to construct a hybrid correlation model by integrating the correlation analysis results.

[0011] The heat dissipation execution module is used to perform reverse analysis on the predicted temperature change trend in combination with the operating parameters, determine the specific operating parameters affecting the current temperature change, and locate the specific heat-generating device or part based on the operating parameters for heat dissipation treatment.

[0012] In the sensor module, the preprocessing includes data cleaning, data conversion, data alignment and synchronization, and data tagging.

[0013] In the correlation analysis module, before performing correlation analysis, it is necessary to first extract features from the preprocessed data to extract more physically meaningful features from the original sensor data; determine the analysis object, and determine the target variable as the temperature parameters of the key equipment of the air compressor station, and the independent variables as various operating parameters.

[0014] The correlation analysis module includes a statistical correlation analysis unit, a causal correlation analysis unit, and a time correlation analysis unit;

[0015] The statistical correlation analysis unit is used to analyze the statistical correlation between various operating parameters and temperature;

[0016] The causal correlation analysis unit is used to analyze the causal correlation between various operating parameters and temperature, establish a causal diagram, and obtain the chain conduction path;

[0017] The time correlation analysis unit is used to analyze the time correlation between various operating parameters and temperature.

[0018] The statistical correlation analysis unit includes the following steps:

[0019] When analyzing the correlation between numerical parameters and temperature, the correlation coefficient is calculated and the calculated value is compared with a preset threshold to determine the strength of the correlation between the data.

[0020] When analyzing the correlation between the categorical parameters and temperature, the method of analysis of variance is used. By comparing the mean temperature differences under different operating modes, the significance of the influence of mode type on temperature is determined.

[0021] Based on the results of the statistical correlation analysis, the operating parameters that are strongly correlated with temperature are retained, and a statistical correlation analysis model based on linear regression is constructed to analyze the statistical relationship between various operating parameters and temperature.

[0022] Based on the linear relationship between the current operating parameters and temperature, the current operating parameters are input into the corresponding linear model. The model outputs the predicted temperature. If the predicted temperature exceeds the preset threshold, the heat dissipation strategy is adjusted. At the same time, new data is collected periodically, and the model parameters are updated online using the stochastic gradient descent algorithm.

[0023] In the causal correlation analysis unit, the causal correlation between various operating parameters and temperature is determined by combining physical mechanisms, including the following steps:

[0024] The operating parameters that are strongly correlated with temperature, as selected by statistical correlation analysis, are denoted as the feature set, and the target variable is set as temperature;

[0025] Based on a preliminary analysis of the physical mechanism of the air compressor station, the direction of potential causal relationships is clarified, an initial causal hypothesis diagram is established, and direct and indirect causal relationships are marked.

[0026] A causal correlation analysis model based on causal random forest is constructed and trained to determine the causal effect strength of each operating parameter;

[0027] The significance test is used to determine whether there is a direct causal relationship between operating parameters and temperature, and the physical consistency test is used to verify whether the direction of the causal effect is reasonable.

[0028] Further analysis of the results of direct causal relationship analysis is conducted to determine whether there are intermediate variables, and indirect causal relationships are uncovered through recursive analysis.

[0029] Based on direct and indirect causal relationships, a causal relationship graph is constructed, and the causal directions between each node are marked.

[0030] Validate the trained model and causal paths;

[0031] The model and causal graph are optimized through critical path prioritization, intervention strategy design, and model iteration.

[0032] In the causal correlation analysis unit, a causal correlation analysis model based on causal random forest is constructed and trained, including the following steps:

[0033] Given the intervention variable, the running parameters are divided into intervention and control groups, and the individual causal effect values ​​are calculated;

[0034] Strongly correlated parameters are treated as intervention variables, and the causal effect of one operational parameter is analyzed each time, while the remaining operational parameters are treated as covariates.

[0035] For each intervention variable, construct a binary treatment variable;

[0036] The dataset is randomly divided into training and test sets; a causal random forest model is fitted using the training set, with the optimization objective being to minimize the mean squared error between the estimated causal effect and the true effect; the model estimates the average causal effect within each leaf node by recursively partitioning the feature space.

[0037] For each sample, output the estimated conditional causal effect of the intervention variable on temperature;

[0038] The causal effect strength of each operating parameter can be assessed by ranking importance or using the SHAP value.

[0039] The time correlation analysis unit includes the following steps:

[0040] The input is time series data, which includes a sequence of operating parameters and a temperature sequence;

[0041] Perform a stationarity test on the input data;

[0042] Construct and train a time correlation analysis model based on a vector autoregression model;

[0043] The time relationship between each operating parameter and temperature is determined based on the sign and magnitude of the regression coefficients;

[0044] Verification was performed using a time-series causal chain, and dynamic response analysis was conducted using an impulse response function.

[0045] In the correlation analysis module, after performing correlation analysis, the analysis results of statistical correlation, causal correlation and temporal correlation are fused and analyzed, and a hybrid correlation model is constructed by weighted fusion method.

[0046] In the heat dissipation execution module, the execution process includes the following steps:

[0047] Obtain the predicted temperature output by the hybrid correlation model, as well as the predicted temperature change trend over a future period, and identify areas of abnormal temperature rise;

[0048] Quantify the contribution of each operating parameter to the current temperature change using SHAP values ​​or feature importance ranking;

[0049] The operating parameters with the highest contribution were selected as key influencing parameters, and pseudo-correlated parameters were eliminated in combination with the air compressor station process logic.

[0050] Based on the established causal graph, path tracing is performed. Through impulse response analysis, the time leader relationship between operating parameters is determined. The operating parameter with the highest contribution and a clear causal chain is taken as the dominant parameter of the current temperature change.

