Transformer fan PLC control alarm system integrating soft start and speed regulation

Abstract

The invention relates to the field of industrial automation control, and discloses a transformer fan PLC control alarm system integrating soft start and speed regulation, and the system comprises a physical perception and execution module which is used for collecting the real-time operation parameters of a transformer; the state estimation and optimization control module is used for calculating optimal internal thermal state estimation including winding hot-spot temperature which cannot be directly measured, and constructing and solving a multi-objective optimization problem to generate the control instruction; the system diagnosis and self-adaption module is used for determining the running state of the system; and the man-machine interaction and communication module is used for displaying the optimal internal thermal state estimation calculated by the state estimation and optimization control module. According to the method, a state estimation algorithm based on a dynamic thermodynamic model and Kalman filtering is adopted, accurate tracking of the winding hot-spot temperature is achieved, the fundamental problems that control lags and core temperature control indexes cannot be touched are solved, transformer core insulation is accurately protected, and the service life of equipment is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation control, and in particular to a transformer fan PLC control alarm system integrating soft start and speed regulation. Background Art

[0002] Currently, large oil-immersed power transformers are core components of power transmission and distribution networks. During operation, they generate significant heat due to winding copper losses and core iron losses. This heat must be dissipated promptly and effectively, directly impacting the equipment's safe and stable operation and determining the aging rate of its internal insulation system and ultimately its service life. Forced air cooling systems are the primary heat dissipation method for large-capacity transformers, and the quality of their fan control strategies plays a decisive role in the transformer's overall performance.

[0003] In light of the aforementioned issues, existing technologies commonly utilize programmable logic controllers (PLCs) to build automatic control systems for transformer fans. These systems typically operate by measuring the top oil temperature using a temperature sensor installed on the transformer body and comparing it with several fixed temperature thresholds pre-set within the PLC. When the measured oil temperature reaches the start threshold, the PLC outputs a switching signal, controlling an intermediate relay or AC contactor to start one or all fans. When the oil temperature drops below the stop threshold, the PLC issues a command to shut down the fans.

[0004] However, the control schemes widely adopted in the above-mentioned prior art have some inherent defects that are difficult to overcome.

[0005] First, it relies entirely on a single, delayed physical quantity: top oil temperature. In reality, the temperature at the internal winding hotspot is crucial for transformer insulation life. Oil temperature changes much more slowly than winding temperature, resulting in a significant dynamic delay between the two. This often results in the cooling system's adjustments being slow to react, with the fan activating only after the winding has been exposed to excessively high temperatures for a period of time, potentially posing a risk of overheating and damage to the equipment.

[0006] Secondly, the control logic and parameters of existing technologies are completely static. A set of factory-set "start and stop thresholds" cannot adapt to the dynamic changes in the cooling system's physical characteristics. For example, over time, radiator fins become clogged due to dust accumulation, and fan blades become less efficient due to wear, all of which lead to reduced heat dissipation performance. However, the rigid PLC control logic is "unaware" of this, and its control effectiveness and accuracy will gradually deteriorate over time, making it impossible to guarantee long-term optimal operation.

[0007] Furthermore, traditional start-stop or simple hierarchical control strategies make decisions based on a single metric: "whether the temperature exceeds the limit." They are completely incapable of comprehensively evaluating the economic viability of operations. For example, they fail to factor in the wind turbine's own operating energy consumption in their control, nor do they quantify the long-term impact of varying temperatures on transformer insulation life. Furthermore, they operate independently, acting more like an "information island," completely unable to perceive or respond to external grid economic or dispatch signals, such as real-time electricity prices and demand response instructions. Consequently, they miss valuable opportunities to collaborate with the entire grid system and achieve higher levels of energy conservation and cost reduction. Summary of the Invention

[0008] In response to the shortcomings of the existing technology, the present invention provides a transformer fan PLC control alarm system with integrated soft start and speed regulation, which solves the problem that the existing fan control strategy relies on a single, lagging temperature signal for extensive control, and its control logic is rigid and cannot adapt to system performance degradation.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a transformer fan PLC control alarm system with integrated soft start and speed regulation, comprising: A physical sensing and execution module, used to collect real-time operating parameters of the transformer and drive the fan to operate according to the generated control instructions; a state estimation and optimization control module, which calculates an optimal internal thermal state estimate, including a winding hot spot temperature that cannot be directly measured, based on a preset dynamic thermodynamic model of the transformer and the real-time operating parameters, and constructs and solves a multi-objective optimization problem based on the optimal internal thermal state estimate to generate the control instructions; a system diagnosis and adaptation module that determines an operating state of the system based on a model residual formed during the calculation of the optimal internal thermal state estimate; A human-computer interaction and communication module is used to display the optimal internal thermal state estimation calculated by the state estimation and optimization control module, and the system operating state determined by the system diagnosis and adaptation module.

