A multi-source parameter monitoring-based gas generator group intelligent deployment system
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AMICO GAS POWER CO LTD
- Filing Date
- 2026-04-15
- Publication Date
- 2026-06-19
Smart Images
Figure CN122052335B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of gas generator set technology, and in particular to an intelligent dispatching system for a group of gas generator sets based on multi-source parameter monitoring. Background Technology
[0002] Gas turbine generator sets are widely used in distributed energy, emergency power supply, and grid peak shaving due to their high efficiency, cleanliness, and flexible start-stop capabilities. In practical applications, they are often operated in the form of generator sets in a network. Currently, the dispatch and control of gas turbine generator sets mostly adopt traditional fixed strategies or independent control methods for individual units, which have the following technical shortcomings: First, control relies solely on single-dimensional operating parameters of the units, failing to integrate global operating parameters of the entire generator set group, making it difficult to achieve group-level collaborative optimization; second, the control rules are mostly statically set, only responding to current operating conditions and unable to predict parameter change trends based on historical operating data, lacking proactive control capabilities; third, the performance and health differences between units are not considered, and a uniform dispatch strategy can easily lead to uneven load distribution and over-operation of some units, reducing the overall operating efficiency and lifespan of the generator set group; fourth, the dispatch rules lack a closed-loop iterative mechanism, making continuous optimization based on actual operating results impossible, resulting in poor adaptability.
[0003] To address the aforementioned issues, there is an urgent need for an intelligent dispatching system for gas generator sets based on multi-source parameter monitoring to improve the overall operating performance, stability, and energy conservation and emission reduction levels of gas generator sets. Summary of the Invention
[0004] In view of this, this application provides an intelligent dispatching system for gas generator sets based on multi-source parameter monitoring to address the shortcomings of existing technologies.
[0005] The first aspect of this application provides an intelligent dispatching system for a group of gas generator sets based on multi-source parameter monitoring, including a multi-source parameter monitoring unit, a data processing unit, a rule fusion training unit, and an intelligent dispatching unit that are connected in sequence via communication.
[0006] The multi-source parameter monitoring unit is used to collect and store real-time operating parameters of the gas generator set in all dimensions, real-time operating parameters of the gas generator set group globally, and all historical operating parameters. The real-time operating parameters of the generator set in all dimensions include electrical parameters, mechanical operating parameters, and gas control parameters. The real-time operating parameters of the generator set group globally include power distribution diagrams, power generation curves per unit time, energy saving and emission reduction statistics, and equipment safety operation index.
[0007] The data processing unit outputs the performance parameters of the gas generator set and the health assessment results of the unit based on the real-time operating parameters of the unit in all dimensions and the real-time operating parameters of the unit group in the whole. It also generates real-time dispatch rules simultaneously. Based on the statistical analysis of the full historical operating parameters, it filters key influencing parameters, outputs parameter trend characteristics by combining the parameter change trends represented by the curvature of the corresponding curves, and generates predictive constraint rules.
[0008] The rule fusion training unit takes the overall performance of the gas generator group as the optimal goal, performs joint decision-making and fusion optimization on the real-time dispatch rules and the predictive constraint rules, and outputs the optimal dispatch rules that take into account both the real-time response and trend prediction of the gas generator group.
[0009] Based on the optimal allocation rules, the intelligent dispatching unit completes the selection of gas generator sets, the planning of unit start-up and shutdown and grid connection sequence, grid connection tripping control, load distribution and unit combination optimization of the gas generator set group.
[0010] In one possible implementation of the first aspect, the multi-source parameter monitoring unit integrates a device sensing acquisition module, a unit group data acquisition module, and a historical parameter storage module;
[0011] The equipment sensing and acquisition module is used to realize the real-time acquisition of all dimensions of the unit's operating parameters; the electrical parameters include active power, reactive power, power factor, voltage, battery voltage, power curve, and power current curve; the mechanical operating parameters include speed, engine water temperature, and engine oil pressure; the gas control parameters include throttle opening parameters and MAP parameters.
[0012] The generator group data acquisition module is used to summarize the real-time operating parameters of all gas generators under the gas generator group in all dimensions, and generate global real-time operating parameters of the generator group based on the statistical analysis of the real-time operating parameters in all dimensions.
[0013] The historical parameter storage module is used to store and backtrack all historical operating parameters.
[0014] In one possible implementation of the first aspect, the data processing unit performs noise reduction, outlier removal and validity verification on the real-time operating parameters of a single gas generator set, and then combines the global real-time operating parameters of the generator group to calculate and output the performance parameters and health assessment results of the single gas generator set through a multi-dimensional quantitative evaluation model.
[0015] The performance parameters include power generation efficiency, power fluctuation rate, and operating condition adaptability index.
[0016] The unit health assessment result is obtained by a comprehensive evaluation of multiple parameters, including the electrical parameters, the mechanical operating parameters, and the gas control parameters.
[0017] In one possible implementation of the first aspect, the data processing unit, based on the significance analysis and correlation mining of the full historical operating parameters, filters out multiple key influencing parameters that have a significant impact on the performance and health of the gas generator set.
[0018] By calculating and monitoring the curvature changes of the dynamic curves corresponding to key influencing parameters, the rate and magnitude of change of the corresponding parameters are characterized, and the deterioration state, abrupt change state, and normal operation state are identified, thus forming the parameter trend characteristics.
[0019] In one possible implementation of the first aspect, the real-time dispatching rule is generated based on the performance parameters of the gas generator set and the unit health assessment results, and is used to characterize the unit dispatching strategy under the current operating conditions. The unit dispatching strategy includes unit priority division, real-time load allocation, and grid connection tripping control logic. The predictive constraint rule is formed based on the parameter trend characteristics and is used to characterize the forward control strategy under trend prediction, including standby unit preparation, forward load adjustment, and unit combination optimization constraints.
