Combined control method and system for multiple fuel generator sets

Abstract

The invention belongs to the technical field of equipment control, and discloses a combined control method and system for multiple fuel generator sets. Comprising the steps of collecting operation data of each fuel generator and performing data cleaning to obtain a generator operation data set; performing oscillation output analysis on the generator operation data set to obtain oscillation data characteristics; offset compensation is carried out on the generator operation data set, and an offset suppression data set is output; performing response delay correction on the offset suppression data set to generate a delay adjustment data set; designing a hot start penalty strategy based on the unit load information and the delay adjustment data set; performing excitation instability early warning based on the unit load information, and generating a magnetic field early warning log; constructing a combined control strategy; sending the combined control strategy to a preset fuel generator set control terminal; and the comprehensive efficiency of a plurality of fuel generating set systems in the aspects of intelligent control and safe operation is obviously improved.

Description

Technical Field

[0001] This invention relates to the field of equipment control technology, and more specifically, to a method and system for the joint control of multiple fuel-powered generator sets. Background Technology

[0002] In parallel power generation systems consisting of multiple fuel-fired generators, achieving coordinated control among multiple units has become a key issue, especially in operating scenarios represented by fuel-fired generator sets. Although some progress has been made in centralized or distributed management of the units by introducing data acquisition systems and control algorithms, the control accuracy is still significantly insufficient and it is difficult to meet the requirements of efficient scheduling and real-time optimization under complex operating conditions. Existing technologies still face many challenges. Because existing traditional control strategies rely heavily on single-point power or frequency data for judgment, they struggle to identify internal hidden instability phenomena caused by slight parameter deviations or oscillatory behavior. For example, when different generators with similar frequencies operate in parallel, undetectable micro-oscillations at different frequencies may occur, causing frequent slight voltage drops at the load end and affecting the stable power supply to voltage-sensitive equipment. Furthermore, existing traditional control strategies often employ a unified command issuance mode for multiple generator units, failing to consider the potential differences in startup response among individual generators. This can easily lead to delayed responses, leading interference, or even misjudged faults. For instance, when the load suddenly increases to a certain threshold, the control system may require all generators in the unit to reduce power, but the faster-responding generators may not be able to do so. One unit may have already completed its operation, while other slow-response units have not yet entered the operating range. This could lead the control system to believe that some units have not received the instruction and send a second instruction, resulting in control logic confusion. At the same time, current traditional control strategies lack assessment of thermal conditions and fail to identify and limit the thermal fatigue and wear that may be caused by high-frequency start-stop, exacerbating the overall operational risks of the unit. On the other hand, in some low-load operating scenarios, such as at night or during emergency standby, some units may experience excitation weakening or even instability in a short period of time because the excitation coil fails to maintain stable energy. However, the control system only alarms when it detects that the magnetic field has completely failed, lacking early warning and protection measures.

[0003] In view of this, the present invention proposes a method and system for joint control of multiple fuel generator sets to solve the above problems. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for joint control of multiple fuel-powered generator sets, comprising: S1. Collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset; S2. Perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; perform offset compensation on the generator operation dataset based on the oscillation data characteristics to output an offset-suppressed dataset; S3. Perform response delay correction on the offset suppression dataset to generate a delay adjustment dataset; S4. Obtain unit load information in real time, and design a hot start penalty strategy based on unit load information and delay adjustment dataset; S5. Based on unit load information, perform excitation instability early warning and generate magnetic field early warning log; S6. Construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning log, and delay adjustment dataset; send the joint control strategy to the preset fuel generator set control terminal.

[0005] Furthermore, the method for performing oscillation output analysis includes: Obtain the generator number, divide the generator operation dataset based on the generator number to obtain individual generator operation data; extract the power output, voltage value and power generation frequency from each generator individual operation data, and construct the power sequence, voltage sequence and frequency sequence respectively; The average rate of change for each power sequence is calculated, and a unit output period is set. A power sliding window of the unit output period size is constructed to traverse the power sequences. The number of power change direction changes within the same power sliding window in each power sequence is counted, and the average number of direction changes for the corresponding power sequence is calculated. The average rate of change and the average number of direction changes for each power sequence are integrated to obtain the power coordinated fluctuation feature. The voltage difference between the maximum and minimum voltage values ​​within the same unit output period in each voltage sequence is calculated. The timestamp corresponding to the maximum voltage value in the voltage sequence is identified, and the average voltage synchronization time deviation is calculated based on the timestamps of reaching the maximum voltage value in all voltage sequences. The maximum voltage difference and the average voltage synchronization time deviation are integrated as the voltage fluctuation feature. The frequency difference of each frequency sequence within the same unit output period is calculated, and the frequency difference of all unit output periods is counted to construct a frequency deviation matrix. The power coordinated fluctuation feature, voltage fluctuation feature, and frequency deviation matrix are concatenated to obtain the oscillation data feature.

[0006] Furthermore, the method for performing offset compensation includes: If the average number of direction changes in the oscillation data features is higher than the preset number threshold, and the average rate of change is higher than the preset rate of change fluctuation threshold; at the same time, the maximum voltage difference is higher than the preset voltage difference threshold, and the average time deviation of voltage synchronization is less than the preset deviation threshold, it is determined that the oscillation type is amplitude fluctuation-dominated oscillation; if the average frequency difference in the frequency offset matrix is ​​higher than the preset frequency difference threshold, and the average number of direction changes is less than or equal to the preset number threshold, it is determined that the oscillation type is phase deviation-dominated oscillation. When the oscillation type is amplitude fluctuation-dominated oscillation, an abnormal segment is identified in the power sequence where the direction of power change changes continuously and the number of power change direction changes exceeds a preset local number threshold. A direction change pause operation is performed on the abnormal segment, and the power output in a single direction is restricted to update continuously at a fixed step size. The local voltage difference of the abnormal segment is calculated, and the ratio of this local voltage difference to the maximum voltage difference in all voltage sequences is calculated to obtain the voltage response coefficient. The voltage response coefficient is used to weight the voltage value of the unit output cycle to which the abnormal segment belongs. When the oscillation type is phase deviation-dominated oscillation, the parameter compensation amount is constructed to adjust the corresponding dimension parameters in the generator operation dataset; the offset-compensated offset suppression dataset is output.

