Air compressor group real-time load distribution method based on equal power change rate

Abstract

The present application relates to the technical field of industrial automation control, and particularly relates to a real-time load distribution method for air compressor groups based on equal power change rate, comprising: obtaining multi-source data, checking and fault-tolerant processing the obtained multi-source data to obtain input data; constructing a nonlinear programming optimization model and solving it through a numerical solution algorithm to obtain an initial load distribution scheme; performing unit safety boundary constraint checking on the initial load distribution scheme, processing over-limit load distribution until all units meet the safety constraints to obtain an optimal load distribution scheme; performing load fluctuation collaborative smoothing processing on the optimal load distribution scheme, performing constraint through a preset maximum load change rate threshold of the unit, performing difference calibration on over-limit load distribution, and combining with the load fluctuation trend to perform dynamic adaptive adjustment to obtain a final load distribution scheme. The present application optimizes real-time load distribution of air compressor groups through equal power change rate theory, and improves system energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, and in particular to a real-time load distribution method for air compressor groups based on equal power change rate. Background Technology

[0002] In industrial production, air compressor stations, as core energy supply units, require multiple air compressors operating in parallel to meet real-time air load demands. The rationality of load allocation directly affects system energy consumption, operational stability, and equipment lifespan. Current mainstream load allocation strategies mainly include two types: load sharing and primary / standby peak shaving. The former distributes the total load evenly across all operating units, without considering individual unit efficiency differences and performance drift (such as changes in energy consumption characteristics due to manufacturing deviations, operational wear, and different maintenance conditions), resulting in higher total system power consumption. The latter fixes some units to full load operation, relying solely on peak shaving by a single unit, resulting in insufficient real-time response to load fluctuations and a tendency for excessive wear and tear on individual units.

[0003] Meanwhile, existing solutions lack sophisticated mechanisms for controlling safety boundaries such as compressor surge and blockage, which can easily lead to equipment failures. They also fail to adequately adapt to the differences in physical characteristics between variable frequency and fixed frequency units, resulting in delayed response or overload adjustments during coordinated operation. In maintenance mode, the contradiction between unit load limits and sudden increases in real-time load is not effectively resolved, which may cause power supply gaps. In addition, traditional solutions have weak fault tolerance for data anomalies and transmission interruptions, and their model parameter update mechanisms are rigid, making it difficult to adapt to the complex operating conditions in industrial scenarios.

[0004] Therefore, there is an urgent need for a method for allocating air compressor loads that takes into account energy efficiency, safety, real-time performance, and adaptability, in order to solve the core pain points of existing strategies, such as low efficiency, poor stability, and insufficient adaptability. Summary of the Invention

[0005] This invention optimizes the real-time load distribution of air compressor groups through the theory of equal power change rate and multi-scenario adaptation mechanism, thereby improving system energy efficiency, operational stability and real-time power supply.

[0006] The technical solution proposed in this invention is: a real-time load distribution method for air compressor groups based on equal power change rate, the method comprising: Acquire multi-source data, perform verification and fault tolerance processing on the acquired multi-source data, and obtain input data; Based on the theory of equal power change rate, with the goal of minimizing the total power consumption of operating units, a nonlinear programming optimization model is constructed by combining the quadratic power consumption model in the input data and the set of operating units, and the initial load allocation scheme is obtained by solving it through a numerical solution algorithm. The initial load allocation scheme is checked for unit safety boundary constraints, and overload allocation is handled until all units meet the safety constraints, thus obtaining the optimal load allocation scheme. The optimal load allocation scheme is processed by load fluctuation coordination and smoothing. The scheme is constrained by the preset maximum load change rate threshold of the unit, the excess load allocation is differentially calibrated, and the load fluctuation trend is dynamically adapted and adjusted to obtain the final load allocation scheme.

[0007] Preferably, the specific process for acquiring the multi-source data is as follows: Data is collected from the pipeline network and each air compressor equipment of the air compressor station to obtain collected data including total air production, equipment power consumption, and operating status. Based on the air compressor station's operating logic, the collected data is associated with devices and encapsulated into events. The encapsulated data is then uploaded to a pre-set central data platform through the standardized interface of the air compressor station's energy management system. Real-time data messages pushed by the central data platform are received via message subscription, and the format of the real-time data messages is parsed to obtain parsed data. The parsed data is correlated with the device and operating condition context to obtain multi-source data.

[0008] Preferably, the specific process for obtaining the input data is as follows: For core data from multiple sources, multi-source cross-verification of primary and backup sensors is adopted. A data deviation threshold is set. When the deviation between primary and backup data exceeds the data deviation threshold, the predicted value derived from the secondary power consumption model is used for replacement. If the power consumption model parameters of a single unit are missing or abnormal, extract the historical data of the same operating conditions of the unit in the most recent period, combine them with the parameter mapping relationship of the same type of unit, and generate a temporary power consumption model through transfer learning. When data transmission is interrupted, the optimal load allocation scheme of the previous cycle is cached and linearly fine-tuned based on the load change trend of the past 5 cycles until the data is restored and the normal data processing flow is switched. The data processed above is then subjected to integrity verification to remove invalid and redundant data, resulting in the input data.

[0009] Preferably, the specific process for obtaining the initial load allocation scheme is as follows: Based on the quadratic power consumption model in the input data, the power change rate function of each unit is derived. The power change rate function is the first derivative of the quadratic power consumption model with respect to the gas production. A nonlinear programming optimization model is constructed with the objective function of minimizing the total power consumption of the operating units and the constraints of equal power change rate of all operating units and total gas production meeting the real-time total load demand. Determine the set of equations corresponding to the nonlinear programming optimization model. The set of equations includes equations that ensure the equal rate of change of power of each unit and equations that balance the total gas production. The Newton-Raphson method is used to iteratively solve the system of equations. An iterative convergence threshold is set, and the solution is stopped when the iterative result meets the iterative convergence threshold, thus obtaining the initial load allocation scheme.

[0010] Preferably, the specific process for obtaining the optimal load allocation scheme is as follows: Determine the unit safety boundary for each air compressor. The unit safety boundary includes the minimum flow boundary for surge and the maximum flow boundary for blockage. The boundary values ​​are calibrated based on the equipment's factory parameters and historical operating data. Verify one by one whether the allocated load of each unit in the initial load allocation scheme is within its corresponding safety boundary range; If a unit's allocated load exceeds the safety boundary, the unit's load is fixed at the corresponding boundary value, and this boundary value is deducted from the real-time total load demand to obtain the remaining total load. Using the remaining total load as the new load demand, a new nonlinear programming optimization model is reconstructed and solved for the remaining operating units to obtain a new load allocation scheme. Repeat the above verification and solution steps until the allocated load of all units meets the safety boundary constraints, and obtain the optimal load allocation scheme.