[0051] Establish a mapping relationship between operating parameters and equipment, and locate specific equipment based on spatial positioning algorithms and heat conduction path analysis;

[0052] For different devices identified, targeted heat dissipation strategies are adopted and continuously monitored.

[0053] The beneficial effects of this invention are as follows:

[0054] This invention provides an intelligent heat dissipation system for oil-free digital energy air compressor stations. It utilizes multiple types of sensors to collect and preprocess operating parameters. By analyzing the statistical, causal, and temporal correlations between these parameters and temperature, it introduces physical mechanisms and constructs a hybrid correlation model. Combined with predicted temperature change trends, it analyzes the specific operating parameters affecting the current temperature change and locates the specific equipment based on these parameters, executing corresponding heat dissipation strategies. Through a closed loop of model prediction, parameter attribution, equipment location, and precise heat dissipation, it combines data-driven correlation analysis with causal reasoning based on physical mechanisms, achieving rapid location from temperature anomalies to specific heat sources, avoiding the inefficiency of traditional equipment-by-equipment troubleshooting. Attached Figure Description

[0055] Figure 1 This is a block diagram of an oil-free digital energy air compressor station intelligent heat dissipation system according to the present invention. Detailed Implementation

[0056] Please see Figure 1 As shown, the present invention relates to an intelligent heat dissipation system for an oil-free digital energy air compressor station, comprising a sensor module, a correlation analysis module, a prediction model construction module, and a heat dissipation execution module that are connected in sequence via communication.

[0057] The sensor module is used to collect various operating parameters of the air compressor station equipment through multiple types of sensors and to preprocess the collected data.

[0058] High-precision temperature sensors are installed at key heat-generating components of the air compressor, such as the motor, compression chamber, and cooler inlet and outlet, as well as at locations prone to heat accumulation, such as the air tank and pipelines, to monitor temperature changes in various parts of the equipment in real time. Pressure sensors are deployed at the air compressor's inlet and outlet, as well as at key nodes in the pipelines, to monitor gas pressure; the pressure measurement range can be set according to the actual operating pressure range of the air compressor station. Flow sensors are installed at the inlet and outlet pipelines of the cooling medium (such as water in a water-cooled system or air in an air-cooled system) to monitor the flow rate of the cooling medium in real time. For rotating equipment such as the air compressor motor and fan motor, speed sensors are installed to monitor motor speed in real time, providing a basis for analyzing equipment operating conditions and energy consumption; the speed measurement range is determined based on the motor's rated speed.

[0059] All data collected by sensors is transmitted to a data center via wired (e.g., industrial Ethernet) or wireless (e.g., LoRa, NB-IoT, and other low-power wide-area network technologies) methods, where it undergoes preprocessing. This preprocessing includes data cleaning, data transformation, data alignment and synchronization, and data tagging.

[0060] The data cleaning includes outlier detection and processing, missing value imputation, and noise filtering.

[0061] Outlier detection and handling: By calculating the mean and standard deviation, outliers that significantly deviate from the normal range are identified using the 3σ principle (under a normal distribution, approximately 99.7% of the data falls within the mean ± 3 times the standard deviation). For parameters such as temperature and pressure that change continuously over time, a sliding window or exponential smoothing method is used to detect abrupt changes in values ​​between adjacent time points (e.g., a sudden temperature jump exceeding the reasonable range). If outliers are caused by transient sensor malfunctions or transmission interference (e.g., transient impulse noise), they can be corrected using the mean or median of adjacent time points (e.g., replacing the outlier with the average temperature of the previous and next minute). For outliers that cannot be corrected (e.g., persistent errors caused by sensor damage), the data point is directly removed and ignored in subsequent interpolation.

[0062] Missing value imputation: For scenarios where data is missing for a short period and the trend is linear (such as a brief absence of cooling water flow rate), the missing values ​​are filled by linear fitting using values ​​from before and after the missing data point. For parameters with large fluctuations, such as temperature, missing values ​​are filled by constructing smooth spline curves to avoid abrupt changes in linear interpolation.

[0063] Noise filtering: For high-frequency noise (such as sensor signal fluctuations caused by compressor vibration), Butterworth low-pass filters or moving average filters are used to smooth the signal curve and preserve low-frequency trend characteristics. For non-stationary signals (such as abrupt data during startup / shutdown), the signal is decomposed by wavelet transform, and the high-frequency noise components are removed before reconstructing the signal.

[0064] After data cleaning, the data is normalized or standardized to ensure it has the same dimensions and scale. Normalization maps the data to the [0,1] interval, while standardization converts the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, thus eliminating the influence of dimensions and facilitating subsequent model processing.

[0065] Since different types of sensors may have different sampling frequencies (e.g., a temperature sensor samples once per second, while a flow sensor samples once every 5 seconds), it is necessary to align timestamps to unify the data into the same time coordinate system. That is, high-frequency data is downsampled (e.g., averaged per minute) to match the sampling frequency of low-frequency data; low-frequency data is upsampled (e.g., by interpolating to generate data at intermediate time points) to match the timestamps of high-frequency data.

[0066] By combining the start / stop signals and operating mode switching signals of the air compressor station equipment (such as the compressor switching from full-load operation to no-load operation), the data is labeled with status tags. For example, the data is marked with states such as "start-up stage", "normal operation", and "shutdown cooling" to avoid data from different operating stages from being mixed, which could lead to biases in model training.

[0067] The correlation analysis module is used to analyze the statistical correlation, causal correlation and temporal correlation between various operating parameters and temperature, and to construct a hybrid correlation model by integrating the correlation analysis results.

