Control method of multi-set compressor unit coordinated with gas injection system

Abstract

The application discloses a kind of control methods of multiple sets of compressor unit collaborative gas injection system, comprising: collecting unit and gas storage static characteristic model data, system real-time state data and system gas injection demand data;Equivalent energy storage data of gas storage and gas storage storage state data are calculated, and marginal energy storage price data of gas storage are generated based on preset pricing function;Equivalent constraint domain data of unit gas storage is constructed to each unit and the connection combination of gas storage;The relative position of system real-time state data in equivalent constraint domain data of unit gas storage is calculated, and comprehensive equivalent margin index data representing the comprehensive safety degree of distance multidimensional constraint boundary is obtained;Collaborative calculation unit gas injection allocation and gas storage matching flow, generate system control execution data.The application solves the problem of decoupling of physical safety and economic value in traditional control, realizes the optimal collaboration of system gas injection efficiency and energy storage value under the premise of ensuring multidimensional constraint safety.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of compressed air energy storage control, and particularly relates to a control method of a multi-compressor unit coordinated air injection system. BACKGROUND

[0002] With the construction of new power systems, large-scale compressed air energy storage technology has become a key means to solve new energy consumption and grid peak regulation due to its large capacity and long service life. The coordinated air injection system with multiple sets of compressor units and multiple air storage libraries operating in parallel can improve the regulation flexibility and operation reliability of the system, and is the mainstream technical route for current large-scale energy storage power station construction.

[0003] The existing multi-unit coordinated control mainly adopts a load distribution strategy based on a single index, such as equal flow distribution, equal surge margin control or equal efficiency optimization control. In terms of air storage library management, a simple pressure balancing strategy is usually adopted, that is, the pressure of each air storage library is kept consistent through valve adjustment. These methods decouple the complex physical constraints (such as surge boundary and efficiency interval) and the state change of the air storage library, and usually control a single variable through an independent PID loop, without establishing a unified system-level coordination model.

[0004] However, the existing technology has defects in dealing with the deep coupling of physical safety boundaries and energy storage economic value. The main problem is that the existing single-index control cannot perceive the nonlinear compression of the unit feasible working condition domain due to the dynamic increase of the air storage library back pressure, and the linear pressure balancing strategy ignores the sharp increase in air injection cost under high filling state. SUMMARY

[0005] The application aims to provide a control method of a multi-compressor unit coordinated air injection system to solve the above problems existing in the prior art.

[0006] The technical scheme is a control method of a multi-compressor unit coordinated air injection system, comprising:

[0007] Collecting static characteristic model data of the units and the air storage libraries, system real-time state data containing operation parameters of the units and the air storage libraries, and system air injection demand data;

[0008] Using the static characteristic model data of the air storage libraries and the system real-time state data to calculate air storage library equivalent energy storage data and air storage library storage state data, and generating air storage library marginal energy storage price data based on a preset pricing function;

[0009] Based on the static characteristic model data of the units, the system real-time state data and the air storage library marginal energy storage price data, constructing unit-air storage library equivalent constraint domain data for each connection combination of the units and the air storage libraries;

[0010] The relative position of the real-time state data of the computing system in the equivalent constraint domain data of the unit gas storage is obtained, and comprehensive equivalent margin index data representing the comprehensive safety degree of the distance from the multi-dimensional constraint boundary is obtained.

[0011] According to the system gas injection demand data, the comprehensive equivalent margin index data and the gas storage marginal energy storage price data, the unit gas injection distribution quantity and the gas storage matching flow are cooperatively calculated, and system control execution data is generated.

[0012] Beneficial effects, the present application solves the problem of decoupling of physical safety and economic value in traditional control, and realizes the optimal cooperation of system gas injection efficiency and energy storage value under the premise of ensuring multi-dimensional constraint safety. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 A step flow chart of a control method of a multi-unit compressor set cooperative gas injection system provided by the embodiment of the present application.

[0014] Figure 2 A step flow chart of generating physical constraint sub-domain data provided by the embodiment of the present application.

[0015] Figure 3 A step flow chart of constructing unit gas storage equivalent constraint domain data provided by the embodiment of the present application.

[0016] Figure 4 A step flow chart of obtaining comprehensive equivalent margin index data provided by the embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable personnel in the art to better understand the present application scheme, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a 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 should belong to the scope of protection of the present application.

[0018] It should be noted that the terms include and have as well as any variations thereof are intended to cover inclusive rather than exclusive inclusion, for example, a process, method, system, product or apparatus that includes a list of steps or units need not be limited to those steps or units explicitly listed, but can include other steps or units not expressly listed or inherent to such process, method, product or apparatus.

[0019] In the study, it is found that the traditional control method regards efficiency, surge and pressure as isolated constraints, lacks a unified geometric space integrating multi-dimensional constraints, and causes the unit to easily touch the dynamically changing surge or overpressure boundary when pursuing high efficiency, triggering unplanned shutdown. At the same time, the trimming logic based on pressure difference alone cannot reflect the nonlinear characteristics of the safety risk and the decline in regulation flexibility when the gas storage approaches full capacity, resulting in that the gas injection load cannot be optimally allocated according to the real energy storage value, limiting the overall energy charging efficiency and life balance of the system.

[0020] As shown in Figure 1 , a control method for a multi-compressor unit coordinated gas injection system is proposed, comprising the following steps:

[0021] Collecting unit static characteristic model data, gas storage static characteristic model data, system real-time state data containing unit and gas storage operating parameters, and system gas injection demand data.

[0022] In this embodiment, the unit static characteristic model data is a set of parameters describing the inherent physical properties of the compressor unit. Specifically, this data can be obtained by analyzing the historical operating data of the compressor unit or through factory tests. For example, the unit static characteristic model data includes the efficiency curve and the surge boundary curve of the unit under different operating conditions. The efficiency curve describes the relationship between the unit efficiency and the flow rate and outlet pressure, while the surge boundary curve defines the minimum flow rate limit for stable operation of the unit. In addition, it also includes the rated parameters of the unit, such as rated power, rated speed, and allowed maximum outlet pressure range, etc.

[0023] The gas storage static characteristic model data covers the design parameters of each gas storage. Exemplarily, it at least includes the geometric volume V j , the design allowed maximum operating pressure P max_j , the minimum operating pressure P min_j , and the allowed maximum pressure change rate dP max_j of each gas storage. These static parameters constitute the basic physical constraints for subsequent calculation of equivalent energy storage and judgment of safety boundary.

[0024] The system real-time state data refers to the actual operating parameters of the field devices collected in the current control period. Specifically, it includes the real-time outlet pressure, inlet pressure, current flow rate, speed and guide vane opening degree of each unit, as well as the real-time pressure P res_j , real-time temperature T j of each gas storage and the current opening state of various valves connected thereto. It is usually collected by pressure sensors, temperature sensors and flow meters arranged in the field in real time, and transmitted to the control system through industrial Ethernet or field bus.

[0025] The system's gas injection demand data reflects the total energy demand of the upper-level scheduling system for the current time period. For example, it clarifies the total gas injection flow target Q that the system needs to achieve within this control cycle. total This includes information on whether the start-up or shutdown of a specific unit is currently permitted. This serves as the basis for subsequent load allocation.

[0026] Using static characteristic model data of the gas storage facility and real-time system status data, the equivalent energy storage data and storage status data of the gas storage facility are calculated, and the marginal energy storage price data of the gas storage facility is generated based on a preset pricing function.

[0027] In this embodiment, the equivalent energy storage data E of the gas storage facility j This refers to uniformly converting the gas state within a gas storage facility into an energy value representing its effective work potential. Specifically, the system utilizes the thermodynamic equation of state to convert the real-time collected pressure P... res_j and temperature T j Combined with the volume V of the gas storage facility j This calculates the current gas mass or available exergy value stored in the gas storage facility. This unifies the gas storage facility status under different pressure and temperature conditions to the energy dimension, making different gas storage facilities comparable. (Gas Storage Status Data SoC) j The equivalent energy storage capacity was further normalized. Specifically, the system normalized the energy storage capacity E based on the maximum storable energy E defined in the static model of the gas storage facility. max_j and minimum effective energy storage E min_j Calculate the current equivalent energy storage data E of the gas storage facility. j The relative position within this range. For example, SoC j The value typically ranges from 0 to 1. A value closer to 1 indicates the gas storage facility is close to full capacity, while a value closer to 0 indicates it is close to empty. This reflects the storage facility's level of fullness. Marginal energy storage price data for gas storage facilities (L) j Based on SoC j The generated economic evaluation indicators. In specific implementation, the system presets a non-linear pricing function, which will SoC j This is mapped to a price value. The pricing function is designed as a convex function, meaning that when the SoC... j When it is low, the price L j It remains at a low level and grows slowly; while when SoC j When L approaches 1, the price L j It will rise sharply. By imposing a high virtual price to punish the continued injection of gas into the high-state-of-memory (SoC) storage tank, the risk of overcharging is automatically suppressed at the control algorithm level.