[0011] Preferably, the specific process for obtaining the final load allocation scheme is as follows: Based on the equipment model, years of operation and performance parameters of the unit, a maximum load change rate threshold is preset for each unit, and the maximum load change rate threshold is dynamically adjusted according to the equipment operating status. Calculate the difference between the allocated load of each unit in the optimal load allocation scheme and the actual load of the previous cycle, and determine whether the difference exceeds the preset maximum load change rate threshold. If the difference exceeds the maximum load change rate threshold, the load allocated to the unit in this cycle will be calibrated to the sum of the actual load and the maximum load change rate threshold of the previous cycle, and the load difference after calibration will be calculated. The load difference is evenly distributed to other units, and the load distribution of other units is fine-tuned in accordance with the principle of equal power change rate. The periodic fluctuation trend of real-time load is statistically analyzed. When the real-time load fluctuates in the same direction for three or more consecutive periods, the maximum load change rate threshold is relaxed by a preset ratio. Based on the above constraints, calibration, and adaptation results, the final load allocation scheme is obtained.

[0012] Preferably, after obtaining the final load allocation scheme, the final load allocation scheme is sent to the local regulator of each air compressor, and the actual operating status and output parameters of each air compressor are collected in real time. The actual operating status and output parameters are compared with the preset parameters in the final load allocation scheme to determine whether there are any operating deviations exceeding the standard. If there are operational deviations exceeding the standard, the final load allocation scheme will be dynamically fine-tuned based on the degree of deviation, and a revised load allocation scheme will be generated and redistributed. The deviation events and correction process are recorded synchronously to form a deviation handling log.

[0013] Preferably, during the verification and fault-tolerant processing of multi-source data, the maintenance trigger status of each air compressor is monitored simultaneously. The maintenance trigger status includes manually issued maintenance instructions, operating parameters reaching preset maintenance thresholds, or cumulative running time meeting the requirements for regular maintenance. When any air compressor is detected to be in a maintenance-triggered state, the unit to be maintained is marked as a non-priority operating unit and its load allocation weight is reduced when constructing the nonlinear programming optimization model. After checking the safety boundary constraints of the initial load allocation scheme, the load allocation ratio of the units to be maintained is further verified to ensure that their load does not exceed the preset maintenance period load limit. During the load fluctuation smoothing phase, a stricter maximum change rate threshold is adopted for the load adjustment of the units to be maintained. At the same time, the load assignment strategy is optimized based on the marginal power consumption and safety margin of other operating units to ensure optimal total power consumption and stable operation. Once the maintenance unit has completed maintenance and resumed normal operation, its non-priority operation flag will be automatically removed.

[0014] The present invention also provides a computer-readable storage medium storing a computer program that is executed by a processor to implement the aforementioned method for real-time load allocation of air compressor groups based on equal power change rate.

[0015] The beneficial effects of this invention are: 1. By using the theory of equal power change rate, the marginal cost of each operating unit is balanced, allowing units with better energy efficiency to take on more load, significantly reducing the total power consumption of the air compressor station and significantly improving energy utilization efficiency; coupled with the online update mechanism of the model parameters using the 72-hour sliding window recursive least squares method, it can adapt to unit performance drift in real time (such as changes in energy consumption characteristics caused by operating wear and environmental changes), ensuring that the power consumption model always maintains high fitting accuracy; combined with the high-frequency iterative solution of the Newton-Raphson method (1-5 seconds / cycle), it can quickly respond to real-time load fluctuations, ensure dynamic matching between energy supply demand and allocation scheme, and avoid the energy waste and response lag problems of traditional equal distribution strategies.

[0016] 2. To address the differences in physical characteristics between heterogeneous units, differentiated allocation rules are designed to fully utilize the flexibility of variable frequency drive speed adjustment and avoid the problem of lag in IGV opening adjustment of fixed frequency drive. By setting dynamic load change rate thresholds according to equipment type and years of operation, combined with a threshold relaxation mechanism during continuous fluctuations and a gradual rollback mechanism after stabilization, the mechanical impact of sudden load changes on the units is effectively suppressed, reducing the risk of equipment failure. After marginal cost balancing and fine-tuning, the balance between optimal total power consumption and system stability during collaborative operation is ensured, adapting to the complex industrial scenarios of hybrid unit operation.

[0017] 3. By calibrating surge and blockage boundaries using factory parameters and historical data, and combining this with unit-by-unit safety verification and over-limit fixing mechanisms, operational risks are mitigated at the source, extending unit lifespan. In maintenance mode, load weight reduction, upper limit restrictions, and temporary adjustment strategies during sudden loads not only reserve safety margins for maintenance operations but also ensure continuous power supply through backup unit linkage, resolving the core contradiction between maintenance and power supply. Coupled with closed-loop management of graded deviation correction and log traceability, operational deviations can be dynamically corrected, providing data support for model parameter optimization and maintenance strategy adjustment, improving the industrial adaptability and long-term operational reliability of the solution. Attached Figure Description

[0018] Figure 1 A flowchart of a real-time load distribution method for an air compressor group based on equal power change rate; Figure 2 This is a flowchart illustrating the load allocation process of a real-time load allocation method for air compressor groups based on equal power change rate. Detailed Implementation

[0019] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0020] It is understood that the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0021] like Figure 1 and Figure 2 As shown, this invention discloses a real-time load allocation method for air compressor groups based on equal power change rate (equal marginal cost), the method comprising the following steps: The process involves acquiring multi-source data, performing verification and fault tolerance processing to obtain input data, and constructing a nonlinear programming optimization model based on the theory of equal power change rate, with the goal of minimizing the total power consumption of operating units. This model combines the secondary power consumption model in the input data with the set of operating units and is solved using a numerical algorithm to obtain the initial load allocation scheme. The initial load allocation scheme is then checked for unit safety boundary constraints, and overload allocation is handled until all units meet the safety constraints, resulting in the optimal load allocation scheme. The optimal load allocation scheme undergoes load fluctuation smoothing processing to obtain the final load allocation scheme. Finally, the final load allocation scheme is distributed to the local regulators of each air compressor and operating deviations are dynamically corrected, while simultaneously adapting to the unit maintenance mode, thus completing the real-time load allocation closed loop for the air compressor group.

[0022] Specifically, the algorithm is deployed on the field edge controller (such as an IPC or high-performance PLC) of the air compressor station. This edge controller sits between the cloud platform and the local regulator of the air compressor, achieving low-latency transmission through high-bandwidth data interaction to ensure high real-time performance. The algorithm is called cyclically at a high frequency (every 1 to 5 seconds) to quickly respond to the on-site gas load. The fluctuations provide support for the real-time nature of load allocation.

[0023] The specific details of obtaining multi-source data are as follows: Data is collected from the air compressor station pipeline network and each air compressor device to obtain collected data including total air production, equipment power consumption, and operating status. Based on the air compressor station's operating logic, the collected data is associated with devices and encapsulated into events to obtain encapsulated data. The encapsulated data is uploaded to a preset central data platform through the standardized interface of the air compressor station energy management system (Open Platform Communication Unified Architecture OPCUA interface). Real-time data messages pushed by the central data platform are received via message subscription (Message Queue Telemetry Transmission MQTT protocol), and the format of the real-time data messages is parsed. Device identifiers and data types are extracted from the header, and core information is extracted from the payload to obtain parsed data. The parsed data is associated with device and operating condition contexts to obtain multi-source data.