[0010] Preferably, the dynamic thermodynamic model used by the state estimation and optimization control module is a discrete time state space model, which is constructed in the following mathematical form: ; ; in, Indicates the The system state vector at time t, including the temperature of the hot node and top oil temperature ; Indicates the The input signal at the moment includes the total power loss in real-time operation , ambient temperature And the heat dissipation power of the air cooling system determined by the controller ; Represents a measurable output; 、 、 and are all system parameter matrices.

[0011] Preferably, the state estimation and optimization control module calculates the optimal internal thermal state estimate through a Kalman filter state observation unit, and the calculation process of the unit is constructed as follows: First, the state space model is used to calculate the prior state estimate ; Then, the measured values ​​of the real-time operating parameters of the transformer are subtracting the predicted output of the model from the predicted output of the model to construct the model residual; Finally, the model residual and a Kalman gain are used The prior state estimate is modified to generate the optimal state estimate , and its generation formula is: ; in, is the system parameter matrix.

[0012] Preferably, the multi-objective optimization problem is solved by constructing a cost function To achieve this, the cost function The construction method is: The cost function is composed of a state tracking term, a control increment term, and an insulation aging term; : No. The future predicted based on the current state The system state prediction value of the step; : No. The desired reference state at the moment; : weight matrix of state tracking error; : Weight matrix that controls the increment; : No. The input increment prediction value of the step; : Weight coefficient of insulation aging penalty term; : insulation aging damage function; : No. Always predict the future The temperature of the key thermal nodes of the step; : Prediction time domain length; : Control the time domain length; : cost function value.

[0013] Preferably, the insulation aging term is constructed by constructing a relative aging rate function To calculate, the construction formula of the function is: ; in, is the winding hot spot temperature generated by the optimal internal thermal state estimation.

[0014] Preferably, the system diagnosis and adaptation module includes an online diagnosis unit, the workflow of which is constructed as follows: First, obtaining the model residual constructed by the Kalman filter state observation unit in the process of generating the optimal state estimate; Then, the statistical characteristics of the time series of the model residuals are analyzed; Finally, when the statistical characteristic deviates from a preset benchmark, it is determined that the system has experienced performance degradation.

[0015] Preferably, the system diagnosis and adaptation module further includes a model self-correction unit, which is triggered after the online diagnosis unit determines that the system has experienced performance degradation. The workflow is constructed as follows: constructing and solving a system identification problem using a set of recent real-time operating parameters; Finally, the solution of the system identification problem is used to generate the system parameter matrix in the dynamic thermodynamic model. The correction value is used to update the dynamic thermodynamic model.

[0016] Preferably, the system diagnosis and self-adaptation module workflow is constructed as follows: First, record the system parameter matrix generated by the model self-correction unit and The cumulative amount of corrections over the years; Then, a normalization function is applied to the Frobenius norm of the cumulative correction To generate a quantitative system health index , and its generation formula is: ; in, No. In subsystem identification, the system parameter matrix The correction amount, that is, the difference between the current estimate and the initial model value; No. In subsystem identification, the system parameter matrix The correction amount, that is, the difference between the current estimate and the initial model value; : Frobenius norm, used to measure the overall correction degree of a matrix, that is, the square root of the sum of the squares of the matrix elements; Indicates that all historical correction steps The sum of the Frobenius norms of the corrections, i.e. the total cumulative deviation of all corrections; Indicates that all historical correction steps The Frobenius norm of the correction quantity is summed.

[0017] Preferably, the workflow of the human-computer interaction and communication module is constructed as follows: First, receiving the optimal internal thermal state estimate generated by the state estimation and optimization control module and the system operating state determined by the system diagnosis and adaptation module, and graphically displaying them on its human-machine interface; Secondly, it is configured to receive an external power grid signal through its communication interface and providing the signal to the state estimation and optimization control module.