[0007] Furthermore, the method for compensating the construction parameters includes: Calculate the mean value of the row elements corresponding to each generator in the frequency offset matrix, and use the mean value of the row elements as the frequency correction reference value for that generator; construct a time offset function based on the frequency correction reference value and the unit time resolution, and use the time offset function to calculate the time index compensation amount; Identify the timestamp corresponding to the maximum voltage value in the voltage sequence. In the power sequence, extract time segments of fixed length before and after the timestamp position to form a power local response window. Calculate the maximum power change rate within the power local response window, and simultaneously calculate the ratio of the maximum power change rate to the maximum power change rate in the same power local response window of other generators, and obtain the average of the ratios. Weight the maximum power change amplitude within the power local response window to the average of the ratios to obtain the power response offset. Integrate the time index offset and the power response offset as a parameter compensation amount.

[0008] Furthermore, the method for performing response latency correction includes: Extract the power sequence corresponding to each generator in the offset suppression dataset, detect the time stamp when the power sequence first shows a continuous positive change after the control command is issued, and calculate the time interval between this time stamp and the control command issuance time stamp to obtain the response interval; compare the response interval of each generator with the response deviation of the standard response interval in the historical normal response record. If the response deviation is higher than the preset response stability upper limit, the corresponding generator is identified as a lagging response individual; if the response deviation is lower than the preset response stability lower limit, the corresponding generator is identified as an early response individual. For individuals with delayed responses, control commands are not executed immediately. Instead, control commands are redistributed to fast-response generator individuals whose response interval is higher than the preset lower limit of response stability but less than or equal to the average response interval. The task allocation ratio of the fast-response generator group composed of fast-response generator individuals is calculated, and the task load of the fast-response generator group is adjusted based on the task allocation ratio. For individuals that respond early, the response lead between the response interval and the preset response stability lower limit is calculated, and a reduction factor is constructed based on the response lead. The power sequences of the corresponding individuals that respond early are weighted using the reduction factor. The power sequences of the adjusted generator individuals are integrated and used to cover the corresponding power sequences in the offset suppression dataset to obtain the delay adjustment dataset.

[0009] Furthermore, the design of the hot-start penalty strategy includes the following methods: Extract the operating timestamps corresponding to the unit load information, calculate the time interval between the operating timestamp and the last shutdown timestamp of each generator, and obtain the idle interval; calculate the load fluctuation value and load average value per unit output cycle based on the unit load information, and construct the load behavior characteristics; construct the thermal state index based on the idle interval and load behavior characteristics. The process involves calculating the ratio of the unit-time load value to the unit-time standard load to obtain the load loading ratio of the corresponding generator; acquiring the maximum power output and minimum stable power output of the corresponding generator in multiple historical operating cycles, and calculating the compensable load based on the maximum power output and minimum stable power output; calculating the ratio of the thermal state index to the compensable load to obtain the unit-time load penalty factor; calculating the average penalty factor based on all unit-time load penalty factors, and selecting generators with unit-time load penalty factors less than the average penalty factor as scheduling candidate units; adjusting the unit-time load value of each generator in the scheduling candidate units to ensure that the load loading ratio reaches the preset load ratio; designating generators with unit-time load penalty factors greater than the preset tolerance threshold as scheduling restricted units and executing a cooling process; and encoding the processing into a hot-start penalty strategy.

[0010] Furthermore, the method of constructing the thermal state index includes: The historical average load of each generator in the unit load information is extracted from the previous operating cycle corresponding to the last shutdown timestamp. When the idle interval is greater than the preset cooling time limit and the historical average load of the previous operating cycle is lower than the preset standard cycle load threshold, the corresponding generator is determined to be of low heat level. If the idle interval is less than or equal to the preset cooling time limit and the load fluctuation value per unit output cycle of the previous operating cycle is higher than the preset load fluctuation threshold, the corresponding generator is determined to be of medium heat level. When the idle interval is lower than the preset cooling time limit and the historical average load of the previous operating cycle is higher than the preset standard cycle load threshold, and the load fluctuation value per unit output cycle is lower than the preset load fluctuation threshold, the corresponding generator is determined to be of high heat level. If it does not meet any of the heat levels, the corresponding heat level is determined to be a fuzzy heat level. By matching the low heat level, medium heat level, high heat level, and fuzzy heat level with the preset thermal state parameter table, the thermal state index of the corresponding level is obtained.

[0011] Furthermore, the methods for providing early warning of excitation instability include: Identify the load value of each generator in the unit load information and construct a load value change sequence; set a critical load power threshold, and extract the continuous time segment in the load value change sequence where the load value is lower than the critical load power threshold, which is the extremely low load segment; Extract the excitation voltage data in the extremely low load section, calculate the change amplitude of the excitation voltage data in the extremely low load section, and determine the main change direction of the excitation voltage. If the main change direction of the excitation voltage is downward and the change amplitude is lower than the preset fluctuation range, the corresponding extremely low load section is determined as the excitation weakening section. Detect the time length occupied by the excitation weakening section. If the time length is higher than the preset instability risk judgment threshold, the corresponding excitation weakening section is designated as an instability warning section. Extract the time boundary range, load value, excitation voltage data, period index, and warning mark of the instability warning section to form a magnetic field warning log.

[0012] Furthermore, the method of constructing the joint control strategy includes: The delay adjustment dataset is grouped based on the generator number. Each sub-delay adjustment dataset is combined with the hot start penalty strategy and magnetic field warning log to form an individual generator control data unit. All individual generator control data units are combined according to the preset fuel generator set control template and obtained through structured coding to obtain a joint control strategy.

[0013] A joint control system for multiple fuel-fired generator sets, used to implement a joint control method for multiple fuel-fired generator sets, characterized in that it includes: The data acquisition module is used to collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset. The data adjustment module is used to perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; based on the oscillation data characteristics, offset compensation is performed on the generator operation dataset to output an offset-suppressed dataset. The delay correction module is used to correct the response delay of the offset suppression dataset and generate a delay adjustment dataset. The penalty strategy generation module is used to acquire unit load information in real time and design a hot start penalty strategy based on the unit load information and delay adjustment dataset. The magnetic field early warning module is used to provide early warning of excitation instability based on unit load information and generate a magnetic field early warning log. The joint control module is used to construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning log, and delay adjustment dataset; send the joint control strategy to the preset fuel generator set control terminal; and connect the various modules through wired and / or wireless means.