[0024] Specifically, comprehensive data collection is conducted on the air compressor station pipeline network and each air compressor unit. High-precision sensors are selected for the data collection to ensure the accuracy and stability of the data acquisition. The total air production is collected using an electromagnetic flowmeter with a measurement range of 0. 500 Accuracy ±0.8%FS, sampling frequency 10Hz, installed in the middle section of the main pipeline, horizontally with straight pipe sections before and after ≥10D and ≥5D (D is the inner diameter of the pipe) respectively, away from valves, elbows and other turbulent areas; power consumption of the equipment is collected by a power sensor, measurement range 0. 1000kW, accuracy ±0.5%FS, sampling frequency 10Hz, connected in series at the main circuit input terminal of the air compressor, ≥3 meters away from the frequency converter to avoid electromagnetic interference; operating status data is collected collaboratively by multiple types of sensors, including a motor temperature sensor (platinum resistance PT100, accuracy ±0.1℃, measurement range -50℃). 200℃), exhaust pressure sensor (accuracy ±0.01MPa, measurement range 0). 2.5MPa), motor speed sensor (Hall effect, accuracy ±1r / min), fault code acquisition module (compatible with GB / T2681-2017 standard fault codes), sampling frequency ≥5Hz, sensor installation must comply with industrial equipment installation specifications, and shockproof, dustproof and waterproof treatment must be done.

[0025] Based on the air compressor station's operating logic, the collected data is associated with devices and encapsulated as events: First, a unique device identifier is assigned to each device (format: "station number-device type-number", e.g., "S01-AC-005"). Data collected from the same device at the same timestamp (accurate to milliseconds) is then associated and bound. Next, events are encapsulated. The encapsulated data uses a JSON structured format, containing a 16-bit unique data gene tag (encoding rule: "module identifier-data type-timestamp-random sequence", where the module identifier is 2 bits (DC = data acquisition module), the data type is 2 bits (01 = device operating data, 02 = pipeline load data, 03 = environmental auxiliary data), the timestamp is 8 bits (format: YYYYMMDD), and the random sequence is 4 bits (hexadecimal characters)), the device unique identifier, the acquisition timestamp (format: YYYY-MM-DDHH:MM:SS.sss), the data type identifier, the payload (including the name, value, and unit of each acquired parameter), and a CRC32 checksum. This ensures the uniqueness, integrity, and traceability of the encapsulated data.

[0026] Encapsulated data is uploaded to a pre-defined central data platform via the standardized interface of the air compressor station energy management system (OPCUAV1.0 interface of the open platform communication unified architecture, with a publish-subscribe transmission mode). Interface communication parameters are set as follows: transmission timeout = 3 seconds, retransmission count = 3 times, and data compression uses the LZ77 algorithm (compression ratio ≥ 3:1) to reduce network bandwidth usage. During the upload process, a dual guarantee of encrypted channel and identity verification is employed. The encrypted channel is built based on the TLS 1.3 protocol, and identity verification is achieved through a unique device identifier and a dynamic token. The dynamic token is updated every 10 minutes to prevent theft or tampering during data transmission. After receiving the encapsulated data, the central data platform performs data storage, format standardization conversion, and redundant data removal to form a unified real-time data resource pool.

[0027] Real-time data messages pushed by the central data platform are received via message subscription (MQTTV 3.1.1 protocol, QoS level 2, ensuring messages are delivered only once): the edge controller subscribes to the corresponding topic in advance to ensure accurate reception of target data; after receiving the data, the real-time data message is immediately parsed. The parsing process is as follows: First, the CRC32 checksum is verified. If the verification fails, the message is discarded and logged; if the verification passes, the next step is performed. Second, the data gene tag, device identifier, data type, and collection timestamp are extracted from the message header. Third, the core information is extracted from the payload, string parameters are converted to numeric types, and the unit format is standardized (e.g., pressure is standardized to MPa, temperature to ℃, flow rate to MPa). The fourth step is to perform anomaly filtering, using 3... Outliers (i.e. values ​​exceeding the mean ± 3 standard deviations) are removed according to the criteria. Linear interpolation is used to fill in missing single-period data. Data missing for more than 3 periods is dynamically estimated using the Kalman filter algorithm. For data without a rated range, the threshold is determined according to the GB / T13926-2008 industrial standard to obtain the analytical data.

[0028] The parsed data is associated with equipment and operating condition contexts to obtain multi-source data. The association rules are timestamp alignment, equipment identifier matching, and operating condition dimension supplementation. Timestamp alignment adopts a ±300ms deviation fault tolerance mechanism to associate equipment operation data and pipeline load data within the same time window. Equipment identifier matching supports fuzzy matching (matching threshold ≥90%, compatible with case sensitivity of equipment numbers, redundant spaces, etc.). Operating condition dimension supplementation associates the associated equipment data with static data (such as equipment model parameters, production order information, gas usage priority, and ambient temperature and humidity data) stored in the central data platform, giving the parsed data complete operating condition background information. For example, the gas production data of a certain air compressor is associated with the gas usage demand of the corresponding production order and the current ambient temperature. Finally, multi-source data covering equipment operation data, pipeline load data, and operating condition auxiliary data is formed, providing a comprehensive and reliable data foundation for subsequent verification and optimization solutions.

[0029] Multi-source data contains three key dynamic inputs: Real-time total load demand The total gas production capacity that the air compressor station needs to meet at the current moment is obtained through real-time measurement and calculation using pipeline flow meters and pressure sensors. The latest power consumption model : Receive and retrieve the latest power consumption model parameters generated from the cloud platform ( The model expression is: ( For the first The operating power consumption of the air compressor. For the first The model parameters (air compressor output) are updated online using recursive least squares method and sliding window, with a fixed data window length of 72 hours. Parameter identification is performed every hour, and new data replaces old data according to the first-in-first-out principle. The model validation index is set as goodness of fit. ≥0.95, when When the value is less than 0.9, parameter retraining is triggered to ensure model accuracy; Operating unit set : Receives the set of units that should be running in the current time period from the start-stop combination optimization module on the cloud platform.

[0030] Specifically, the multi-source data is validated and fault-tolerant to obtain the input data, the details of which are as follows: For core data (real-time total load demand, unit power consumption parameters, and operating status data) from multiple sources, multi-source cross-verification using primary and backup sensors is employed. A data deviation threshold is set, and when the primary / backup data deviation exceeds the threshold, the predicted value derived from the secondary power consumption model is used for replacement. If the power consumption model parameters of a single unit are missing or abnormal, historical data of the same operating conditions for that unit in the most recent period is extracted. Combined with the parameter mapping relationship of units of the same model, a temporary power consumption model is generated through transfer learning. When data transmission is interrupted, the optimal load allocation scheme of the previous period is cached, and linear fine-tuning is performed based on the load change trend of the past 5 periods until the data is restored and the normal data processing flow is switched to. The data processed in the above way is subjected to integrity verification, and invalid and redundant data is removed to obtain the input data.