[0020] In one possible implementation of the first aspect, the rule fusion training unit has a built-in dual-rule fusion model;
[0021] The dual-rule fusion model assigns corresponding weights to the real-time dispatch rules and the predictive constraint rules. Combining the global real-time operating parameters of the generator group, it conducts multiple rounds of simulation training with the overall performance of the gas generator group as the optimal goal. Based on actual operating feedback data, iteratively optimizes the rule parameters to achieve coordination and unification between the real-time dispatch rules and the predictive constraint rules. The evaluation indicators of the fusion optimization include the start-up efficiency, load distribution balance, operating stability, and comprehensive energy consumption and emission optimization of the gas generator group. The comprehensive energy consumption and emission optimization is quantitatively evaluated based on energy conservation and emission reduction statistics.
[0022] In one possible implementation of the first aspect, the intelligent allocation unit has a built-in unit difference adaptation module;
[0023] The unit difference adaptation module is used to quantify individual differences based on the performance parameters and unit health assessment results of each gas generator set to obtain quantified difference results. The quantified difference results include difference quantification value, difference level and difference core impact dimension.
[0024] The generator set difference adaptation module characterizes the degree of difference between each gas generator set based on the difference quantification value, determines the difference impact weight based on the difference level, locates the root cause of the difference based on the core impact dimension of the difference, and establishes a correlation model between the generator set start-up and shutdown and grid connection sequence and the overall operating performance of the gas generator set group.
[0025] In one possible implementation of the first aspect, the unit difference adaptation module is configured with a difference compensation mechanism;
[0026] The difference compensation mechanism is used to determine the compensation priority based on the correlation model and the difference level in the quantitative difference results, formulate targeted compensation strategies based on the core impact dimension of the difference and adjust the compensation range with reference to the quantitative difference value, determine the unit start-up and grid connection sequence that optimizes the overall operating performance of the gas generator group, and make adaptive corrections to the unit start-up and shutdown sequence, grid connection sequence and load distribution ratio.
[0027] In one possible implementation of the first aspect, the difference quantification value is used to characterize the degree of difference in performance and health among the gas generator sets; the difference level is divided into four levels according to the magnitude of the difference quantification value; the larger the difference quantification value, the higher the corresponding difference level, and the four levels are no difference, level one difference, level two difference, and level three difference; the core impact dimensions of the difference include electrical parameter dimension, mechanical operating parameter dimension, and gas control parameter dimension.
[0028] In one possible implementation of the first aspect, an execution unit and a monitoring and display unit are also included;
[0029] The execution unit includes a load adjustment module, a start / stop control module, and a grid-connected tripping execution module, which are used to execute the optimal allocation instructions output by the intelligent allocation unit and synchronously transmit the execution status, operation effect data, and actual operation feedback data back to the rule fusion training unit for iterative optimization of rule parameters.
[0030] The monitoring and display unit is used to visualize the real-time operating parameters of the unit in all dimensions, the real-time operating parameters of the unit group globally, and the unit status. It also supports receiving manual intervention commands and sending the manual intervention commands to the intelligent dispatching unit to adjust the dispatching strategy.
[0031] Its beneficial effects are as follows: This invention discloses an intelligent dispatching system for gas generator sets based on multi-source parameter monitoring, including a multi-source parameter monitoring unit, a data processing unit, a rule fusion training unit, and an intelligent dispatching unit. The multi-source parameter monitoring unit collects and stores real-time operating parameters of all units, real-time global parameters of the entire generator set group, and all historical operating parameters. The data processing unit combines the real-time parameters of the units and the entire generator set group to output performance parameters and health assessment results and generate real-time dispatching rules. It also filters key factors based on historical parameters and generates predictive constraint rules based on curve curvature. The rule fusion training unit optimizes the fusion of the two rules and outputs the optimal dispatching rules. The intelligent dispatching unit completes unit selection, start-up and shutdown grid connection timing planning, grid connection and generator tripping control, load allocation, and combination optimization based on the optimal dispatching rules. This invention realizes intelligent dispatching of generator sets driven by multi-source parameters, improving the stability, balance, and energy efficiency of generator set operation. Attached Figure Description
[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0033] Figure 1 This is a schematic diagram of the composition of an intelligent dispatching system for a gas generator group based on multi-source parameter monitoring, provided in an embodiment of this application.
[0034] Figure 2 This is a schematic diagram of the multi-source parameter monitoring unit in an embodiment of this application;
[0035] Figure 3 This is a schematic diagram of the execution unit composition in an embodiment of this application;
[0036] Figure 4 This is a schematic diagram of the workflow of an intelligent dispatching system for a gas generator group based on multi-source parameter monitoring, provided in an embodiment of this application. Detailed Implementation
[0037] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0038] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.
[0039] Example
[0040] This invention discloses an intelligent dispatching system for gas generator sets based on multi-source parameter monitoring, such as... Figure 1 As shown, the gas generator set group consists of five 500KW gas generator sets (corresponding to numbers #1 to #5) in the industrial park, of which #1 and #2 have been in use for 1 year, #3 and #4 for 3 years, and #5 for 5 years. There are significant differences in performance and health among the five gas generator sets. The five gas generator sets correspond to the industrial park's production load, living load, and emergency response, and are connected to the park's load end through a grid connection cabinet.
[0041] The system comprises a multi-source parameter monitoring unit, a data processing unit, a rule fusion training unit, and an intelligent allocation unit, all connected in sequence. It also includes an execution unit and a monitoring and display unit connected to the intelligent allocation unit. Data exchange between the units is achieved via industrial Ethernet, CAN bus, or wireless communication modules, ensuring real-time and stable communication. All units are deployed in the energy station control room of the industrial park for centralized management. The operating principles of each unit are as follows:
[0042] I. Working Principle of Multi-Source Parameter Monitoring Unit
[0043] like Figure 2 As shown, the multi-source parameter monitoring unit is the data acquisition layer. Its core function is to achieve full-dimensional, all-time, and high-precision parameter acquisition and storage. It integrates equipment acquisition modules, unit group acquisition modules, and historical parameter storage modules. The specific working principles of each module are as follows:
[0044] 1. Equipment sensor acquisition module
[0045] A distributed sensing architecture is adopted, with industrial-grade sensors deployed at corresponding measuring points of each gas generator set to directly collect real-time operating parameters across all dimensions. The sampling frequency is 1Hz (meeting the requirements of industrial field control). The sensor types and their corresponding measuring points are as follows:
[0046] Electrical parameter measurement points: Active power, reactive power, power factor, voltage and battery voltage are collected by a three-phase multi-function power meter, and power curves and power current curves are generated by a power recorder and deployed in the unit's distribution cabinet;
[0047] Mechanical operating parameter measurement points: The unit speed is collected by a speed sensor, the engine water temperature is collected by a temperature sensor, and the engine oil pressure is collected by a pressure sensor. All of these are deployed on the unit engine body.