[0068] In the correlation analysis module, before performing correlation analysis, feature extraction is first performed on the preprocessed data to extract more physically meaningful features from the original sensor data. For example, the temperature change rate (∆T / ∆t) can reflect the heating rate of the equipment; the pressure fluctuation coefficient (standard deviation / mean) can characterize the influence of airflow stability on heat generation; and the cooling efficiency index (temperature difference between inlet and outlet of cooling medium × flow rate) can quantify the real-time performance of the heat dissipation system to enhance the data's ability to characterize temperature changes.

[0069] When determining the analysis object, the target variable (dependent variable) is determined to be the temperature parameters of the key equipment of the air compressor station (such as compressor motor temperature, bearing temperature, inlet and outlet temperatures of cooling medium, etc.); the independent variables are various operating parameters, such as equipment operating status parameters (compressor speed, intake pressure, exhaust flow, motor power, load rate, etc.), environmental parameters (ambient temperature, humidity, cooling water flow rate, air volume, etc.), and time parameters (running time, start-stop cycle, etc.).

[0070] The correlation analysis module includes a statistical correlation analysis unit, a causal correlation analysis unit, and a time correlation analysis unit;

[0071] The statistical correlation analysis unit is used to analyze the statistical correlation between various operating parameters and temperature, specifically including the following steps:

[0072] When analyzing the correlation between numerical parameters and temperature, the correlation coefficient is calculated and compared with a preset threshold to determine the strength of the correlation between the data.

[0073] The Pearson correlation coefficient is applicable to linear relationships. It calculates the correlation coefficient r between operating parameters and temperature, with a value ranging from -1 to 1. The closer the absolute value is to 1, the stronger the linear correlation. A threshold range for the correlation coefficient is set: |r|>0.7 indicates a strong correlation (e.g., motor power and motor temperature); 0.3≤|r|≤0.7 indicates a moderate correlation (e.g., cooling water flow rate and bearing temperature); and |r|<0.3 indicates a weak correlation (e.g., non-critical pipeline pressure). Parameters strongly correlated with temperature are retained, such as compressor inlet pressure, outlet flow rate, motor current, cooling water flow rate, and ambient humidity.

[0074] Spearman Rank Correlation is suitable for nonlinear relationships or nonnormally distributed data (such as the piecewise linear relationship between equipment loading rate and temperature). It calculates the correlation based on the parameter rank, sets an appropriate threshold, judges the correlation strength, and retains parameters that are strongly correlated with temperature.

[0075] When analyzing the correlation between categorical parameters and temperature, the analysis of variance (ANOVA) method is used. This method compares the mean temperature differences under different operating modes (such as full load / half load) to determine the significance of the influence of mode type on temperature (for example, if the p value of the F test is <0.05, then it is considered that there is a significant correlation between operating mode type and temperature).

[0076] Based on the results of the statistical correlation analysis, the operating parameters that are strongly correlated with temperature are retained, and a statistical correlation analysis model based on linear / nonlinear regression is constructed to analyze the statistical correlation between various operating parameters and temperature.

[0077] A multiple linear regression (MLR) model is constructed based on the linear association. The model formula is as follows:

[0078] ;

[0079] Among them, T t The temperature at time t (target variable) represents the temperature. β represents the i-th running parameter at time t. i The regression coefficients, estimated using the least squares method, represent the degree of influence of the independent variable on the dependent variable. n is the number of operating parameters, and ϵ t Let N(0, σ) represent the random error term, which follows the order N(0, σ).2 (Through the coefficient of determination R) 2 To evaluate the model, R 2 The closer to 1, the better the model fit.

[0080] A multinomial regression model is constructed to address the nonlinear association. The model formula is as follows:

[0081] ;

[0082] Where, β i and γ i Let Z and X represent the regression coefficients, respectively. A quadratic term is introduced to capture nonlinear relationships (such as the rate of temperature decrease slowing down after the cooling airflow exceeds a critical value, exhibiting a quadratic function relationship). This is achieved through variable substitution (letting Z=X). 2 The problem is transformed into a multiple linear regression problem and then solved.

[0083] The preprocessed and feature-extracted data is divided into training and testing sets. The constructed linear / nonlinear model is trained using the training set data, and validated using the testing set data. After the model is trained, evaluation metrics are calculated to assess the model, such as root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²). 2 If R 2 If the residual value is greater than 0.8 and the residual distribution is random, then the model is qualified. According to Occam's razor principle, models with simple structures should be selected first.

[0084] In practical applications, based on the linear / nonlinear relationship between the current operating parameters and temperature, the current operating parameters are input into the corresponding linear / nonlinear model. The model outputs a predicted temperature. If the predicted temperature exceeds a preset threshold, the heat dissipation strategy is adjusted. Simultaneously, new data is collected periodically, and the model parameters are updated online using the stochastic gradient descent (SGD) algorithm.

[0085] For multicollinearity treatment, calculate the variance inflation factor (VIF). i If the value is greater than 10, remove the variable or use principal component analysis. Plot a scatter plot of residuals and predicted values ​​to check for heteroscedasticity; if the residuals show a regularity, a higher-order polynomial or variable transformation is needed. Collinearity can be mitigated through regularization methods.

[0086] For example, when predicting the temperature of an air compressor, a linear model is established:

[0087] T = 65 + 0.03P + 0.15I - 0.2Q + 0.5T env ;

[0088] Where P is pressure (bar), I is current (A), and Q is cooling airflow (m³ / s). 3 / min), Tenv The ambient temperature is shown in °C. The regression coefficients preceding each operating parameter are standardized values. For example, the regression coefficient between pressure and temperature is 0.03, meaning that for every unit increase in pressure, the temperature is expected to increase by 0.03 units, indicating that pressure has a positive effect on temperature.

[0089] R on the test set 2 =0.87, RMSE=1.8℃, can predict the risk of temperature exceeding the threshold 30 minutes in advance.

