Load balancing distribution control method and system for parallel operation of multiple energy hosts

Abstract

The application discloses a load balancing distribution control method and system in parallel operation of multiple energy main machines, relates to the technical field of load distribution, and comprises the following steps: collecting real-time operation parameter sets of each energy main machine in the system; calculating real-time health degree coefficients of each energy main machine; solving energy efficiency deviation degrees of each energy main machine under current load demand, and constructing a load redistribution optimization function; solving the load redistribution optimization function, dynamically generating and issuing individualized load setting instructions of each energy main machine; monitoring overall energy efficiency improvement rates of the system and operation stability indexes of each energy main machine, and dynamically backtracking and adjusting parameter weights of the main machine health degree attenuation model according to the overall energy efficiency improvement rates and the operation stability indexes. The technical problem that the existing load distribution method ignores the service life and maintenance state of the main machine, resulting in overall performance attenuation, is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of load distribution, in particular to a load balancing distribution control method and system in parallel operation of multiple energy host machines. BACKGROUND

[0002] With the deepening of energy structure transformation, the demand for comprehensive utilization of multiple energy forms in the industrial field is increasing, and the system architecture of multiple types and different specifications of energy host machines in parallel operation gradually becomes the mainstream mode of energy supply.

[0003] In the prior art, the traditional load distribution method mostly adopts a static strategy based on rated capacity proportional distribution or average distribution, which fails to fully consider the dynamic changes of the actual energy efficiency characteristics of each energy host machine due to differences in operation years and maintenance conditions, thereby causing the total energy consumption of the parallel system to be relatively high.

[0004] At the same time, the existing control method lacks a dynamic evaluation mechanism for the health state of the host machine, and it is difficult to adaptively adjust the load distribution scheme according to the real-time performance degradation and maintenance recovery of the host machine, which leads to the system running in a non-optimal energy efficiency range for a long time, and may accelerate the wear rate of some host machines, affecting the overall operation stability and service life. SUMMARY

[0005] The embodiments of the present application provide a load balancing distribution control method and system in parallel operation of multiple energy host machines, which solves the technical problem that the existing load distribution method ignores the host machine years and maintenance state, leading to overall performance degradation.

[0006] The technical solution of the present application to solve the above technical problems is as follows:

[0007] In a first aspect, the present application provides a load balancing distribution control method in parallel operation of multiple energy host machines, the method comprising:

[0008] Collecting real-time operation parameter sets of each energy host machine in the parallel system, wherein the real-time operation parameter set at least includes current output load, energy consumption rate and historical maintenance mark;

[0009] Based on the real-time operation parameter set, the real-time health degree coefficient of each energy host machine is dynamically calculated through a host health degree degradation model, wherein the host health degree degradation model couples an operation year cumulative factor and a maintenance effect recovery factor;

[0010] According to the real-time health degree coefficient and a preset standard energy efficiency reference curve, the energy efficiency deviation of each energy host machine under the current load demand is solved, and a load redistribution optimization function with the target of minimizing the total energy consumption of the parallel system is constructed;

[0011] The particle swarm optimization algorithm is used to solve the load redistribution optimization function, and individual load setting instructions of each energy host are dynamically generated and issued;

[0012] After the individual load setting instructions are executed, the overall energy efficiency improvement rate of the parallel system and the operation stability index of each energy host are monitored, and the parameter weight of the host health degree attenuation model is dynamically backtracked and adjusted according to the overall energy efficiency improvement rate and the operation stability index.

[0013] In the second aspect, the application provides a load balancing distribution control system in parallel operation of multiple energy hosts, comprising:

[0014] An information acquisition module is configured to acquire a real-time operation parameter set of each energy host in the parallel system, wherein the real-time operation parameter set at least includes a current output load, an energy consumption rate and a historical maintenance mark;

[0015] A data processing module is configured to dynamically calculate a real-time health degree coefficient of each energy host by a host health degree attenuation model based on the real-time operation parameter set, wherein the host health degree attenuation model is coupled with an operation life cumulative factor and a maintenance effect recovery factor;

[0016] An energy efficiency optimization module is configured to solve an energy efficiency deviation of each energy host under a current load demand according to the real-time health degree coefficient and a preset standard energy efficiency reference curve, and construct a load redistribution optimization function with the minimum total energy consumption of the parallel system as a target;

[0017] An instruction generation module is configured to solve the load redistribution optimization function by using a particle swarm optimization algorithm, and dynamically generate and issue individual load setting instructions of each energy host;

[0018] An instruction execution module is configured to monitor the overall energy efficiency improvement rate of the parallel system and the operation stability index of each energy host after the individual load setting instructions are executed, and dynamically backtrack and adjust the parameter weight of the host health degree attenuation model according to the overall energy efficiency improvement rate and the operation stability index.

[0019] The application provides one or more technical solutions, at least having the following technical effects or advantages:

[0020] The embodiments of the present application provide a load balancing distribution control method and system in parallel operation of multiple energy hosts, first, the current output load, energy consumption rate and historical maintenance mark and other parameters of each energy host are acquired in real time, which provides a comprehensive data basis for subsequent health degree evaluation and energy efficiency analysis. Secondly, the host health degree attenuation model is introduced, which innovatively couples the running time cumulative factor and the maintenance effect recovery factor to dynamically calculate the real-time health degree coefficient of each host. Thirdly, according to the real-time health degree coefficient and the preset standard energy efficiency reference curve, the standard curve is scaled by the running time cumulative factor and the maintenance effect recovery factor, and the theoretical energy consumption rate and the two estimated energy consumption rates are obtained by combining the current load demand query, and then the weight is determined according to the ratio of the current load to the rated load for weighted summation, and finally the accurate energy efficiency deviation is obtained, which provides a precise correction basis for the optimization of load distribution. Then, the particle swarm optimization algorithm is used to solve the above optimization function, and in the iteration process, the inertia weight is dynamically adjusted according to the running time cumulative factor and the maintenance effect recovery factor of each host, the speed and position vector of the particle are updated, and strict constraint processing is performed, so that the algorithm can better consider the characteristics of hosts with different health states when searching for the optimal solution, and the optimization efficiency and solution accuracy are improved. Finally, the overall energy efficiency improvement rate of the system and the operation stability index of each host are continuously monitored, and the parameter weight of the host health degree attenuation model is dynamically traced back and adjusted accordingly, forming a closed-loop adaptive optimization control mechanism, which continuously learns and adapts to the changes of host performance, and continuously optimizes the load distribution strategy.

[0021] Through the above technical solutions, the present application maximizes the energy utilization efficiency of the entire parallel system under the premise of ensuring the stability of the system operation, effectively solves the problem of overall performance degradation and high energy consumption caused by ignoring the age and maintenance state difference of the host in the prior art, and provides strong technical support for efficient and stable operation of the multiple energy host parallel system. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating any creative labor.

[0023] Figure 1 is a flowchart of the load balancing distribution control method of multiple energy hosts in parallel operation provided by the embodiments of the present application;

[0024] Figure 2 is a structural schematic diagram of the load balancing distribution control system of multiple energy hosts in parallel operation provided by the embodiments of the present application.

[0025] In the drawings, the components represented by the respective reference numerals are explained as follows:

[0026] The information collection module 11, the data processing module 12, the energy efficiency optimization module 13, the instruction generation module 14, and the instruction execution module 15. DETAILED DESCRIPTION

[0027] The embodiments of the present application provide a load balancing distribution control method and system in parallel operation of multiple energy host machines, which are used to solve the technical problem that the existing load distribution method ignores the service life and maintenance state of the host machine, resulting in overall performance degradation.

[0028] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0029] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0030] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. In order to enable any person skilled in the art to implement and use the present application, the following description is given. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed in the present application.

[0031] Embodiment one, as shown in the present application, provides a load balancing distribution control method in parallel operation of multiple energy host machines, comprising: Figure 1

[0032] S10: Collecting real-time running parameter sets of each energy host in the system, wherein the real-time running parameter set at least includes current output load, energy consumption rate and historical maintenance mark;​

[0033] In the embodiments of the present application, first, the real-time running parameter set of each energy host in the parallel system is collected and associated. Specifically, data collection is performed on each energy host running in parallel at a preset sampling frequency, for example, once per second or once every five seconds.

[0034] Further, the real-time running parameter set at least includes the current output load, energy consumption rate and historical maintenance marker. The current output load can be directly obtained by the power sensor installed at the output end of each host, and the unit is usually kilowatt or megawatt.

[0035] The energy consumption rate is calculated by collecting the fuel consumption of the host in unit time, such as cubic meters / hour for natural gas host, liters / hour for diesel generator or electricity consumption, such as kilowatt-hour / hour for electric host, combined with the current output load, and the unit is generally kilogram of standard coal / kilowatt-hour or kilowatt-hour / kilowatt.