[0048] Gas control parameter measurement points: Throttle opening parameters are collected by an angle sensor, and intake manifold pressure is collected by a MAP sensor, which are deployed in the unit's gas control loop.
[0049] The analog signals collected by the sensors are converted into standard signals by a signal isolator and then uploaded to the core controller of the equipment's sensor acquisition module via a remote data acquisition module, completing the acquisition and preprocessing of all dimensions of real-time operating parameters for a single unit.
[0050] 2. Unit Group Data Acquisition Module
[0051] The core controller uses a Siemens S7-1500 PLC, which communicates with the S7-1200 PLCs of units #1 to #5 via industrial Ethernet to achieve the aggregation of real-time operating parameters of all units across all dimensions, and generates global real-time operating parameters for the entire unit group based on engineering statistical algorithms. The specific generation principle is as follows:
[0052] Power distribution map: Based on the real-time active power of each unit, a pie chart statistical algorithm is used to calculate the proportion of the power of a single unit to the total power of the unit group, and generate a real-time power distribution chart to reflect the load distribution of the units.
[0053] Power generation curve per unit time: The total power generation of the unit group is statistically analyzed in 15-minute time units. A line graph fitting algorithm is used to generate a power generation curve, which reflects the matching relationship between load fluctuations in the park and power generation of the units.
[0054] Energy conservation and emission reduction statistics: Based on the gas consumption data (m³ / h) collected by the gas flow meter of each unit, combined with the power generation efficiency of the unit, the gas consumption rate per unit power generation (m³ / kWh) and pollutant emissions (NOx, g / kWh) are calculated and summarized on a daily / monthly basis.
[0055] Equipment safety operation index: Based on the threshold range of electrical parameters and mechanical operating parameters (such as the normal range of engine water temperature 80-95℃), a weighted scoring method is used (the weight of each parameter is calibrated by the industrial park operation and maintenance personnel) to calculate an index value of 0-100. An index <60 indicates that there is a safety hazard in the unit.
[0056] 3. Historical parameter storage module
[0057] An industrial cloud server is used as the storage medium, equipped with a MySQL database. It communicates with the unit group data acquisition module via industrial Ethernet to achieve full and timely storage and backtracking of operating parameters. The database establishes a three-dimensional index by unit number-parameter type-time stamp, with a storage period of 5 years. It supports fast queries by unit, by parameter type, and by time range (hour / day / month / year), with a query response time of less than 1 second, providing complete historical data support for subsequent data processing and trend analysis.
[0058] II. Working Principle of Data Processing Unit
[0059] The data processing unit is the core computing layer. Its core function is to perform in-depth processing on the collected parameters to generate real-time allocation rules and predictive constraint rules. The specific working principle is divided into two parts: real-time parameter processing and historical parameter trend analysis. The specific implementation principle is as follows:
[0060] (a) Real-time parameter processing: generating real-time allocation rules
[0061] 1. Parameter preprocessing
[0062] The real-time operating parameters of a single unit undergo a three-step preprocessing process: noise reduction, outlier removal, and validity verification, to ensure parameter accuracy. The specific algorithm is as follows:
[0063] For noise reduction, a moving average filtering algorithm (window size of 5) is used to eliminate random noise collected by the sensor.
[0064] Outlier removal, using 3 In principle, calculate the mean of the parameters. and standard deviation Remove The outliers are identified and supplemented by interpolation using parameter values from adjacent time points;
[0065] Validity verification uses parameter consistency verification, such as active power = The system checks the consistency of electrical parameters by measuring voltage, current, and power factor. If any parameters are inconsistent, they are marked as invalid, triggering a sensor fault alarm.
[0066] 2. Multi-dimensional quantitative evaluation model: Calculating performance parameters and unit health assessment results
[0067] The multi-dimensional quantitative assessment model is a weighted comprehensive assessment model that combines real-time parameters of individual units and real-time parameters of the entire unit group to calculate performance parameters and unit health assessment results. The model input includes preprocessed electrical parameters, mechanical operating parameters, gas control parameters, a global power distribution map of the unit group, and equipment safety operation index. The output includes performance parameters (power generation efficiency, power fluctuation rate, and operating condition adaptability index) and a health score of 0-100. The specific calculation principle is as follows:
[0068] 2a. Performance Parameter Calculation
[0069] Power generation efficiency: ( For real-time active power, This refers to gas consumption. (lower heating value of gas);
[0070] Power volatility: (Take the active power at 10 consecutive moments,) (mean)
[0071] Operating condition adaptability index: ( For the actual value of the parameter, The upper and lower limits of the parameter rating, The weights are for parameters, such as power generation efficiency (0.4), power fluctuation rate (0.3), and throttle opening (0.3). The weights are determined by the park's operation and maintenance personnel. The values range from 0 to 1, with the closer to 1 indicating better adaptability to operating conditions.
[0072] 2b. Calculation of Unit Health Assessment Results
[0073] Based on the normal threshold ranges of electrical parameters, mechanical operating parameters, and gas control parameters, a health score of 0-100 is calculated using a deduction system. The base score is 100 points, and points are deducted for parameters that deviate from the normal range. The more serious the deviation, the more points are deducted. For example, if the engine coolant temperature is >95℃, 2 points are deducted for every 1℃ exceeding the limit; if the oil pressure is <0.2MPa, 3 points are deducted for every 0.01MPa below the limit. At the same time, the overall equipment safety operation index of the unit group is combined. If the index is <60, an additional 10 points are deducted. The final score is the health assessment result. The higher the score, the better the health status of the unit.