[0028] Based on the unit static characteristic model data, the system real-time state data and the gas storage marginal energy storage price data, the unit gas storage equivalent constraint domain data is constructed for each unit and gas storage combination. The unit gas storage equivalent constraint domain data represents the feasible working condition set that integrates the physical characteristics of the unit and the energy storage value constraint.

[0029] In the embodiment, the unit gas storage equivalent constraint domain data D ij is a multi-dimensional geometric space description. Specifically, the constraint domain is not only a physically feasible region, but also an operable space that integrates economic value. The construction process is to define the physically feasible region that meets the efficiency requirements and is away from the surge boundary based on the static characteristics of the unit in the two-dimensional plane composed of flow and outlet pressure. On this basis, the marginal energy storage price data L j of the gas storage is introduced to deform and correct the physical region. Specifically, when the marginal price L j of a certain gas storage is high, it indicates that the cost of injecting gas into it is high, and the system will accordingly shrink the feasible range of the combination in the constraint domain, for example, reduce the upper limit of the allowed maximum injection flow. Conversely, for the gas storage with a lower marginal price, the system will maintain or appropriately relax its feasible region. The originally abstract economic value is converted into a specific geometric constraint boundary, which limits the search space of the subsequent allocation algorithm.

[0030] The relative position of the system real-time state data in the unit gas storage equivalent constraint domain data is calculated to obtain the comprehensive equivalent margin index data representing the overall safety distance from the multi-dimensional constraint boundary.

[0031] Specifically, the comprehensive equivalent margin index data M ij is a quantitative index that can evaluate the safety distance of the current operating point from various constraint boundaries. In a preferred implementation, the system calculates the normalized distance of the current operating point with respect to the efficiency lower limit boundary, the surge safety boundary, the gas storage pressure upper limit boundary and the value boundary defined based on the price. In order to ensure the overall safety of the system, the calculation logic of the short board effect is adopted. Specifically, the system selects the minimum value of all the above-mentioned normalized distances as the final comprehensive equivalent margin M ij . For example, even if the current surge margin of the unit is large, if the corresponding gas storage pressure is close to the upper limit, resulting in a very small pressure margin, the final comprehensive equivalent margin M ij will also be judged as a minimum value. This makes the control system always focus on the weakest link and avoids the safety hazards caused by the neglect of a single dimension.

[0032] According to the system gas injection demand data, the comprehensive equivalent margin index data and the gas storage marginal energy storage price data, the unit gas injection allocation quantity and the gas storage matching flow are calculated in coordination to generate system control execution data to drive the field equipment.

[0033] In this embodiment, the unit gas injection allocation refers to the allocation of the total gas injection demand Q. total The target flow rate is broken down to the specific flow rate of each unit. Specifically, the comprehensive equivalent margin M is used. ij As a weighting factor, priority is given to allocating more gas injection load to combinations of generating units and gas storage facilities with greater margins, i.e., higher overall safety and economic value. If the initial allocation exceeds the equivalent constraint domain boundary of a combination, the system will automatically perform a limit adjustment and redistribute the remaining unfulfilled load to other combinations with remaining margins until the total demand is met. The gas storage facility balance flow rate refers to the amount of gas transferred between different gas storage facilities through connecting pipelines. Specifically, the system does not pursue absolutely equal pressure in all gas storage facilities, but rather calculates the price difference between them based on marginal energy storage price data. Based on the magnitude and direction of the price difference, control commands are generated to drive the balance valves, guiding gas from high-priced (high-filling) gas storage facilities to low-priced (low-filling) gas storage facilities, thereby achieving energy storage value balance across the entire system. The final generated system control execution data, including target flow rates for generating units and target valve openings, is sent to the field actuators, completing the control loop for this cycle.

[0034] In one possible implementation, the equivalent energy storage data and storage status data of the gas storage facility are calculated, including:

[0035] By using the thermodynamic equation of state, the pressure and temperature in the real-time system state data and the gas storage volume parameters in the static characteristic model data of the gas storage are converted into equivalent energy storage data of the gas storage, which characterizes the effective work potential.

[0036] In this embodiment, the equivalent energy storage data E of the gas storage facility j It is a physical quantity that reflects the actual energy level of the gas storage facility. It can be calculated using the actual gas law or a lookup table method. In one exemplary implementation,

[0037] E j = (P res_j * V j ) / (Z * R gas * T j );

[0038] Where E j For the equivalent energy storage data of the gas storage facility, P res_j V represents the real-time pressure of the gas storage facility. j Let Z be the gas storage volume, Z be the gas compressibility factor, and R be... gas T is the gas constant. jThe real-time temperature of the gas storage. Compared with using only the pressure indicator, this embodiment can more accurately reflect the actual work capacity of the gas in the gas storage under different temperature and pressure combinations.

[0039] The minimum effective energy storage and the maximum energy storage in the static characteristic model data of the gas storage are obtained, and the equivalent energy storage data of the gas storage is used for normalization calculation to generate the storage state data of the gas storage representing the filling degree of the gas storage.

[0040] In this embodiment, the storage state data SoC j is used to standardize the energy state of the gas storage to the interval of 0 to 1. Preferably, the system reads the design parameters of the gas storage to determine the minimum effective energy storage E min_j and the maximum energy storage E max_j allowed to operate, and the storage state data of the gas storage is:

[0041] SoC j = (E j - E min_j ) / (E max_j - E min_j );

[0042] wherein E j is the equivalent energy storage data of the gas storage. Further, in order to ensure the physical meaning of the data, the result is usually truncated in actual calculation, that is, if the calculation result is less than 0, 0 is taken, and if it is greater than 1, 1 is taken, so that SoC j is always located in the closed interval [0, 1].

[0043] In a further embodiment, the marginal energy storage price data of the gas storage is generated, including:

[0044] A marginal pricing model in the form of a predetermined convex function is constructed or called, and the marginal pricing model is configured to have a nonlinear growth characteristic that the slope of the function value increases with the increase of the value of the independent variable, so as to generate an increased penalty value when the independent variable approaches the saturation value.

[0045] Specifically, the marginal pricing model adopts a convex function structure, and one preferred mathematical expression form is:

[0046] L j = a j / (1 - SoC j +ε) + b j * SoC j ;

[0047] wherein L j is the original marginal energy storage price value, SoC j is the storage state data of the gas storage, aj b is the weight coefficient to control the steepness of the non-linear growth j ε is a very small positive number, for example 0.001, to prevent the denominator from being zero. When SoC j approaches 1, the denominator term (1 - SoC j ) approaches zero, causing the first term to increase sharply in value, thereby simulating the effect of the marginal cost tending to infinity. It can effectively form a soft wall at the algorithm level, generate an inhibition signal when the gas storage is about to be full, forcing the control system to reduce or stop injecting gas into the gas storage.

[0048] The storage state data of the gas storage is substituted into the marginal pricing model for calculation to generate gas storage marginal energy storage price data, so as to quantify the cost of each gas storage to accept further injection operation under the current state.

[0049] In this embodiment, the SoC j calculated in real time is substituted into the above convex function model to obtain the original marginal price value. In order to facilitate subsequent comprehensive calculation with other physical quantities such as efficiency and surge margin, the system further normalizes the original price. Specifically, the maximum value L max and the minimum value L min of the price of all gas storages at the current time are counted, and the original price L j is linearly mapped to the interval of 0 to 1 to obtain the normalized marginal price L norm_j . In addition, in order to eliminate the dramatic fluctuations of the price data caused by sensor noise or system disturbance, the system also introduces a smoothing processing link before generating the final output. Preferably, a first-order exponential smoothing algorithm is adopted to calculate the smoothed price L smooth (t) at the current time:

[0050] L smooth (t) = γ * L norm (t) + (1 - γ) * L smooth (t-1);

[0051] wherein γ is a smoothing coefficient, usually ranging from 0.1 to 0.3; L norm (t) is the normalized marginal price at the current time; and L smooth (t-1) is the smoothed price at the previous time. The final output of the gas storage marginal energy storage price data is more stable, which is conducive to improving the stability and robustness of the control system.