[0036] The historical maintenance marker is a structured data record, which contains the time of the last maintenance of each host, the maintenance type such as preventive maintenance, fault repair, overhaul, replacement of key parts, etc., the maintenance time and the performance test data before and after maintenance, etc. The real-time running parameter set is the original data for subsequent health degree evaluation and energy efficiency analysis.

[0037] S20: Based on the real-time running parameter set, the real-time health degree coefficient of each energy host is dynamically calculated by a host health degree attenuation model, wherein the host health degree attenuation model couples an operation time cumulative factor and a maintenance effect recovery factor;

[0038] In the embodiments of the present application, after obtaining the real-time running parameter set, the real-time health degree coefficient of each energy host is dynamically calculated based on this by the host health degree attenuation model. The host health degree attenuation model innovatively couples the operation time cumulative factor and the maintenance effect recovery factor to more comprehensively and dynamically reflect the actual health status of the energy host.

[0039] Specifically, the operation time cumulative factor is a monotonically increasing function with the increase of the running time of the host, and its initial value is 0, which gradually approaches to 1 with the increase of the operation time. The maintenance effect recovery factor is used to quantify the recovery effect of maintenance behavior on the health degree of the host, and its value range is between 0 and 1. When the host is maintained, the value of the factor will be adjusted according to the type and depth of the maintenance.

[0040] The construction steps of the host health degree attenuation model include:

[0041] According to the equipment model and design parameters of each energy host in the parallel system, the corresponding standard energy efficiency reference curve and rated life data are retrieved;

[0042] Based on the historical operation database, sample data sets of each energy host in different operation stages are collected, wherein the sample data sets at least include cumulative operation life, type of each maintenance record, and energy efficiency measured data in the corresponding period;

[0043] According to the rated life data and the cumulative operation life, a reference curve representing performance attenuation with running time is generated by fitting, and a quantitative calculation rule of the operation life accumulation factor is defined;

[0044] The maintenance records are analyzed, the contribution weight and time limit of different maintenance types to the energy host energy efficiency recovery are distinguished, and a quantitative calculation rule of the maintenance effect recovery factor is defined;

[0045] The cumulative operation life, the maintenance records, and the energy efficiency measured data are taken as input samples, the operation life accumulation factor and the maintenance effect recovery factor calculated by the quantitative calculation rule are taken as intermediate supervision labels, and a model framework of the host health degree attenuation model is constructed;

[0046] The input samples and the intermediate supervision labels are used to supervise the training of the model framework until the error converges to the allowable range, and the construction of the model is completed.

[0047] In the embodiment of the application, the host health degree attenuation model outputs the real-time health degree coefficient by nonlinearly weighting and fusing the operation life accumulation factor and the maintenance effect recovery factor.

[0048] Firstly, according to the equipment model and design parameters of each energy host, the standard energy efficiency reference curve corresponding to the host of the model is called from the technical manual provided by the equipment manufacturer or the preset equipment parameter database. The curve usually takes the load rate, i.e. the percentage of the current load and the rated load, as the horizontal coordinate, and takes the energy efficiency value, such as thermal efficiency and power generation efficiency, as the vertical coordinate, reflecting the energy efficiency performance of the host at different load points in the ideal state. At the same time, the rated life data, such as the design running hours of 80000 hours or the design running life of 20 years, are called.

[0049] Secondly, based on the historical operation database of the system, sample data sets of each energy host in different operation stages since being put into operation are collected. The sample data sets at least cover the cumulative operation life, i.e. the total running time from the host being put into operation to the present, accurate to hours; the type of each maintenance record, which is divided in detail, such as daily inspection, preventive maintenance, fault repair, overhaul, and replacement of key components; and the energy efficiency measured data in the corresponding period, i.e. the actual energy efficiency value in a specific time period before and after the above cumulative operation life node and maintenance, such as one week before maintenance and one week after maintenance, which is obtained by calculating the real-time energy consumption rate and output load.

[0050] Thirdly, a benchmark curve representing the performance decay with running time is generated by fitting the rated life data and the accumulated running time. For example, assuming the rated life is L years and the accumulated running time is t years, the benchmark calculation formula of the running time accumulation factor F can be initially set as F = 1 - e^(-kt / L), where k is the decay coefficient, which is determined by fitting the decay proportion of the actual energy efficiency of the historical samples corresponding to different t relative to the standard energy efficiency benchmark curve, so that when t = 0, F = 0, indicating that the state is brand new and there is no decay, and when t tends to L, F tends to 1, indicating that the performance has seriously decayed. And according to this, the quantitative calculation rule of the running time accumulation factor is defined, and its value range is [0, 1], and the larger the value, the more serious the performance decay caused by the running time.

[0051] Then, the maintenance records are analyzed to distinguish the contribution weight and time efficiency period of different maintenance types to the energy host energy efficiency recovery. For example, the contribution weight of daily inspection is small, which can be set to 0.05, and the time efficiency period is short, such as 3 days; the contribution weight of preventive maintenance is moderate, which is set to 0.2, and the time efficiency period is 1 month; the contribution weight of fault repair is between 0.1-0.3 according to the fault severity, and the time efficiency period is 2 weeks; the contribution weight of overhaul is large, which is set to 0.6-0.8, and the time efficiency period is 6 months to 1 year; the contribution weight of replacing key components is the largest, which can reach 0.8-0.95, and the time efficiency period is set according to the component life, such as 2 years after replacing new blades. The quantitative calculation rule of the maintenance effect recovery factor F1 is defined as follows: for each maintenance, the initial recovery value R and the time efficiency period T are determined according to its type, F1 starts from R at the time when the maintenance is completed, and gradually decreases to 0 according to a certain decay law, such as exponential decay F1 = Re^(-λt / T), where λ is the decay rate constant. When there are multiple maintenances, the current maintenance effect recovery factor is the superposition of the recovery factors generated by the previous maintenances, but the total recovery factor does not exceed 1.

[0052] Subsequently, the accumulated running time and the maintenance records are used as time stamps and type identifiers, and the energy efficiency measured data are used as input samples to construct the model framework of the host health decay model, with the running time accumulation factor and the maintenance effect recovery factor calculated by the above quantitative calculation rule as intermediate supervision labels. The model framework can be a multi-layer perceptron, the input layer neurons correspond to each dimension of the above input samples, the hidden layer is used to learn the non-linear mapping relationship, and the output layer outputs the final real-time health coefficient H. The loss function of the model is defined as the mean square error between the output real-time health coefficient H and the target health coefficient derived from the actual decay degree of the energy efficiency measured data relative to the standard energy efficiency benchmark curve.

[0053] Finally, the model framework is supervised trained by using the input samples and the intermediate supervision labels. During the training process, the weights and biases of each layer of the model are constantly adjusted, and the back propagation algorithm is used to minimize the loss function until the prediction error of the model, i.e., the error between the real-time health degree coefficient output by the model and the health degree reference value calculated based on the contemporaneous energy efficiency measured data, converges to a preset allowable range. For example, when the root mean square error (RMSE) on the validation set converges to a preset allowable range, such as RMSE < 0.05, it is considered that the model can accurately calculate the real-time health degree coefficient by coupling the running age cumulative factor and the maintenance recovery factor, and the construction of the host health degree attenuation model is completed.

[0054] Further, the calculation formula of the real-time health degree coefficient H can be expressed as H = 1 - (F x (1 - F1)), that is, the attenuation caused by the running age and the recovery brought by the maintenance are comprehensively considered, and the value range thereof is (0, 1], and the closer the value is to 1, the better the current health condition of the host is.

[0055] Specifically, the quantitative calculation rule of the running age cumulative factor comprises:

[0056] The performance attenuation inflection point year is set according to the rated life data, when the cumulative running age is less than the performance attenuation inflection point year, the running age cumulative factor is calculated by using a linear mild attenuation function;

[0057] When the cumulative running age is greater than or equal to the performance attenuation inflection point year, the running age cumulative factor is calculated by using an exponential accelerated attenuation function.

[0058] In the embodiments of the present application, first, in the quantitative calculation rule of the running age cumulative factor, the performance attenuation inflection point year is introduced. Since the performance attenuation trend of the energy host in the whole life cycle is not a simple linear or single exponential mode, generally, the attenuation is relatively slow at the beginning of operation, and the attenuation rate will significantly accelerate after a certain age. The performance attenuation inflection point year is determined based on the change rate of the attenuation ratio of the energy efficiency measured data in the historical sample data relative to the standard energy efficiency benchmark curve, when the change rate exceeds a preset threshold, for example, from an annual average attenuation of 1% to an annual average attenuation of 3%, the corresponding cumulative running age is set as the performance attenuation inflection point year of the host of this type. For example, the performance attenuation inflection point year = rated life x a, a is a calibration coefficient, and the value is 0.6-0.8.