[0018] Preferably, the state estimation and optimization control module further includes a collaborative strategy and dynamic weighting unit, the workflow of which is constructed as follows: First, the external power grid signal provided by the human-computer interaction and communication module is obtained Then the external grid signal Mapping to the cost function Medium weight matrix 、 or weight coefficient A new set of values ​​for Finally, the cost function is reconstructed using the new value , to adjust the optimization direction of the multi-objective optimization problem.

[0019] The present invention provides a transformer fan PLC control alarm system that integrates soft start and speed regulation. It has the following beneficial effects: 1. This invention utilizes a state estimation algorithm based on a dynamic thermodynamic model and Kalman filtering to accurately track winding hotspot temperatures. Compared to existing solutions that rely solely on top-layer oil temperature thresholds for coarse start-stop control, this solution overcomes the fundamental drawbacks of control lag and inability to achieve core temperature control indicators. This enables proactive intervention before hotspot temperatures exceed limits, precisely protecting transformer core insulation and extending equipment life.

[0020] 2. This invention establishes an online diagnostic and adaptive correction mechanism driven by model residuals. This solution monitors the consistency of the control model with physical reality in real time. Unlike traditional fixed-logic PLC control, which deteriorates performance if radiator dust accumulates or fan fails, this solution overcomes its shortcomings of being "unaware" of system performance degradation. It proactively detects problems and automatically corrects the control model, ensuring that the control system always operates in optimal condition. This shift from "passive maintenance" to "predictive maintenance" significantly improves operational convenience.

[0021] 3. This invention uses model predictive control to directly quantify the transformer insulation aging rate and fan energy consumption and incorporates them into optimization objectives. This addresses the lack of consideration of operational economics in existing control strategies. While ensuring safety, it meticulously balances temperature control, energy conservation, and equipment depreciation, ultimately achieving a comprehensive cost reduction for the transformer's entire life cycle.

[0022] 4. This invention also incorporates a collaborative strategy with external grid signals. Through a human-machine interaction and communication module, it receives electricity prices or dispatch instructions and dynamically adjusts the "baton" (i.e., the cost function weights) for optimal control. This overcomes the existing limitations of transformer cooling systems, which function as isolated information islands and remain oblivious to grid conditions. This transforms transformers from mere energy-consuming devices into "smart" grid assets capable of participating in demand response, enabling them to operate in tandem with grid economics. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a system framework structure diagram of the present invention; Figure 2 This is a schematic diagram of the process architecture of the physical perception and execution module of the present invention; Figure 3 Schematic diagram of the process architecture of the state estimation and optimization control module of the present invention; Figure 4 Schematic diagram of the process architecture of the system diagnosis and self-adaptation module of the present invention; Figure 5 Schematic diagram of the process architecture of the human-computer interaction and communication module of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] Please see the attached Figure 1 -Attached Figure 5 The embodiment of the present invention provides a transformer fan PLC control alarm system integrating soft start and speed regulation, comprising: A physical sensing and execution module, used to collect real-time operating parameters of the transformer and drive the fan to operate according to the generated control instructions; In this embodiment, the physical perception and execution module, serving as an interface for information exchange and physical intervention between the system and the physical transformer device, is configured to perform underlying real-time data acquisition and physical implementation of high-level control instructions.

[0026] Through a set of integrated sensor units, the transformer's real-time operating parameters are continuously collected, forming the only source of raw data required for all subsequent state estimation, control optimization, and diagnostic functions.

[0027] The sensor unit specifically includes a current transformer installed on the secondary side of the transformer, which is configured to accurately measure the real-time load current flowing through the transformer. This measurement is the basis for calculating the heat generation rate of the transformer. It is the input vector required to construct the subsequent dynamic thermodynamic model. , the module will use the collected load current Further calculate the total power loss of the transformer The calculation process follows the following formula: ; in, The no-load loss of the transformer is a fixed parameter preset according to the transformer design nameplate data; is the rated load loss of the transformer, which is also a preset design parameter.

[0028] Load factor By real-time load current and the rated current of the transformer Jointly determined, the calculation method is: ; in, : Load factor, which represents the ratio of the actual load current to the rated current of the transformer. This factor helps to assess the load level of the transformer during operation.

[0029] : Actual load current refers to the actual current value that the transformer bears during the operation of the equipment.

[0030] : The rated current of the transformer indicates the maximum operating current value specified when the transformer is designed.