[0074] 3. Real-time allocation rule generation
[0075] Based on performance parameters, health assessment results, and real-time global operating parameters of the unit group, combined with the compliance requirements of the industrial park, a unit scheduling strategy under the current operating conditions is generated, namely, real-time dispatch rules. This strategy comprises three parts, which are then implemented in the actual control of units #1 to #5:
[0076] Unit priority classification: Based on a comprehensive score of health level (weight 0.6) + power generation efficiency (weight 0.4), units are classified into three levels from high to low: Level I, Level II, and Level III. Level I units have priority in grid connection and load allocation. For example, Unit #1 (health level 95, power generation efficiency 42%) is Level I, and Unit #5 (health level 65, power generation efficiency 35%) is Level III.
[0077] Real-time load allocation: Based on the real-time load of the park, an equal power margin allocation algorithm is adopted to allocate higher loads (such as 500kW full load) to Class I units, medium loads (such as 300-400kW) to Class II units, and low loads (such as 100-200kW) to Class III units, ensuring that the total load of the group matches the load of the park and that the load allocation does not exceed the power margin of the units.
[0078] Grid-connected unit tripping control logic: When the park load is <1000kW, only 1-2 Class I units are put into operation; when the load is 1000-2000kW, 2-3 (Class I+II) units are put into operation; when the load is >2000kW, 4-5 (Class I+II+III) units are put into operation; when the unit health score is <50, the unit is immediately tripped and shut down for maintenance.
[0079] (ii) Historical parameter trend analysis: generating predictive constraint rules
[0080] 1. Selection of key influencing parameters
[0081] Based on the complete historical operating parameters (operating data of units #1 to #5 over the past year) in the historical parameter storage module, a combination algorithm of significance analysis (t-test) and correlation mining (Pearson correlation coefficient) is used to screen out key parameters that have a significant impact on unit performance and health. The specific principle is as follows:
[0082] Significance analysis: With unit power generation efficiency and health score as dependent variables and each operating parameter as independent variables, a t-test was conducted, and parameters with P < 0.05 were considered to have a significant impact.
[0083] Correlation mining: Calculate the Pearson correlation coefficient between the significantly influential parameters and the dependent variable, and select parameters with |r|>0.7 as key influential parameters.
[0084] For the five 500kW gas generator sets in the industrial park, the key influencing parameters identified were: engine coolant temperature, engine oil pressure, throttle opening, MAP parameters, and active power, providing core parameters for subsequent trend analysis.
[0085] 2. Parameter trend feature generation: Curvature calculation method
[0086] For the key influencing parameters identified through screening, their historical dynamic change curves are extracted. The curvature of these curves is calculated to characterize the rate and magnitude of parameter change, identifying three states: trend deterioration, trend abrupt change, and normal operation, thus forming parameter trend characteristics. The specific curvature calculation principle is as follows:
[0087] Let the curvature of the historical dynamic change of the parameter be... ( For time, (where k is a parameter value), the curvature power of the curve at a certain point is k (k is calculated using the curvature formula), and the larger k is, the faster the parameter changes and the greater the change amplitude;
[0088] Based on the unit operation data of the industrial park, the curvature thresholds are calibrated as follows: k < 0.05 indicates normal operation (parameters change smoothly); 0.05 ≤ k < 0.1 indicates a deteriorating trend (parameters slowly deviate from the normal range); and k ≥ 0.1 indicates a sudden change in trend (parameters deviate sharply from the normal range).
[0089] For example, the curvature of the engine water temperature curve of Unit #5 is 0.08, which is judged as a trend of deterioration, indicating that the water temperature is slowly rising and there is a potential fault.
[0090] 3. Generation of Prediction Constraint Rules
[0091] Based on parameter trend characteristics and combined with the production load forecast of the industrial park (such as the load rising to 2000kW during the morning peak from 8-10 am), a pre-control strategy under trend forecast is generated, namely the forecast constraint rule. This sets pre-constraint conditions for the real-time dispatch rules to avoid unit failures and load mismatches. Specifically, it includes three parts:
[0092] Backup unit preparation: When the key influencing parameters of a certain unit are in a state of deterioration, prepare a backup unit of the same level in advance. For example, if the water temperature of Unit #4 (Level II) is deteriorating, Unit #2 (Level I) will be set as a backup in advance and can be connected to the grid at any time.
[0093] Preemptive load adjustment: Based on the load change trend and parameter trend characteristics of the park, adjust the unit load in advance, such as gradually increasing the load of Class I units 30 minutes before the morning peak to avoid sudden load increases during peak hours that cause unit fluctuations.
[0094] Unit combination optimization constraints: It is prohibited to combine units in a deteriorating / abrupt state with other units to avoid the spread of faults; for example, if Unit #5 (Level III) is in a deteriorating state, it is constrained to operate alone and the load shall not exceed 100kW.
[0095] III. Working Principle of the Rule Fusion Training Unit
[0096] The rule fusion training unit is a rule optimization layer. Its core function is to fuse and optimize real-time allocation rules and predictive constraint rules, and output the optimal allocation rule that takes into account both real-time response and trend prediction. The built-in dual-rule fusion model runs on an industrial computer. The specific working principle is as follows:
[0097] 1. Dual-rule fusion model structure
[0098] The model is a combination of weighted fusion and simulation training. The inputs are real-time allocation rules, prediction constraint rules and global real-time operating parameters of the unit group. The output is the optimal allocation rule. The core of the model is dynamic weight allocation and multi-round simulation training, which is adapted to the operating characteristics of the unit group in the industrial park.