[0014] The technical effects and advantages of the joint control method and system for multiple fuel-fired generator sets of the present invention are as follows: By collecting operating data from multiple fuel-fired generator sets and performing oscillation analysis, offset compensation, and response delay correction on this data, a hot-start penalty strategy and an excitation instability early warning system were designed and integrated into a complete joint control strategy to achieve joint control of multiple fuel-fired generator sets. Compared with existing experience, the introduction of oscillation analysis and offset compensation actively identifies and optimizes minor disturbances, significantly improving the sensitivity of anomaly identification. Response delay correction effectively solves the scheduling mis-triggering problem caused by inconsistent response speeds of multiple generators. By constructing thermal state parameters and penalty factors to guide the unit scheduling strategy, frequent starts of high-heat units are avoided, reducing unnecessary energy consumption. Early warning based on excitation voltage conditions prevents magnetic field instability and alarm lag, effectively extending the protection response time of individual generators. The overall solution enhances the stable control capability under complex parallel operation conditions, avoids misjudgments, improves anti-disturbance capability, and extends unit life, significantly improving the overall efficiency of the multi-fuel-fired generator set system in intelligent control and safe operation. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a joint control method for multiple fuel-powered generator sets according to the present invention; Figure 2 This is a schematic diagram of a combined control system for multiple fuel-powered generator sets according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 Please see Figure 1 As shown in this embodiment, a method for joint control of multiple fuel-powered generator sets includes: S1. Collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset; S2. Perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; perform offset compensation on the generator operation dataset based on the oscillation data characteristics to output an offset-suppressed dataset; S3. Perform response delay correction on the offset suppression dataset to generate a delay adjustment dataset; S4. Obtain unit load information in real time, and design a hot start penalty strategy based on unit load information and delay adjustment dataset; S5. Based on the unit load information, perform excitation instability early warning on the delay adjustment dataset and generate a magnetic field early warning log; S6. Construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning log, and delay adjustment dataset; send the joint control strategy to the preset fuel generator set control terminal.

[0018] The operating data of the fuel generators includes parameters or fields that reflect the operating status of each fuel generator, such as power output, voltage value, generation frequency, and digital identifiers for operating, standby, and fault status. By aligning the timestamp data of each sampling point in the operating data according to time sequence, removing missing data that may be due to communication interruptions, and using moving average filtering to eliminate data noise, the operating data is cleaned to obtain a higher quality generator operating dataset.

[0019] Methods for performing oscillation output analysis include: The generator number is obtained, and the generator operation dataset is divided based on the generator number to obtain individual generator operation data. In particular, by obtaining the number corresponding to each generator, the individual generator operation data belonging to the corresponding generator is separated from the generator operation dataset, which reduces the mixed impact in the case of multiple machines operating in parallel.

[0020] The power output, voltage value, and generation frequency are extracted from each generator's individual operating data, and power, voltage, and frequency sequences are constructed respectively. The power output, voltage, and frequency sequences are constructed by extracting the power output, voltage value, and generation frequency from each generator's individual operating data, and each sequence has the same time length and the same sampling interval. It should be noted that only power output, voltage value, and generation frequency are extracted here because these three types of parameters have the most representative oscillation prediction and feedback capabilities under standard generator control scenarios, and are used for subsequent oscillation analysis.

[0021] Calculate the average rate of change for each power sequence, set a unit output period, and construct a power sliding window of unit output period size to traverse the power sequence; count the number of power change direction changes in each power sequence within the same power sliding window and calculate the average number of direction changes for the corresponding power sequence.

[0022] The average rate of change is obtained by summing the rates of change of each sampling point in each power sequence and taking the average value. The rate of change also refers to the slope between adjacent sampling points. Based on historical operating experience, the unit output cycle of each generator is set during operation. Each power sequence is traversed using a power sliding window of unit output cycle size. The number of power change direction changes in the same power sliding window in the power sequence corresponding to each generator is counted. The power change direction refers to whether the power increases or decreases. The average number of direction changes is obtained by calculating the average number of power change direction changes in all power sliding windows in each power sequence.

[0023] The power coordinated fluctuation feature is obtained by integrating the average rate of change and the average number of direction changes of each power sequence. The average rate of change and the average number of direction changes are combined into an array and transformed into row vectors. Each row vector represents the fluctuation feature of an individual generator. All row vectors are superimposed to form the power coordinated fluctuation feature. This power coordinated fluctuation feature is used to represent the specific values ​​of the average rate of change and the average number of direction changes of each individual generator, reflecting the intensity and stability of power changes.

[0024] Calculate the voltage difference between the maximum and minimum voltage values ​​within the same unit output cycle in each voltage sequence. The voltage difference is used to reflect the drastic degree of voltage variation within the unit output cycle.

[0025] The timestamp corresponding to the maximum voltage value in the voltage sequence is identified. The average voltage synchronization time deviation is calculated based on the timestamps of the maximum voltage values ​​in all voltage sequences. Specifically, by identifying the timestamp of the maximum voltage value in each voltage sequence, the deviation between any two timestamps of the maximum voltage values ​​in the voltage sequences is calculated and the average value is taken to obtain the average voltage synchronization time deviation. This value is used to quantify the synchronization of the voltage response in the entire parallel fuel generator set.

[0026] The maximum voltage difference and the average time deviation of voltage synchronization are integrated as voltage fluctuation features, wherein the maximum voltage difference and the average time deviation of voltage synchronization for each voltage sequence are combined into an array as voltage fluctuation features.

[0027] Calculate the frequency difference of each frequency sequence within the same unit output cycle, and construct a frequency deviation matrix by statistically analyzing the frequency differences of all unit output cycles. The frequency difference refers to the difference in the average frequency of different frequency sequences within the same unit output cycle. Combine these frequency differences into column vectors and concatenate the vectors to obtain the frequency deviation matrix. The number of columns is equal to the number of unit output cycles, and the number of rows is equal to the number of frequency sequences, which is the number of individual generators.

[0028] The power-coordinated fluctuation characteristics, voltage fluctuation characteristics, and frequency deviation matrix are concatenated to obtain the oscillation data characteristics. Specifically, the oscillation data characteristics are obtained by combining the power-coordinated fluctuation characteristics, voltage fluctuation characteristics, and frequency deviation matrix into a single array.

[0029] Methods for offset compensation include: If the average number of direction changes in the oscillation data characteristics is higher than the preset number threshold and the average rate of change is higher than the preset rate of change fluctuation threshold, the preset number threshold and the preset rate of change fluctuation threshold are set based on historical operating experience, and the average number of direction changes and the average rate of change being higher than the corresponding thresholds are used as indicators to judge the amplitude fluctuation-dominant oscillation type.