[0052] In a preferred embodiment, the process of constructing or invoking the marginal pricing model in the form of a preset convex function can also be: presetting multiple model variants adapted to different control objectives. Specifically, the system constructs a life balance priority strategy model and an efficiency priority strategy model. For the life balance priority strategy model, the number of charge-discharge cycles and the depth of each gas reservoir are kept consistent to prevent overuse of individual gas reservoirs. To this end, in this model variant, the convex function formula is: L j =a j_life / (1 - SoC j +ε) + b j_life * SoC j ; wherein the weight coefficient a j_life is set to a large value. For example, a j_life can take a value of 0.5 or higher, meaning that when SoC j rises, the growth rate of price L j is extremely fast, thus triggering the high price suppression signal earlier, forcing the system to shift the gas injection load to other low SoC gas reservoirs, achieving alternating use between the reservoirs. For the efficiency priority strategy model, those gas reservoirs with short connecting pipelines, small flow resistance, or thermodynamic state more conducive to efficient gas injection are preferentially used. In this model variant, the system adjusts the linear term coefficient b j_eff according to the inherent characteristics of the gas reservoir. For example, for a gas reservoir with smaller flow resistance, a smaller b j_eff is set, which has a lower base price at the same SoC, thus attracting more gas injection flow.

[0053] Further, the process of generating gas reservoir marginal energy storage price data can also be: parallel computing of candidate price data under the above multiple strategies. Specifically, for each gas reservoir, the system calculates the corresponding life balance price L j_life and efficiency priority price L j_eff . Read the current system operation strategy instruction to execute the preferred decision. Specifically, when the system is in the night energy storage period and the grid load is low, the system often pays more attention to the long-term health of the equipment, at this time the system automatically selects the life balance price L j_life as the final output of the marginal energy storage price data. And in the system in the stage of rush charging or the price sensitive period, in order to pursue the ultimate energy conversion efficiency, the system switches to select the efficiency priority price L j_eff as the output.

[0054] In addition, in some optional embodiments, the system can also generate the final price in a weighted fusion manner. For example, set a dynamic weight coefficient w strategy , and calculate the final price as:

[0055] L final= w strategy * L j_life + (1 - w strategy ) * L j_eff ;

[0056] Where L final For the final marginal energy storage price after integration, L j_life L is the price calculated based on the lifetime equilibrium strategy. j_eff The price is calculated based on an efficiency-first strategy. This is achieved by adjusting w. strategy The value can enable a smooth transition from focusing on lifespan to focusing on efficiency, allowing the control system to more flexibly respond to complex engineering needs.

[0057] like Figure 2 As shown, according to one aspect of this application, constructing equivalent constraint domain data for a gas storage unit includes generating physical constraint subdomain data, specifically:

[0058] Efficiency functions and surge margin functions are extracted from the static characteristic model data of the unit. Based on the efficiency functions and surge margin functions, regions that meet the preset lower efficiency limit and regions that meet the preset minimum surge margin are delineated in the flow-outlet pressure plane.

[0059] In this embodiment, the system uses flow rate Q as the horizontal axis and outlet pressure P as the horizontal axis. out A basic physical framework is constructed in a two-dimensional plane coordinate system for the vertical axis. Specifically, the efficiency function η(Q, P) of the unit is read. out And a preset efficiency lower limit coefficient α, for example, 0.9. Traverse all points in the coordinate plane and filter out all points that satisfy the efficiency η≥α·η ref The set of points, where η ref For rated efficiency; these point sets constitute the efficiency feasible region, which typically exhibits a closed or semi-closed shape located within the unit's high-efficiency region. Simultaneously, the unit's surge margin function SM(Q, P) is read. out and minimum surge margin threshold SM min For example, 10%. The system marks the surge boundary line on the coordinate plane and filters out those located to the right of the boundary line that satisfy SM≥SM. min The set of points constitutes the surge safety region, which keeps the unit away from unstable fluid conditions during operation.

[0060] Find the intersection of the region that satisfies the preset lower efficiency limit and the region that satisfies the preset minimum surge margin, and generate physical constraint subdomain data that characterizes the physical safety boundary.

[0061] Exemplarily, the system performs a Boolean intersection operation on the two regions described above. Specifically, only those operating points that are both within the efficiency feasible region and within the surge safety region are retained. The intersection region forms the physical constraint sub-domain data, representing the range of flow and pressure combinations that the unit is allowed to operate at when only its own physical performance and safety limits are considered. Geometrically, this sub-domain often appears as an irregular polygonal region enclosed by multiple curves and straight lines.

[0062] As shown in FIG. 1, in further embodiments, constructing the unit gas storage equivalent constraint domain data further comprises: Figure 3

[0063] normalizing the gas storage marginal energy storage price data to obtain a normalized marginal price, and calculating a region adjustment coefficient for representing the scaling degree of the operating condition region based on the normalized marginal price.

[0064] In the present embodiment, the system introduces the energy storage value factor to intervene in the physical region. Specifically, the system reads the gas storage marginal energy storage price data L j , and normalizes it to a value between 0 and 1, L norm_j . A calculation formula for the adjustment coefficient is constructed, for example, factor j = 1 - k shrink * L norm_j ; where factor j is the region adjustment coefficient; k shrink is a preset contraction strength coefficient, usually ranging from 0.2 to 0.5; L norm_j is the normalized gas storage marginal energy storage price. When the marginal price L norm_j of the gas storage is higher, i.e., closer to full storage or greater gas injection cost, the calculated region adjustment coefficient factor j is smaller; conversely, when the price is lower, factor j is close to 1. For example, if k shrink is 0.4, when L norm_j is 1, factor j is 0.6, meaning that the feasible region will be compressed to 60% of the original.

[0065] The region adjustment coefficient is used to geometrically contract or expand the operating condition range covered by the physical constraint sub-domain data, generating unit gas storage equivalent constraint domain data that integrates energy storage value constraints.

[0066] Specifically, the system uses the calculated region adjustment coefficient factor j to perform geometric transformation on the physical constraint sub-domain. Exemplarily, including contraction of the flow upper limit. The maximum allowed flow after correction is: ​

[0067] Q max_final = Q max_physical * factor j ;

[0068] where Q max_physical is the maximum flow allowed in the physical constraint sub-domain. In the flow- outlet pressure plane, it manifests as the right boundary of the constraint domain shifting left or inwards, directly reducing the range of acceptable injection flow for the high price gas storage.

[0069] In some alternative embodiments, the dilation deformation can also be an overall scaling of the area. For example, the system contracts the boundary of the entire feasible region by a factor of factor j with respect to the geometric center or minimum flow point of the constraint domain. This makes the otherwise abstract injection cost materialized as a physical limit of the unit operation space, so that the subsequent load dispatch algorithm naturally avoids the high cost region when searching for feasible solutions, achieving a deep integration of economy and physical feasibility.

[0070] In yet further embodiments, constructing the equivalent constraint domain data for the gas storage unit also includes:

[0071] Using the system real-time state data and the gas storage static characteristic model data, the system calculates the current pressure change rate of the gas storage and predicts the pressure evolution trajectory within a preset time window.

[0072] In this embodiment, the system can calculate the current pressure change rate dP res / dt of the gas storage based on the principle of mass conservation or a simplified gas state equation. Specifically, the system real-time statistics the total injection flow Q in_total of all compressor units injecting gas into the gas storage, and the total outflow Q out_total of the gas storage to the expander or other discharge channels. According to the equivalent volume characteristic parameter C j of the gas storage, the pressure change rate is calculated as:

[0073] dP res / dt≈(Q in_total -Q out_total ) / C j .

[0074] where the equivalent volume characteristic parameter C j reflects the sensitivity of the gas storage pressure to the change of gas mass, which is stored in the gas storage static characteristic model data. After obtaining the current pressure change rate, the system sets an effective prediction time window Δ t , for example, 5 minutes or 10 minutes. Using linear extrapolation or higher order prediction algorithm, the predicted pressure P pred at the end of the time window is calculated. For example,

[0075] P pred = P res +(dP res / dt)*Δ t ;

[0076] Where P res This provides the current real-time pressure, enabling the control system to anticipate whether the gas storage facility will reach its safety threshold in the near future under the current injection intensity.

[0077] Based on the pressure evolution trajectory and the preset pressure safety threshold, dynamic safety margin data of gas storage pressure is generated, and the dynamic safety margin data of gas storage pressure is mapped to the dynamic limit boundary of injection flow and outlet pressure, thereby correcting the equivalent constraint domain data of unit gas storage.