[0090] When the operating parameter is cooling power, establish the cooling power P. cool Nonlinear model with flow velocity v:

[0091] ;

[0092] Among them, T in Let v be the inlet water temperature (°C). Find the extreme point v by differentiation. * =5.8m / s, at which point the cooling efficiency is highest, guiding the control of the variable frequency water pump.

[0093] The causal correlation analysis unit is used to analyze the causal correlation between various operating parameters and temperature to obtain the chain conduction path;

[0094] Correlation only indicates the degree of association between variables; causal relationship analysis requires further verification to determine whether the operating parameters are driving factors of temperature changes. For strongly correlated operating parameters, the causal relationship is determined by combining the physical mechanism of the air compressor station, including the following steps:

[0095] The operating parameters strongly correlated with temperature, selected through statistical correlation analysis, are denoted as the feature set X = {X1, X2, ..., X...}. n The target variable is set as temperature Y;

[0096] Based on a preliminary analysis of the physical mechanism of the air compressor station, the direction of potential causal relationships is clarified, an initial causal hypothesis diagram is established, and direct and indirect causal relationships are marked.

[0097] For example, increased compressor power leads to heat generation from mechanical friction, which in turn raises the cylinder block temperature; increased coolant temperature leads to decreased heat dissipation efficiency, resulting in increased gas temperature after cooling; insufficient intake airflow leads to an increased compression ratio, which in turn raises the exhaust temperature. The direct causal relationship can be denoted as X. i →Y, the indirect causal relationship can be represented as X. i →X j →Y.

[0098] A causal correlation analysis model based on causal random forest is constructed and trained to determine the causal effect strength of each operating parameter;

[0099] Given intervention variable Di (Indicates whether to adjust the running parameter X) i (To conduct intervention), the operating parameters are divided into intervention and control groups, D i =1 represents the intervention group, D i =0 represents the control group, and the individual causal effect (ATE) is:

[0100] τ(x)=E[Y(1)-Y(0)|X=x], where Y(1) is the potential outcome after intervention (temperature), Y(0) is the potential outcome without intervention, and X=x represents the value of the given feature vector X.

[0101] The strongly correlated parameter X i Each parameter was treated as an intervention variable, and the causal effect of one operating parameter was analyzed at a time, with the remaining operating parameters treated as covariates.

[0102] For each intervention variable X i Construct binary processing variables, for example, if X i If the threshold is exceeded, then D i =1, otherwise D i =0.

[0103] The dataset was randomly divided into training and test sets to ensure that the intervention and control groups were balanced in terms of covariate distribution. A causal random forest model was fitted using the training set, with the optimization objective being to minimize the mean squared error between the estimated causal effect and the true effect. The model estimated the average causal effect within each leaf node by recursively partitioning the feature space. .

[0104] For each sample x, output the intervention variable X. i Estimated conditional causal effect of temperature Y .

[0105] The causal effect strength of each operating parameter can be assessed by ranking importance or using the SHAP value.

[0106] The significance test is used to determine whether there is a direct causal relationship between operating parameters and temperature, and the physical consistency test is used to verify whether the direction of the causal effect is reasonable.

[0107] Effect significance testing: Use bootstrap or random permutation tests to determine the estimated causal effect. Is it significantly different from zero? If the threshold is met and the p-value is less than 0.05, then X is considered... i There is a direct causal relationship with Y.

[0108] Physical consistency verification: Based on the mechanism of the air compressor station, determine whether the direction of the causal effect is reasonable. For example, the increase in cooling water temperature should lead to the increase in gas temperature, and the causal effect is positive.

[0109] Further analysis of the results of direct causal relationship analysis is conducted to determine whether there are intermediate variables, and indirect causal relationships are explored through recursive analysis and comparison of conditional causal effects.

[0110] Recursive analysis: For a confirmed direct causal relationship X i →Y, further analyze whether there is an intermediate variable X. j Make X i →X j →Y. For example, if the compressor power X1 directly affects the cylinder temperature Y, and the increase in power leads to an increase in the equipment load rate X2, and the load rate X2 in turn affects the temperature Y through the heat dissipation efficiency, then there exists a chain path X1→X2→Y.

[0111] Comparison of Conditional Causal Effects: Fixing the Intermediate Variable X j Comparing intervention variable X i If the effect weakens due to changes in temperature Y before and after the change, it indicates the existence of an indirect path. Mathematically, this can be expressed as: if... fixed Original, then X i →X j →Y may be true.

[0112] Based on direct and indirect causal relationships, a causal relationship diagram is constructed, and the causal direction (effect intensity) between each node (operational parameter) is marked.

[0113] For example, if there is an indirect causal relationship between compressor power X1 and cylinder temperature Y, and equipment load rate X2, then the causal effect strength between compressor power X1 and cylinder temperature Y is the sum of the effect strength between compressor power X1 and equipment load rate X2, the effect strength between equipment load rate X2 and cylinder temperature Y, and the effect strength between compressor power X1 and cylinder temperature Y.

[0114] Validate the trained model and causal paths;

[0115] Model validity verification: The prediction error and effect estimation stability of the causal random forest are evaluated by k-fold cross-validation.

[0116] Counterfactual testing: Design virtual intervention scenarios (such as forcibly reducing cooling water temperature) to verify whether the temperature changes predicted by the model conform to the physical laws of the air compressor station.

[0117] Physical verification of causal paths: Combining thermodynamic laws (such as energy conservation) and fluid mechanics principles, verify the physical rationality of each step in the chain path. For example, increased equipment load leading to insufficient heat dissipation area conforms to the physical logic of "load - heat dissipation efficiency - temperature". Invite domain experts to review the cause-effect graph and correct unreasonable relationships caused by data bias or model assumptions.

[0118] The model and causal graph are optimized through critical path prioritization, intervention strategy design, and model iteration.