[0059] For example, the rated life of a gas turbine host is 20 years, through the analysis of the historical energy efficiency data, it is found that the performance attenuation is relatively gentle in the first 8 years, and the annual average attenuation is about 0.8%, while the attenuation rate accelerates to an annual average of 2.5% after the 8th year, and 8 years is set as the performance attenuation inflection point year thereof.

[0060] When the accumulated running time t is less than the inflection point t0, i.e. t < t0, the running time accumulation factor F is calculated using a linear light attenuation function, which can be set as F = a x (t / t0), where a is a proportional coefficient less than 1, for example 0.3, to ensure that at the inflection point, the F value calculated by the linear function is smoothly connected with the value at this point calculated by the subsequent exponential function.

[0061] When the accumulated running time t is greater than or equal to the inflection point t0, i.e. t ≥ t0, then switch to an exponential accelerated attenuation function, which can be set as F = F0 + (1-F0) x (1-e^(-b x (t-t0) / (L-t0))), where F0 is the running time accumulation factor value calculated by the linear function when t = t0, b is an accelerated attenuation coefficient, and L is the rated life time.

[0062] Specifically, if t0 = 8 years and L = 20 years, when t = 8, F0 = a x (8 / 8) = a = 0.3. When t = 10 years, substitute into the exponential accelerated attenuation function, F = 0.3 + (1-0.3) x (1-e^(-b x (10-8) / (20-8))) = 0.3 + 0.7 x (1-e^(-2b / 12)). By analyzing the actual energy efficiency decay data at t = 10 years in the historical data, assuming that the actual decay ratio at this time corresponds to an F target value of 0.45, then the b value can be calculated by back calculation: 0.45 = 0.3 + 0.7 x (1-e^(-b / 6)) → 0.15 / 0.7 = 1-e^(-b / 6) → e^(-b / 6) = 1-0.15 / 0.7 ≈ 0.7857 → -b / 6 = ln(0.7857) ≈ -0.241, then b ≈ 1.446.

[0063] Through the design of the piecewise function, the performance decay characteristics of the energy host in different running stages are simulated, so that the calculation of the running time accumulation factor is more in line with the actual situation, laying a foundation for the accurate evaluation of the subsequent health degree coefficient.

[0064] Further, the quantitative calculation rule of the maintenance effect recovery factor comprises:

[0065] According to the maintenance type of the historical maintenance record, a preset contribution weight coefficient is matched;

[0066] According to the interval between the completion time point of each maintenance and the current time, and in combination with the preset time limit period of the corresponding maintenance type, the residual performance contribution value of single maintenance is calculated;

[0067] The residual performance contribution values corresponding to all the historical maintenance records are weighted and summed to obtain the maintenance effect recovery factor.

[0068] In the embodiments of the present application, first, for each maintenance record in the historical maintenance records, a corresponding coefficient value is matched from a preset maintenance contribution weight coefficient table according to the maintenance type, such as preventive maintenance, fault repair, overhaul, replacement of key components, etc. The coefficient table is constructed based on industry standard maintenance guidelines and historical energy efficiency recovery data statistics. For example, the contribution weight coefficient of replacing key components is set to 0.9, the contribution weight coefficient of overhaul is set to 0.7, the contribution weight coefficient of preventive maintenance is set to 0.3, the contribution weight coefficient of daily inspection is set to 0.05, and the contribution weight coefficient of fault repair is set to float between 0.2 and 0.5 according to the fault severity, with a high value for severe fault repair and a low value for slight fault repair.

[0069] Secondly, the residual performance contribution value of each maintenance is calculated. Residual performance contribution value = contribution weight coefficient x time decay coefficient. If the time interval ≥ time effectiveness period, the time decay coefficient = 0, indicating that the maintenance effect is completely invalid; if the time interval < time effectiveness period, the time decay coefficient = 1-(time interval / time effectiveness period), the closer the time is, the closer the coefficient is to 1, and the more the effect is retained.

[0070] For example, a certain energy host performed an overhaul three months ago, the preset contribution weight coefficient of the overhaul is 0.7, and the corresponding time effectiveness period is 180 days. The interval between the current time and the completion time of the maintenance is 90 days, since 90 days < 180 days, the time decay coefficient = 1-(90 / 180) = 0.5. Then the residual performance contribution value of the overhaul = 0.7 x 0.5 = 0.35.

[0071] If the host also performed a preventive maintenance six months ago, the contribution weight coefficient is 0.3, and the time effectiveness period is 90 days. The current interval is 180 days ≥ 90 days, the time decay coefficient = 0, so the residual performance contribution value of the preventive maintenance = 0.3 x 0 = 0.

[0072] If the host performed a daily inspection one month ago, the contribution weight coefficient is 0.05, the time effectiveness period is 15 days, and the current interval is 30 days ≥ 15 days, the time decay coefficient = 0, and the residual performance contribution value = 0.05 x 0 = 0.

[0073] At this time, the residual performance contribution values of the historical maintenance records are weighted and summed to obtain the maintenance effect recovery factor.

[0074] S30: According to the real-time health coefficient and the preset standard energy efficiency baseline curve, the energy efficiency deviation of each energy host under the current load demand is solved, and a load redistribution optimization function with the minimum total energy consumption of the parallel system as the target is constructed.

[0075] In the embodiments of the present application, first, the actual energy efficiency curve under the current health state is calculated according to the real-time health coefficient H of each energy host and the corresponding standard energy efficiency baseline curve.

[0076] Specifically, multiplying the energy efficiency value of each load rate point on the standard energy efficiency benchmark curve by the real-time health coefficient H, the actual energy efficiency value of the host under the current health is obtained, that is, the actual energy efficiency curve = standard energy efficiency benchmark curve x H. The energy efficiency deviation is defined as the percentage of the difference between the energy efficiency value of the standard energy efficiency benchmark curve and the energy efficiency value of the actual energy efficiency curve under the same load rate to the energy efficiency value of the standard energy efficiency benchmark curve, which is used to quantify the degree of energy efficiency loss caused by health degradation.

[0077] Further, the energy efficiency deviation of each energy host under the current load demand is solved, that is, for the load rate interval that each host may undertake under the current total system load demand, the average deviation degree of the actual energy efficiency curve and the standard energy efficiency benchmark curve in the interval is calculated.

[0078] Secondly, a load redistribution optimization function is constructed, taking the minimum total system energy consumption as the optimization objective, and the independent variables of the function are the distributed loads of each energy host, and the constraint conditions include the maximum load limit, the minimum load limit, the load adjustment rate limit and the system total load demand balance of each host.

[0079] Among them, according to the real-time health coefficient and the preset standard energy efficiency benchmark curve, the energy efficiency deviation of each energy host under the current load demand is solved, including:

[0080] The standard energy efficiency benchmark curve is scaled by using the running time cumulative factor to obtain a first equivalent energy efficiency curve;

[0081] The standard energy efficiency benchmark curve is scaled by using the maintenance effect recovery factor to obtain a second equivalent energy efficiency curve;

[0082] Based on the current load demand, the theoretical energy consumption rate, the first estimated energy consumption rate and the second estimated energy consumption rate are obtained by querying on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve respectively;

[0083] The first difference between the first estimated energy consumption rate and the theoretical energy consumption rate is calculated relative to the percentage of the theoretical energy consumption rate, as a first energy efficiency deviation component;

[0084] The second difference between the second estimated energy consumption rate and the theoretical energy consumption rate is calculated relative to the percentage of the theoretical energy consumption rate, as a second energy efficiency deviation component;

[0085] According to the ratio of the current load demand to the rated load corresponding to the standard energy efficiency benchmark curve, the first weight and the second weight are determined;

[0086] The first energy efficiency deviation component and the second energy efficiency deviation component are summed with the first weight and the second weight to obtain an energy efficiency deviation of each energy host.

[0087] In the embodiments of the present application, first, the energy consumption rate value of the standard energy efficiency benchmark curve is adjusted by a "1 / operation life cumulative factor" to scale the standard energy efficiency benchmark curve, and a first equivalent energy efficiency curve is obtained. For example, if the operation life cumulative factor F=0.4, which indicates that the energy efficiency is attenuated by 40% due to the operation life, the theoretical energy consumption rate of a certain load rate point on the standard energy efficiency benchmark curve is 100 kW, and after scaling by the operation life cumulative factor, the first estimated energy consumption rate is 100 / (1-F)=100 / 0.6≈166.67 kW, that is, the energy consumption rate of the first equivalent energy efficiency curve at this point is 166.67 kW, which intuitively reflects the negative impact of the operation life on energy consumption.