[0078] In this embodiment, the dynamic safety margin data of the gas storage tank pressure is calculated. Specifically, the maximum allowable operating pressure P of the gas storage tank is calculated. max Subtract the predicted pressure P pred To obtain the upper limit of pressure safety margin P_high Simultaneously, the maximum permissible rate of pressure change dP can also be calculated. max_rate Subtract the current rate of change of pressure dP res The absolute value of / dt yields the rate safety margin. rate Performing a mapping from the pressure domain to the flow domain translates the aforementioned safety margin into direct constraints on the unit's operating conditions. In practice, if the margin... P_high If the pressure is below the preset warning threshold, it indicates that if the current injection rate continues, the gas storage facility will experience overpressure within the predicted time window. At this point, the system uses the pressure-flow coupling relationship to inversely calculate the maximum allowable injection flow rate Q. allowable .For example,

[0079] Q allowable =(P max - P res ) * C j / Δ t + Q out_total ;

[0080] Q allowable P is the maximum allowable injection flow rate based on pressure dynamic constraints. res For the current real-time pressure, C j Q is the equivalent volume characteristic parameter of the gas storage facility. out_total This represents the current total venting flow rate. Using the calculated Q... allowableThe equivalent constraint domain data of the unit gas storage is corrected. Exemplarily, a vertical limiting line is added on the flow coordinate axis of the equivalent constraint domain, that is, the gas injection flow rate Q i of the unit is required to be less than or equal to Q allowable . By intersecting the dynamic limiting boundary with the original physical boundary and value boundary, the feasible working condition region is further cut. The gas injection instruction assigned to the unit is not only safe at the current moment, but also will not cause the gas storage pressure to exceed the limit in the future predicted time window, realizing the double safety constraints in time and space dimensions.

[0081] As Figure 4 shown, in a preferred implementation, the comprehensive equivalent margin index data is obtained, including:

[0082] The distances of the current operating point with respect to the efficiency boundary, the surge boundary and the pressure dynamic boundary in the unit gas storage equivalent constraint domain data are calculated respectively, and the local margins in each dimension are generated in combination with the gas storage marginal energy storage price data.

[0083] Exemplarily, for each unit and gas storage combination, the local margins in four key dimensions are calculated respectively. Including: efficiency dimension local margin margin η , the calculation method is:

[0084] margin η =(η curr -η min ) / (η ref -η min );

[0085] Where η curr is the current operating point efficiency, η min is the lower limit of efficiency, and η ref is the rated efficiency; the efficiency dimension local margin reflects the distance of the unit from the low efficiency area. Surge dimension local margin margin SM , the calculation method is:

[0086] margin SM = (SM curr - SM min ) / (SM ref - SM min );

[0087] Where SM curr is the current surge margin, SM min is the minimum surge margin threshold, and SM ref is the reference surge margin; the surge dimension local margin reflects the safety space of the unit from the surge instability boundary. Pressure dimension local margin margin P, the system directly utilizes the pressure dynamic safety margin, which is normalized to the interval of 0 to 1; the pressure dimension local margin reflects the remaining capacity of the gas storage away from the overpressure alarm. The value dimension local margin margin val The index can be constructed based on the marginal energy storage price data of the gas storage, and the specific calculation formula is:

[0088] margin val = 1 - L norm_j ;

[0089] Wherein L norm_j is the normalized marginal energy storage price of the gas storage; the lower the marginal price of the gas storage, that is, the smaller the cost of gas injection, the greater the value dimension margin, indicating that the gas storage is more suitable for receiving gas injection.

[0090] The minimum value of all normalized dimension local margins is extracted as the comprehensive equivalent margin index data to represent the safety degree of the unit and gas storage combination in the current most constrained dimension.

[0091] In this embodiment, the system applies the principle of the wooden barrel effect to make a comprehensive decision. Specifically, the local margins margin η , margin SM , margin P and margin val of each dimension calculated above are collected into a set, and a minimum value extraction operation is performed, that is, the comprehensive equivalent margin index data M ij =min(margin η , margin SM , margin P , margin val ), wherein min represents the minimum value operation. For example, in a specific calculation scenario, a unit operates in the high efficiency zone, and its efficiency margin is 0.9; far away from the surge line, the surge margin is 0.8; and the pressure of the gas storage is low, and the pressure margin is 0.7. However, due to the system strategy adjustment, the marginal price of the gas storage is extremely high, so that the value margin is only 0.1. In this case, the system determines that the final comprehensive equivalent margin M ij is 0.1. Reflects that the current combination is limited by economic value constraints, and is not suitable to bear more load, although its physical state seems very healthy.

[0092] Further, before generating the final output, in order to avoid the jump of the margin index caused by the sensor fluctuation, and thus causing the shock of the control allocation, the system performs time smoothing processing on the comprehensive equivalent margin index data M ij . Preferably, an exponential smoothing algorithm is adopted, that is:

[0093] M smooth (t) =β* M ij + (1 -β) * M smooth (t-1);

[0094] wherein M smooth (t) is the current time smoothed comprehensive equivalent margin, β is the smoothing coefficient, M smooth (t-1) is the smoothed value of the previous time. The smoothed comprehensive equivalent margin index data will be input as a stable weight parameter into the subsequent collaborative allocation link.

[0095] According to an aspect of the present application, the collaborative computer group gas injection allocation amount is calculated, comprising:

[0096] The comprehensive equivalent margin index data is normalized to construct an initial allocation weight, and the total gas injection target in the system gas injection demand data is decomposed into the initial allocation amount of each unit and gas storage combination using the initial allocation weight.

[0097] Specifically, the system aggregates the comprehensive equivalent margin index M smooth_ij of all available units and gas storage combinations. The sum ∑ M of all available combination margin values is calculated. The allocation weight w ij of each combination is calculated, and the formula is:

[0098] w ij = M smooth_ij / ∑ M ;

[0099] wherein w ij is the initial allocation weight, and M smooth_ij is the smoothed comprehensive equivalent margin of the combination. The initial allocation weight reflects the ability of each combination to bear the load, and the greater the margin, the greater the weight. The total gas injection target Q total is read, and the initial allocation amount Q init_ij of each combination is calculated, Q init_ij = Q total * w ij . Through the margin-based allocation method, the system naturally directs the main gas injection task to high-efficiency, far-from-surge, low-gas-storage pressure, and cheap marginal price high-quality combinations, achieving performance optimization of the entire system.

[0100] The initial allocation amount is mapped to the unit and gas storage equivalent constraint domain data for boundary checking. When the initial allocation amount exceeds the constraint domain boundary, the excess part is truncated and the remaining unallocated demand is calculated.

[0101] In this embodiment, the system performs a feasibility check on the initial allocation result. Specifically, for each combination, it determines whether its initial allocation quantity Q init_ij falls within the final equivalent constraint domain. In particular, it checks whether the initial allocation quantity Q init_ij exceeds the maximum flow Q max_final allowed by the constraint domain. If Q init_ij is greater than Q max_final , the system performs a truncation operation, forcibly setting the actual allocation quantity Q assign_ij of the combination to Q max_final , and calculates the overflow quantity Δ Qij = Q init_ij - Q max_final that the combination fails to bear. If Q init_ij is less than or equal to Q max_final , the allocation quantity is directly accepted, and the overflow quantity is zero. The system traverses all combinations, accumulates all overflow quantities, and obtains the remaining unallocated demand Q remain of the system.

[0102] Based on the comprehensive equivalent margin index data of the remaining unit and gas storage combination, the remaining unallocated demand is iteratively redistributed until all allocation quantities fall within the corresponding unit and gas storage equivalent constraint domain data range, generating combination gas injection allocation result data containing unit gas injection allocation quantities.

[0103] In this embodiment, if the remaining unallocated demand Q remain is greater than a preset convergence threshold, for example, 1 cubic meter per hour, the system will start the iterative redistribution process. Remove those combinations that have reached the boundary limit in the last round of allocation, i.e., those that have reached the saturation limit. Only non-saturated combinations that still have remaining capacity are retained. The weights of non-saturated combinations are recalculated. Specifically, the new weight w new_ij is equal to the margin M smooth_ij of the combination divided by the sum of the margins of all non-saturated combinations. The remaining demand Q remain is allocated to these non-saturated combinations according to the new weights, obtaining incremental allocation values. Again check whether the total flow after adding the increment is out of bounds, and if it is, truncate it again and accumulate the new remaining demand. This process is repeated until the remaining demand is zero or all combinations are saturated. The final determined unit flow targets constitute the combination gas injection allocation result data, which is directly converted into speed setting values or guide vane opening commands to accurately control the unit to perform gas injection tasks.

[0104] In one embodiment of the present application, the gas storage balancing flow is calculated in coordination, including:

[0105] According to the system real-time state data, identify the physical connection relationship between the gas storages, and extract the price values of the connected pair in the gas storage marginal energy storage price data.

[0106] Specifically, scan real-time pipeline topology data, and parse out gas storage pairs that are currently in a connected state or have a connected condition. For example, if there is a balancing pipeline between gas storage A and gas storage B and the isolation valve in between is in a fault-free state, the system will identify (A, B) as a valid balancing connection pair. A balancing connection list is established, and the normalized price L norm_A of gas storage A and the normalized price L norm_B of gas storage B are extracted from the gas storage marginal energy price data, respectively.