[0119] Critical path prioritization: Based on the causal effect strength |τ| and path length, causal paths are prioritized to identify the chain paths that have the greatest impact on temperature, which are then the focus of optimization.

[0120] Intervention strategy design: Based on causal effect estimation, parameter control schemes are formulated, and the intervention effect is simulated using historical data.

[0121] Model iteration: Regularly update the data, incorporate new operating parameters, and retrain the causal random forest to adapt to equipment aging or changes in operating conditions.

[0122] The time correlation analysis unit is used to analyze the time correlation between various operating parameters and temperature;

[0123] Analyzing the time correlation between operating parameters and temperature requires considering the temporal dependence and dynamic correlation characteristics of time series data, as well as the physical mechanisms of the air compressor station. The focus should be on the lag effects and dynamic causal chains between variables. By quantifying the intensity of lag effects and the leading relationship, the correlation mechanism between operating parameters and temperature over time can be revealed, providing a basis for system optimization and prediction. This embodiment uses a vector autoregression (VAR) model for analysis, including the following steps:

[0124] The input is time series data, which includes a sequence of operating parameters and a temperature sequence;

[0125] Perform a stationarity test on the input data, such as using the unit root test (ADF test) to determine whether the sequence is stationary. If it is not stationary, convert it into a stationary sequence through differencing.

[0126] A time correlation analysis model based on a vector autoregression model is constructed and trained. The mathematical expression of the model is as follows:

[0127] ;

[0128] Among them, Y t Y represents the temperature at the current moment. t-i X represents the hysteresis value of temperature. t-i α represents the lag value of the operating parameter.i β i ϵ represents the regression coefficients, used to measure the influence of the lag term on the current temperature; p represents the lag order, which is determined by the AIC / BIC criterion. t This represents the random error term, which must satisfy the white noise assumption.

[0129] By iterating through different p-values, the lag order p that minimizes the AIC / BIC ratio is selected. The regression coefficient α is then estimated using the least squares (OLS) method. i β i The quantitative relationship between each hysteresis term and temperature was obtained. The model was tested by residual test (to check whether the residual is white noise, using Ljung-Box test) and stability test (if the eigenvalues ​​are inside the unit circle, the model is stable, using AR root test).

[0130] Based on the regression coefficient β i The sign and magnitude of the parameters determine the time relationship between the operating parameters and the temperature;

[0131] Specifically, the regression coefficient β i The sign of the coefficient β reflects the direction of the correlation (positive / negative correlation), and the absolute value reflects the strength of the influence. A significantly non-zero regression coefficient β... i This indicates that the operating parameters of the corresponding hysteresis step are time-dependent with temperature.

[0132] Verification was performed using a time-series causal chain, and dynamic response analysis was conducted using an impulse response function.

[0133] Verification of temporal causal chain: For example, in an air compressor station, "intake flow rate X1" and "compressor power X2" may change before "exhaust temperature Y". By analyzing the significance of the lag coefficient in the time correlation analysis model, it can be determined whether the lag term of intake flow rate X1 or compressor power X2 has a significant impact on exhaust temperature Y. The rationality of the correlation can be verified by combining physical logic (such as power increase → increased compression work → temperature increase).

[0134] Dynamic response analysis: Based on the model, the dynamic impact path of a unit impact on temperature of a certain operating parameter is analyzed by impulse response function (IRF). If the temperature rises significantly in the next k periods after the operating parameter is increased, it indicates that there is a positive time series correlation with a lag time of k.

[0135] In the correlation analysis module, after performing correlation analysis, the analysis results of statistical correlation, causal correlation and temporal correlation are fused and analyzed, and a hybrid correlation model is constructed by weighted fusion method.

[0136] For ease of understanding and calculation, the temperature output from statistical correlation, causal correlation, and time correlation analyses will be uniformly represented by the letter T. The model formula is as follows:

[0137] ;

[0138] in, This represents the overall forecast result. This represents the predicted value from the statistical correlation analysis model. The temperature response (T) of the causal correlation analysis model t +ATE), The values ​​represent the predicted values ​​from the time correlation analysis model, where ω1, ω2, and ω3 represent the weight coefficients, satisfying ω1 + ω2 + ω3 = 1, determined through cross-validation. The weight values ​​are updated using an adaptive update strategy.

[0139] The hybrid association model is trained based on preprocessed and feature-extracted data, and the weight coefficients are optimized. When validating the model, for regression models, the root mean square error (RMSE) can be used to reflect the average deviation between predicted and true values; the mean absolute error (MAE) can avoid the excessive influence of outliers on the evaluation; and the coefficient of determination (R²) can be used. 2 The value can be 0-1, and the closer it is to 1, the better the model fit. For time series models, the mean absolute percentage error (MAPE) can also be used, which can measure the relative error of the predicted value (e.g., MAPE=5% means that the average prediction error is 5% of the true value).

[0140] The real-time prediction process is as follows:

[0141] Input the current time's operating parameters X into the statistical correlation analysis model, causal correlation analysis model, and time correlation analysis model, respectively. t ;

[0142] Output of statistical correlation analysis model ;

[0143] The causal correlation analysis model calculates and sums the causal effects to obtain ;

[0144] The time correlation analysis model is based on historical series predictions. ;

[0145] The final prediction is obtained by weighted fusion of statistical association, causal association, and temporal association results. ;

[0146] like If the threshold is reached, the heat dissipation strategy will be adjusted.

[0147] For example, suppose the following parameters are collected at a certain moment:

[0148] Compressor speed X1 = 3000 rpm, cooling water flow rate X2 = 5 m / s, ambient temperature X3 = 30℃.