[0031] In detail, based on the theory of equal power change rate, and with the objective of minimizing the total power consumption of operating units, a nonlinear programming optimization model is constructed by combining the quadratic power consumption model in the input data and the set of operating units. This model is then solved using a numerical algorithm to obtain the initial load allocation scheme, as detailed below: Calculating the marginal cost function: Based on the quadratic power consumption model in the input data, the power change rate function (i.e., the marginal cost function) for each unit is derived. This function is the first derivative of the quadratic power consumption model with respect to gas production, and its expression is: This represents the additional power required for the unit to produce one more cubic meter of gas at the current load point. Construct a nonlinear programming model: Objective function: (Minimize the total power consumption of the operating units); Constraint 1: (Total output equals total demand); Constraint 2: (The output of each unit must be within the safe operating range.) For surge or minimum unloading flow rate, (For blocked or maximum load flow). Determine the system of equations: The necessary and sufficient condition for satisfying the above optimization problem is that all units operating within the constraint interval have the same rate of power change (marginal cost), i.e. ( (These are Lagrange multipliers, i.e., instantaneous marginal costs), which, combined with constraint 1, form a system of equations: (For all operating units) ); ; The Newton-Raphson method is used to iteratively solve the equation system. An iterative convergence threshold is set, and the solution is stopped when the iterative result meets the iterative convergence threshold, thus obtaining the initial load allocation scheme. The specific process is as follows: The rationality of the initial value directly affects whether the iteration gets stuck in a local optimum and the convergence efficiency. The optimal load allocation scheme of the previous cycle is preferred as the initial value of the gas production of each operating unit. This initial value has a small deviation from the true optimal solution and can significantly shorten the convergence time. If it is a scenario without historical data, such as the first operation or the access of a new unit, the initial value is allocated according to the proportion of the unit's maximum gas production. That is, the initial gas production of a single unit is equal to the real-time total load demand multiplied by the ratio of the unit's maximum gas production to the sum of the maximum gas production of all operating units. This ensures that the initial value is within the safe operating range. For the initial value of the instantaneous marginal cost (Lagrange multiplier), the corresponding marginal cost is calculated through the initial gas production of each unit, and the average value of the marginal cost of all units is taken as the initial value. The equation system is transformed into residual form, where the residual is the difference between the calculated result of each equation and the target value (zero). Specifically, it includes two parts: first, the difference between the marginal cost and instantaneous marginal cost of each unit; and second, the difference between the total gas production of all units and the real-time total load demand. The Jacobian matrix is ​​a matrix composed of the partial derivatives of the residuals with respect to each variable to be determined (gas production per unit and instantaneous marginal cost). The construction rules are as follows: the partial derivative of the gas production of a single unit with respect to its own marginal cost residual is equal to twice the nonlinear power consumption coefficient of that unit; the partial derivative of the gas production of a single unit with respect to the total load residual is 1; the partial derivative of the instantaneous marginal cost with respect to the marginal cost residual of each unit is -1; and the partial derivatives between other variables are all 0. Using this rule to construct the matrix can significantly improve computational efficiency and has better accuracy than the numerical difference method. Calculate the residual vector corresponding to the current iteration value, and then solve the linear equation system composed of the Jacobian matrix and the residual vector using the LU decomposition method to obtain the increment values ​​of each variable to be solved. Superimpose the current iteration value with the increment value to obtain the new iteration value. To avoid iteration oscillation, the increment step size can be dynamically adjusted according to the convergence situation (the initial step size is 1, and if the residual increases, it is adjusted to 0.5). The convergence threshold is set as the relative error between two adjacent iterations. The relative error is the ratio of the absolute value of the difference between the old and new iteration values ​​to the new iteration value. In practical applications, the threshold is usually set to one ten-thousandth, which balances computational accuracy and efficiency. When the relative error is less than or equal to this threshold, the iteration is considered to have converged, the calculation is stopped, and the gas production of each unit at this time is extracted as the initial load allocation scheme. If the iteration count reaches 10 and convergence is still not achieved, a linear approximation fallback strategy is used to simplify the quadratic power consumption model to a linear form and then solve it again to ensure that an effective allocation scheme can always be output.

[0032] The optimization principle is based on the "equal marginal cost" theory in economics (derived through the Lagrange multiplier method). This theory proves that in a multi-unit parallel energy supply system, the total energy consumption of the system is minimized when the marginal costs of all operating units are equal. The specific details are as follows: The core objective is to minimize the total power consumption of all operating units while meeting the real-time total load demand. The core constraint is to ensure that the total gas production of all operating units equals the real-time total load demand to guarantee energy supply matching. To solve this constrained optimization problem, a Lagrange multiplier, representing instantaneous marginal cost, is introduced to construct a Lagrange function (i.e., the product of the total power consumption of all operating units minus the Lagrange multiplier and "the product of the total gas production of all units minus the real-time total load demand"). This transforms the constrained problem into an unconstrained one, simultaneously considering both the objective and the constraint. Taking the partial derivatives of this function with respect to the gas production of each unit and the Lagrange multiplier, and setting the partial derivatives to zero, we can obtain the equation: "the product of the total gas production of all operating units minus the Lagrange multiplier and "the product of the total gas production of all units minus the real-time total load demand". The optimality conditions of "the marginal cost (power change rate) of a unit equals the Lagrange multiplier" and "the sum of the gas production of all units equals the real-time total load demand" directly lead to the core conclusion that the marginal costs of all operating units must be equal. Combined with the constraint of total gas production, this ensures that the energy supply meets the demand. This derivation also confirms the scientific validity of the theory that the total energy consumption reaches its minimum when the marginal costs of all operating units in a multi-unit parallel energy supply system are at the same level. If there are units with lower marginal costs, increasing their load and reducing the load of units with higher marginal costs can further reduce the total energy consumption until the marginal costs of all units are equal. At this point, adjusting the load can no longer achieve energy consumption optimization.