[0107] Based on the price values, the marginal price difference of the two connection pairs is calculated, and the marginal price difference is mapped to the target balancing flow and the corresponding valve opening target to generate gas storage balancing control instruction data containing the gas storage balancing flow to drive gas to flow from the gas storage with a high marginal price to the gas storage with a low marginal price or to achieve value balance through shunt adjustment.

[0108] In this embodiment, the marginal price difference AL AB of each pair of connected gas storages is calculated,

[0109] AL AB = L norm_A - L norm_B .

[0110] If the absolute value of AL AB is greater than a preset dead zone threshold, for example, 0.05, it indicates that there is a significant energy value imbalance between the two, and balancing is needed. The target balancing flow Q balance_target is determined according to the size of AL AB . Specifically, a proportional control strategy is adopted, Q balance_target = K balance * | AL AB |; where K balance is a proportional coefficient. The greater the value difference, the stronger the balancing strength. The flow direction is determined by the sign of AL AB : if it is positive, it means that the price of A storage is high (the filling degree is high or the gas injection cost is large), and gas should be supplied to B storage; otherwise, gas is supplied from B to A. The target flow is converted into the opening instruction of the physical valve. The flow characteristic curve or inverse formula of the balancing valve can be called. For example, using Q = K valve (open) * sqrt(P high - P low ), in the case where the target flow Q balance_target and the current pressure difference are known, the required target opening open target is inversely solved by numerical iteration or table lookup. The target opening constitutes the gas storage balancing control instruction data. Where Q is the flow used to match the target balancing; K valve(open) is a flow coefficient function related to the valve opening degree open, P high is the high-pressure side pressure, P low is the low-pressure side pressure;

[0111] In further embodiments, the coordinated calculation of the gas storage balancing flow further comprises:

[0112] Using the system real-time state data and the gas storage static characteristic model data, the pressure change and the storage state change of the gas storage after executing the gas storage balancing control instruction data are simulated and deduced.

[0113] Specifically, the system performs a virtual rehearsal before issuing the instruction. It is assumed that the balancing valve has been actuated to the target opening degree open target and maintained for a short time window (for example, 1 minute). Based on the current pressure state and the balancing flow to be executed, the pressure P pred and the storage state SoC pred of the two gas storages at the end of the time window are predicted using the gas state equation. For example, for the outflow side gas storage, the predicted pressure will decrease; for the inflow side, the predicted pressure will rise. The pressure change rate dP pred / dt resulting therefrom is also calculated.

[0114] It is determined whether the deduction result meets the preset pressure safety threshold and the storage state change rate constraint. If not, the valve opening degree target in the gas storage balancing control instruction data is recalled and corrected, and the corrected gas storage balancing control instruction data is generated.

[0115] In the present embodiment, the system compares the deduction result with the safety boundary. Exemplarily, the verification conditions include: the predicted pressure P pred must be located in the interval [P min , P max ]; the predicted pressure change rate dP pred / dt must not exceed the allowed maximum rate dP max_rate ; and the growth rate of SoC must not cause the inflow side gas storage to directly enter the overcharged state in a very short time. If any condition is not met, the system determines that the initial instruction is risky. The recall and correction strategy is executed, for example, the target opening degree open target is multiplied by a decay coefficient, such as 0.5, or the opening degree is directly limited to the maximum boundary value that can meet the safety constraint. The corrected opening degree instruction is re-verified for safety until all constraint conditions are met. The finally determined safe opening degree value constitutes the gas storage balancing control instruction data and is sent to the field execution mechanism, thereby realizing the dynamic balance of the system energy storage value on the premise of ensuring safety.

[0116] According to an aspect of the present application, the marginal energy storage price data of the gas storage is generated, and the data can also be read as follows: static characteristic model data of the gas storage, containing model parameters such as volume, upper and lower limits of design pressure P min_j , P max_j , minimum effective energy storage E min_j , maximum energy storage E max_j of each gas storage. Read control parameter data, including parameters a j , b j and possible price normalization parameters for constructing the marginal energy storage price function. Read real-time gas storage state data, which at least contains the current pressure P res_j , current temperature T j and corresponding gas storage number of each gas storage. Index alignment is performed on the static characteristic model data of the gas storage, the control parameter data and the real-time gas storage state data to construct the gas storage basic input data for each gas storage j. Read the gas storage basic input data, and for each gas storage j, extract its pressure P res_j , temperature T j , volume parameter and the like. According to the thermodynamic state equation or lookup table model established in advance in the static characteristic model data of the gas storage, an energy calculation function is called to calculate the equivalent energy E j of each gas storage. The calculation can be represented in the form of the following linear formula: E j =f energy (P res_j , T j , V j , other model parameters); wherein V j is the volume parameter of the gas storage j, E j represents the equivalent energy of the gas storage j that can be used for subsequent gas release and power generation under the current state, and f energy represents the energy calculation function determined through modeling. The calculation results of E j of all gas storages are summarized according to the gas storage number to form the equivalent energy storage data of the gas storage, which is a scalar array indexed by the gas storage number. Read the equivalent energy storage data of the gas storage to obtain E j of each gas storage j. E min_j and E max_j of the corresponding gas storage are extracted from the gas storage basic input data for determining the available energy storage range thereof. For each gas storage j, its storage state SoC j is calculated according to the following linear formula: SoC j =(E j - E min_j ) / (E max_j - E min_j ); wherein, SoC jThe value range is typically between 0 and 1, used to represent the current level of gas storage capacity. When E j Approaching E max_j Time SoC j When E approaches 1 j Approaching E min_j Time SoC j Approaching 0. For the calculated SoC j Perform boundary processing when the SoC j When the value is less than 0, it is truncated to 0. When the SoC j Data greater than 1 is truncated to 1 to ensure the physical validity of the data. All gas storage SoCs are processed. j The results are summarized by gas storage unit number to generate gas storage unit storage status data, which is an array of SoC values ​​indexed by the gas storage unit number.

[0117] From the basic input data of the gas storage facilities, read the identifier and E of each gas storage facility. min_j E max_j Design pressure range and the basic parameter 'a' for the price function. j b j Initial value or recommended range. Extract global control terms related to the price function from the control parameter data, including a list of optional price function types and the price ceiling L. max_global Price floor L min_global And weight parameters applicable to different operating strategies, etc. For each gas storage j, the local parameter range from the gas storage's basic input data is merged with the global constraints from the control parameter data to form the price function configuration record for that gas storage. For example, the fields include: function type. j (e.g., convex function type, piecewise linear type), parameter upper and lower bound intervals, global price saturation value, etc. All gas storage configuration records are aggregated by gas storage number to generate structured price function configuration data. The current SoC of each gas storage is read from the gas storage storage status data. j Read the price function configuration data; for each gas storage unit j, obtain its configured price function type (type). j And the parameter range. Based on type j Depending on the type, a corresponding price function expression is constructed for each gas storage facility. For example: if type j If it is a convex function, then use a form of L. j (SoC j ) = a j / (1 - SoC j + ε) + b j * SoC j The expression; if type j For piecewise linear types, use an L-shaped form.j (SoC j ) =k1 j * SoC j + c1 j (SoC) j (low range) or L j (SoC j ) = k2 j * SoC j + c2 j (SoC) j Expressions such as (high interval), where k1 j c1 represents the slope of the lower interval. j k2 is the intercept of the lower interval. j c2 is the slope of the high interval. j This is the intercept of the high interval. When constructing the expression, based on the parameter range in the price function configuration data, a... j b j k1 j k2 j c1 j c2 j Initially select or interpolate equal coefficients to satisfy: when SoC j When L approaches 1 j Significantly increased, when SoC j At lower L j Smaller and depends on the SoC j Monotonically increasing. The price function type, specific parameter values, and current SoC of each gas storage j are recorded. j They are packaged together to form instance data of the gas storage price function.