[0031] Through the above two formulas, the system converts the easily measurable electrical quantities into the heat input parameters required by the subsequent thermodynamic model. .

[0032] The sensor unit further includes a high-precision platinum resistance thermometer, preferably a Pt100 temperature sensor with good linearity and stability, which is installed in a dedicated temperature measuring sleeve on the top of the transformer tank to accurately measure the top oil temperature. .

[0033] Measured top oil temperature The real-time value is directly used to construct the measurable output vector in the dynamic thermodynamic model This vector is the only measured benchmark for subsequent Kalman filter state observation units to perform model prediction corrections.

[0034] The sensor unit also includes an ambient temperature sensor, which is reasonably set at the air inlet position of the forced air cooling cooler to ensure that the initial temperature of the unheated cooling medium entering the cooling system, that is, the ambient temperature, can be measured. .

[0035] The physical perception and execution module also includes an actuator unit, which drives the fan in response to the control instructions generated by the state estimation and optimization control module, thereby accurately converting the control decisions of the digital world into heat dissipation actions in the physical world.

[0036] In a specific embodiment, the actuator unit is mainly one or more high-performance frequency converters (VFDs), and the three-phase AC output terminals thereof are electrically connected to the fan cooling motor of the transformer.

[0037] The control instruction received by the frequency converter is specifically the target frequency value calculated by the model prediction control unit. The inverter adjusts the frequency of the AC power it outputs to the fan motor through pulse width modulation (PWM) technology, achieving stepless and smooth speed regulation of the fan.

[0038] The inverter's inherent soft start function is used to smoothly increase the output frequency and voltage through a preset ramp function during the wind turbine startup phase, effectively limiting the motor's starting current to within a specific multiple of the rated current, thereby avoiding the voltage drop impact on the power grid caused by direct startup and significantly reducing the mechanical stress on the motor windings and mechanical transmission chain.

[0039] The input vector for building a dynamic thermodynamic model , the control instructions need to be Converted into actual physical heat dissipation power This conversion process is based on the fluid mechanics characteristics of the fan and follows the fan law. The relationship can be expressed as follows: ; in, : Current cooling power refers to the actual cooling power of the fan system at a specific frequency; :Fan at rated frequency The rated cooling power under this condition is the maximum cooling capacity specified when the equipment is designed; : The current operating frequency of the variable frequency drive.

[0040] : Rated operating frequency of the motor.

[0041] : Exponential parameter used to describe the relationship between cooling power and frequency. A value of 3 indicates a cubic relationship between cooling power and frequency.

[0042] in, The fan is at rated frequency The rated heat dissipation power corresponding to the operation under the condition of the UPS is stored in the system as a preset parameter; is the rated operating frequency of the motor; index is the power exponential relationship coefficient between heat dissipation power and frequency. Its theoretical value is 3. In practical applications, it can be calibrated according to the measured performance curve of the fan to improve the model accuracy.

[0043] final, as well as Together they form a complete input vector for driving the next module to perform state estimation and control optimization. , thus forming a complete technical chain from physical perception to data construction.

[0044] a state estimation and optimization control module, which calculates an optimal internal thermal state estimate, including a winding hot spot temperature that cannot be directly measured, based on a preset dynamic thermodynamic model of the transformer and the real-time operating parameters, and constructs and solves a multi-objective optimization problem based on the optimal internal thermal state estimate to generate the control instructions; The state estimation and optimization control module, as the core computing and decision-making center of the system of the present invention, is configured to perform state estimation and control optimization based on a preset dynamic thermodynamic model of the transformer and real-time operating parameters provided by the physical perception and execution module.

[0045] In a specific embodiment, the dynamic thermodynamic model is constructed as a discrete time state space model, which is a mathematical abstraction of the physical process of heat generation, transfer, and dissipation within the transformer. The model has the following standard mathematical form: ; ; in, Indicates the The system state vector at time t, including the temperature of the hot node and top oil temperature ; Indicates the The input signal at the moment includes the total power loss in real-time operation , ambient temperature And the heat dissipation power of the air cooling system determined by the controller ; Represents a measurable output; 、 、 and are all system parameter matrices, where is the state transition matrix, which describes the characteristics of the system's internal state evolving freely over time; is the input matrix, which describes how external input affects the change of internal state; is the output matrix, which describes how the internal states are mapped to measurable outputs; and The values ​​of these matrices are preset during system initialization according to the design parameters of the transformer or through offline system identification experiments.