[0030] Simultaneously, if the maximum voltage difference is higher than the preset voltage difference threshold and the average voltage synchronization time deviation is less than the preset deviation threshold, the oscillation type is determined to be amplitude fluctuation-dominated oscillation. The preset voltage difference threshold and preset deviation threshold are set based on historical operating experience. If the maximum voltage difference is higher than the preset voltage difference threshold, it indicates that the voltage change amplitude is too large. However, if the average voltage synchronization time deviation is less than the preset deviation threshold, it indicates that although the voltage change amplitude is large, the response timing is not much different from the standard case. This indicates that the oscillation originates from a large fluctuation amplitude. The average number of direction changes, the average rate of change, the maximum voltage difference, and the average voltage synchronization time deviation are used as indicators to determine amplitude fluctuation-dominated oscillation.

[0031] If the average frequency difference in the frequency offset matrix is ​​higher than the preset frequency difference threshold, and the average number of direction changes is less than or equal to the preset number of times threshold, the oscillation type is determined to be phase deviation-dominated oscillation. The average frequency difference refers to the average of the row elements in the frequency offset matrix, which is the average of the frequency difference of all unit output cycles belonging to a single generator. At the same time, a preset frequency difference threshold is set based on historical operating experience. If the average frequency difference is higher than the preset frequency difference threshold and the average number of direction changes indicates that the frequency of the corresponding generator cannot be synchronized with the unit, but the average number of direction changes is low, it indicates that the operating state is relatively stable. This is an oscillation caused by frequency drift, so it is determined to be phase deviation-dominated oscillation.

[0032] It should be noted that if the parameters of the oscillation data characteristics do not fall under the above conditions, the corresponding generator individual will not be judged to have oscillation. Oscillation is only considered to exist if the conditions of the above two oscillation types are met.

[0033] When the oscillation type is amplitude fluctuation-dominated oscillation, an abnormal segment is identified in the power sequence where there are consecutive power change direction changes and the number of power change direction changes is higher than a preset local number threshold. If there are two or more consecutive power change direction changes in the power sequence and the number of power change direction changes is higher than the preset local number threshold for a local area in a complete operating cycle, which is set based on historical experience, the corresponding segment is determined to be an abnormal segment.

[0034] The system performs a direction change pause operation on abnormal sections and restricts the continuous updating of power in a single direction according to a fixed step size. This is achieved by stopping the abnormal response logic for the power change direction in the abnormal section and selecting a single direction for power update based on the direction of the preceding local area of ​​the abnormal section. For example, if the current local area is in a power increase phase, the direction of the abnormal section is restricted to the power increase direction; otherwise, it is set to the power decrease direction. The power update is performed by setting a fixed step size. To ensure that there are no large power changes, the specific value of the fixed step size is set based on the specific operating conditions, thereby improving operational stability.

[0035] The voltage response coefficient is obtained by calculating the local voltage difference of the abnormal section and the ratio of this voltage difference to the maximum voltage difference in all voltage sequences. The local voltage difference refers to the difference between the maximum and minimum voltage values ​​in the abnormal section. The voltage response coefficient is obtained by calculating the ratio of the local voltage difference to the maximum voltage difference in all voltage sequences. It is used to quantify the response intensity ratio of the voltage change in the abnormal section in the unit.

[0036] The voltage response coefficient is used to weight the voltage value of the unit output cycle to which the abnormal section belongs. By identifying the unit output cycle to which the abnormal section is located and using the voltage response coefficient to multiply all the voltage values ​​in the unit output cycle, the intensity of the voltage response value of the abnormal power fluctuation section is reduced, and the impact of abnormal voltage response caused by high-frequency power fluctuation on the operational stability is reduced.

[0037] When the oscillation type is phase deviation-dominated oscillation, the constructed parameter compensation amount adjusts the corresponding dimension parameters in the generator operation dataset. Specifically, the constructed parameter compensation amount adjusts the value of the corresponding parameter so that the corresponding dimension parameter converges to the corresponding dimension standard parameter set based on the generator set operation theory. For example, if a certain dimension parameter is higher than the standard parameter of that dimension, the corresponding dimension compensation amount is used to reduce that dimension parameter; the offset-compensated offset suppression dataset is output.

[0038] The methods for constructing parameter compensation include: The mean value of the row elements corresponding to each generator in the frequency offset matrix is ​​calculated, and this mean value of the row elements is used as the frequency correction reference value for that generator. The mean value of the row elements corresponding to each generator in the frequency offset matrix is ​​used to reflect the frequency offset trend of that generator during the overall operation, and is used as the frequency correction reference value for subsequent compensation.

[0039] A time offset function is constructed based on the frequency correction reference value and the unit time resolution. The time offset function is then used to calculate the time index compensation. The formula for calculating the time offset function is as follows: ;in, Indicates the amount of time index compensation; This indicates the standard operating frequency set based on operational experience; Indicates the frequency correction reference value; This represents the behavior regulation factor, which is set to a constant of 0.5 in this embodiment; represents the resolution per unit time, and represents the sampling interval per unit time; the above formula is a dimensionless calculation.

[0040] Identify the timestamp corresponding to the maximum voltage value in the voltage sequence. In the power sequence, extract a time segment of fixed length before and after the timestamp position to form a local response window for the power. The fixed time length is set based on historical operating experience. The timestamp corresponding to the maximum voltage value is used as the reference. Time segments of fixed length are extracted before and after the timestamp, and the time segments of this region are used as the local response window.

[0041] Calculate the maximum power change rate within the power local response window. Simultaneously, calculate the ratio of the maximum power change rate to the maximum power change rate within the same power local response window in the power sequence of other generators, and obtain the mean of these ratios. The maximum power change rate refers to the largest slope among the slopes of the power change rates of consecutive sampling points within the power local response window. Also, calculate the ratio of the current maximum power change rate to the maximum power change rate in the region where the same power local response window is located in the power sequence of other generators, and calculate the mean of these ratios.

[0042] The power response offset is obtained by weighting and summing the maximum power change amplitude and the mean ratio within the local power response window. The maximum power change amplitude refers to the difference between the maximum and minimum power in the local response window of the power sequence of the current generator individual. The maximum power change amplitude and the mean ratio are weighted and summed using two weighting factors to obtain the power response offset. In this embodiment, the weighting factor for the maximum power change amplitude is set to 0.6, and the weighting factor for the mean ratio is set to 0.4, which can be adjusted based on specific operating conditions. The time index offset and the power response offset are integrated as parameter compensation amounts.

[0043] Methods for correcting response latency include: The power sequence corresponding to each generator in the offset suppression dataset is extracted. The time stamp when the power sequence first shows a continuous positive change after the control command is issued is detected. At the same time, the time interval between this time stamp and the control command issuance time stamp is calculated to obtain the response interval. The response interval refers to the time interval between the time point when the control command is issued and the time point when the power sequence first shows two or more consecutive power increases. It quantifies the degree of delay from receiving the command to actually making an output response.