[0033] Specifically, the initial load allocation scheme is checked for unit safety boundary constraints, and overload allocation is handled until all units meet safety constraints, resulting in the optimal load allocation scheme, including: Determine safety boundaries: Determine the unit safety boundaries for each air compressor, including the minimum flow boundary for surge (or minimum unloading flow) and the maximum flow boundary for blockage (or maximum full load flow). The boundary values ​​are calibrated based on the equipment's factory parameters and historical operating data. Constraint checks: Verify one by one whether the allocated load of each unit in the initial load allocation scheme is within its corresponding safety boundary range; Over-limit handling: If a unit's allocated load exceeds the safety boundary, the unit's load will be "pinned" to the corresponding boundary value (if it exceeds the minimum boundary, it will be fixed at a certain value). If it exceeds the maximum boundary, it will be fixed as ), and remove the unit from the equal marginal cost optimization group; Update total load: During the solution process, if any unit Calculation load Exceeding its constraint boundaries (e.g.) If the load of the unit is fixed at its boundary value (i.e., ...), then the load of the unit is fixed at its boundary value. This unit Remove it from the equal marginal cost optimization group, and subtract the boundary load value of the unit from the real-time total load demand to obtain the remaining total load. ; Iterative solution: Taking the remaining total load (i.e., the remaining gas volume after deducting the overload of units that have been fixed at the safety boundary from the real-time total load demand) as the new core load demand, firstly, the set of remaining operating units is defined, and the overloaded units that have been removed from the marginal cost optimization group are excluded, and only the units whose load is still within the safety boundary and can participate in dynamic load adjustment are retained. For the remaining set of operating units, the complete optimization solution process is re-executed: First, based on the latest power consumption model parameters (no-load power consumption coefficient, linear power consumption coefficient, and nonlinear power consumption coefficient) of each remaining unit, its marginal cost function (i.e., the first derivative of the quadratic power consumption model with respect to gas production, calculated as the linear power consumption coefficient plus twice the product of the nonlinear power consumption coefficient and gas production) is calculated for each unit, ensuring that the marginal cost function accurately reflects the current energy consumption characteristics of the remaining units. Second, the nonlinear programming optimization model is reconstructed, and the objective function is updated to "minimize the total power consumption of the remaining operating units" (i.e., the sum of the power consumption of each remaining unit). Constraint 1 is adjusted to "the sum of the gas production of the remaining units equals the total remaining load," and constraint 2 remains "the gas production of each remaining unit is still within its own safety boundary" (minimum surge). The third step is to reconstruct the equation system based on the principle of equal marginal cost. The marginal cost of all remaining operating units is equal (all equal to the new instantaneous marginal cost, i.e., the Lagrange multiplier). At the same time, the total gas production of the remaining units is equal to the total remaining load, forming a new nonlinear equation system. The fourth step is to use the Newton-Raphson method to iteratively solve the new equation system. The initial value setting prioritizes the load allocation results of the remaining units in the previous iteration (if it is the first reconstruction, it is allocated according to the maximum gas production ratio of the remaining units). The iteration convergence threshold still adopts the standard that the relative error of two adjacent iterations does not exceed one ten-thousandth. When the iteration result meets the convergence requirement, the solution is stopped, and a new load allocation scheme for the remaining total load is obtained, ensuring that the scheme can both meet the remaining total load demand and minimize the total power consumption of the remaining units. Cyclic verification: Repeat the above constraint checks, over-limit handling, total load updates and iterative solution steps until the allocated load of all units meets the safety boundary constraints and the optimal load allocation scheme is obtained.

[0034] In detail, the optimal load allocation scheme is subjected to load fluctuation collaborative smoothing processing to obtain the final load allocation scheme, as shown below: Based on the unit's equipment model (variable frequency converter / fixed frequency converter), operating years, and core performance parameters (motor rated efficiency, inverter response speed, historical fault frequency), a differentiated maximum load change rate threshold is preset for each unit. The basic threshold for variable frequency converters is set at 5% per cycle (i.e., the load change within a single algorithm execution cycle does not exceed 5% of the current load), and the basic threshold for fixed frequency converters is set at 2% per cycle (fixed frequency converters have less adjustment flexibility, so the threshold is more stringent). For each additional year of operating years, the threshold decreases by 0.5% (for example, for a variable frequency converter that has been operating for 4 years, the threshold = 5% - 4 × 0.5% = 3%). For units that do not meet the core performance parameters (motor efficiency below 85%, inverter response delay > 100ms) or have a fault frequency of ≥ 3 times in the past 6 months, the threshold is further reduced by 1% to ensure the operational stability of old, inefficient, or frequently faulty units. The maximum load change rate threshold is dynamically adjusted according to the equipment's operating status. The adjustment logic is as follows: real-time monitoring of the unit's operating temperature, vibration value, and energy consumption deviation. When the above parameters exceed the normal range for three consecutive cycles, it is determined as "deterioration of operating status," and the threshold is immediately lowered by 0.3%. When the parameters return to normal after equipment maintenance, they are gradually raised back to the initial threshold at a rate of 0.1% per cycle.

[0035] Calculate the absolute difference between the allocated load of each unit in the optimal load allocation scheme and the actual load of the previous cycle. The calculation rule for the difference is "the absolute value of the current allocated load minus the actual load of the previous cycle" to avoid misjudgment caused by positive and negative offsetting. Compare this absolute difference with the preset maximum load change rate threshold (threshold = actual load of the previous cycle × preset ratio) to determine whether it exceeds the limit. If the difference exceeds the maximum load change rate threshold, calibrate the allocated load of the unit in this cycle according to the load adjustment direction: when the optimal scheme load is higher than the actual load of the previous cycle, the calibration value = actual load of the previous cycle + threshold (calibrated according to the maximum allowable increase limit); when the optimal scheme load is lower than the actual load of the previous cycle, the calibration value = actual load of the previous cycle - threshold (calibrated according to the maximum allowable decrease limit). After calibration, calculate the load difference of the unit (i.e., the absolute difference between the optimal scheme load and the calibrated load), and summarize the load differences of all units exceeding the limit to form the total difference to be allocated.

[0036] The total load difference to be allocated is distributed to other units that have not exceeded the limits using the marginal cost weighting method. During allocation, the coordination and adaptation logic between variable frequency and fixed frequency units is strengthened: 70% of the load to be allocated is prioritized to variable frequency units (leveraging their high speed regulation flexibility to avoid untimely response caused by lag in IGV opening adjustment in fixed frequency units). The remaining 30% of the load is allocated to fixed frequency units according to the marginal cost weighting, and the single allocation adjustment amount for fixed frequency units must not exceed 50% of their maximum load change rate threshold to prevent overload adjustment. The reciprocal of the marginal cost of the units that have not exceeded the limits is used as the weight (the lower the marginal cost, the higher the weight). The allocation ratio for a single unit is calculated (single unit weight ÷ sum of weights of all units that have not exceeded the limits). Then, the load adjustment amount for each unit that has not exceeded the limits is determined by "allocation ratio × total load difference to be allocated," ensuring that the allocation process takes into account the optimal goal of total power consumption. After allocation, the load distribution of units not exceeding limits is fine-tuned based on the principle of equal power change rate: the marginal cost of each unit after fine-tuning is calculated, and if the difference in marginal cost exceeds... If necessary, further small adjustments will be made (each adjustment ≤ 0.5%) until the marginal costs of all units that have not exceeded the limits tend to be balanced, and the adjusted loads do not exceed their respective safety boundaries (from minimum surge flow to maximum blockage flow), so as to avoid new over-limit problems caused by sharing.