[0088] Secondly, the energy consumption rate value of the standard energy efficiency benchmark curve is adjusted by a "1 / maintenance effect recovery factor" to obtain a second equivalent energy efficiency curve. If the maintenance effect recovery factor F1=0.35, which indicates that the maintenance brings 35% of energy efficiency recovery, the second estimated energy consumption rate is the theoretical energy consumption rate / (1+F1)=100 / (1+0.35)≈74.07 kW, that is, the maintenance makes the energy consumption rate recover from the attenuated 166.67 kW to 74.07 kW, and the second equivalent energy efficiency curve reflects the improvement of maintenance on energy consumption.

[0089] Thirdly, for the current total load demand of the system, it is assumed that a certain energy host needs to bear 60% of the rated load, then the theoretical energy consumption rate E0, the first estimated energy consumption rate E1 and the second estimated energy consumption rate E2 corresponding to 60% load rate are respectively queried on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve.

[0090] Then, the first energy efficiency deviation component D1=(|E1-E0| / E0)×100% is calculated, which reflects the energy consumption deviation proportion only considering the operation life; the second energy efficiency deviation component D2=(|E0-E2| / E0)×100% is calculated, which reflects the energy consumption improvement proportion only considering the maintenance effect, which is a positive value indicating energy consumption reduction.

[0091] Subsequently, according to the ratio R of the current load demand to the rated load corresponding to the standard energy efficiency benchmark curve, that is, according to the load rate, the weight is determined. When the current load / rated load≤50%, the first weight=1-(current load / rated load), and the second weight=current load / rated load; when the current load / rated load>50%, the first weight=current load / rated load, and the second weight=1-(current load / rated load).

[0092] Specifically, when R = current load / rated load ≤ 50%, it indicates that the host is in a low load operation state, at this time the influence of the basic decay caused by the running age is more significant, so the first weight takes 1-R, and the second weight takes R; when R > 50%, the host is in a higher load operation state, and the performance recovery caused by maintenance is more prominent under high load, so the first weight takes R, and the second weight takes 1-R.

[0093] Exemplarily, if the current load rate R = 60% > 50%, the first weight = 0.6, the second weight = 0.4, and the energy efficiency deviation D = 0.6 * D1 + 0.4 * D2. By dynamically adjusting the weights, the energy efficiency deviation calculation is more in line with the actual energy efficiency influencing factor primary and secondary relationship under different load conditions.

[0094] Further, based on the current load demand, the theoretical energy consumption rate, the first estimated energy consumption rate and the second estimated energy consumption rate are obtained by querying on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve respectively, comprising:

[0095] According to the current load demand, matching search is performed in the discrete load-energy consumption rate data points of the standard energy efficiency benchmark curve;

[0096] If a data point completely matched with the current load demand is found, the corresponding energy consumption rate is directly read as the theoretical energy consumption rate;

[0097] If a data point completely matched with the current load demand is not found, the two nearest data points are selected, and the theoretical energy consumption rate under the current load demand is obtained by linear interpolation calculation;

[0098] According to the ratio of the current load demand to the rated load of each energy host, the load rate interval is determined;

[0099] If the load rate is lower than the preset low load threshold, the first estimated energy consumption rate is obtained on the first equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and the second estimated energy consumption rate is directly obtained on the second equivalent energy efficiency curve by using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve;

[0100] If the load rate is in the preset medium-high load interval, the second estimated energy consumption rate is obtained on the second equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and the first estimated energy consumption rate is obtained on the first equivalent energy efficiency curve by using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve.

[0101] In the embodiment of the present application, first, the discrete data points of the standard energy efficiency reference curve are matched and searched. If the load rate corresponding to the current load demand is 25%, that is, no completely matched data point is found, two adjacent data points of 20% and 30% are selected, and it is assumed that the theoretical energy consumption rates corresponding to the two adjacent data points are E20 and E30 respectively. The theoretical energy consumption rate E under the load rate of 25% is calculated by the linear interpolation formula E=E20+(E30-E20)×(25%-20%) / (30%-20%). If the current load rate is 30%, the energy consumption rate corresponding to the data point of the load rate of 30% is directly read as the theoretical energy consumption rate E0.

[0102] Secondly, the interval in which the current load rate is located is judged. If the preset low load threshold is 30% rated load, when the calculated current load rate is 25%, that is, lower than 30%, the first estimated energy consumption rate E1 is obtained on the first equivalent energy efficiency curve by using the linear interpolation method; and on the second equivalent energy efficiency curve, since the improvement effect of maintenance on energy efficiency is limited under the low load condition, the energy consumption rate of the data point corresponding to the load rate on the standard energy efficiency reference curve is directly used as the second estimated energy consumption rate E2, that is, it is considered that the maintenance effect can be ignored under the low load.

[0103] If the current load rate is 60%, which is in the medium and high load interval of 30%~100%, the second estimated energy consumption rate E2 is obtained on the second equivalent energy efficiency curve by linear interpolation; on the first equivalent energy efficiency curve, considering that the base decay of the running time under the medium and high load has been fully reflected by the previous function calculation, the energy consumption rate of the corresponding data point on the standard energy efficiency reference curve is directly used as the first estimated energy consumption rate E1, so as to avoid repeated calculation of the influence of the running time. Through the differential processing mode in different intervals, the accuracy and rationality of the energy consumption rate query under different load conditions are further improved, so that the calculation basis of the subsequent energy efficiency deviation is more reliable.

[0104] S40: using a particle swarm optimization algorithm to solve the load redistribution optimization function, and dynamically generating and issuing a personalized load setting instruction to each energy host;

[0105] In the embodiment of the present application, first, the particle swarm parameters are initialized, and the position vector of each particle represents a set of distributed load values of each energy host, which needs to meet that the distributed load of each host is between the minimum load limit and the maximum load limit, and the sum of the components of all particle position vectors is equal to the current system total load demand.

[0106] Secondly, the fitness function is constructed. Taking the minimum total energy consumption of the parallel system as the target, the distributed load of each energy host is substituted into its actual energy efficiency curve, the energy consumption of a single host is calculated based on the curve corrected by the health degree coefficient, that is, the distributed load x energy consumption rate, and the sum is obtained to obtain the total energy consumption of the system. The total energy consumption value is taken as the particle fitness value, and the smaller the fitness value is, the better the particle position is.

[0107] Again, the personalized load setting instructions dynamically generated and issued to each energy host enable the host with high energy efficiency to undertake more load proportion, and the host with low energy efficiency to undertake less load.

[0108] The load redistribution optimization function with the minimum total energy consumption of the parallel system as the target is constructed, and includes:

[0109] The to-be-allocated load values of the energy hosts in the parallel system are taken as optimization variables;

[0110] The total load demand in the parallel system is taken as an equation constraint, and the sum of the to-be-allocated load values of the energy hosts is taken as an inequality constraint;

[0111] The allowed load range of each energy host is taken as an inequality constraint, wherein the lower limit of the allowed load range is determined based on the operation age cumulative factor, and the upper limit of the allowed load range is determined based on the maintenance effect recovery factor;

[0112] The energy efficiency deviation of each energy host is taken as a correction coefficient for calculating the estimated energy consumption of the corresponding host under the to-be-allocated load value;

[0113] The load redistribution optimization function is constructed with the minimum sum of the estimated energy consumptions of all energy hosts as the target.

[0114] In the embodiments of the present application, first, the to-be-allocated load values of the energy hosts in the parallel system are taken as optimization variables, and the to-be-allocated load value of the i th energy host is Pi, i = 1, 2,..., n, and n is the total number of energy hosts, so the optimization variable vector is [P1, P2,..., Pn].

[0115] Secondly, the equation constraint is the balance of the total load demand of the system, and the inequality constraint is the allowed load range of each host, that is, the allowed load lower limit ≤ Pi ≤ the allowed load upper limit. The allowed load lower limit is dynamically adjusted based on the operation age cumulative factor. The longer the operation age, the smaller the cumulative factor, and the more serious the performance degradation of the host, and the minimum stable operation load is correspondingly increased. The allowed load lower limit is set to be rated load × (1- operation age cumulative factor × 0.3);

[0116] The allowed load upper limit is adjusted based on the maintenance effect recovery factor. The larger the maintenance effect recovery factor, the better the performance recovery of the host after maintenance, and the stronger the maximum output capacity. The allowed load upper limit is rated load × (1+ maintenance effect recovery factor × 0.2). Wherein, 0.3, 0.2 are basic coefficients, which can be calibrated according to the type of energy host and combined with industry standards.