[0099] 2. Dynamic weight allocation
[0100] Dynamic weights are assigned to the real-time deployment rule (W1) and the predictive constraint rule (W2) to satisfy W1+W2=1. The weight values are dynamically adjusted based on the operating status of the unit group. The specific calibration rules are applied to the industrial park scenario.
[0101] All units in the generating group are operating normally, and the load in the park is stable: W1=0.7, W2=0.3 (priority response to real-time operating conditions).
[0102] Some units are in a state of trend deterioration, and the load in the park fluctuates slightly: W1=0.5, W2=0.5 (taking into account both real-time response and trend prediction).
[0103] Some units are in a state of sudden trend change, and the load in the park fluctuates greatly: W1=0.3, W2=0.7 (prioritize following the forecast constraints to ensure the safety of the units).
[0104] 3. Multi-round simulation training
[0105] Based on the digital twin model of five 500kW gas generator sets in the industrial park, and with the goal of optimizing the overall performance of the generator group, multiple rounds of simulation training were conducted. The specific principle is as follows:
[0106] Optimization objectives: The optimization function is to achieve the highest startup efficiency, the greatest load distribution balance, the highest operational stability, and the lowest overall energy consumption and emissions optimization. The optimal solution is obtained by using the particle swarm optimization algorithm (PSO).
[0107] Simulation training: Input dual rules with different weight combinations into the digital twin model to simulate the actual operating conditions of the industrial park (load fluctuations, unit status changes), and output the fusion optimization evaluation index value for each round of training;
[0108] Optimal solution selection: Select the weight combination that satisfies all evaluation index thresholds as the optimal weight under the current working condition. Based on this weight, fuse the two rules to obtain the initial optimal allocation rule.
[0109] 4. Iterative optimization of closed-loop rules
[0110] Based on the actual operational feedback data (execution status, operational effect data) returned by the execution unit, the initial optimal allocation rule is iteratively optimized, and the model parameters are updated. The specific principle is as follows:
[0111] Error calculation: Calculate the error between the evaluation index value of the simulation training and the evaluation index value of the actual operation. For example, if the simulation startup efficiency is 90% and the actual startup efficiency is 85%, the error is 5%.
[0112] Parameter update: Gradient descent is used to update the weight parameters of the dual-rule fusion model based on the error value, so that the error is gradually reduced;
[0113] Rule iteration: Substitute the updated parameters into the model, re-integrate the two rules, and obtain the optimal allocation rule after iteration to ensure that the rule continuously adapts to the actual operating conditions of the industrial park.
[0114] 5. Quantitative calculation of integrated and optimized evaluation indicators
[0115] All evaluation indicators are engineered and quantified to assess the actual operational effectiveness of the unit clusters in the industrial park. The specific calculation principles are as follows:
[0116] Startup efficiency: The closer the value is to 1, the higher the startup efficiency.
[0117] Load balancing: ( For the load of each unit, (This represents the average load); the closer it is to 1, the more balanced the load distribution.
[0118] Operational stability: The fluctuation rate of the unit's active power is used as an indicator. A fluctuation rate of <5% is considered stable. Operational stability = stable operating time / total operating time;
[0119] Comprehensive energy consumption and emission optimization degree: Based on energy conservation and emission reduction statistics, the comprehensive cost per unit of electricity generation (gas cost + environmental protection cost) is calculated. The lower the cost, the better the optimization degree. This indicator is quantitatively evaluated based on the energy conservation and emission reduction statistics of industrial parks.
[0120] IV. Working Principle of Intelligent Dispatch Unit
[0121] The intelligent dispatching unit is the dispatching execution layer. Its core function is to complete the selection of units #1 to #5 in the industrial park, start-up and shutdown and grid connection timing planning, grid connection tripping control, load distribution, and unit combination optimization based on the optimal dispatching rules. It has a built-in unit difference adaptation module to fully consider the individual differences of units #1 to #5 due to their service life, and realize differentiated dispatching. The working principle of each module is as follows:
[0122] 1. Core working principle of the unit difference adaptation module
[0123] Based on performance parameters and health assessment results, the module quantifies individual differences among units #1 to #5, obtaining quantified difference results (difference quantification value, difference level, and core impact dimension of the difference), and establishes a correlation model to provide a basis for difference compensation. The specific principle is as follows:
[0124] (1) Quantification of individual differences of generating units: landing on #1~#5 generating units
[0125] Difference Quantification Value: Characterizes the degree of difference in performance and health among generating units. It is calculated using the Euclidean distance method, with four dimensions: power generation efficiency, power fluctuation rate, operating condition adaptability index, and health score. The Euclidean distance between a single generating unit and the group average is calculated using the following formula: , The larger the value, the greater the difference. A single dimension value corresponding to a single unit. The mean of a single dimension value for the corresponding unit group; such as Unit #1. Unit #5 This indicates that Unit #5 differs significantly from the group average.
[0126] Difference Levels: Divided into four levels based on the magnitude of the difference quantification value. The larger the difference quantification value, the higher the level. Based on the actual difference situation of units #1-#5 in the industrial park, the thresholds are set as follows: No difference: D<10 (e.g., units #1-#2); Level 1 difference: 10≤D<20 (e.g., unit #3); Level 2 difference: 20≤D<30 (e.g., unit #4); Level 3 difference: D≥30 (e.g., unit #5).
[0127] Core impact dimensions of differences: The contribution rate analysis method is used to calculate the contribution rate of each parameter dimension to the quantitative difference value. The contribution rate = (Euclidean distance of a single dimension / total Euclidean distance) × 100%. The dimension with a contribution rate > 50% is the core impact dimension of differences. For example, the contribution rate of the mechanical operating parameters of Unit #5 is 75%, indicating that its core impact dimension of differences is the mechanical operating parameters (due to the long service life and aging of engine parts).
[0128] (2) Correlation model: The mapping relationship between unit start-up and shutdown and grid connection timing and overall group performance.