[0037] The periodic fluctuation trend of real-time load is statistically analyzed: if the direction of the month-on-month change of the total real-time load is consistent (both increasing or both decreasing) within three consecutive periods, and the single month-on-month change is ≥1%, it is determined to be a continuous fluctuation. At this time, the maximum load change rate threshold is relaxed by a preset ratio of 10%-20%. For example, for a frequency converter with an original threshold of 5%, the maximum relaxation should not exceed 6% (20% increase) to avoid sudden load changes exceeding the mechanical bearing limit of the equipment. At the same time, an upper limit constraint is set on the relaxation threshold. The maximum relaxation threshold for frequency converters should not exceed 6%, and for fixed frequency converters, it should not exceed 3% to prevent excessive relaxation from causing equipment shock.

[0038] When the real-time load remains stable for two consecutive cycles (cycle-to-cycle change ≤ 0.5%) or the fluctuation direction is reversed, the threshold recovery mechanism is activated: the load gradually reverts to the initial threshold by 5% per cycle (for example, relaxing the threshold to 6% for frequency converters, decreasing by 0.3% per cycle until it recovers to 5%), while ensuring that the load change rate does not exceed the allowable range of the equipment during the reversion process. Based on the above constraints such as threshold preset and dynamic adjustment, load difference calibration and scientific allocation, fluctuation trend adaptation and threshold recovery, and combined with the fine-tuning results of the equal power change rate principle, a final load allocation scheme that balances operational stability and optimal total power consumption is obtained.

[0039] Specifically, the final load distribution plan will be sent to the local regulators of each air compressor and the operating deviations will be dynamically corrected. The details are as follows: The allocated load for each unit in the final load allocation scheme is converted into a real-time flow setpoint (SP) based on the equipment's rated parameters, with the unit standardized as follows: The accuracy is retained to two decimal places, and the conversion rule is "Set value = Distributed load × Equipment rated flow correction factor" (the correction factor is determined based on the equipment's flow calibration results over the past 3 months, with a range of 0.98). 1.02); Different command formats are adopted for variable frequency drives and fixed frequency drives. For variable frequency drives, the command is converted into a speed command signal (unit: r / min, corresponding to a linear mapping relationship between flow rate setpoint and speed), while for fixed frequency drives, it is converted into an inlet guide vane IGV opening command signal (unit: %). (100% corresponds to minimum to maximum flow). A collaborative synchronization mechanism is added when issuing commands: Variable frequency drive (VFD) commands are issued first, followed by a 50ms delay before the fixed frequency drive commands are issued, ensuring that the VFD responds to load adjustments first and avoiding network pressure fluctuations caused by simultaneous adjustments of both types of units. During deviation correction, the VFD uses a fast fine-tuning strategy (correcting once per cycle), while the fixed frequency drive uses a slow correction strategy (correcting once every two cycles), adapting to the physical response characteristics of both types of units. Commands are issued to the corresponding local controller via industrial communication protocols (preferably Profinet or Modbus TCP). Communication parameters are set as follows: Profinet transmission level RT (real-time level), Modbus TCP port 502, data frame timeout 1 second, and 3 retransmissions. Before issuing commands, the online status of the equipment is verified; commands are only issued to units that are online and have no fault alarms. After issuance, the command confirmation frame from the equipment is received. If no confirmation is received, a retry is triggered. If the retry fails after 3 attempts, an alarm log is recorded.

[0040] Operation status and output parameter acquisition: at a frequency consistent with the algorithm execution cycle (1 The system collects the operating status and output parameters of each air compressor in real time (within 5 seconds). The collected parameters include: actual air output (collected by the flow sensor at the equipment end, with an accuracy of ±1%FS), exhaust pressure (±0.01MPa), operating frequency (variable frequency machine, ±1Hz), motor stator temperature (±0.1℃), lubricating oil temperature (±0.1℃), IGV opening degree (fixed frequency machine, ±1%), and equipment vibration value (±0.1mm / s). After the collected data is pre-processed by the sensor (filtering, temperature compensation), it is uploaded to the edge controller through the same industrial communication protocol to ensure the timing consistency of data acquisition and command issuance.

[0041] Deviation Calculation and Exceedance Judgment: The collected actual parameters are compared with the preset parameters in the final load allocation scheme. Differentiated deviation calculation methods are used according to parameter type: actual gas production, operating frequency, and IGV opening are calculated using relative deviation (relative deviation = |actual value - preset value| / preset value × 100%); exhaust pressure, motor temperature, lubricating oil temperature, and vibration value are calculated using absolute deviation (absolute deviation = |actual value - preset value|). Graded deviation exceedance thresholds are set based on equipment operating accuracy and industry standards. Minor deviations: relative deviations of gas production / frequency / IGV opening >2% and ≤3%, absolute deviations of exhaust pressure >0.01MPa and ≤0.02MPa, absolute deviations of temperature parameters >3℃ and ≤5℃, and absolute deviations of vibration values ​​>0.3mm / s and ≤0.5mm / s; Moderate deviation: relative deviation of gas production / frequency / IGV opening >3% and ≤5%, absolute deviation of exhaust pressure >0.02MPa and ≤0.05MPa, absolute deviation of temperature parameters >5℃ and ≤8℃, absolute deviation of vibration value >0.5mm / s and ≤0.8mm / s; Severe deviations: relative deviations of gas production / frequency / IGV opening > 5%, absolute deviations of exhaust pressure > 0.05 MPa, absolute deviations of temperature parameters > 8℃, and absolute deviations of vibration values ​​> 0.8 mm / s.

[0042] Tiered fine-tuning and scheme redistribution: If the operational deviation exceeds the threshold, a tiered fine-tuning strategy is adopted based on the degree of deviation. The fine-tuning priority follows the principle of "core units are corrected first, and gas production deviation is given priority over status parameter deviation." Slight deviation: Only the current unit's flow rate setpoint is finely adjusted, with the adjustment range being 1% of the preset value (relative fine adjustment), and the adjusted setpoint must not exceed the equipment's safe flow rate range; Moderate deviation: Fine-tune the current unit setpoint by 3% (relative fine-tuning), and simultaneously adjust 1... The setpoints for the two associated generating units with the lowest marginal cost (with a fine-tuning increment of 0.5%). (1%), ensuring that total load demand remains unchanged; Severe Deviation: Fine-tune the current unit's setpoint by 5% (relative fine-tuning), re-verify the unit's safety boundary. If the deviation threshold is still exceeded after fine-tuning, suspend the unit's load optimization privileges and temporarily transfer its load to a standby unit (if applicable). Simultaneously, trigger a high-priority alarm to prompt maintenance personnel to check for equipment faults (such as sensor malfunctions or mechanical jamming). After generating the corrected load allocation plan, reissue it to the corresponding units according to the original issuance process to ensure rapid convergence of the deviation.