[0117] Further, the energy efficiency deviation degree D of each energy host is taken as a correction coefficient to correct the estimated energy consumption of the energy host under the to-be-allocated load Pi. Specifically, if the theoretical energy consumption rate of an energy host under the load rate corresponding to the to-be-allocated load Pi is E0, which is obtained by querying the standard energy efficiency benchmark curve, then the actual estimated energy consumption rate E of the energy host after considering the energy efficiency deviation degree is E = E0×(1+D / 100), where D is the energy efficiency deviation degree.

[0118] For example, if D = 20%, then the actual estimated energy consumption rate E is E0×1.2, that is, the energy consumption increases by 20% due to the health degree attenuation. Therefore, the estimated energy consumption of a single energy host is Pi×E = Pi×E0×(1+D / 100), and the total estimated energy consumption of the system is the sum of the estimated energy consumptions of all the energy hosts.

[0119] Finally, a load redistribution optimization function is constructed with the minimization of the sum of the estimated energy consumptions of all the energy hosts as the goal, and the load redistribution optimization function is represented as f(P1, P2,..., Pn) =∑[Pi×E0i×(1+Di / 100)], where i = 1, 2,..., n, and the following conditions are met: ∑Pi = total load demand of the system, allowable load lower limit i≤Pi≤allowable load upper limit i, allowable load lower limit i = rated load i×(1-run age cumulative factor i×0.3), and allowable load upper limit i = rated load i×(1+maintenance effect recovery factor i×0.2). The optimization function directly introduces the energy efficiency deviation degree of each host into the energy consumption calculation, so that the optimization process can accurately reflect the influence of the health degree difference on the actual energy consumption, thereby realizing energy efficiency priority load distribution.

[0120] Specifically, step S40 in the method comprises:

[0121] initializing a particle swarm, wherein a position vector of each particle represents a set of load distribution schemes, each dimension of the position vector corresponds to an energy host, and the initial value of each dimension is randomly generated within the allowable load range of the corresponding energy host;

[0122] taking the load redistribution optimization function as a fitness function of a particle swarm optimization algorithm to calculate the fitness value of each particle;

[0123] in each iteration, updating the velocity vector and the position vector of each particle based on the individual historical optimal position of the particle and the global optimal position of the particle swarm, wherein the inertia weight used in the updating process is dynamically adjusted according to the run age cumulative factor and the maintenance effect recovery factor of the energy host in the corresponding dimension;

[0124] performing constraint processing on the updated particle position vector;

[0125] Determine whether the preset maximum number of iterations has been reached, or whether the improvement of the global optimal fitness value of the particle swarm within a preset number of consecutive iterations is lower than the preset convergence threshold;

[0126] If any condition is met, the iteration is terminated, and the global optimal position vector of the particle swarm is taken as the optimal load allocation scheme.

[0127] The optimal load allocation scheme is decoded into personalized load setting instructions for each energy host and sent to the control system of the corresponding energy host.

[0128] In this embodiment, the particle swarm is first initialized to a population size of 50, a maximum number of iterations of 100, an initial inertia weight of w=0.729, and learning factors c1=c2=1.494. The position vector dimension of each particle is consistent with the number of power hosts n. For example, when n=3 power hosts, the particle position vector is [P1,P2,P3], where P1, P2, and P3 are the allocated loads of the 3 power hosts, respectively. When generating the initial positions, for each power host Pi, a value is randomly selected between its lower and upper allowable load limits, and normalization is performed to ensure that ΣPi=the total system load requirement.

[0129] Next, calculate the particle fitness value. Substitute each Pi in the particle position vector into the actual energy efficiency curve of the corresponding host, and combine it with the energy consumption rate corrected by the health coefficient to calculate the energy consumption of a single unit Pi×Ei. Summing these values ​​gives the total system energy consumption as the fitness value.

[0130] For example, if a particle is located at [400, 300, 300]kW, and the Ei values ​​of the three main units are 0.8kW / kW, 0.75kW / kW, and 0.82kW / kW respectively, then the total energy consumption is 400×0.8+300×0.75+300×0.82=791kW, and the fitness value of the particle is 791.

[0131] Next, the inertia weight is dynamically adjusted. For example, the update formula for the inertia weight w is w = w1 × (1 - α × Fi - β × F1i), where w1 is the basic inertia weight of 0.729, α and β are weight coefficients, α = 0.2 and β = 0.15, Fi is the cumulative factor of the operating years of the i-th host, and F1i is its maintenance effect recovery factor. For a host corresponding to a certain dimension in the position vector, if its cumulative factor of operating years Fi = 0.4, indicating a 40% decay, and its maintenance effect recovery factor F1i = 0.35, indicating a 35% recovery, then the inertia weight correction term for that dimension = 0.2 × 0.4 + 0.15 × 0.35 = 0.1325, and after correction, w = 0.729 × (1 - 0.1325) = 0.729 × 0.8675 ≈ 0.633.

[0132] Further, the longer the running time of the host and the worse the maintenance recovery effect, the smaller the inertia weight of the corresponding dimension, so that the search step of the particle in the dimension is reduced, avoiding unstable load distribution caused by low health degree of the host; on the contrary, the inertia weight of the corresponding dimension of the host with high health degree is larger, which enhances the search flexibility.

[0133] Subsequently, particle speed and position updating is performed, and after updating, the position vector needs to be constrained. If the load Pi allocated to a host exceeds the allowed range, it is truncated to the nearest boundary value, and the loads of other hosts are adjusted in proportion to meet the total load constraint.

[0134] Specifically, each component of the particle position vector is normalized and scaled, so that the sum of the scaled components is equal to the total load demand. For each component of the normalized and scaled particle position vector, boundary correction is performed according to the allowed load range of the corresponding energy host, and the component exceeding the range is adjusted to the corresponding boundary value. The normalized and scaled particle position vector after boundary correction is normalized and scaled again, so that the sum of the scaled components is equal to the total load demand, and boundary correction is performed again. The normalization and scaling and boundary correction process are repeated until the sum of the components of the particle position vector and the absolute error of the total load demand are less than the preset tolerance threshold, and all components are within the allowed load range of the corresponding energy host.

[0135] Finally, the iteration termination condition is judged. When the number of iterations reaches 100 times, or the change amount Δf of the global optimal fitness value in 20 consecutive iterations is less than 0.5kW, i.e. the convergence threshold is reached, the iteration is terminated.

[0136] Finally, the optimal position vector is decoded into personalized load setting instructions to ensure that each host receives and executes load adjustment smoothly, achieving the dynamic optimization goal of minimizing the total energy consumption of the interconnected system.

[0137] wherein, in each iteration, the speed vector and the position vector of each particle are updated based on the individual historical optimal position of the particle and the global optimal position of the particle swarm, comprising:

[0138] An adaptive inertia weight is set for the update of the speed vector and the position vector of each particle, wherein the value of the adaptive inertia weight is positively correlated with the running time cumulative factor of the energy host in the corresponding dimension and negatively correlated with the maintenance effect recovery factor;

[0139] Based on the individual historical optimal position of the particle, the global optimal position of the particle swarm and the adaptive inertia weight, the speed vector of the particle is updated according to a preset basic formula of particle swarm optimization algorithm;

[0140] The updated speed vector and the current particle position vector are added to obtain the updated position vector.

[0141] In the embodiments of the present application, firstly, an adaptive inertia weight is set for the update of the velocity vector and the position vector of each particle, and the adaptive inertia weight = basic inertia weight + running age influence coefficient x running age cumulative factor - maintenance state influence coefficient x maintenance effect recovery factor.

[0142] wherein the basic inertia weight value is an adjustable parameter of the particle swarm algorithm, for example, 0.5-0.7; the running age influence coefficient is a preset constant greater than zero, used to adjust the positive influence degree of the running age on the weight, for example, 0.2; and the maintenance state influence coefficient is a preset constant greater than zero, used to adjust the negative influence degree of the maintenance state on the weight, for example, 0.15.

[0143] Exemplarily, when the running age cumulative factor Fi of a certain energy host is relatively large, for example, Fi = 0.6, indicating that the device is relatively seriously aged, and the maintenance effect recovery factor F1i is relatively small, for example, F1i = 0.2, indicating that the maintenance effect is limited, the adaptive inertia weight = 0.6 + 0.2 x 0.6 - 0.15 x 0.2 = 0.6 + 0.12 - 0.03 = 0.69, at this time, the inertia weight is relatively high, indicating that the corresponding load distribution dimension of the host retains more historical search experience in the particle iteration process.