[0044] The response interval of each generator is compared with the standard response interval in the historical normal response record. If the response deviation is higher than the preset upper limit of response stability, the corresponding generator is identified as a delayed response generator. If the response deviation is lower than the preset lower limit of response stability, the corresponding generator is identified as an early response generator. A response stability time interval is set based on historical operating experience. If the response deviation is higher than the preset upper limit of response stability, it means that the corresponding generator has a delayed response, and the corresponding generator is identified as a delayed response generator. If the response deviation is lower than the preset lower limit of response stability, it means that the corresponding generator is responding too quickly, and the corresponding generator is identified as an early response generator. In other cases, it is considered that there is no delay anomaly.

[0045] For generators with delayed responses, control commands are not executed immediately. Instead, the commands are redistributed to fast-response generators whose response intervals are higher than the preset lower limit of response stability but lower than or equal to the average response interval. For generators identified as having delayed responses, no direct control operation is performed. Instead, the load tasks of these generators are transferred to generators whose response intervals are higher than the preset lower limit of response stability but lower than the average response interval. These generators that meet the requirements of good response capability and whose response time does not exceed the average response interval are considered fast-response generators. The average response interval refers to the average of the response intervals of all generators, which is used to reflect the overall response level of the unit and serves as a condition for determining fast-response generators.

[0046] The task allocation ratio of the fast response generator group composed of individual fast response generators is calculated, and the task load of the fast response generator group is adjusted based on the task allocation ratio. The power generation capacity is evaluated by querying the historical load of each fast response generator in the historical operation record, and the task allocation ratio of each fast response generator is set based on historical operation experience. Thus, the task load of each fast response generator in the fast response generator group is adjusted based on the task allocation ratio.

[0047] For individuals responding early, the response advance is calculated between the response interval and the preset lower limit of response stability. A reduction factor is constructed based on the response advance, where the response advance is the absolute value of the difference between the response interval and the preset lower limit of response stability. The response advance is then normalized to obtain the reduction factor. The power sequence of the corresponding individuals responding early is weighted using the reduction factor. By weighting the power sequence of the corresponding individuals responding early using the reduction factor, the power output value is reduced, thus achieving a delayed equivalent response.

[0048] The power sequences of individual generators after integration and adjustment are used to cover the corresponding power sequences in the offset suppression dataset to obtain the delayed adjustment dataset. The adjusted power sequences corresponding to the lagging response individuals and the early response individuals cover the power sequences of the corresponding generators in the offset suppression dataset. The adjusted power sequences and the normal power sequences that have not been adjusted are integrated into the delayed adjustment dataset.

[0049] The methods for designing hot start penalty strategies include: Extract the operating timestamps corresponding to the unit load information, calculate the time interval between the operating timestamps and the last shutdown timestamps of each generator, and obtain the idle interval. The unit load information refers to the time-series data generated by each generator to represent its operating power and operating status, including information such as the latest collected real-time output power and operating status identifiers. The operating timestamp refers to the first time point in the unit load information that indicates the generator started performing its task, and the last shutdown timestamp refers to the timestamp corresponding to the last shutdown of the generator during its previous operation. The time interval between the operating timestamps and the last shutdown timestamps is the idle interval.

[0050] Based on the unit load information, the load fluctuation value and load average value per unit output cycle are calculated to form the load behavior characteristics. The load fluctuation value refers to the difference between the maximum and minimum real-time output power within a unit output cycle in the unit load information. The load average value refers to the average real-time output power within a unit output cycle. It should be noted that the load fluctuation value and load average value represent power-related values, but they are different from the power in the previous power output or power sequence. Here, the power refers to the latest real-time collected output power and is used to represent the load information. The load fluctuation value and load average value are combined into an array to obtain the load behavior characteristics.

[0051] A thermal state index is constructed based on idle intervals and load behavior characteristics. The thermal state index is used to represent the thermal state level per unit output cycle and is used to construct a thermal penalty strategy.

[0052] The load ratio of the unit time load value to the unit time standard load is calculated to obtain the load loading ratio of the corresponding generator. The unit time load value refers to the average of the maximum real-time output power and the minimum real-time output power within a unit output cycle. The unit time standard load refers to the standard power of the generator when it is running stably within a unit output cycle, which is set based on historical operating experience. The load loading ratio is obtained by calculating the ratio of the two. This load loading ratio is used to represent the proportion of the load of the current generator to the standard load within a unit output cycle.

[0053] The maximum power output and minimum stable power output of the corresponding generator in multiple historical operating cycles are obtained, and the compensable load is calculated based on the maximum power output and minimum stable power output. The compensable load is obtained by multiplying the difference between the maximum power output and the minimum stable power output in multiple historical operating cycles with the adjustment parameter. In this embodiment, the adjustment parameter is set to 0.5, which can be adjusted according to specific operating conditions. The compensable load represents the additional load that the corresponding generator can bear in the current unit output cycle.

[0054] The ratio of thermal state index to compensable load is used to obtain the unit time load penalty factor, which represents the thermal fatigue cost required for a generator to bear additional load in a unit output cycle. The larger the value, the worse the thermal state and the weaker the load capacity.

[0055] The average penalty factor is calculated based on the load penalty factor of all units of time. Generators with a load penalty factor of less than the average penalty factor are selected as scheduling candidate units. Selecting generators with a load penalty factor of less than the average penalty factor indicates that these generators have a more suitable thermal state and a stronger load-bearing capacity. Therefore, they have a higher scheduling priority and are selected as scheduling candidate units. Other units are not prioritized as scheduling candidate units.

[0056] The unit-time load value of each generator in the scheduling candidate unit is adjusted to ensure that the load loading ratio reaches the preset load ratio. The preset load ratio is set based on historical operating experience. By adjusting the unit-time load value, the maximum and minimum real-time output power of the corresponding generator is limited, so as to ensure that the adjusted load loading ratio converges to the preset load ratio.

[0057] Generators with a unit-time load penalty factor greater than a preset tolerance threshold are designated as restricted units and subjected to a cooling process. If the unit-time load factor is greater than the preset tolerance threshold set based on historical operating experience, it indicates that the corresponding generator has an excessively high thermal state and poor availability, requiring immediate cooling to ensure that the generator returns to normal operation. If the unit-time load penalty factor is less than or equal to the preset tolerance threshold, it indicates that the thermal state is acceptable and has not reached the point where a cooling process is required. The processing is encoded into a hot-start penalty strategy, whereby the above processing is encoded into a system-recognizable hot-start penalty strategy.