[0118] Read the instance data of the gas storage price function. For each gas storage j, obtain its price function expression L. j (SoC j ) and current SoC j For each gas storage cell j, the SoC will be... j Substitute L j (SoC j In the calculation, the uncorrected original price value L is obtained. j_raw Handling gas storage facilities in extreme conditions: When SoC j Very close to 1 (e.g., greater than a certain threshold SoC) high_thr When calculating L, to prevent the denominator from approaching zero in the calculation formula, j_raw Numerical anomalies, L j_raw Limit to the upper limit L set based on control parameter data. max_global Or dynamically adjusted price ceiling L for gas storage facility j max_j Within; when SoC jVery close to 0 (e.g. less than a certain threshold SoC low_thr ) to avoid infinite price leading to over-concentration of subsequent allocation on this reservoir, limit L j_raw to be above the lower bound of gas reservoir j's price L min_global or the dynamically adjusted lower bound of gas reservoir j's price L min_j . Processed L j_raw and corresponding SoC j and gas reservoir number are one-to-one corresponding, forming the original gas reservoir marginal energy storage price data. Read L j_raw of each gas reservoir from the original gas reservoir marginal energy storage price data. From the control parameter data, read the configurations related to the operation strategy, such as: whether to prefer the life balance priority strategy; whether to prefer the efficiency priority strategy; whether to use aggressive or conservative charging strategy under the current electricity price or system state, etc. For each operation strategy, construct the corresponding price variant based on L j_raw : when life balance priority is preferred, L j of gas reservoirs with higher SoC j_raw can be further amplified, obtaining the price variant L j_life_variant under the life balance priority strategy; when efficiency priority is preferred, L j_raw of gas reservoirs in the lower efficiency segment can be slightly amplified or L j of gas reservoirs with lower SoC j_raw can be slightly compressed, obtaining the price variant L j_eff_variant under the efficiency priority strategy; other strategies can be linearly combined or segmented adjusted based on the above two variants, obtaining the price variant L j_strategy_variant under other operation strategies. Encapsulate the original price L j_raw of each gas reservoir j and the price values of the above strategy variants according to a unified structure to form the candidate gas reservoir marginal energy storage price data, wherein each record includes the gas reservoir number and the corresponding multiple candidate price fields.

[0119] Read the candidate gas reservoir marginal energy storage price data to obtain the original price and the price of each strategy variant of each gas reservoir. From the control parameter data and the real-time system state data, read the actual enabled operation strategy identifier, such as marking the current stage of energy storage period, power generation period, maintenance period, etc. According to the operation strategy identifier, select the appropriate price version for each gas reservoir j: in the energy storage period and prefer life balance, prefer L j_life_variant ; in the energy storage period and prefer efficiency, prefer L j_eff_variant ; in other scenarios without special preference, L j_raw or the compromise version can be directly selected. Perform uniform upper and lower limit check on the selection result, so that the selected price value is still within the global allowed interval [L min_global , L max_globalThe final selected price value of each gas reservoir is summarized according to the gas reservoir number to generate the gas reservoir marginal energy storage price data available to the outside. The gas reservoir marginal energy storage price data is read to obtain the L j value set of all gas reservoirs. The global statistics of all L j , such as the minimum value L min , the maximum value L max , the average value L avg , etc., are calculated for subsequent normalization and outlier analysis. Based on L min and L max , each L j is normalized to obtain the normalized price value L norm_j , which can adopt a linear normalization form: L norm_j = (L j - L min ) / (L max - L min + ε norm ); wherein the theoretical value range of L norm_j is between 0 and 1, and ε norm is a small amount set to prevent the denominator from being zero. Normalization is beneficial to include price variables and other dimensionally different indicators such as efficiency and surge margin into the same equivalent constraint domain and comprehensive margin calculation. The normalized L norm_j is smoothed, and if necessary, the L norm_j of adjacent control periods can be smoothed by exponential smoothing or moving average method to obtain more stable smoothed marginal price data to reduce the impact of short-term sharp fluctuations on control allocation. The original L j , the normalized L norm_j and the smoothed L norm_j are stored one by one to form structured normalized gas reservoir marginal energy storage price data. The normalized gas reservoir marginal energy storage price data and the gas reservoir storage state data are checked for consistency, such as verifying whether the L j of the gas reservoir with a larger SoC norm_j is generally higher than that of the gas reservoir with a smaller SoC j . If abnormal conditions contrary to the monotonic trend are found, they can be marked for subsequent diagnosis or slight correction.

[0120] According to another aspect of the present application, obtaining comprehensive equivalent margin index data can also be: reading the efficiency function η i (Q i , P out_i ) and the surge margin function SM i (Q i , P out_i of each compressor under different flow and outlet pressure conditions from the unit static characteristic model data.), and the corresponding rated efficiency and rated surge margin. From the control parameter data, read the efficiency lower bound coefficient a, the minimum surge margin threshold SM min , the upper pressure limit P max_j , the lower pressure limit P min_j , the upper pressure rate of change dP max_j , and other control thresholds. From the system real-time state data, extract the actual operating point of each compressor at the current time and the real-time pressure, flow rate, and temperature data of each gas storage. From the real-time unit state data and the real-time gas storage state data, combined with the field valve connection topology, generate unit-gas storage connection state data, which identifies which gas storages are in a gas injection connection state with each unit. From the normalized gas storage marginal energy storage price data, read the normalized price L norm_j of each gas storage. Correlate the above data according to the unit number and the gas storage number in multiple dimensions to construct equivalent constraint basic input data for each unit-gas storage combination (i, j), where each combination data contains the static characteristic model of the unit, the current operating point, the control parameters, the real-time state of the corresponding gas storage, and the normalized marginal price L norm_j of the gas storage. Read the equivalent constraint basic input data, and for each unit-gas storage combination (i, j), extract the pressure P res_j of the current gas storage j, the current total gas injection flow rate, and the possible gas release flow rate. From the unit-gas storage connection state data, count the current gas injection flow rate of all units connected to gas storage j, superimpose these flow rates to obtain the current total gas injection flow rate Q in_j . If the gas storage has a current gas release or discharge path, read the corresponding gas release flow rate from the system real-time state data to obtain the current total gas release flow rate Q out_j . Using a pressure dynamic approximation model, estimate the change in gas storage pressure within a short prediction time window in the future, for example, using the following approximation form: dP res_j / dt ≈ (Q in_j - Q out_j ) / C j ; where C j is the equivalent volume-compressibility parameter of gas storage j, which is given in the gas storage static characteristic model data; using the above slope estimation, the change in gas storage pressure can be estimated within a time window length At. Calculate the upper and lower boundary estimates of the pressure at the end of the prediction time window: P res_j_pred_max = P res_j + dP res_j / dt * At max ; P res_j_pred_min = P res_j + dP res_j / dt * At min ; where At maxand Δt min This represents the defined prediction time range boundary. For each gas storage facility j, according to P... min_j P max_j dP max_j Based on the safety threshold, calculate the acceptable pressure safety space within the prediction time window. For example, the pressure safety margin of gas storage j within the prediction time window. margin_P_j = min(P max_j - P res_j_pred_max P res_j_pred_min - P min_j ); Safety margin of pressure change rate in gas storage j margin_dP_j = dP max_j - |dP res_j / dt|. These safety margin results are compiled into dynamic safety margin data for gas storage pressure.

[0121] From the equivalent constraint-based input data, read the efficiency function η of each unit i. i (Q i P out_i Rated efficiency η ref_i And the efficiency lower limit coefficient α. Using α and η ref_i Lower limit of computational efficiency η min_i For example, η min_i = α * η ref_i In the two-dimensional space of flow rate and outlet pressure for each unit i, the injection flow rate Q is adjusted according to the preset discrete step length. i and export pressure P out_i Grid sampling is performed to generate a set of candidate operating conditions for the unit. For each candidate operating condition, Q is... i and P out_i Substitute into the efficiency function η i (Q i P out_i ) to obtain the corresponding efficiency η i_current And determine η i_current Is it greater than or equal to η? min_i Record all operating point sets that satisfy the efficiency constraints. For each unit-gas storage combination (i, j), obtain the operating point set filtered only by efficiency constraints, and summarize them to form the efficiency constraint subdomain data. From the equivalent constraint basic input data, read the surge margin function SM for each unit i. i (Q i P out_i and minimum surge margin threshold SM min Read the efficiency constraint subdomain data, and for each unit-gas storage combination (i, j), obtain the set of candidate operating conditions that have passed the efficiency screening. For each candidate operating condition, Q...i and P out_i Substitute SM i (Q i , P out_i ), calculate current surge margin SM i_current , and determine whether SM i_current is greater than or equal to SM min . The working condition points that meet both the efficiency constraint and the surge margin constraint are retained to form a surge safety sub-domain record for each (i, j) combination. The results of all (i, j) combinations are summarized in a unified structure to form surge constraint sub-domain data, which describes the flow-outlet pressure area allowed under the dual efficiency and surge conditions.