[0149] The prediction results of each model are as follows:

[0150] Statistical correlation analysis model: T 1 =75.3℃;

[0151] Causal correlation analysis model: If the flow velocity increases to 6 m / s, ATE = -2.1℃, the predicted temperature T 2 =73.2℃;

[0152] Time-correlation analysis model: Predicting temperature T by considering historical trends. 3 =76.8℃;

[0153] Fusion results (weights ω1=0.4, ω2=0.3, ω3=0.3)

[0154] T 0 =0.4×75.3+0.3×73.2+0.3×76.8=75.06℃;

[0155] System decision: The current temperature is close to the threshold (76℃). The cooling water flow rate will be automatically increased to 6m / s, and the temperature is expected to drop to 73.2℃, thus avoiding the risk of equipment overheating.

[0156] The heat dissipation execution module is used to perform reverse analysis on the predicted temperature change trend in combination with the operating parameters, determine the specific operating parameters affecting the current temperature change, and locate the specific heat-generating device or part based on the operating parameters for heat dissipation treatment.

[0157] The execution process includes the following steps:

[0158] Obtain the predicted temperature output by the hybrid correlation model, as well as the predicted temperature change trend over a future period, and identify areas of abnormal temperature rise;

[0159] Quantify the contribution of each operating parameter to the current temperature change using SHAP values ​​or feature importance ranking;

[0160] When calculating the SHAP value, it is first necessary to set a temperature baseline value. In this embodiment, the baseline temperature is set to 60°C.

[0161] The operating parameters with the highest contribution were selected as key influencing parameters, and pseudo-correlated parameters were eliminated in combination with the air compressor station process logic.

[0162] For example, if sensor measurement delay causes a strong correlation between "exhaust temperature" and "downstream valve opening", but the valve opening does not directly generate heat physically, this parameter should be removed, and parameters that conform to causal logic should be retained.

[0163] Based on the established causal graph, path tracing is performed. Through impulse response analysis, the time leader relationship between operating parameters is determined. For example, if the power change leads the temperature change by 5-10 minutes, the operating parameter with the highest contribution and a clear causal chain is taken as the dominant parameter of the current temperature change.

[0164] Establish a mapping relationship between operating parameters and equipment / parts, and locate specific equipment / parts based on spatial positioning algorithms and heat conduction path analysis;

[0165] Establish a parameter ledger for "equipment / part - monitoring parameters - sensor location - physical heat generation mechanism", for example, "compressor cylinder - cylinder temperature, compression power - cylinder surface - gas compression work, mechanical friction heat generation". Based on the sensor location, if the key parameter is "compressor power", the ledger shows that this parameter corresponds to the compressor cylinder and motor. Further compare the predicted temperature values ​​of these two parts. If the cylinder temperature rises more, it indicates insufficient heat dissipation in the compressor cylinder; if the motor winding temperature is abnormal, it indicates a motor fault (such as bearing wear or winding short circuit).

[0166] Based on the air compressor station layout diagram, analyze the heat transfer path. For example, a decrease in cooling water flow rate leads to a decrease in cooler heat exchange efficiency, which in turn leads to an increase in compressor exhaust temperature. Heat is then conducted to the downstream storage tank through pipelines, causing the tank wall temperature to rise. In this case, the dominant parameter is "cooling water flow rate," but the direct heat-generating components are the cooler and compressor cylinder.

[0167] Targeted heat dissipation strategies are adopted for different devices / parts located, and continuous monitoring is conducted.

[0168] For example, for coolers (equipment), clean the heat exchange tube bundles, increase the cooling water flow rate, and replace aging cooling media. After implementing heat dissipation measures, continuously monitor the temperature change trend, compare the predicted values ​​with the actual values, and calculate the error rate. If the error rate is >10%, it indicates inaccurate positioning or ineffective measures, requiring a re-examination of the mapping relationship between operating parameters and equipment / parts. Add new fault case data, update the parameter-temperature correlation rules, and improve the accuracy of subsequent reverse analysis.

[0169] Air-cooling system adjustment: If the problem is identified as related to the air-cooling system, such as abnormal fan speed leading to insufficient heat dissipation, increase the fan speed by controlling the fan motor's inverter when the temperature rises. This increases airflow and enhances heat dissipation.

[0170] Water cooling system adjustment: For water cooling systems, if a decrease in cooling medium flow is detected, the frequency converter of the water pump can be controlled to increase the pump speed and water flow. Simultaneously, check the status of pipeline valves; if the valve opening is insufficient, automatically adjust the valve opening to the appropriate position. If a decrease in cooler heat exchange efficiency is found, it may be due to scale buildup inside the cooler. In this case, an online cleaning system (such as a chemical or physical cleaning device) can be activated to clean the cooler and restore its heat exchange capacity.

[0171] Equipment operation adjustment: If the temperature change is caused by excessive air compressor load pressure, the air compressor's operating mode can be adjusted according to the actual air demand. For example, some air compressors can be switched to unloaded operation to reduce air production, lower the load pressure, and thus reduce heat generation. At the same time, check for leaks or abnormal air usage in the air-consuming equipment, repair leaks promptly, or adjust the operating parameters of the air-consuming equipment to optimize air balance and reduce the overall load on the air compressor.

[0172] Intelligent linkage control: All heat dissipation devices and actuators can work collaboratively. For example, when a temperature rise is detected, the air-cooled and water-cooled systems can simultaneously adjust their heat dissipation according to their respective regulation strategies, coordinating with the air compressor's operating mode adjustment to form an organic whole. Through intelligent linkage, rapid and efficient heat dissipation is achieved, ensuring that the air compressor station equipment temperature remains within a safe and stable range.