[0129] The correlation model is a multiple linear regression model. The inputs are the unit start-up, shutdown, and grid connection timing parameters (e.g., start-up time t1 and grid connection time t2 for Unit #1) and the quantified difference results. The output is the overall performance indicators of the unit group (start-up efficiency, load distribution balance, etc.). The model formula is: ( For the overall performance indicators of the group, For timing parameters, To quantify the differences, , , The regression coefficients are fitted using historical operational data of the industrial park to obtain the final correlation model, which can accurately characterize the mapping relationship between time series and overall group performance.
[0130] 2. Working principle of the difference compensation mechanism
[0131] The difference compensation mechanism is the core of the unit difference adaptation module. Based on the correlation model and quantified difference results, it formulates targeted compensation strategies for units #1 to #5, which are then implemented in actual start-up, shutdown, grid connection, and load distribution control to avoid the overall performance degradation of the group due to individual differences. The specific working principle is as follows:
[0132] Compensation priority: sorted from highest to lowest according to the level of difference, Level 3 difference (Unit #5) > Level 2 difference (Unit #4) > Level 1 difference (Unit #3) > No difference (Units #1~#2). The higher the priority, the greater the compensation.
[0133] Targeted compensation strategies: Strategies are formulated based on the core impact dimensions of differences. For differences in electrical parameters: adjust the load distribution ratio to reduce the load on the different units; for differences in mechanical operating parameters: adjust the start-up, shutdown, and grid connection sequences, extend the preheating time, and avoid frequent start-ups and shutdowns; for differences in gas control parameters: adjust the throttle opening to optimize gas supply. For example, for Unit #5 (with three levels of differences in mechanical operating parameters), extend its preheating time to 5 minutes, lag its grid connection sequence behind other units, and ensure its load distribution does not exceed 100kW.
[0134] Compensation range adjustment: The compensation range is adjusted with reference to the differential quantification value. The larger the differential quantification value, the higher the compensation range. For example, the load adjustment range for Unit #4 (D=25) is 30%, and the load adjustment range for Unit #5 (D=28.6) is 60%.
[0135] Final control: Based on the compensation strategy, select the start-up, shutdown and grid connection sequence that optimizes the overall performance of the group, and make adaptive adjustments to the start-up and shutdown order, grid connection sequence and load distribution ratio of units #1 to #5. For example, unit #1 (no difference) is started and connected to the grid first, and operates at full load of 500 kW; unit #5 (three-level difference) is started last and connected to the grid last, and operates at low load of 100 kW.
[0136] 3. Implementation of core intelligent allocation actions
[0137] Based on optimal allocation rules and difference compensation strategies, the five core allocation actions of Units #1 to #5 were completed, and the actual operation was implemented in the industrial park.
[0138] Unit selection: Select units with a health score of ≥60 and performance parameters that meet the standards. For example, select units #1 to #4 to put them into operation, and unit #5 to standby.
[0139] Start-up, shutdown, and grid connection sequence planning: Start-up should be in the order of Level I → Level II → Level III, with a 2-minute interval between grid connection times to avoid grid fluctuations caused by simultaneous grid connection; for example, the start-up, shutdown, and grid connection sequence of #1 → #2 → #3 → #4.
[0140] Grid connection and tripping control: Automatic grid connection / tripping is achieved through the circuit breaker of the grid connection cabinet. Soft grid connection technology is used during grid connection to avoid inrush current. Load transfer technology is used during tripping to transfer the load of the tripped unit to other units to ensure that the load of the park is stable.
[0141] Load allocation: Based on the differential compensation strategy, 500kW is allocated to Unit #1, 500kW to Unit #2, 400kW to Unit #3, and 300kW to Unit #4, with a total group load of 1700kW, matching the real-time load of the park.
[0142] Unit combination optimization: adopt a combination of Class I + Class II units, such as the combination of #1 + #2 + #3 units, to meet the 1200kW load demand of the park and avoid performance mismatch caused by the combination of Class III units and Class I / II units.
[0143] V. Working Principle of the Execution Unit
[0144] like Figure 3 As shown, the execution unit is the field execution layer. Its core function is to convert the optimal allocation instructions from the intelligent allocation unit into actions of the field equipment. It seamlessly connects with the existing control loops of units #1 to #5 in the industrial park without requiring modification of the original equipment. It includes load adjustment modules, start-stop control modules, and grid-connected unit switching execution modules. Each module is based on an industrial controller. The specific working principle is as follows:
[0145] The load adjustment module controls the throttle actuator of the gas generator set through analog signals to adjust the throttle opening and achieve precise adjustment of the load distribution ratio. For example, when it receives the instruction "#1 unit load 500kw", the controller outputs a 20mA signal to adjust the throttle opening to 100% to achieve full-load operation.
[0146] The start-stop control module controls the start motor and stop solenoid valve of the unit through switch signals to achieve automatic start-stop. For example, when it receives the command "#1 Unit Start", the controller outputs a closed signal, the start motor works, and the unit starts; when it receives the command "#5 Unit Stop", it outputs a disconnect signal, the stop solenoid valve is activated, and the unit stops.
[0147] The grid-connected and tripping execution module controls the vacuum circuit breaker of the grid-connected cabinet through switch signals to achieve automatic grid connection / tripping. Before grid connection, synchronization detection (voltage, frequency, phase matching) is performed. After the detection is qualified, the circuit breaker is closed to achieve grid connection. Before tripping, the load is transferred first, and then the circuit breaker is opened to achieve tripping.
[0148] Data feedback involves the execution unit collecting the execution status of each module (such as the actual value of throttle opening and the on / off status of the circuit breaker) and operational performance data (such as the actual load and actual grid connection time). This actual operational feedback data is then transmitted back to the rule fusion training unit via industrial Ethernet for iterative optimization of rule parameters, forming a closed loop of scheduling-execution-feedback.
[0149] VI. Working Principle of the Monitoring and Display Unit
[0150] The monitoring and display unit is the human-computer interaction layer. Its core function is to realize visual monitoring and manual intervention, adapting to the operating habits of industrial park maintenance personnel. It is based on an industrial touch screen and a host computer and is deployed in the energy station control room. The specific working principle is as follows:
[0151] 1. Visual presentation
[0152] A configuration interface is used to realize the real-time visualization of various data and unit status, which is applied to the operation and maintenance needs of industrial parks.