[0144] Secondly, the velocity vector of the particle is updated, and the motion inertia of the particle at present is considered, which is controlled by the adaptive inertia weight, that is, the inertia weight adjusted dynamically according to the running age cumulative factor and the maintenance effect recovery factor of the corresponding energy host is multiplied by the current speed of the particle, so as to reflect the characteristic that the particle maintains the original motion trend.

[0145] Meanwhile, the individual cognitive component and the social cognitive component are introduced, the individual cognitive component is obtained by multiplying the difference between the current position of the particle and the individual historical optimal position by a learning factor, to guide the particle to approach the optimal solution found by itself; and the social cognitive component is obtained by multiplying the difference between the current position of the particle and the global optimal position of the particle swarm by another learning factor, to promote the particle to move in the direction of the global optimal solution found by the whole group. The three components are superimposed, and the updated velocity vector of the particle is obtained.

[0146] Finally, the updated velocity vector and the current position vector of the particle are directly added, so as to obtain the new position of the particle in the solution space, and realize the dynamic exploration and optimization of the load distribution scheme.

[0147] S50: After executing the personalized load setting instruction, the overall energy efficiency improvement rate of the parallel system and the operation stability index of each energy host are monitored, and the parameter weight of the host health degree attenuation model is dynamically traced back and adjusted according to the overall energy efficiency improvement rate and the operation stability index.

[0148] In the embodiments of the present application, first, the calculation formula of the energy efficiency improvement rate is set as (total energy consumption of the system before optimization - total energy consumption of the system after optimization) / total energy consumption of the system before optimization x 100%, and the energy consumption data before and after optimization is calculated through real-time collection. The operation stability index includes the load fluctuation coefficient of each energy host, the start-stop frequency and the over-temperature alarm frequency.

[0149] Among them, the load fluctuation coefficient needs to be controlled within ± 5%, the start-stop frequency is not more than 1 time per hour, and the over-temperature alarm frequency is not more than 3 times per 24 hours. If any index exceeds the threshold value for 3 consecutive sampling periods, it is determined that the operation stability of the host decreases.

[0150] Secondly, a parameter weight backtracking adjustment mechanism is established. The weighted combination value of the overall energy efficiency improvement rate and the operation stability index is used as the model correction trigger condition. When the energy efficiency improvement rate is lower than the preset target value, such as 5%, or the stability index comprehensive score is lower than 80 points, the parameter weight adjustment process is started.

[0151] Specifically, a multi-objective optimization model is constructed, in which the running time cumulative factor weight in the health degree attenuation model, the maintenance effect recovery factor weight and the energy efficiency deviation weight are the parameters to be adjusted. The maximum energy efficiency improvement rate and the maximum stability index comprehensive score are the optimization objectives, and the sum of the weights of the three is 1 as the constraint condition.

[0152] The model is solved by gradient descent method. If the energy efficiency improvement rate is insufficient, the weight of the maintenance effect recovery factor is increased to strengthen the positive influence of the maintenance effect on the load upper limit, so that the host with better performance recovery after maintenance can bear higher load. If the stability decreases, the weight of the running time cumulative factor is increased to strictly limit the load lower limit of the old host, so as to avoid excessive pursuit of energy efficiency leading to excessive load fluctuation.

[0153] In summary, compared with the prior art, the present application deeply integrates the health degree characteristic parameters of the energy host into the whole process of load distribution optimization, and constructs a closed-loop control mechanism of "health degree perception-dynamic constraint-energy efficiency correction-adaptive optimization".

[0154] In summary, the embodiments of the present application have at least the following technical effects:

[0155] This application provides a load balancing distribution control method for multiple energy hosts operating in parallel. First, it acquires parameters such as the current output load, energy consumption rate, and historical maintenance records of each energy host in real time, providing a comprehensive data foundation for subsequent health assessment and energy efficiency analysis. Second, it introduces a host health decay model, innovatively coupling the accumulated factor of operating years and the maintenance effect recovery factor to dynamically calculate the real-time health coefficient of each host. Third, based on the real-time health coefficient and a preset standard energy efficiency benchmark curve, the standard curve is scaled using the accumulated factor of operating years and the maintenance effect recovery factor. Combined with current load demand queries, the theoretical energy consumption rate and two estimated energy consumption rates are obtained. Then, weighted summation is performed based on the ratio of the current load to the rated load to obtain an accurate energy efficiency deviation, providing a precise correction basis for load distribution optimization. Subsequently, a particle swarm optimization algorithm is used to solve the aforementioned optimization function. During the iteration process, the inertia weight is dynamically adjusted based on the cumulative factor of the operating years of each host and the recovery factor of the maintenance effect, updating the velocity and position vectors of the particles, and strict constraint processing is applied. This allows the algorithm to better consider the characteristics of hosts in different health states when searching for the optimal solution, improving optimization efficiency and solution accuracy. Finally, the overall energy efficiency improvement rate of the system and the operational stability indicators of each host are continuously monitored, and the parameter weights of the host health decay model are dynamically adjusted accordingly, forming a closed-loop adaptive optimization control mechanism. This mechanism continuously learns and adapts to changes in host performance and continuously optimizes the load allocation strategy. Through the above technical solution, this application maximizes the energy utilization efficiency of the entire parallel system while ensuring system operational stability. It effectively solves the problems of overall performance degradation and high energy consumption caused by ignoring the age and maintenance status differences of hosts in existing technologies, providing strong technical support for the efficient and stable operation of parallel systems with multiple energy hosts.

[0156] Example 2, as Figure 2 As shown, based on the same inventive concept as the load balancing distribution control method for multiple energy hosts operating in parallel provided in Embodiment 1, this application also provides a load balancing distribution control system for multiple energy hosts operating in parallel, including:

[0157] The information acquisition module 11 is used to acquire the real-time operating parameter set of each energy host in the parallel system, wherein the real-time operating parameter set includes at least the current output load, energy consumption rate and historical maintenance markers;

[0158] Data processing module 12 is used to dynamically calculate the real-time health coefficient of each energy host based on the real-time operating parameter set and through the host health decay model, wherein the host health decay model is coupled with the operating years accumulation factor and the maintenance effect recovery factor.

[0159] An energy efficiency optimization module 13 is configured to calculate an energy efficiency deviation of each energy host under a current load demand according to the real-time health degree coefficient and a preset standard energy efficiency benchmark curve, and construct a load redistribution optimization function with a target of minimizing the total energy consumption of the parallel system;

[0160] An instruction generation module 14 is configured to solve the load redistribution optimization function by using a particle swarm optimization algorithm, and dynamically generate and issue individualized load setting instructions to each energy host;

[0161] An instruction execution module 15 is configured to monitor an overall energy efficiency improvement rate of the parallel system and an operation stability index of each energy host after executing the individualized load setting instructions, and dynamically backtrack and adjust parameter weights of the host health degree attenuation model according to the overall energy efficiency improvement rate and the operation stability index.

[0162] Further, in an application embodiment, the construction step of the host health degree attenuation model comprises:

[0163] According to the equipment model and design parameters of each energy host in the parallel system, the corresponding standard energy efficiency benchmark curve and rated service life data are retrieved;

[0164] Based on a historical operation database, a sample data set of each energy host in different operation stages is collected, wherein the sample data set at least includes cumulative operation years, types of previous maintenance records, and energy efficiency measured data in the corresponding period;

[0165] According to the rated service life data and the cumulative operation years, a benchmark curve representing performance attenuation with operation time is fitted and generated, and a quantitative calculation rule of the operation year accumulation factor is defined;

[0166] The previous maintenance records are analyzed to distinguish the contribution weights and time efficiency periods of different maintenance types to the energy host efficiency recovery, and a quantitative calculation rule of the maintenance effect recovery factor is defined;

[0167] The cumulative operation years, previous maintenance records, and energy efficiency measured data are taken as input samples, and the operation year accumulation factor and the maintenance effect recovery factor calculated by the quantitative calculation rule are taken as intermediate supervision labels, and a model framework of the host health degree attenuation model is constructed;

[0168] The model framework is supervised and trained by using the input samples and intermediate supervision labels until the error converges to an allowable range, and the construction of the model is completed.

[0169] Further, in an application embodiment, the quantitative calculation rule of the operation year accumulation factor comprises:

[0170] According to the rated life data, a performance attenuation inflection point year limit is set, when the cumulative running year limit is less than the performance attenuation inflection point year limit, a linear light attenuation function is used to calculate a running year limit cumulative factor;

[0171] When the cumulative running year limit is greater than or equal to the performance attenuation inflection point year limit, an exponential accelerated attenuation function is used to calculate the running year limit cumulative factor.