[0173] Utilize data visualization platforms such as Tableau, Power BI, or Industrial Internet of Things (IIoT) to display the contribution of operating parameters, temperature field distribution, and equipment location results in real time. Deploy lightweight models locally to achieve real-time early warning and rapid response to temperature anomalies (e.g., completing parameter analysis, equipment location, and alarm push within 5 minutes). Through a closed loop of model prediction, parameter attribution, equipment location, and precise heat dissipation, data-driven correlation analysis is combined with causal reasoning based on physical mechanisms to achieve rapid location of specific heat sources from temperature anomalies, avoiding the inefficiency of traditional device-by-device troubleshooting.

[0174] The following example illustrates the situation where the compressor cylinder in an oil-free digital energy air compressor station in a food processing workshop experiences an abnormal temperature rise during operation.

[0175] The temperature sensor collected the compressor cylinder surface temperature as 92℃ (safe threshold is 95℃), the pressure sensor collected the inlet pressure as 1.2MPa and the outlet pressure as 8.5MPa, the flow sensor collected the cooling water inlet flow rate as 50L / min and the outlet flow rate as 48L / min (abnormal difference > 5%), and the speed sensor collected the compressor rotor speed as 2900rpm.

[0176] Extract characteristic parameters, including temperature change rate, cooling efficiency index, and pressure fluctuation coefficient.

[0177] Statistical correlation analysis revealed a strong negative correlation between cylinder block temperature and cooling water flow rate (r = -0.82), a strong positive correlation between cylinder block temperature and compressor speed (r = 0.75), and a moderate correlation between cylinder block temperature and exhaust pressure (r = 0.3). Retaining the strongly correlated parameters, a linear regression model was used to predict a temperature of 91.5℃, which deviated from the measured value by 0.5℃.

[0178] Using cooling water flow rate as the intervention variable, intervention and control groups were divided, and the average causal effect ATE was calculated to be -3.2℃ (i.e., for every 10L / min increase in flow rate, the temperature decreases by 3.2℃). Validation using a causal diagram showed that the direct causal chain of cooling water flow rate → cooling efficiency → cylinder block temperature had no intermediate variables.

[0179] In the time correlation analysis, the VAR model lag order was set to p=2. The regression coefficients showed that the influence coefficient of cooling water flow rate lag-1 period on temperature was β=-0.45 (p<0.01), the influence coefficient of engine speed lag-2 period on temperature was β=0.32, and the influence coefficient of exhaust pressure lag-2 period on temperature was β=0.32 (p<0.05). This indicates that the effects of pressure and engine speed changes on temperature lag behind those of cooling water flow rate (lag-1 period), demonstrating that their influence on temperature is delayed and not an immediate dominant factor. Through impulse response analysis, after a 10% reduction in cooling water flow rate, the cylinder block temperature significantly increased by 1.5℃ after 5 minutes.

[0180] The weight coefficients of the hybrid correlation model were set, and the final weighted calculation yielded a current temperature prediction of 91.8℃.

[0181] Based on the final temperature predicted by the hybrid model, combined with the rate of temperature change, the rate of temperature rise due to insufficient flow in the causal effect, and the prediction of the hysteresis effect by the time series model, it is predicted that the temperature will rise to 95.3℃ in 10 minutes, exceeding the safety threshold of 95℃.

[0182] The current measured temperature of 92℃ has a mixed deviation of 0.2℃ from T, which is within the normal range. However, the predicted trend indicates that it will soon exceed the limit, triggering the reverse analysis process.

[0183] The contribution of each parameter to the difference between the current temperature of 92℃ and the baseline value (assuming 60℃) was analyzed by using SHAP values. The SHAP values ​​were calculated and sorted, and the order was: cooling water flow rate (SHAP=2.1) > speed (SHAP=1.8) > exhaust pressure (SHAP=0.9). The dominant parameter was determined to be cooling water flow rate, which should be analyzed first.

[0184] Analysis using a causal correlation model revealed that for every 10 L / min increase in cooling water flow rate, the temperature decreased by 3.2 °C. The causal diagram confirmed that the chain of "flow rate ↓ → cooling efficiency ↓ → temperature ↑" is a direct causal chain with no intermediate variables. A time correlation model showed that the β value for cooling water flow rate lags by one period (5 minutes ago) was -0.45, indicating that the current temperature is sensitive to changes in flow rate 5 minutes prior, confirming that a decrease in flow rate is a leading factor in the temperature increase.

[0185] Based on the established mapping relationship between operating parameters and equipment / parts, and combined with the physical layout analysis of the air compressor station, the heat conduction path was determined: flow rate ↓ → cylinder water jacket temperature ↑ → cylinder surface temperature ↑. The measured cooling efficiency index of 343 L・℃ / min < normal average of 400 L・℃ / min confirmed insufficient heat dissipation.

[0186] Cooling strategy execution: Start the backup cooling water pump and increase the flow rate to 60L / min (inverter frequency increased from 40Hz to 50Hz). Check the pipeline valves and find that the outlet valve opening is only 50%. Automatically adjust it to 80%. Turn on the axial fan above the cylinder block (speed increased from 1500rpm to 2000rpm) to enhance air convection. At the same time, reduce the compressor load rate from 85% to 75% and reduce the speed to 2800rpm to reduce heat generation.

[0187] Performance verification: Real-time monitoring showed that after 30 minutes, the cylinder temperature dropped to 88℃, the cooling water flow rate stabilized at 58L / min, and the cooling efficiency index rebounded to 410L・℃ / min. The deviation between the predicted value and the measured value of the hybrid model was 0.8℃, with an error rate of <10%, indicating that the strategy was effective.

[0188] System optimization: The fault data (92℃ temperature rise case) was added to the training set, the causal random forest model was updated, the flow-temperature correlation weight was strengthened, and the abnormal cooling water flow threshold was lowered from 5% to 3% based on the response data to trigger an early warning.