[0153] 2. Artificial intervention
[0154] It supports industrial park operation and maintenance personnel in manually issuing intervention commands to deal with emergencies (such as sudden power outages or unit failures). The intervention commands are issued through the configuration interface and, after being parsed by the intelligent dispatch unit, the dispatch strategy is adjusted. Specific intervention methods include:
[0155] Manual start / stop / grid connection / shutdown: Maintenance personnel can manually click on the equipment icon to issue start, stop, grid connection and shutdown commands for units #1 to #5;
[0156] Manual load adjustment: Maintenance personnel can manually input the unit load value and adjust the load distribution ratio of a single unit;
[0157] Manual rule modification: Operation and maintenance personnel can manually modify the parameters of the real-time dispatch rules (such as unit priority and load allocation threshold) according to the actual working conditions.
[0158] VII. Complete Workflow of this Embodiment in the Industrial Park
[0159] Based on the actual application scenario of a group of 5 500kW gas generator sets in an industrial park, the complete workflow of this invention is as follows (e.g. Figure 4 As shown), it achieves fully automated intelligent allocation without human intervention:
[0160] Data acquisition: The multi-source parameter monitoring unit collects real-time operating parameters of all dimensions of units #1-#5 through sensing devices, summarizes and generates group-wide real-time operating parameters, and stores them in the historical parameter storage module;
[0161] Data processing and rule generation: The data processing unit preprocesses real-time parameters, calculates performance parameters and health assessment results through a multi-dimensional quantitative evaluation model, and generates real-time allocation rules; at the same time, it screens key influencing factors based on historical parameters, calculates curvature to obtain parameter trend characteristics, and generates predictive constraint rules.
[0162] Rule fusion optimization: The dual-rule fusion model of the rule fusion training unit dynamically assigns weights to real-time and prediction rules, combines the digital twin model for multi-round simulation training, iterates and optimizes based on the actual feedback data of the execution unit, and outputs the optimal allocation rule;
[0163] Intelligent dispatching and differential compensation: Based on the optimal dispatching rules, the intelligent dispatching unit performs differential quantification and correlation modeling of units #1-#5 through the unit differential adaptation module, formulates targeted compensation strategies through the differential compensation mechanism, and completes core dispatching actions such as unit selection, start-up and shutdown grid connection timing planning, and load allocation.
[0164] On-site execution and data feedback: The execution unit translates dispatch instructions into on-site equipment actions to realize automatic start-up and shutdown of the unit, grid connection, and load adjustment, and sends the execution status and operation effect data back to the rule fusion training unit to realize closed-loop rule iteration;
[0165] Visualized monitoring and manual intervention: The monitoring and display unit shows various data and the allocation process in real time. Operation and maintenance personnel can issue manual intervention commands based on emergencies to adjust allocation strategies and ensure stable power supply in the park.
[0166] This embodiment provides an intelligent dispatching system for gas generator sets based on multi-source parameter monitoring, which makes at least the following technical contributions compared to existing technologies:
[0167] 1. Through the multi-source parameter monitoring unit, the gas generator set can be fully collected and stored in real time, the group as a whole in real time, and all historical operating parameters. This breaks through the limitation of traditional control relying only on the single-dimensional parameters of a single unit, and provides a comprehensive and accurate parameter basis for the intelligent allocation of the group.
[0168] 2. The data processing unit is set to generate real-time dispatch rules and predictive constraint rules respectively, which can not only respond to the actual operating conditions of the current unit and unit group, but also realize forward control based on the trend analysis of historical parameters, thereby improving the timeliness and predictability of the system's control.
[0169] 3. The two rules are fused and optimized through the rule fusion training unit, and the rules are closed-loop iteratively implemented by combining actual operation feedback data, so that the optimal allocation rule can continuously adapt to the operation changes of the unit group, thereby improving the adaptability and optimization effect of the rule.
[0170] 4. Set up a unit difference adaptation module in the intelligent dispatching unit to fully consider the performance and health differences of each gas generator set. Through difference quantification, correlation modeling and difference compensation, realize differentiated dispatching strategies, avoid uneven load distribution, improve the overall operating efficiency of the unit group and extend the service life of the units.
[0171] 5. The system is equipped with execution and monitoring units, which can accurately execute optimal allocation commands, visualize parameters and status, and support manual intervention, thus balancing automated operation and manual controllability and improving the system's usability.
[0172] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computing software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0173] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.
[0174] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. An intelligent dispatching system for a gas generator set group based on multi-source parameter monitoring, wherein the gas generator set group comprises multiple gas generator sets, characterized in that, It includes a multi-source parameter monitoring unit, a data processing unit, a rule fusion training unit, and an intelligent allocation unit that are connected in sequence via communication. The multi-source parameter monitoring unit is used to collect and store real-time operating parameters of the gas generator set in all dimensions, real-time operating parameters of the gas generator set group globally, and all historical operating parameters. The real-time operating parameters of the generator set in all dimensions include electrical parameters, mechanical operating parameters, and gas control parameters. The real-time operating parameters of the generator set group globally include power distribution diagrams, power generation curves per unit time, energy saving and emission reduction statistics, and equipment safety operation index. The data processing unit outputs the performance parameters of the gas generator set and the health assessment results of the unit based on the real-time operating parameters of the unit in all dimensions and the real-time operating parameters of the unit group in the whole. It also generates real-time dispatch rules simultaneously. Based on the statistical analysis of the full historical operating parameters, it filters key influencing parameters, outputs parameter trend characteristics by combining the parameter change trends represented by the curvature of the corresponding curves, and generates predictive constraint rules. The rule fusion training unit takes the overall performance of the gas generator group as the optimal goal, performs joint decision-making and fusion optimization on the real-time dispatch rules and the predictive constraint rules, and outputs the optimal dispatch rules that take into account both the real-time response and trend prediction of the gas generator group. Based on the optimal allocation rules, the intelligent dispatching unit completes the selection of gas generator sets, the planning of unit start-up and shutdown and grid connection sequence, grid connection tripping control, load distribution and unit combination optimization of the gas generator set group; The rule fusion training unit has a built-in dual-rule fusion model; The dual-rule fusion model assigns corresponding weights to the real-time dispatch rules and the predictive constraint rules. Combining the global real-time operating parameters of the generator group, it conducts multiple rounds of simulation training with the overall performance of the gas generator group as the optimal goal. Based on actual operating feedback data, iteratively optimizes the rule parameters to achieve coordination and unification between the real-time dispatch rules and the predictive constraint rules. The evaluation indicators of the fusion optimization include the start-up efficiency, load distribution balance, operating stability, and comprehensive energy consumption and emission optimization of the gas generator group. The comprehensive energy consumption and emission optimization is quantitatively evaluated based on energy conservation and emission reduction statistics.