[0046] In order to make the model work, each vector in the state space model is constructed in real time from physical quantities in the following way: The elements of are composed of the internal thermal state of the system, specifically, , which includes the winding hot spot temperature which cannot be directly measured and plays a decisive role in insulation aging , and the top oil temperature can be measured .

[0047] The input vector The elements of are composed of real-time operating parameters from the physical perception and execution modules and the control decisions of the previous cycle. Specifically, .in and is the real-time disturbance input, is the controlled input of the previous cycle.

[0048] The measurable output vector The elements are composed of the top oil temperature measurement value in the real-time operating parameters collected by the physical perception and execution module. Specifically, This vector is the physical reality benchmark for subsequent state estimation corrections.

[0049] Based on the constructed mathematical model, the state estimation and optimization control module calculates the optimal internal thermal state estimate, including unmeasurable states, through a Kalman filter state observation unit. The calculation process of this unit is constructed as a prediction-correction recursive process.

[0050] First, the unit uses the state space equation and the optimal state estimate at the previous moment , calculate the prior state estimate at the current moment . Then, by taking the real-time measured values Subtract the model's predicted output from , to construct the model residuals.

[0051] Finally, the unit uses the model residuals with the real-time calculated Kalman gain Perform weighted correction on the prior state estimate to generate the optimal state estimate at the current moment The generation process follows the following formula: ; ; in, is the posterior state estimate, that is, the optimal estimate at the current moment; is the prior state estimate; is the Kalman gain, Its value is updated in real time according to the process noise covariance matrix and the measurement noise covariance matrix of the model, and is used to dynamically balance the weights of model predictions and actual measurements in the estimation results.

[0052] To obtain the optimal internal thermal state estimate After that, the module generates control instructions to the actuator through a model prediction control unit (MPC). As the initial condition of the prediction model, the control instructions are generated by online constructing and solving a multi-objective optimization problem with the future optimal control sequence as the optimization variable.

[0053] The multi-objective optimization problem is solved by constructing a cost function To achieve this, the cost function The construction method is: ; in, To predict the length of the horizon, it determines the degree of foresight of the controller optimization; To control the time domain length, the number of solutions for future control actions is determined; is the state reference trajectory; To control the increment.

[0054] The first term of the cost function is the state tracking term, where is a positive definite weight matrix used to penalize the deviation between the predicted state and the reference value. The second term is the control increment term, where is a positive definite weight matrix, which is used to penalize the drastic change of the control quantity to ensure stability. The third term is the insulation aging term, where is the scalar penalty weight.

[0055] The insulation aging term in the cost function is constructed by constructing a relative aging rate function To quantify the loss of transformer insulation life. The formula for constructing this function is: ; in, is estimated by the optimal internal thermal state The generated predicted value of the winding hotspot temperature within the prediction time domain; 15000 is the activation energy-related constant of cellulose insulation paper; 110 is the reference hotspot temperature (°C). The introduction of this term enables control decisions to proactively consider the long-term health of the equipment.

[0056] In a control cycle, the MPC unit solves the minimization problem of the cost function (7) under the conditions of satisfying input and state constraints through numerical optimization algorithms such as quadratic programming, thereby obtaining an optimal control increment sequence, and only the first element of the sequence is applied to the current control quantity, which is issued as the control instruction of the current cycle.

[0057] Furthermore, the state estimation and optimization control module also includes a collaborative strategy and dynamic weighting unit. The workflow of this unit is constructed as follows: first, the external power grid signal is obtained through the human-computer interaction and communication module. , the signal may be time-of-use electricity price information or demand response instruction.

[0058] The unit then converts the external grid signal Through a preset mapping rule, such as a piecewise function or a lookup table, it is converted into a weight matrix in the cost function 、 or weight coefficient A new set of values.

[0059] Finally, the unit reconstructs the cost function using the new values For example, when a high electricity price signal is received, the weight matrix is ​​automatically increased. Medium and fan cooling power The corresponding element values ​​are used to make the system more inclined to generate energy-saving control instructions when solving the optimization problem, so as to dynamically adjust the optimization direction of the multi-objective optimization problem.

[0060] a system diagnosis and adaptation module that determines an operating state of the system based on a model residual formed during the calculation of the optimal internal thermal state estimate; The system diagnosis and adaptation module is configured to monitor the system's operating status online and enable the system model to self-adjust to changes in physical properties. Its entire workflow is driven by the "model residuals" generated during the calculation process of the state estimation and optimization control module.