[0058] Methods for constructing the thermal state index include: Extract the historical average load of each generator in the unit load information for the previous operating cycle corresponding to the last shutdown timestamp. When the idle interval is greater than the preset cooling time limit and the historical average load of the previous operating cycle is lower than the preset standard cycle load threshold, the corresponding generator is judged as low heat level. The historical average load refers to the historical average output power of the operating cycle in which the generator's last shutdown timestamp is located, which is the historical average output power of the previous operating cycle of the current operating cycle. Based on historical operating experience, a preset cooling time limit and a preset standard cycle load threshold are set. When the idle interval is greater than the preset cooling time limit and the historical average load of the previous operating cycle is lower than the preset standard cycle load threshold, it indicates that the idle interval time experienced by the corresponding generator from shutdown to the current operating cycle is relatively long, and the historical average load of the previous operating cycle is low. It is not operating under high load, so it can be considered that the natural cooling process has been completed and the thermal state has little impact, and it is judged as low heat level.

[0059] If the idle interval is less than or equal to the preset upper limit of cooling time and the load fluctuation value per unit output cycle of the previous operating cycle is higher than the preset load fluctuation threshold, the corresponding generator is determined to be of intermediate heat level. In this embodiment, the preset load fluctuation threshold is set based on historical operating experience. The idle interval being equal to the preset upper limit of cooling time and the load fluctuation value per unit output cycle of the previous operating cycle being higher than the preset load fluctuation threshold means that the length of the idle interval is just not greater than the preset upper limit of cooling time, but there is obvious load fluctuation in the previous operating cycle. Therefore, it may be possible that the generator is not completely cooled at this time and cannot be evaluated as a cold machine state. Therefore, it is determined to be of intermediate heat level.

[0060] If the idle interval is lower than the preset cooling time limit and the historical average load of the previous operating cycle is higher than the preset standard cycle load threshold, while the load fluctuation value per unit output cycle is lower than the preset load fluctuation threshold, then the corresponding generator is judged to be of high heat level. The fact that the idle interval is lower than the preset cooling time limit and the historical average load of the previous operating cycle is higher than the preset standard cycle load threshold, while the load fluctuation value per unit output cycle is lower than the preset load fluctuation threshold, indicates that the idle interval is short and the generator was in a high load but relatively stable state for a long time before shutdown in the previous operating cycle, resulting in significant internal heat accumulation. Therefore, generators in this case have extremely high thermal inertia and are judged to be of high heat level.

[0061] If any of the heat levels is not met, the corresponding heat level is determined to be a fuzzy heat level. Specifically, when the idle interval, historical average load, and load fluctuation value per unit output cycle do not meet the conditions of any of the above heat levels, the heat level at this time is determined to be a fuzzy heat level. By identifying the specific values ​​of the idle interval, historical average load, and load fluctuation value per unit output cycle at this fuzzy heat level state, a parameter distance is set. Based on the parameter distance between each parameter of the fuzzy heat level state and the corresponding dimension parameter in the judgment condition, it is determined which one or more parameters are close to the preset corresponding threshold. Then, the corresponding parameters are weighted and summed to obtain the score of the fuzzy heat level. In this embodiment, the weights in the above weighting are set based on historical operating experience and can be adjusted according to specific operating conditions.

[0062] By matching low thermal level, intermediate thermal level, high thermal level and fuzzy thermal level with a preset thermal state parameter table, the corresponding thermal state index is obtained. The preset thermal state parameter table is a parameter table designed based on historical operating experience and is used to match thermal state level and thermal state index.

[0063] Methods for early warning of excitation instability include: The load value of each generator in the unit load information is identified and a load value change sequence is constructed. The load value refers to the latest real-time instantaneous output power in the unit load information and is sorted according to the timestamp of the unit load information to obtain the load value change sequence.

[0064] A critical load power threshold is set, and continuous time segments in the load value change sequence where the load value is lower than the critical load power threshold are extracted, which are the extremely low load segments. In this embodiment, the critical load power threshold is set based on the generator operation theory, and continuous time segments in the load value change sequence where the load value is lower than the critical load power threshold are identified. In this embodiment, the continuous time segment specifically refers to a time segment composed of five or more consecutive timestamps. This type of time segment is determined to be the extremely low load segment.

[0065] The excitation voltage data in the extremely low load section is extracted, the variation amplitude of the excitation voltage data in the extremely low load section is calculated, and the main change direction of the excitation voltage is determined. The excitation voltage data refers to the voltage data around the magnetic field coil in the generator. The variation amplitude of the excitation voltage data is obtained by calculating the difference between the maximum and minimum values ​​of the excitation voltage data in the base load section. At the same time, the sum of all increases and the sum of all decreases of the excitation voltage data in the extremely low load section are counted. The main change direction of the excitation voltage is determined based on the magnitude of the sum of increases and the sum of decreases. If the sum of increases is greater than the sum of decreases, it is an upward direction. If the sum of increases is less than or equal to the sum of decreases, it is a downward direction.

[0066] If the main direction of change of the excitation voltage is downward and the amplitude of the change is lower than the preset fluctuation range, the corresponding extremely low load section is identified as the excitation attenuation section. The preset fluctuation range is set based on historical operating experience. If the main direction of change of the excitation voltage is downward and the amplitude of the change is lower than the preset fluctuation range, it indicates that the excitation voltage is in a continuous downward process and the amplitude of the change is small, indicating that no effective control has been generated and excitation attenuation may occur. Therefore, the corresponding extremely low load section is identified as the excitation attenuation section. If the direction of change is upward or the amplitude of the change is higher than the preset fluctuation range, it indicates that it may be in the excitation voltage enhancement stage or a rapid response state, and it cannot be identified as excitation attenuation.

[0067] The system detects the duration of the excitation weakening segment. If this duration exceeds a preset instability risk assessment threshold, the corresponding excitation weakening segment is designated as an instability warning segment. The preset instability risk assessment threshold is set based on existing theoretical knowledge. If the duration of the excitation weakening segment exceeds the preset instability risk assessment threshold, it indicates that the excitation cannot recover for too long, indicating a strong instability situation. Therefore, the corresponding excitation weakening segment is identified as an instability warning segment. The system extracts the time boundary range, load value, excitation voltage data, cycle index, and warning flag of the instability warning segment to form a magnetic field warning log.

[0068] The ways to construct a joint control strategy include: The delay adjustment dataset is grouped based on the generator number. Each sub-delay adjustment dataset is combined with the hot start penalty policy and the magnetic field warning log to form a generator individual control data unit. The sub-delay adjustment dataset, the hot start penalty policy and the magnetic field warning log belonging to each generator individual constitute the control data unit belonging to the corresponding generator individual.