[0043] Deviation event recording and log generation: Each deviation event and correction process is recorded synchronously. The deviation handling log includes fixed fields: event number (equipment number and timestamp), occurrence time (accurate to milliseconds), involved unit number and equipment type, deviation parameter name, preset value, actual value, deviation type (relative / absolute), deviation value / proportion, deviation level, fine-tuning strategy (fine-tuning parameters, correction range, linked units), actual value after correction, correction effect (whether the deviation has converged), and recorder (automatically labeled "automatic correction" or manually intervened account). Log data is archived daily on the edge controller's local storage (retained for 90 days) and simultaneously synchronized daily to the cloud platform, providing data support for subsequent optimization of power consumption model parameters and adjustment of load distribution strategies (such as optimizing threshold settings and fine-tuning range), achieving closed-loop optimization of deviation handling.

[0044] In detail, the synchronous adaptation of unit maintenance modes is as follows: During the verification and fault tolerance processing of multi-source data, the frequency is consistent with the algorithm execution cycle (1 (5 seconds) Synchronously monitor the maintenance trigger status of each air compressor to ensure real-time response to maintenance status changes. Maintenance trigger statuses are specifically divided into three scenarios, with clearly defined trigger thresholds and judgment rules: Manual maintenance commands: Issued through the cloud platform or the local operation interface of the edge controller. The commands include the maintenance type (routine maintenance / medium maintenance / deep overhaul), planned maintenance duration, and maintenance priority. The maintenance mode is triggered immediately after issuance, and the commands must be verified (administrator permission authentication and dynamic verification code verification) to prevent accidental operation. Operating parameter over-limit trigger: Real-time monitoring of core operating parameters of the unit (motor stator temperature, lubricating oil temperature, exhaust pressure fluctuation value, equipment vibration value, and cumulative number of inverter fault codes). When any parameter meets the preset maintenance threshold (motor temperature ≥110℃, lubricating oil temperature ≥85℃, exhaust pressure fluctuation value >0.05MPa / cycle, vibration value ≥4.5mm / s, cumulative fault codes ≥3 times / 24 hours) and fails to return to normal for 2 consecutive cycles, the maintenance mode is automatically triggered. Cumulative runtime trigger: Set the graded maintenance time threshold according to the equipment maintenance manual. Routine maintenance corresponds to 5,000 hours of cumulative operation, medium maintenance corresponds to 10,000 hours, and deep overhaul corresponds to 20,000 hours. When the unit's cumulative runtime reaches 95% of the corresponding threshold, a maintenance warning is triggered. When it reaches 100%, the maintenance mode is automatically triggered. The runtime is calculated to the minute and deducts downtime and standby time (only the actual runtime is counted).

[0045] When any air compressor is detected to be in a maintenance-triggered state, during the nonlinear programming optimization model construction phase, the unit to be maintained is automatically marked as a non-priority operating unit, and the load allocation weight adjustment rules are clearly defined: during normal operation, the weight of each unit is "1 / total number of operating units", and the weight of the unit to be maintained is reduced to 50% of the normal weight (i.e., "0.5 / total number of operating units"). The core purpose of reducing the weight is to reduce its load share and reserve a safety margin for maintenance operations. At the same time, a "maintenance period load limit constraint" is added to the model constraints, clarifying that the load allocation of the unit to be maintained must not affect the safety of maintenance operations.

[0046] After checking the safety boundary constraints of the initial load allocation scheme, a further specific verification of the load allocation ratio of the units to be maintained is conducted to ensure that their load does not exceed the preset maintenance period load limit. The maintenance period load limit is dynamically set according to the maintenance type and is based on the equipment maintenance manual, historical maintenance data, and manufacturer recommendations: for routine maintenance (cleaning, tightening, and oil checks), the limit is 70% of the maximum flow boundary of normal operation blockage; for medium maintenance (filter replacement and oil replacement), the limit is 60%; and for deep overhaul (component disassembly and fault repair), the limit is 50%. This avoids excessive load on the units during maintenance, which could lead to increased component wear or obstruction of maintenance operations. If the verification finds that the load exceeds the limit, the load is immediately reduced to the limit value, and the excess load is distributed to other non-maintenance units according to the "marginal cost weighting method".

[0047] When the real-time total load suddenly increases (the month-on-month increase is ≥10% / cycle), and the load of non-maintenance units has reached 90% of their safety boundary (close to full load) and cannot handle the additional load, an emergency load adjustment strategy is activated. The load limit of the unit to be maintained is temporarily relaxed (80% of normal congestion flow during routine maintenance, 70% during medium maintenance, and no relaxation during deep overhaul). The relaxation period only lasts until the next algorithm cycle. At the same time, the standby unit is automatically started. After the standby unit is started (≤30 seconds), the original load limit of the unit to be maintained is immediately restored. If no standby unit can be called up, the load limit of the unit to be maintained is continuously relaxed until the total load meets the demand, but the maximum relaxation period does not exceed 5 minutes. During this period, a high-priority alarm is pushed to the administrator once every 30 seconds to ensure that the maintenance operation is not seriously affected.

[0048] During the load fluctuation smoothing phase, a stricter maximum change rate threshold is applied to the load adjustment of the units to be maintained. The threshold is refined according to the maintenance type: 80% of the normal threshold for routine maintenance, 65% for medium maintenance, and 50% for deep overhaul. By reducing the magnitude of load changes, the fluctuation of the unit's operating status during maintenance is reduced, thereby lowering maintenance risks. Simultaneously, based on the real-time marginal power consumption and safety margin of other operating units, the load assignment strategy is optimized—first, units with marginal power consumption lower than the system average marginal power consumption are selected; then, units with a load redundancy ≥ 20% (load redundancy = (maximum gas production of the unit - current allocated load) / maximum gas production of the unit × 100%) are prioritized to accept the reduced load of the units to be maintained. If multiple units meet the criteria, they are assigned according to the principle of "the lower the marginal power consumption and the higher the load redundancy, the higher the acceptance ratio," ensuring optimal total power consumption and stable system operation.

[0049] After maintenance is completed, the maintenance mode can only be lifted if two conditions are met: first, a manually issued "maintenance completed" confirmation command is received; second, the core operating parameters (temperature, pressure, vibration, etc.) of the unit are within the normal range for three consecutive algorithm cycles, and no new fault codes are added. After confirmation of recovery, its non-priority operation flag is automatically removed, the optimal load allocation scheme is regenerated according to the core process, and a "gradient recovery strategy" is used to gradually restore its load share—in the first cycle, it is restored to 30% of the normal weight, and then increased by 20% in each subsequent cycle, and fully restored to the normal weight after 5 cycles. This achieves a smooth switch between maintenance mode and normal mode, avoiding the impact of sudden load increases on the unit and system.

[0050] In addition, during maintenance mode operation, the system automatically retains 10%-15% of the total load redundancy (reserved by non-maintenance units) to cope with sudden increases in gas load; it also records unit operation data, load allocation adjustment records, and energy consumption changes during maintenance, forming a special maintenance log to provide data support for subsequent optimization of maintenance cycles and adjustment of maintenance period load strategies.

[0051] The processes described above with reference to the flowcharts in the embodiments disclosed in this invention can be implemented as computer software programs. The embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wire segments, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical fibers, RF, etc., or any suitable combination thereof.