[0172] Further, in one application embodiment, the quantitative calculation rule of the maintenance effect recovery factor includes:

[0173] According to the maintenance type of the historical maintenance record, a preset contribution weight coefficient is matched;

[0174] According to the interval between the completion time point of each maintenance and the current time, combined with the preset time limit period of the corresponding maintenance type, the residual performance contribution value of single maintenance is calculated;

[0175] The weighted sum of all residual performance contribution values corresponding to the historical maintenance records is performed to obtain the maintenance effect recovery factor.

[0176] Further, according to the real-time health degree coefficient and a preset standard energy efficiency benchmark curve, the energy efficiency deviation of each energy host under the current load demand is solved, including:

[0177] The running year limit cumulative factor is used to scale the standard energy efficiency benchmark curve to obtain a first equivalent energy efficiency curve;

[0178] The maintenance effect recovery factor is used to scale the standard energy efficiency benchmark curve to obtain a second equivalent energy efficiency curve;

[0179] Based on the current load demand, the theoretical energy consumption rate, the first estimated energy consumption rate and the second estimated energy consumption rate are obtained by querying on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve respectively;

[0180] The first difference value between the first estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate is calculated as a first energy efficiency deviation component;

[0181] The second difference value between the second estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate is calculated as a second energy efficiency deviation component;

[0182] According to the ratio of the current load demand to the rated load corresponding to the standard energy efficiency benchmark curve, a first weight and a second weight are determined;

[0183] The first energy efficiency deviation component and the second energy efficiency deviation component are summed with the first weight and the second weight to obtain an energy efficiency deviation of each energy host.

[0184] Further, based on the current load demand, a theoretical energy consumption rate, a first estimated energy consumption rate and a second estimated energy consumption rate are obtained on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve respectively, including:

[0185] According to the current load demand, a matching search is performed in discrete load-energy consumption rate data points of the standard energy efficiency benchmark curve;

[0186] If a data point completely matching the current load demand is found, the corresponding energy consumption rate is directly read as the theoretical energy consumption rate;

[0187] If a data point completely matching the current load demand is not found, the nearest two data points are selected, and the theoretical energy consumption rate under the current load demand is obtained through linear interpolation calculation;

[0188] According to the ratio of the current load demand to the rated load of each energy host, a load rate interval is determined;

[0189] If the load rate is lower than a preset low load threshold, the first estimated energy consumption rate is obtained on the first equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and the second estimated energy consumption rate is directly obtained on the second equivalent energy efficiency curve by using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve;

[0190] If the load rate is in a preset medium-high load interval, the second estimated energy consumption rate is obtained on the second equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and the first estimated energy consumption rate is obtained on the first equivalent energy efficiency curve by using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve.

[0191] Further, in one application embodiment, a load redistribution optimization function with the minimum total energy consumption of the parallel system as the target is constructed, including:

[0192] Taking the to-be-allocated load values of each energy host in the parallel system as optimization variables;

[0193] Taking the total load demand in the parallel system equal to the sum of the to-be-allocated load values of each energy host as an equality constraint;

[0194] Taking the allowable load range of each energy host as an inequality constraint, wherein the lower limit of the allowable load range is determined based on the running age cumulative factor, and the upper limit is determined based on the maintenance effect recovery factor;

[0195] The energy efficiency deviation of each energy host is used as a correction coefficient for calculating the estimated energy consumption of the corresponding host under the to-be-allocated load value.

[0196] A load re-allocation optimization function is constructed with the sum of the estimated energy consumptions of all energy hosts being minimized as a target.

[0197] In an embodiment, the instruction generation module 14 is specifically configured to:

[0198] The load re-allocation optimization function is solved by using a particle swarm optimization algorithm, and individual load setting instructions of each energy host are dynamically generated and issued, including:

[0199] A particle swarm is initialized, wherein a position vector of each particle represents a set of load allocation schemes, each dimension of the position vector corresponds to an energy host, and the initial value of each dimension is randomly generated within the allowed load range of the corresponding energy host;

[0200] The load re-allocation optimization function is used as a fitness function of the particle swarm optimization algorithm, and the fitness value of each particle is calculated;

[0201] In each iteration, the speed vector and the position vector of each particle are updated based on the individual historical optimal position of the particle and the global optimal position of the particle swarm, wherein the inertia weight used in the updating process is dynamically adjusted according to the operation age cumulative factor and the maintenance effect recovery factor of the energy host in the corresponding dimension;

[0202] Constraint processing is performed on the updated particle position vector;

[0203] It is determined whether a preset maximum number of iterations is reached, or the improvement amplitude of the global optimal fitness value of the particle swarm within a continuous preset number of iterations is lower than a preset convergence threshold;

[0204] If any condition is met, the iteration is terminated, and the global optimal position vector of the particle swarm is used as an optimal load allocation scheme;

[0205] The optimal load allocation scheme is decoded into individual load setting instructions of each energy host, and is issued to the control system of the corresponding energy host.

[0206] Further, in an application embodiment, in each iteration, the speed vector and the position vector of each particle are updated based on the individual historical optimal position of the particle and the global optimal position of the particle swarm, including:

[0207] An adaptive inertia weight is set for the update of the speed vector and the position vector of each particle, wherein the value of the adaptive inertia weight is positively correlated with the operation age cumulative factor of the energy host in the corresponding dimension, and is negatively correlated with the maintenance effect recovery factor.

[0208] Based on the individual optimal position of the particle, the global optimal position of the particle group and the adaptive inertia weight, a velocity vector of the particle is updated according to a preset basic formula of the particle swarm optimization algorithm;

[0209] The updated velocity vector is added to a current position vector of the particle to obtain an updated position vector.

[0210] It should be noted that the above sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0211] The above description is only the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0212] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application intends to include these modifications and changes.

Claims

1. A load balancing distribution control method in parallel operation of multiple energy hosts, characterized by, The method comprises: collecting and associating real-time operation parameter sets of each energy host in the system, wherein the real-time operation parameter sets at least include current output load, energy consumption rate and historical maintenance marker; based on the real-time operation parameter sets, dynamically calculating real-time health degree coefficients of each energy host through a host health degree attenuation model, wherein the host health degree attenuation model couples operation life cumulative factor and maintenance effect recovery factor; according to the real-time health degree coefficients and preset standard energy efficiency benchmark curve, solving energy efficiency deviation degrees of each energy host under current load demand, and constructing a load redistribution optimization function with the minimum total energy consumption of the parallel system as the target, wherein according to the real-time health degree coefficients and preset standard energy efficiency benchmark curve, solving energy efficiency deviation degrees of each energy host under current load demand comprises: scaling the standard energy efficiency benchmark curve by using the operation life cumulative factor to obtain a first equivalent energy efficiency curve; scaling the standard energy efficiency benchmark curve by using the maintenance effect recovery factor to obtain a second equivalent energy efficiency curve; based on the current load demand, respectively querying theoretical energy consumption rate, first estimated energy consumption rate and second estimated energy consumption rate on the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve and the second equivalent energy efficiency curve, comprising: according to the current load demand, matching and searching in discrete load-energy consumption rate data points of the standard energy efficiency benchmark curve; if a data point completely matched with the current load demand is found, directly reading the corresponding energy consumption rate as the theoretical energy consumption rate; if a data point completely matched with the current load demand is not found, selecting the nearest two data points to obtain the theoretical energy consumption rate under the current load demand through linear interpolation calculation; determining a load rate interval according to the ratio of the current load demand to the rated load of each energy host; if the load rate is lower than a preset low load threshold, obtaining the first estimated energy consumption rate on the first equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and directly using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve as the second estimated energy consumption rate on the second equivalent energy efficiency curve; if the load rate is in a preset medium-high load interval, obtaining the second estimated energy consumption rate on the second equivalent energy efficiency curve by using the same method as querying the theoretical energy consumption rate, and using the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve as the first estimated energy consumption rate on the first equivalent energy efficiency curve; calculating the first difference value between the first estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate as a first energy efficiency deviation component; calculating the second difference value between the second estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate as a second energy efficiency deviation component; determining a first weight and a second weight according to the ratio of the current load demand to the rated load corresponding to the standard energy efficiency benchmark curve; summing the first energy efficiency deviation component and the second energy efficiency deviation component weighted by the first weight and the second weight to obtain an energy efficiency deviation of each energy host; solving the load redistribution optimization function by using a particle swarm optimization algorithm to dynamically generate and issue a personalized load setting instruction to each energy host; monitoring an overall energy efficiency improvement rate of the parallel system and an operation stability index of each energy host after executing the personalized load setting instruction, and dynamically backtracking and adjusting a parameter weight of the host health degree attenuation model according to the overall energy efficiency improvement rate and the operation stability index.