[0069] All individual generator control data units are combined according to a preset fuel generator set control template, and a joint control strategy is obtained through structured coding. The preset fuel generator set control template refers to a suitable multi-generator combination scheme designed based on historical operating experience. By combining the control data units of each individual generator according to the preset fuel generator set control template, a joint control data unit is obtained. The joint control data unit is then transformed into a transmittable and executable joint control strategy using structured coding.

[0070] This embodiment collects operating data from multiple fuel-fired generator sets and performs oscillation analysis, offset compensation, and response delay correction on this data. A hot-start penalty strategy is then designed, and an excitation instability early warning system is added, integrating these into a complete joint control strategy to achieve joint control of multiple fuel-fired generator sets. Compared with existing experience, by introducing oscillation analysis and offset compensation, minor disturbances are actively identified and optimized, significantly improving the sensitivity of anomaly identification. Response delay correction effectively solves the scheduling mis-triggering problem caused by inconsistent response speeds among multiple generators. By constructing thermal state parameters and penalty factors to guide the unit scheduling strategy, frequent starts of high-heat units are avoided, reducing unnecessary energy consumption. Early warning based on excitation voltage conditions prevents magnetic field instability and alarm lag, effectively extending the protection response time of individual generators. The overall solution enhances the stable control capability under complex parallel operation conditions, avoids misjudgments, improves anti-disturbance capabilities, and extends unit lifespan, significantly improving the overall efficiency of the multi-fuel-fired generator set system in intelligent control and safe operation.

[0071] Example 2 Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A joint control system for multiple fuel generator sets is provided, including: The data acquisition module is used to collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset. The data adjustment module is used to perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; based on the oscillation data characteristics, offset compensation is performed on the generator operation dataset to output an offset-suppressed dataset. The delay correction module is used to correct the response delay of the offset suppression dataset and generate a delay adjustment dataset. The penalty strategy generation module is used to acquire unit load information in real time and design a hot start penalty strategy based on the unit load information and delay adjustment dataset. The magnetic field early warning module is used to provide early warning of excitation instability based on unit load information and generate a magnetic field early warning log. The joint control module is used to construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning log, and delay adjustment dataset; send the joint control strategy to the preset fuel generator set control terminal; and connect the various modules through wired and / or wireless means.

[0072] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0073] It should be noted that, in this document, 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. Unless otherwise specified, 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 said element.

[0074] In the description of this invention, it should be understood that the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0075] In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0076] In the description of this invention, "several" means one or more, and "a large number" means two or more.

[0077] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0078] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0079] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for joint control of multiple fuel-fired generator sets, characterized in that, include: S1. Collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset; S2. Perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; Based on the characteristics of oscillating data, offset compensation is performed on the generator operation dataset to output an offset-suppressed dataset. S3. Perform response delay correction on the offset suppression dataset to generate a delay adjustment dataset; S4. Obtain unit load information in real time, and design a hot start penalty strategy based on unit load information and delay adjustment dataset; S5. Based on unit load information, perform excitation instability early warning and generate magnetic field early warning log; S6. Construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning log, and delay adjustment dataset; send the joint control strategy to the preset fuel generator set control terminal.

2. The method for joint control of multiple fuel-fired generator sets according to claim 1, characterized in that, The methods for performing oscillation output analysis include: Obtain the generator number, divide the generator operation dataset based on the generator number to obtain individual generator operation data; extract the power output, voltage value and power generation frequency from each generator individual operation data, and construct the power sequence, voltage sequence and frequency sequence respectively; The average rate of change for each power sequence is calculated, and a unit output period is set. A power sliding window of the unit output period size is constructed to traverse the power sequences. The number of power change direction changes within the same power sliding window in each power sequence is counted, and the average number of direction changes for the corresponding power sequence is calculated. The average rate of change and the average number of direction changes for each power sequence are integrated to obtain the power coordinated fluctuation feature. The voltage difference between the maximum and minimum voltage values ​​within the same unit output period in each voltage sequence is calculated. The timestamp corresponding to the maximum voltage value in the voltage sequence is identified, and the average voltage synchronization time deviation is calculated based on the timestamps of reaching the maximum voltage value in all voltage sequences. The maximum voltage difference and the average voltage synchronization time deviation are integrated as the voltage fluctuation feature. The frequency difference of each frequency sequence within the same unit output period is calculated, and the frequency difference of all unit output periods is counted to construct a frequency deviation matrix. The power coordinated fluctuation feature, voltage fluctuation feature, and frequency deviation matrix are concatenated to obtain the oscillation data feature.

3. The method for joint control of multiple fuel-fired generator sets according to claim 2, characterized in that, The methods for performing offset compensation include: If the average number of direction changes in the oscillation data features is higher than the preset number threshold, and the average rate of change is higher than the preset rate of change fluctuation threshold; at the same time, the maximum voltage difference is higher than the preset voltage difference threshold, and the average time deviation of voltage synchronization is less than the preset deviation threshold, it is determined that the oscillation type is amplitude fluctuation-dominated oscillation; if the average frequency difference in the frequency offset matrix is ​​higher than the preset frequency difference threshold, and the average number of direction changes is less than or equal to the preset number threshold, it is determined that the oscillation type is phase deviation-dominated oscillation. When the oscillation type is amplitude fluctuation-dominated oscillation, an abnormal segment is identified in the power sequence where the direction of power change changes continuously and the number of power change direction changes exceeds a preset local number threshold. A direction change pause operation is performed on the abnormal segment, and the power output in a single direction is restricted to update continuously at a fixed step size. The local voltage difference of the abnormal segment is calculated, and the ratio of this local voltage difference to the maximum voltage difference in all voltage sequences is calculated to obtain the voltage response coefficient. The voltage response coefficient is used to weight the voltage value of the unit output cycle to which the abnormal segment belongs. When the oscillation type is phase deviation-dominated oscillation, the parameter compensation amount is constructed to adjust the corresponding dimension parameters in the generator operation dataset; the offset-compensated offset suppression dataset is output.