[0061] This module includes an online diagnostic unit for determining the operating status of the system in real time. The workflow of this unit is constructed as follows: first, in real time and cycle by cycle, the model residual is obtained from the Kalman filter state observation unit in the state estimation and optimization control module. : ; in, is the actual measurement value of the top oil temperature collected by the physical perception and execution module, and is the predicted output based on outdated model parameters. When the system is healthy and the model is accurate, Statistically, it should appear as a white noise sequence with zero mean.

[0062] Then, the online diagnosis unit performs the model residual In a specific embodiment, the mean of the residual sequence is detected using the Cumulative Sum (CUSUM) control chart algorithm. With negative cumulative sum To achieve: ; ; in, is the residual vector scalar elements of , is the amount of change allowed to detect a mean shift of a certain size.

[0063] When the statistical characteristics deviate from a preset benchmark, or The value exceeds a preset control limit threshold , the online diagnostic unit determines that the system has experienced performance degradation, such as reduced heat dissipation efficiency caused by dust accumulation on the cooler fins, and immediately generates a trigger signal.

[0064] The system diagnosis and adaptation module further includes a model self-correction unit, which is activated after receiving a trigger signal generated by the online diagnosis unit. Its workflow is constructed as follows: First, a set of recent real-time operating parameters stored in the system's circular buffer is used, specifically a historical input vector sequence and the output vector sequence , construct and solve a system identification problem.

[0065] In a specific embodiment, the recursive least square method with forgetting factor (RLS-FF) is used to solve the system identification problem online to obtain the system parameter matrix in the dynamic thermodynamic model. and The estimated value of the element to be corrected.

[0066] Finally, the solution of the system identification problem is used to generate the system parameter matrix in the dynamic thermodynamic model and Correction value and and use it to update the model used in the state estimation and optimization control module, that is, , thereby achieving adaptive correction of the model.

[0067] The system diagnosis and adaptation module is further configured to quantitatively evaluate the long-term health status of the transformer cooling system. This function is constructed as follows: first, in its internal non-volatile memory, it records and accumulates the system parameter matrix generated by the model self-correction unit. and The amount of correction.

[0068] Then, a normalization function is applied to the Frobenius norm of the cumulative correction To generate a quantitative system health index The generation process follows the following formula: ; in, Represents the Frobenius norm of the matrix, which is defined as the square root of the sum of the squares of all matrix elements and is used to quantify the magnitude of each model correction.

[0069] In a specific embodiment, the normalization function Can be constructed as a linear decay function to provide an intuitive readout from 100% to 0%: in, is the norm of the total cumulative correction; It is a preset total correction threshold that represents a complete deterioration of the health status. Its value is set based on offline simulation or empirical data, indicating that the model parameters have drifted to the point where physical maintenance is required.

[0070] A human-computer interaction and communication module for displaying the optimal internal thermal state estimate calculated by the state estimation and optimization control module and the system operating state determined by the system diagnosis and adaptation module The human-computer interaction and communication module serves as the only window for information exchange between the system and the outside world, including operation and maintenance personnel and upper-level management systems. It is configured to realize the graphical presentation of internal operation data and the reception and translation of external collaborative instructions.

[0071] The working process of this module is constructed as follows: it is first configured to continuously receive the optimal internal thermal state estimation including unmeasurable states generated in real time by the state estimation and optimization control module through the internal high-speed data bus. .

[0072] Furthermore, in a specific embodiment, the module includes an industrial grade embedded human machine interface (HMI), such as a color touch screen with sufficient resolution and processing power. The HMI is configured to estimate the optimal internal thermal state. The various elements in the are displayed graphically.

[0073] For example, the top oil temperature and the critical winding hot spot temperature are simultaneously displayed in the form of a real-time trend curve on a main monitoring screen.

[0074] This module is also configured to receive the system operating status determined by the system diagnosis and adaptation module and also visually display it on the human-machine interface. In one embodiment, when the system diagnosis and adaptation module determines that the system has experienced performance degradation, this module will issue an alarm on its human-machine interface with a striking red alarm pop-up window or sound prompt; at the same time, the system health index Displayed in a gauge style from 0% to 100% of the scale.