2. The multi-source parameter monitoring based intelligent deployment system for gas generator groups according to claim 1, characterized in that, The multi-source parameter monitoring unit integrates a device sensor acquisition module, a unit group data acquisition module, and a historical parameter storage module. The equipment sensing and acquisition module is used to realize the real-time acquisition of all dimensions of the unit's operating parameters; the electrical parameters include active power, reactive power, power factor, voltage, battery voltage, power curve, and power current curve; the mechanical operating parameters include speed, engine water temperature, and engine oil pressure; the gas control parameters include throttle opening parameters and MAP parameters. The generator group data acquisition module is used to summarize the real-time operating parameters of all gas generators under the gas generator group in all dimensions, and generate global real-time operating parameters of the generator group based on the statistical analysis of the real-time operating parameters in all dimensions. The historical parameter storage module is used to store and backtrack all historical operating parameters.
3. The multi-source parameter monitoring based intelligent deployment system for gas generator groups according to claim 1, characterized in that, After performing noise reduction, outlier removal, and validity verification on the real-time operating parameters of a single gas generator set, the data processing unit combines the global real-time operating parameters of the generator group and calculates and outputs the performance parameters and health assessment results of the single gas generator set through a multi-dimensional quantitative evaluation model. The performance parameters include power generation efficiency, power fluctuation rate, and operating condition adaptability index. The unit health assessment result is obtained by a comprehensive evaluation of multiple parameters, including the electrical parameters, the mechanical operating parameters, and the gas control parameters.
4. The multi-source parameter monitoring based intelligent deployment system for gas generator groups according to claim 1, characterized in that, Based on the significance analysis and correlation mining of the full historical operating parameters, the data processing unit screens out several key parameters that have a significant impact on the performance and health of the gas generator set. By calculating and monitoring the curvature changes of the dynamic curves corresponding to key influencing parameters, the rate and magnitude of change of the corresponding parameters are characterized, and the deterioration state, abrupt change state, and normal operation state are identified, thus forming the parameter trend characteristics.
5. The intelligent dispatching system for gas generator sets based on multi-source parameter monitoring according to claim 1, characterized in that, The real-time dispatching rules are generated based on the performance parameters of the gas generator sets and the health assessment results of the units, and are used to characterize the unit dispatching strategy under the current operating conditions. The unit dispatching strategy includes unit priority division, real-time load allocation, and grid connection tripping control logic. The predictive constraint rules are formed based on the parameter trend characteristics and are used to characterize the forward control strategy under trend prediction, including standby unit preparation, forward load adjustment, and unit combination optimization constraints.
6. The intelligent dispatching system for gas generator sets based on multi-source parameter monitoring according to claim 1, characterized in that, The intelligent allocation unit has a built-in unit difference adaptation module; The unit difference adaptation module is used to quantify individual differences based on the performance parameters and unit health assessment results of each gas generator set to obtain quantified difference results. The quantified difference results include difference quantification value, difference level and difference core impact dimension. The generator set difference adaptation module characterizes the degree of difference between each gas generator set based on the difference quantification value, determines the difference impact weight based on the difference level, locates the root cause of the difference based on the core impact dimension of the difference, and establishes a correlation model between the generator set start-up and shutdown and grid connection sequence and the overall operating performance of the gas generator set group.
7. The intelligent dispatching system for gas generator sets based on multi-source parameter monitoring according to claim 6, characterized in that, The unit difference adaptation module is equipped with a difference compensation mechanism; The difference compensation mechanism is used to determine the compensation priority based on the correlation model and the difference level in the quantitative difference results, formulate targeted compensation strategies based on the core impact dimension of the difference and adjust the compensation range with reference to the quantitative difference value, determine the unit start-up and grid connection sequence that optimizes the overall operating performance of the gas generator group, and make adaptive corrections to the unit start-up and shutdown sequence, grid connection sequence and load distribution ratio.
8. The multi-source parameter monitoring based intelligent deployment system for gas generator groups according to claim 6, characterized in that, The difference quantification value is used to characterize the degree of difference in performance and health among various gas generator sets; the difference level is divided into four levels according to the size of the difference quantification value; the larger the difference quantification value, the higher the corresponding difference level, and the four levels are no difference, level one difference, level two difference, and level three difference; the core impact dimensions of the difference include electrical parameter dimension, mechanical operating parameter dimension, and gas control parameter dimension.
9. The multi-source parameter monitoring based intelligent deployment system for gas generator groups according to claim 1, characterized in that, It also includes an execution unit and a monitoring and display unit; The execution unit includes a load adjustment module, a start / stop control module, and a grid-connected tripping execution module, which are used to execute the optimal allocation instructions output by the intelligent allocation unit and synchronously transmit the execution status, operation effect data, and actual operation feedback data back to the rule fusion training unit for iterative optimization of rule parameters. The monitoring and display unit is used to visualize the real-time operating parameters of the unit in all dimensions, the real-time operating parameters of the unit group globally, and the unit status. It also supports receiving manual intervention commands and sending the manual intervention commands to the intelligent dispatching unit to adjust the dispatching strategy.