2. The load balancing and distribution control method in parallel operation of multiple energy hosts according to claim 1, characterized in that, The construction steps of the host health degree attenuation model include: According to the equipment model and design parameters of each energy host in the parallel system, the corresponding standard energy efficiency benchmark curve and rated service life data are retrieved; Based on the historical operation database, a sample data set of each energy host in different operation stages is collected, wherein the sample data set at least includes cumulative operation years, types of historical maintenance records, and energy efficiency measured data in the corresponding period; According to the rated service life data and the cumulative operation years, a benchmark curve representing performance attenuation with operation time is fitted and generated, and a quantitative calculation rule of the operation year accumulation factor is defined; The historical maintenance records are analyzed to distinguish the contribution weight and time efficiency period of different maintenance types to the energy host efficiency recovery, and a quantitative calculation rule of the maintenance effect recovery factor is defined; The model framework of the host health degree attenuation model is constructed by taking the cumulative operation years, historical maintenance records, and energy efficiency measured data as input samples, and taking the operation year accumulation factor and the maintenance effect recovery factor calculated by the quantitative calculation rule as intermediate supervision labels; The model framework is supervised and trained by using the input samples and intermediate supervision labels until the error converges to the allowable range, and the construction of the model is completed.

3. The load balancing and distribution control method in parallel operation of multiple energy hosts according to claim 2, characterized in that, The quantitative calculation rule of the operation year accumulation factor includes: According to the rated service life data, a performance attenuation inflection point year is set, when the cumulative operation years are less than the performance attenuation inflection point year, a linear light attenuation function is used to calculate the operation year accumulation factor; When the cumulative operation years are greater than or equal to the performance attenuation inflection point year, an exponential accelerated attenuation function is used to calculate the operation year accumulation factor.

4. The load balancing and distribution control method in parallel operation of multiple energy hosts according to claim 2, characterized in that, The quantitative calculation rule of the maintenance effect recovery factor includes: According to the maintenance type of the historical maintenance records, a preset contribution weight coefficient is matched; According to the interval between the completion time point of each maintenance and the current time, combined with the preset time efficiency period of the corresponding maintenance type, the residual performance contribution value of single maintenance is calculated; All residual performance contribution values corresponding to the historical maintenance records are weighted and summed to obtain the maintenance effect recovery factor.

5. The load balancing and distribution control method for parallel operation of multiple energy hosts according to claim 1, wherein, A load redistribution optimization function with the minimum total energy consumption of the parallel system as the target is constructed, including: Taking the to-be-allocated load value of each energy host in the parallel system as the optimization variable; Taking the total load demand of the parallel system equal to the sum of the to-be-allocated load values of each energy host as the equality constraint; taking the allowed load range of each energy host as an inequality constraint, wherein the lower limit of the allowed load range is determined based on the operation age accumulation factor, and the upper limit is determined based on the maintenance effect recovery factor; taking the energy efficiency deviation degree of each energy host as a correction coefficient for calculating the estimated energy consumption of the corresponding host under the to-be-allocated load value; constructing the load redistribution optimization function with the minimum sum of the estimated energy consumptions of all energy hosts as a target.

6. The load balancing and distribution control method for multiple energy host parallel operation according to claim 1, characterized in that, solving the load redistribution optimization function by using a particle swarm optimization algorithm, and dynamically generating and issuing individualized load setting instructions of each energy host, including: initializing a particle swarm, wherein the position vector of each particle represents a set of load allocation schemes, each dimension of the position vector corresponds to an energy host, and the initial value of each dimension is randomly generated within the allowed load range of the corresponding energy host; taking the load redistribution optimization function as the fitness function of the particle swarm optimization algorithm, and calculating the fitness value of each particle; in each iteration, updating the velocity vector and the position vector of each particle based on the individual historical optimal position and the global optimal position of the particle swarm, wherein the inertia weight used in the updating process is dynamically adjusted according to the operation age accumulation factor and the maintenance effect recovery factor of the energy host in the corresponding dimension; performing constraint processing on the updated particle position vector; determining whether a preset maximum iteration number is reached, or the improvement amplitude of the global optimal fitness value of the particle swarm within a continuous preset iteration number is lower than a preset convergence threshold; if any condition is met, terminating the iteration, and taking the global optimal position vector of the particle swarm as the optimal load allocation scheme; decoding the optimal load allocation scheme into individualized load setting instructions of each energy host, and issuing the instructions to the control system of the corresponding energy host.

7. The load balancing and distribution control method in parallel operation of multiple energy hosts according to claim 6, characterized in that, in each iteration, updating the velocity vector and the position vector of each particle based on the individual historical optimal position and the global optimal position of the particle swarm, including: setting an adaptive inertia weight for the update of the velocity vector and the position vector of each particle, wherein the value of the adaptive inertia weight is positively correlated with the operation age accumulation factor of the energy host in the corresponding dimension, and is negatively correlated with the maintenance effect recovery factor; updating the velocity vector of the particle according to a preset basic formula of the particle swarm optimization algorithm based on the individual historical optimal position, the global optimal position and the adaptive inertia weight; adding the updated velocity vector and the current particle position vector to obtain the updated position vector.

8. A load balancing distribution control system for multiple energy host parallel operation, characterized in that, a load balancing and distribution control method for performing the parallel operation of a plurality of energy hosts according to any one of claims 1-7, including: an information acquisition module, configured to acquire real-time operation parameter sets of each energy host in the system, wherein the real-time operation parameter sets at least include a current output load, an energy consumption rate and a historical maintenance marker; a data processing module, configured to dynamically calculate real-time health degree coefficients of each energy host by a host health degree attenuation model based on the real-time operation parameter sets, wherein the host health degree attenuation model is coupled with an operation age accumulation factor and a maintenance effect recovery factor; The energy efficiency optimization module is used to calculate the energy efficiency deviation of each energy host under the current load demand based on the real-time health coefficient and the preset standard energy efficiency benchmark curve, and to construct a load redistribution optimization function with the objective of minimizing the total energy consumption of the parallel system. Calculating the energy efficiency deviation of each energy host under the current load demand based on the real-time health coefficient and the preset standard energy efficiency benchmark curve includes: The standard energy efficiency benchmark curve is scaled using the accumulated factor of operating years to obtain the first equivalent energy efficiency curve; The standard energy efficiency baseline curve is scaled using the maintenance effect recovery factor to obtain a second equivalent energy efficiency curve; Based on the current load demand, the theoretical energy consumption rate, the first estimated energy consumption rate, and the second estimated energy consumption rate are obtained by querying the standard energy efficiency benchmark curve, the first equivalent energy efficiency curve, and the second equivalent energy efficiency curve, respectively, including: Based on the current load demand, a matching search is performed on the discrete load-energy rate data points of the standard energy efficiency baseline curve; If a data point that perfectly matches the current load demand is found, the corresponding energy consumption rate is directly read as the theoretical energy consumption rate. If no data point is found that perfectly matches the current load demand, the two nearest data points are selected, and the theoretical energy consumption rate under the current load demand is obtained by linear interpolation. The load factor range is determined based on the ratio of the current load demand to the rated load of each energy host. If the load rate is lower than the preset low load threshold, the first estimated energy consumption rate is obtained on the first equivalent energy efficiency curve using the same method as querying the theoretical energy consumption rate, and the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve is directly used as the second estimated energy consumption rate on the second equivalent energy efficiency curve. If the load rate is in the preset medium-high load range, the second estimated energy consumption rate is obtained on the second equivalent energy efficiency curve using the same method as querying the theoretical energy consumption rate, and the energy consumption rate of the corresponding data point on the standard energy efficiency benchmark curve is used as the first estimated energy consumption rate on the first equivalent energy efficiency curve. Calculate the percentage of the first difference between the first estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate, as the first energy efficiency deviation component; Calculate the percentage of the second difference between the second estimated energy consumption rate and the theoretical energy consumption rate relative to the theoretical energy consumption rate, as the second energy efficiency deviation component; The first weight and the second weight are determined based on the ratio of the current load demand to the rated load corresponding to the standard energy efficiency benchmark curve. The first energy efficiency deviation component and the second energy efficiency deviation component are weighted and summed according to the first weight and the second weight to obtain the energy efficiency deviation of each energy host. The instruction generation module is used to solve the load redistribution optimization function using the particle swarm optimization algorithm, and dynamically generate and send personalized load setting instructions to each energy host. The instruction execution module is used to monitor the overall energy efficiency improvement rate of the parallel system and the operational stability indicators of each energy host after executing the personalized load setting instruction, and dynamically adjust the parameter weights of the host health decay model based on the overall energy efficiency improvement rate and the operational stability indicators.

Images