[0189] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. An oil-free digital energy air compressor station intelligent heat dissipation system, characterized in that, This includes a sensor module, a correlation analysis module, and a heat dissipation execution module that are connected in sequence via communication. The sensor module is used to collect various operating parameters of the air compressor station equipment through multiple types of sensors and to preprocess the collected data. The correlation analysis module is used to analyze the statistical correlation, causal correlation and temporal correlation between various operating parameters and temperature, and to construct a hybrid correlation model by integrating the correlation analysis results. The correlation analysis module includes a statistical correlation analysis unit, a causal correlation analysis unit, and a time correlation analysis unit; The statistical correlation analysis unit is used to analyze the statistical correlation between various operating parameters and temperature; The causal correlation analysis unit is used to analyze the causal correlation between various operating parameters and temperature, establish a causal diagram, and obtain the chain conduction path; The time correlation analysis unit is used to analyze the time correlation between various operating parameters and temperature; In the causal correlation analysis unit, the causal correlation between various operating parameters and temperature is determined by combining physical mechanisms, including the following steps: The operating parameters that are strongly correlated with temperature, as selected by statistical correlation analysis, are denoted as the feature set, and the target variable is set as temperature; Based on a preliminary analysis of the physical mechanism of the air compressor station, the direction of potential causal relationships is clarified, an initial causal hypothesis diagram is established, and direct and indirect causal relationships are marked. A causal correlation analysis model based on causal random forest is constructed and trained to determine the causal effect strength of each operating parameter; The significance test is used to determine whether there is a direct causal relationship between operating parameters and temperature, and the physical consistency test is used to verify whether the direction of the causal effect is reasonable. Further analysis of the results of direct causal relationship analysis is conducted to determine whether there are intermediate variables, and indirect causal relationships are uncovered through recursive analysis. Based on direct and indirect causal relationships, a causal relationship graph is constructed, and the causal directions between each node are marked. Validate the trained model and causal paths; The model and causal graph are optimized through critical path prioritization, intervention strategy design, and model iteration. The heat dissipation execution module is used to perform reverse analysis on the predicted temperature change trend in combination with the operating parameters, determine the specific operating parameters affecting the current temperature change, and locate the specific heat-generating device or part based on the operating parameters for heat dissipation treatment.

2. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, In the sensor module, the preprocessing includes data cleaning, data conversion, data alignment and synchronization, and data tagging.

3. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, The statistical correlation analysis unit includes the following steps: When analyzing the correlation between numerical parameters and temperature, the correlation coefficient is calculated and the calculated value is compared with a preset threshold to determine the strength of the correlation between the data. When analyzing the correlation between the categorical parameters and temperature, the method of analysis of variance is used. By comparing the mean temperature differences under different operating modes, the significance of the influence of mode type on temperature is determined. Based on the results of the statistical correlation analysis, the operating parameters that are strongly correlated with temperature are retained, and a statistical correlation analysis model based on linear regression is constructed to analyze the statistical relationship between various operating parameters and temperature. Based on the linear relationship between the current operating parameters and temperature, the current operating parameters are input into the corresponding linear model. The model outputs the predicted temperature. If the predicted temperature exceeds the preset threshold, the heat dissipation strategy is adjusted. At the same time, new data is collected periodically, and the model parameters are updated online using the stochastic gradient descent algorithm.

4. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, In the causal correlation analysis unit, a causal correlation analysis model based on causal random forest is constructed and trained, including the following steps: Given the intervention variable, the running parameters are divided into intervention and control groups, and the individual causal effect values ​​are calculated; Strongly correlated parameters are treated as intervention variables, and the causal effect of one operational parameter is analyzed each time, while the remaining operational parameters are treated as covariates. For each intervention variable, construct a binary treatment variable; The dataset is randomly divided into training and test sets; a causal random forest model is fitted using the training set, with the optimization objective being to minimize the mean squared error between the estimated causal effect and the true effect; the model estimates the average causal effect within each leaf node by recursively partitioning the feature space. For each sample, output the estimated conditional causal effect of the intervention variable on temperature; The causal effect strength of each operating parameter can be assessed by ranking importance or using the SHAP value.

5. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, The time correlation analysis unit includes the following steps: The input is time series data, which includes a sequence of operating parameters and a temperature sequence; Perform a stationarity test on the input data; Construct and train a time correlation analysis model based on a vector autoregression model; The time relationship between each operating parameter and temperature is determined based on the sign and magnitude of the regression coefficients; Verification was performed using a time-series causal chain, and dynamic response analysis was conducted using an impulse response function.

6. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, In the correlation analysis module, after performing correlation analysis, the analysis results of statistical correlation, causal correlation and temporal correlation are fused and analyzed, and a hybrid correlation model is constructed by weighted fusion method.

7. The intelligent heat dissipation system for an oil-free digital energy air compressor station according to claim 1, characterized in that, In the heat dissipation execution module, the execution process includes the following steps: Obtain the predicted temperature output by the hybrid correlation model, as well as the predicted temperature change trend over a future period, and identify areas of abnormal temperature rise; Quantify the contribution of each operating parameter to the current temperature change using SHAP values ​​or feature importance ranking; The operating parameters with the highest contribution were selected as key influencing parameters, and pseudo-correlated parameters were eliminated in combination with the air compressor station process logic. Based on the established causal graph, path tracing is performed. Through impulse response analysis, the time leader relationship between operating parameters is determined. The operating parameter with the highest contribution and a clear causal chain is taken as the dominant parameter of the current temperature change. Establish a mapping relationship between operating parameters and equipment, and locate specific equipment based on spatial positioning algorithms and heat conduction path analysis; For different devices identified, targeted heat dissipation strategies are adopted and continuously monitored.

Images