[0122] From the gas storage pressure dynamic safety margin data, read safety margin_P_j and safety margin_dP_j for each gas storage j, and the current estimated pressure change rate dP res_j_dt_est . From the efficiency constraint sub-domain data and the surge constraint sub-domain data, combine the current working point and the candidate working condition point set for each unit-gas storage combination (i, j). Using the simplified pressure dynamic model of the gas storage, superimpose the unit gas injection flow rate Q i corresponding to each candidate working condition point on the current system total gas injection flow rate to estimate the pressure change rate dP res_j_dt_candidate of the gas storage j under that working condition. For each candidate working condition point, determine whether the following conditions are met simultaneously: after the pressure change within the prediction time window, it is still located within the interval [P min_j , P max_j ]; the absolute value of dP res_j_dt_candidate does not exceed the upper limit of the allowed pressure change rate dP max_j ; safety margin_P_j and safety margin_dP_j are still non-negative after using that working condition point. The working condition points that meet the above conditions are recorded as part of the pressure dynamic safety sub-domain for each unit-gas storage combination (i, j) to form pressure dynamic constraint sub-domain data. From the normalized gas storage marginal energy storage price data, read L norm_j for each gas storage. Read the surge constraint sub-domain data and the pressure dynamic constraint sub-domain data to obtain the working condition point set for each unit-gas storage combination (i, j) that has already met the efficiency and safety conditions. For each gas storage j, according to the numerical value of L norm_j , calculate the area adjustment factor factor j , for example: when L norm_j is close to 1, factor j is close to a scaling coefficient less than 1, which is used to shrink its available area; when L norm_j is close to 0, factorj Close to or slightly greater than 1, used to expand the range of optional operating conditions within its available area. For each combination (i, j), according to factor j Screen and adjust its candidate operating point set, for example: for L norm_j For larger gas storage, delete operating points with large flow rates, or limit operating points to the low injection area; for L norm_j For smaller gas storage, appropriately retain more high-flow operating points. The factor j The adjusted operating point set is recorded to form the value-adjusted constraint sub-domain data, which reflects the feasible operating condition area after superimposing the energy storage value on the basis of efficiency, surge, and safety.

[0123] Read the efficiency constraint sub-domain data, surge constraint sub-domain data, pressure dynamic constraint sub-domain data, and value-adjusted constraint sub-domain data. For each unit-gas storage combination (i, j), perform set intersection operations on the above four sub-domains in the flow-outlet pressure space, and only retain operating points that belong to all four sub-domains. If a combination (i, j) is empty after intersection, record that there is no available operating condition for the combination in the current control period. For combinations (i, j) with non-empty intersection, encapsulate their operating point set and corresponding operating boundary characteristics (such as maximum allowable flow, available outlet pressure range) into structured records. Aggregate the intersection results of all (i, j) combinations to form the final unit-gas storage equivalent constraint domain data.

[0124] From the unit-gas storage equivalent constraint domain data, read the current operating point and equivalent constraint domain boundary information for each combination (i, j). From the equivalent constraint base input data, read the efficiency function, surge margin function, and control threshold value of the corresponding unit, as well as the upper and lower limits of the gas storage pressure. From the gas storage pressure dynamic safety margin data, obtain safety margin_P_j and safety margin_dP_j . From the normalized gas storage marginal energy storage price data, read L norm_j of gas storage j. For each combination (i, j), perform the following calculations: Substitute Q i , P out_i of the current operating condition into the efficiency function to obtain η i_current , and calculate the efficiency dimension margin margin η_i_j based on η min_i , η ref_i ; Substitute the current operating point into the SM i function to obtain SM i_current , and calculate the surge dimension margin margin SM_i_j based on SM min , the reference value; use safety margin_P_jand the current pressure of gas storage j, calculate the pressure margin margin P_i_j ; use safety margin_dP_j and the current dP res_j_dt_est , calculate the pressure rate margin margin dP_i_j ; use L norm_j , calculate the value dimension margin margin value_i_j = 1 - L norm_j . The above five local margin indicators are one-to-one corresponding to the index of combination (i, j) to form the dimension margin indicator data. Read the dimension margin indicator data to obtain margin η_i_j , margin SM_i_j , margin P_i_j , margin dP_i_j , margin value_i_j of all combinations (i, j). From the control parameter data, read the weight coefficients set for different dimensions, such as w η , w SM , w P , w dP , w value , which are used to reflect the importance difference of each dimension in different running stages. Calculate the global minimum and maximum value of each type of dimension margin respectively, and map it to the interval of 0 to 1 through linear scaling or other normalization methods to obtain the normalized margin margin η_norm_i_j , etc. Multiply each dimension margin after normalization by the corresponding weight coefficient to obtain the weighted margin, such as margin η_weighted_i_j =w η * margin η_norm_i_j , etc. Summarize the weighted margin of each dimension with the combination (i, j) index to form the weighted dimension margin indicator data, which contains the weighted margin value of each combination in each constraint dimension.

[0125] Read the weighted dimension margin indicator data to obtain the weighted margin value of each combination (i, j) in the efficiency, surge, pressure, pressure rate and value dimensions. For each combination (i, j), according to the idea of the weakest link determining the overall margin, the following aggregation rule is used to calculate the comprehensive equivalent margin: the minimum value of each dimension weighted margin can be directly taken, such as M ij =min(margin η_weighted_i_j , margin SM_weighted_i_j , margin P_weighted_i_j , margin dP_weighted_i_j , margin value_weighted_i_j ); wherein margin η_weighted_i_j is the weighted margin of the combination of unit i and gas storage j in the efficiency dimension, marginSM_weighted_i_j For the combination of unit i and gas storage j, margin is the weighted margin in the surge margin dimension. P_weighted_i_j For the weighted margin of unit i and gas storage j in the pressure dimension, margin dP_weighted_i_j The margin is the weighted margin of the combination of unit i and gas storage j in terms of the rate of pressure change. value_weighted_i_j This represents the weighted margin in the value dimension for the combination of unit i and gas storage j. Alternatively, a slightly weighted minimum or other non-linear combination form can be introduced based on the above to preserve priority protection for certain key constraints. The overall margin M for each combination (i, j) is... ij The index is stored accordingly to form the comprehensive equivalent margin index data. The comprehensive equivalent margin index data is read to obtain the M for all combinations (i, j). ij Value. Calculate all M values ​​within the current control cycle. ij Minimum value M min and maximum value M max Based on M min and M max For each M ij Normalization is performed to obtain the normalization margin value M. norm_ij M norm_ij = (M ij - M min ) / (M max - M min +ε margin ), where ε margin This is a numerical stability correction term; after normalization, M norm_ij The theoretical range is 0 to 1. For M... norm_ij Smoothing can be performed along the time dimension, for example, using first-order exponential smoothing: M smooth_ij (t) = γ * M norm_ij (t) + (1- γ) * M smooth_ij (t - 1); where γ is the smoothing coefficient, t represents the current control cycle index, and M smooth_ij This represents the smoothed margin value. Smoothing reduces drastic margin fluctuations caused by measurement noise and short-term disturbances. The original M... ij Normalized M norm_ij And the smoothed M smooth_ij The data is stored in a one-to-one correspondence, forming structured, normalized, comprehensive equivalent margin data, where each record contains a combined (i, j) index and three versions of the margin metric.

[0126] In an embodiment of the present application, a multi-set compressor unit coordinated gas injection system is also provided. Specifically, the compression side of the large-scale compressed air energy storage system adopts a multi-set parallel compressor unit system to store air in multiple air storage reservoirs. The parallel compressor units are divided into fixed unit columns and flexible unit columns. The fixed unit columns are designated to store air in a specific air storage reservoir and cannot store air in other air storage reservoirs. The fixed unit columns can be configured with a minimum of one column and a maximum number related to the design volume of a single air storage reservoir, the design duration of energy storage, and the rated power of a single column compressor unit. The flexible unit columns can inject air into all air storage reservoirs through the shunt valve and have the ability to inject air into a single air storage reservoir or multiple air storage reservoirs simultaneously. To ensure safe and stable operation of the system, there can be only one flexible column. To ensure the balance of air injection flow from the flexible column to all air storage reservoirs, the equipment and pipeline layout need to ensure a symmetrical structure design, so that the pipeline resistance of the flexible column compressor unit to all air storage reservoirs is consistent. Each column of compressor units uses a 3-stage / 4-stage compressor in series, and the last stage compressor uses frequency conversion + guide vane regulation. Each stage of compressor is equipped with an anti-surge valve and a backflow bypass. A mixing box is configured before the air storage reservoir to form a parallel structure for multiple compressor units, ensuring that the exhaust pressure of each column of compressor units is consistent. At the same time, each column of compressor units is equipped with a check valve before the mixing box to prevent backflow. An independent flow meter is configured before the box to monitor the air flow output of each column of compressor units. A pressure detection meter is configured on the air storage reservoir main pipe to detect the background pressure of the air storage reservoir. A bypass pipe is configured between the air storage reservoirs, and a pressure equalizing valve is configured on the bypass pipe,

[0127] To achieve capacity matching while minimizing the coupling interference of parallel compressors, a multi-set parallel compressor unit topology is proposed to achieve capacity matching among the compression subsystem, air storage reservoir, and turbine subsystem. Specifically, the first coupling of the multi-set parallel compressor unit appears in the flexible column. When the flexible column determines to inject air into multiple air storage reservoirs, all shunt valves are opened after the fixed column and the flexible column are coupled as a system whole. The pressure vessel has the characteristic of continuously rising back pressure during the air injection process, and flow fluctuations will indirectly cause pressure fluctuations. This poses a challenge to the stable output of all compressor units. The second coupling appears in the fixed unit column, where multiple sets of parallel compressors store air in the same air storage reservoir. The multiple sets of parallel compressors in the fixed unit column operate in parallel, and according to the parallel characteristics, the output pressure of each set is equal. However, the operating characteristics of each set of compressors are not completely consistent, resulting in unequal output flow. Secondly, each set of compressors is connected in series with multiple stages of compressors, and the coupling operation of series and parallel makes it difficult for the single-set CCS control system to cope with this complex scenario.