[0052] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0053] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and explained in the embodiments. Without departing from the stated principles, the implementation of the present invention may have any variations or modifications.

Claims

1. A real-time load distribution method for air compressor groups based on equal power change rate, characterized in that, The method includes: Acquire multi-source data, perform verification and fault tolerance processing on the acquired multi-source data, and obtain input data; Based on the theory of equal power change rate, with the goal of minimizing the total power consumption of operating units, a nonlinear programming optimization model is constructed by combining the quadratic power consumption model in the input data and the set of operating units, and the initial load allocation scheme is obtained by solving it through a numerical solution algorithm. The initial load allocation scheme is checked for unit safety boundary constraints, and overload allocation is handled until all units meet the safety constraints, thus obtaining the optimal load allocation scheme. The optimal load allocation scheme is processed by load fluctuation coordination and smoothing. The scheme is constrained by the preset maximum load change rate threshold of the unit, the excess load allocation is differentially calibrated, and the load fluctuation trend is dynamically adapted and adjusted to obtain the final load allocation scheme.

2. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 1, characterized in that, The specific process for obtaining the multi-source data is as follows: Data is collected from the pipeline network and each air compressor equipment of the air compressor station to obtain collected data including total air production, equipment power consumption, and operating status. Based on the air compressor station's operating logic, the collected data is associated with devices and encapsulated into events. The encapsulated data is then uploaded to a pre-set central data platform through the standardized interface of the air compressor station's energy management system. Real-time data messages pushed by the central data platform are received via message subscription, and the format of the real-time data messages is parsed to obtain parsed data. The parsed data is correlated with the device and operating condition context to obtain multi-source data.

3. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 2, characterized in that, The specific process of obtaining the input data is as follows: For core data from multiple sources, multi-source cross-verification of primary and backup sensors is adopted. A data deviation threshold is set. When the deviation between primary and backup data exceeds the data deviation threshold, the predicted value derived from the secondary power consumption model is used for replacement. If the power consumption model parameters of a single unit are missing or abnormal, extract the historical data of the same operating conditions of the unit in the most recent period, combine them with the parameter mapping relationship of the same type of unit, and generate a temporary power consumption model through transfer learning. When data transmission is interrupted, the optimal load allocation scheme of the previous cycle is cached and linearly fine-tuned based on the load change trend of the past 5 cycles until the data is restored and the normal data processing flow is switched. The data processed above is then subjected to integrity verification to remove invalid and redundant data, resulting in the input data.

4. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 3, characterized in that, The specific process for obtaining the initial load allocation scheme is as follows: Based on the quadratic power consumption model in the input data, the power change rate function of each unit is derived. The power change rate function is the first derivative of the quadratic power consumption model with respect to the gas production. A nonlinear programming optimization model is constructed with the objective function of minimizing the total power consumption of the operating units and the constraints of equal power change rate of all operating units and total gas production meeting the real-time total load demand. Determine the set of equations corresponding to the nonlinear programming optimization model. The set of equations includes equations that ensure the equal rate of change of power of each unit and equations that balance the total gas production. The Newton-Raphson method is used to iteratively solve the system of equations. An iterative convergence threshold is set, and the solution is stopped when the iterative result meets the iterative convergence threshold, thus obtaining the initial load allocation scheme.

5. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 4, characterized in that, The specific process for obtaining the optimal load allocation scheme is as follows: Determine the unit safety boundary for each air compressor. The unit safety boundary includes the minimum flow boundary for surge and the maximum flow boundary for blockage. The boundary values ​​are calibrated based on the equipment's factory parameters and historical operating data. Verify one by one whether the allocated load of each unit in the initial load allocation scheme is within its corresponding safety boundary range; If a unit's allocated load exceeds the safety boundary, the unit's load is fixed at the corresponding boundary value, and this boundary value is deducted from the real-time total load demand to obtain the remaining total load. Using the remaining total load as the new load demand, a new nonlinear programming optimization model is reconstructed and solved for the remaining operating units to obtain a new load allocation scheme. Repeat the above verification and solution steps until the allocated load of all units meets the safety boundary constraints, and obtain the optimal load allocation scheme.

6. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 5, characterized in that, The specific process for obtaining the final load allocation scheme is as follows: Based on the equipment model, years of operation and performance parameters of the unit, a maximum load change rate threshold is preset for each unit, and the maximum load change rate threshold is dynamically adjusted according to the equipment operating status. Calculate the difference between the allocated load of each unit in the optimal load allocation scheme and the actual load of the previous cycle, and determine whether the difference exceeds the preset maximum load change rate threshold. If the difference exceeds the maximum load change rate threshold, the load allocated to the unit in this cycle will be calibrated to the sum of the actual load and the maximum load change rate threshold of the previous cycle, and the load difference after calibration will be calculated. The load difference is evenly distributed to other units, and the load distribution of other units is fine-tuned in accordance with the principle of equal power change rate. The periodic fluctuation trend of real-time load is statistically analyzed. When the real-time load fluctuates in the same direction for three or more consecutive periods, the maximum load change rate threshold is relaxed by a preset ratio. Based on the above constraints, calibration, and adaptation results, the final load allocation scheme is obtained.

7. A real-time load distribution method for air compressor groups based on equal power change rate according to claim 6, characterized in that, After obtaining the final load allocation plan, the final load allocation plan is sent to the local regulators of each air compressor, and the actual operating status and output parameters of each air compressor are collected in real time. The actual operating status and output parameters are compared with the preset parameters in the final load allocation scheme to determine whether there are any operating deviations exceeding the standard. If there are operational deviations exceeding the standard, the final load allocation scheme will be dynamically fine-tuned based on the degree of deviation, and a revised load allocation scheme will be generated and redistributed. The deviation events and correction process are recorded synchronously to form a deviation handling log.

8. The real-time load distribution method for air compressor groups based on equal power change rate according to claim 7, characterized in that, During the verification and fault tolerance processing of multi-source data, the maintenance trigger status of each air compressor is monitored synchronously. The maintenance trigger status includes manually issued maintenance instructions, operating parameters reaching preset maintenance thresholds, or cumulative running time meeting the requirements for regular maintenance. When any air compressor is detected to be in a maintenance-triggered state, the unit to be maintained is marked as a non-priority operating unit and its load allocation weight is reduced when constructing the nonlinear programming optimization model. After checking the safety boundary constraints of the initial load allocation scheme, the load allocation ratio of the units to be maintained is further verified to ensure that their load does not exceed the preset maintenance period load limit. During the load fluctuation smoothing phase, a stricter maximum change rate threshold is adopted for the load adjustment of the units to be maintained. At the same time, the load assignment strategy is optimized based on the marginal power consumption and safety margin of other operating units to ensure optimal total power consumption and stable operation. Once the maintenance unit has completed maintenance and resumed normal operation, its non-priority operation flag will be automatically removed.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement a real-time load distribution method for an air compressor group based on an equal power change rate, as described in any one of claims 1-8.

Images