4. The method for joint control of multiple fuel-fired generator sets according to claim 3, characterized in that, The methods for compensating the construction parameters include: Calculate the mean value of the row elements corresponding to each generator in the frequency offset matrix, and use the mean value of the row elements as the frequency correction reference value for that generator; construct a time offset function based on the frequency correction reference value and the unit time resolution, and use the time offset function to calculate the time index compensation amount; Identify the timestamp corresponding to the maximum voltage value in the voltage sequence. In the power sequence, extract time segments of fixed length before and after the timestamp position to form a power local response window. Calculate the maximum power change rate within the power local response window, and simultaneously calculate the ratio of the maximum power change rate to the maximum power change rate in the same power local response window of other generators, and obtain the average of the ratios. Weight the maximum power change amplitude within the power local response window to the average of the ratios to obtain the power response offset. Integrate the time index offset and the power response offset as a parameter compensation amount.

5. The method for joint control of multiple fuel-fired generator sets according to claim 4, characterized in that, The methods for correcting response latency include: Extract the power sequence corresponding to each generator in the offset suppression dataset, detect the time stamp when the power sequence first shows a continuous positive change after the control command is issued, and calculate the time interval between this time stamp and the control command issuance time stamp to obtain the response interval; compare the response interval of each generator with the response deviation of the standard response interval in the historical normal response record. If the response deviation is higher than the preset response stability upper limit, the corresponding generator is identified as a lagging response individual; if the response deviation is lower than the preset response stability lower limit, the corresponding generator is identified as an early response individual. For individuals with delayed responses, control commands are not executed immediately. Instead, control commands are redistributed to fast-response generator individuals whose response interval is higher than the preset lower limit of response stability but less than or equal to the average response interval. The task allocation ratio of the fast-response generator group composed of fast-response generator individuals is calculated, and the task load of the fast-response generator group is adjusted based on the task allocation ratio. For individuals that respond early, the response lead between the response interval and the preset response stability lower limit is calculated, and a reduction factor is constructed based on the response lead. The power sequences of the corresponding individuals that respond early are weighted using the reduction factor. The power sequences of the adjusted generator individuals are integrated and used to cover the corresponding power sequences in the offset suppression dataset to obtain the delay adjustment dataset.

6. The method for joint control of multiple fuel-fired generator sets according to claim 5, characterized in that, The design of the hot start penalty strategy includes the following methods: Extract the operating timestamps corresponding to the unit load information, calculate the time interval between the operating timestamp and the last shutdown timestamp of each generator, and obtain the idle interval; calculate the load fluctuation value and load average value per unit output cycle based on the unit load information, and construct the load behavior characteristics; construct the thermal state index based on the idle interval and load behavior characteristics. The process involves calculating the ratio of the unit-time load value to the unit-time standard load to obtain the load loading ratio of the corresponding generator; acquiring the maximum power output and minimum stable power output of the corresponding generator in multiple historical operating cycles, and calculating the compensable load based on the maximum power output and minimum stable power output; calculating the ratio of the thermal state index to the compensable load to obtain the unit-time load penalty factor; calculating the average penalty factor based on all unit-time load penalty factors, and selecting generators with unit-time load penalty factors less than the average penalty factor as scheduling candidate units; adjusting the unit-time load value of each generator in the scheduling candidate units to ensure that the load loading ratio reaches the preset load ratio; designating generators with unit-time load penalty factors greater than the preset tolerance threshold as scheduling restricted units and executing a cooling process; and encoding the processing into a hot-start penalty strategy.

7. The method for joint control of multiple fuel-fired generator sets according to claim 6, characterized in that, The methods for constructing the thermal state index include: The historical average load of each generator in the unit load information is extracted from the previous operating cycle corresponding to the last shutdown timestamp. When the idle interval is greater than the preset cooling time limit and the historical average load of the previous operating cycle is lower than the preset standard cycle load threshold, the corresponding generator is determined to be of low heat level. If the idle interval is less than or equal to the preset cooling time limit and the load fluctuation value per unit output cycle of the previous operating cycle is higher than the preset load fluctuation threshold, the corresponding generator is determined to be of medium heat level. When the idle interval is lower than the preset cooling time limit and the historical average load of the previous operating cycle is higher than the preset standard cycle load threshold, and the load fluctuation value per unit output cycle is lower than the preset load fluctuation threshold, the corresponding generator is determined to be of high heat level. If it does not meet any of the heat levels, the corresponding heat level is determined to be a fuzzy heat level. By matching the low heat level, medium heat level, high heat level, and fuzzy heat level with the preset thermal state parameter table, the thermal state index of the corresponding level is obtained.

8. The method for joint control of multiple fuel-fired generator sets according to claim 7, characterized in that, The methods for providing early warning of excitation instability include: Identify the load value of each generator in the unit load information and construct a load value change sequence; set a critical load power threshold, and extract the continuous time segment in the load value change sequence where the load value is lower than the critical load power threshold, which is the extremely low load segment; Extract the excitation voltage data in the extremely low load section, calculate the change amplitude of the excitation voltage data in the extremely low load section, and determine the main change direction of the excitation voltage. If the main change direction of the excitation voltage is downward and the change amplitude is lower than the preset fluctuation range, the corresponding extremely low load section is determined as the excitation weakening section. Detect the time length occupied by the excitation weakening section. If the time length is higher than the preset instability risk judgment threshold, the corresponding excitation weakening section is designated as an instability warning section. Extract the time boundary range, load value, excitation voltage data, period index, and warning mark of the instability warning section to form a magnetic field warning log.

9. The method for joint control of multiple fuel-fired generator sets according to claim 8, characterized in that, The methods for constructing the joint control strategy include: The delay adjustment dataset is grouped based on the generator number. Each sub-delay adjustment dataset is combined with the hot start penalty strategy and magnetic field warning log to form an individual generator control data unit. All individual generator control data units are combined according to the preset fuel generator set control template and obtained through structured coding to obtain a joint control strategy.

10. A joint control system for multiple fuel-fired generator sets, used to implement the joint control method for multiple fuel-fired generator sets as described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to collect the operating data of each fuel generator and perform data cleaning to obtain the generator operating dataset. The data adjustment module is used to perform oscillation output analysis on the generator operation dataset to obtain oscillation data characteristics; Based on the characteristics of oscillating data, offset compensation is performed on the generator operation dataset to output an offset-suppressed dataset. The delay correction module is used to correct the response delay of the offset suppression dataset and generate a delay adjustment dataset. The penalty strategy generation module is used to acquire unit load information in real time and design a hot start penalty strategy based on the unit load information and delay adjustment dataset. The magnetic field early warning module is used to provide early warning of excitation instability based on unit load information and generate a magnetic field early warning log. The joint control module is used to construct a joint control strategy by combining the hot start penalty strategy, magnetic field warning logs, and delay adjustment dataset. The joint control strategy is sent to the preset fuel generator set control terminal; the modules are connected to each other via wired and / or wireless means.

Images