[0128] Further, in order to ensure that the output flow of each set is balanced and that each compressor is operated efficiently and stably, a compressor system energy storage control process and an equivalent domain allocation control strategy are proposed. The compressor system energy storage control process makes each gas storage become a whole through pressure balancing bypass, so that there is no need to set a valve. The compressor unit equivalent domain allocation control strategy is based on the characteristics of parallel unit pressure equality and flow superposition, starts from regulating unit output flow, substitutes into the unit efficiency-flow mapping relationship, realizes stable output of flow, and ensures unit operation efficiency. The fixed unit dynamic flow allocation module calculates the flow and efficiency gain through input flow and unit efficiency data, feeds back to the CCS system of each unit, and realizes stable output of flow.

[0129] In summary, the control method of the multi-set compressor unit coordinated gas injection system acquires real-time state and static characteristic model of the system, calculates equivalent energy storage of the gas storage by using thermodynamic equation and a preset convex function model, and generates gas storage marginal energy storage price data with nonlinear penalty characteristics. On this basis, the unit and gas storage combination multi-dimensional equivalent constraint domain data are constructed by combining unit efficiency, surge margin, gas storage pressure dynamics and marginal price constraints, and comprehensive equivalent margin index data representing the minimum safety distance from the multiple constraint boundaries are calculated. The gas injection load is iteratively allocated based on the margin index, and the marginal price difference is used to drive the gas storage balancing.

[0130] The application adopts multi-dimensional equivalent constraint domain construction technology, and fuses the originally independent efficiency, surge physical limit and abstract marginal energy storage price into a unified feasible working condition domain through geometric intersection and value-driven deformation. The control instruction falls within the common safety zone of all constraints, solving the problem of losing one thing to gain another caused by independent action of each constraint in the traditional method. The marginal energy storage price model based on the convex function is used to describe the mechanism that the price increases nonlinearly and sharply with the increase of SoC, and the price difference instead of the pressure difference is used to drive balancing and allocation. A soft value wall is constructed to automatically inhibit gas injection to high-risk gas storage, solving the problem of overfilling risk and loss of flexibility caused by traditional pressure balancing. The prediction and comprehensive equivalent margin index in the time dimension are introduced, the future pressure trajectory is corrected by predicting the future pressure trajectory, and the minimum value of each dimension margin is taken as the allocation weight, so that the load is always allocated to the unit with the longest current bottleneck and the largest comprehensive safety margin, realizing global optimization at the system level.

[0131] The above describes the preferred embodiments of the application in detail, but the application is not limited to the specific details in the above embodiments, and various equivalent transformations of the technical solutions of the application can be made within the technical concept of the application, and these equivalent transformations all belong to the protection scope of the application.

Claims

1. A control method for a plurality of compressor trains in cooperation with a gas injection system, characterized by, The method comprises the following steps: collecting static characteristic model data of the unit group and the gas storage, real-time state data of the system containing operation parameters of the unit and the gas storage, and gas injection demand data of the system; calculating equivalent energy storage data and storage state data of the gas storage by using the static characteristic model data of the gas storage and the real-time state data of the system, and generating marginal energy storage price data of the gas storage based on a preset pricing function; constructing equivalent constraint domain data of the unit and the gas storage for each connection combination of the unit and the gas storage based on the static characteristic model data of the unit, the real-time state data of the system and the marginal energy storage price data of the gas storage; calculating the relative position of the real-time state data of the system in the equivalent constraint domain data of the unit and the gas storage to obtain comprehensive equivalent margin index data representing the comprehensive safety degree of the distance to the multi-dimensional constraint boundary; cooperatively calculating the unit gas injection allocation and the gas storage balancing flow according to the system gas injection demand data, the comprehensive equivalent margin index data and the marginal energy storage price data of the gas storage to generate system control execution data; constructing the equivalent constraint domain data of the unit and the gas storage, including generating physical constraint sub-domain data, specifically: extracting efficiency function and surge margin function from the static characteristic model data of the unit, and then dividing the region meeting the preset efficiency lower limit and the region meeting the preset minimum surge margin in the flow-outlet pressure plane; calculating the intersection of the region meeting the preset efficiency lower limit and the region meeting the preset minimum surge margin to generate physical constraint sub-domain data representing the physical safety boundary; calculating the equivalent energy storage data and the storage state data of the gas storage, including: using the thermodynamic state equation to convert the pressure, temperature in the real-time state data of the system and the gas storage volume parameter in the static characteristic model data of the gas storage into equivalent energy storage data of the gas storage representing the effective work potential; obtaining the minimum effective energy storage and the maximum energy storage from the static characteristic model data of the gas storage, and performing normalization calculation by using the equivalent energy storage data of the gas storage to generate storage state data representing the filling degree of the gas storage; generating the marginal energy storage price data of the gas storage, including: constructing or calling a marginal pricing model in the form of a preset convex function, the marginal pricing model being configured to have a nonlinear growth characteristic with an increasing slope of the function value as the value of the independent variable increases; putting the storage state data of the gas storage into the marginal pricing model as an independent variable to calculate the marginal energy storage price data of the gas storage.

2. The method of claim 1, wherein, constructing the equivalent constraint domain data of the unit and the gas storage, further comprising: performing normalization processing on the marginal energy storage price data of the gas storage to obtain a normalized marginal price, and calculating a region adjustment coefficient for representing the scaling degree of the working condition region based on the normalized marginal price; performing geometric contraction or expansion deformation on the working condition range covered by the physical constraint sub-domain data by using the region adjustment coefficient to generate the equivalent constraint domain data of the unit and the gas storage integrating the energy storage value constraint.

3. The method of claim 2, wherein, constructing the equivalent constraint domain data of the unit and the gas storage, further comprising: calculating the current pressure change rate of the gas storage by using the real-time state data of the system and the static characteristic model data of the gas storage, and predicting the pressure evolution trajectory within a preset time window; Based on the pressure evolution trajectory and the preset pressure safety threshold, generate the gas storage pressure dynamic safety margin data, and map it as the dynamic limit boundary of the gas injection flow and the outlet pressure, and correct the unit gas storage equivalent constraint domain data.

4. The method of claim 1, wherein, Get comprehensive equivalent margin index data, including: Calculate the distance of the current operating point relative to the efficiency boundary, surge boundary and pressure dynamic boundary in the unit gas storage equivalent constraint domain data respectively, and combine the gas storage marginal storage price data to generate local margin in each dimension; Normalize each dimension local margin, and extract the minimum value of all normalized local margins in each dimension as the comprehensive equivalent margin index data.

5. The method of claim 1, wherein, Collaborative calculation of unit gas injection allocation, including: Normalize the comprehensive equivalent margin index data to build the initial allocation weight, and decompose the total gas injection target in the system gas injection demand data into the initial allocation of each unit and gas storage combination according to the initial allocation weight; Map the initial allocation to the unit gas storage equivalent constraint domain data for boundary checking, and when the initial allocation exceeds the constraint domain boundary, truncate the excess part and calculate the remaining unallocated demand; Based on the comprehensive equivalent margin index data of the remaining unit and gas storage combination, iteratively redistribute the remaining unallocated demand until all allocation falls within the corresponding unit gas storage equivalent constraint domain data range, and generate combination gas injection allocation result data containing unit gas injection allocation.

6. The method of claim 1, wherein, Collaborative calculation of gas storage matching flow, including: According to the system real-time state data, identify the physical connection relationship between the gas storages, and extract the price values of the connected pair in the gas storage marginal storage price data; Based on the price value, calculate the marginal price difference of the connected pair, and map the marginal price difference to the target matching flow and the corresponding valve opening target to generate gas storage matching control instruction data containing gas storage matching flow.

7. The method of claim 6, wherein, Collaborative calculation of gas storage matching flow, further including: Using system real-time state data and gas storage static characteristic model data, simulate and deduce the pressure change and storage state change of the gas storage after executing the gas storage matching control instruction data; Determine whether the deduction result meets the preset pressure safety threshold and storage state change rate constraint, if not, adjust and correct the valve opening target in the gas storage matching control instruction data, and generate the corrected gas storage matching control instruction data.

Images