[0075] Secondly, the module is also configured to exchange data with an external network through its integrated communication interface unit. The communication interface unit can be physically an RJ45 Ethernet port and is configured to support one or more standard industrial communication protocols at the protocol level.

[0076] Through the communication interface unit, the module is able to receive formatted external grid signals In a specific application scenario, the signal It is a message containing a specific data structure, and its payload may include the time-of-use electricity price level price_level (for example: peak, flat, valley) used for economic scheduling, or instructions for responding to grid stability requirements.

[0077] Upon receiving the external grid signal After that, the processor of the module parses it, extracts the valid information and temporarily stores it in the internal buffer, and immediately provides the valid information carried by the signal to the collaborative strategy and dynamic weighting unit in the state estimation and optimization control module through the internal data bus.

[0078] The core function of the collaborative strategy and dynamic weighting unit is to convert the external grid signal Mapping to the model prediction control cost function In a specific embodiment, the mapping relationship can be constructed as a lookup table or a set of conditional judgment rules.

[0079] Exemplarily, the unit adjusts the control increment weight matrix in the cost function formula according to the received electricity price level price_level The adjustment process can be expressed as follows: ; in, is the new weight matrix after adjustment, and the adjustment coefficient The value of The price_level in the decision.

[0080] For example: If price_level is peak, then set (like ) to increase the penalty for energy consumption and make the control strategy more energy-efficient.

[0081] If price_level is flat, set , maintaining benchmark weights.

[0082] If price_level is 'valley', set (like ) to reduce the penalty on energy consumption and allow the system to trade higher energy consumption for better temperature control or life protection when necessary.

[0083] Finally, the collaborative strategy and dynamic weighting unit use this new weight matrix Reconstruct the cost function This completes the closed loop from receiving external signals to adjusting internal optimization targets, enabling the system’s control behavior to dynamically coordinate with external economic or safety goals.

[0084] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A transformer fan PLC control alarm system with integrated soft start and speed regulation, characterized in that: include: A physical sensing and execution module, used to collect real-time operating parameters of the transformer and drive the fan to operate according to the generated control instructions; a state estimation and optimization control module, which calculates an optimal internal thermal state estimate, including a winding hot spot temperature that cannot be directly measured, based on a preset dynamic thermodynamic model of the transformer and the real-time operating parameters, and constructs and solves a multi-objective optimization problem based on the optimal internal thermal state estimate to generate the control instructions; a system diagnosis and adaptation module that determines an operating state of the system based on a model residual formed during the calculation of the optimal internal thermal state estimate; A human-computer interaction and communication module is used to display the optimal internal thermal state estimation calculated by the state estimation and optimization control module, and the system operating state determined by the system diagnosis and adaptation module.

2. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 1 is characterized in that: The dynamic thermodynamic model used by the state estimation and optimization control module is a discrete time state space model, which is constructed in the following mathematical form: ； ； in, Indicates the The system state vector at time t, including the temperature of the hot node and top oil temperature ; Indicates the The input signal at the moment includes the total power loss in real-time operation , ambient temperature And the heat dissipation power of the air cooling system determined by the controller ; Represents a measurable output; 、 、 and are all system parameter matrices.

3. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 1 is characterized in that: The state estimation and optimization control module calculates the optimal internal thermal state estimate through a Kalman filter state observation unit. The calculation process of this unit is constructed as follows: First, the state space model is used to calculate the prior state estimate ; Then, the measured values ​​of the real-time operating parameters of the transformer are subtracting the predicted output of the model from the predicted output of the model to construct the model residual; Finally, the model residual and a Kalman gain are used The prior state estimate is modified to generate the optimal state estimate , and its generation formula is: ； in, is the system parameter matrix.

4. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 1 is characterized in that: The multi-objective optimization problem is solved by constructing a cost function To achieve this, the cost function The construction method is: The cost function is composed of a state tracking term, a control increment term, and an insulation aging term; : No. The future predicted based on the current state The system state prediction value of the step; : No. The desired reference state at the moment; : weight matrix of state tracking error; : Weight matrix that controls the increment; : No. The input increment prediction value of the step; : Weight coefficient of insulation aging penalty term; : insulation aging damage function; : No. Always predict the future The temperature of the key thermal nodes of the step; : Prediction time domain length; : Control the time domain length; : cost function value.

5. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 4 is characterized in that: The insulation aging term is constructed by constructing a relative aging rate function To calculate, the construction formula of the function is: ； in, is the winding hot spot temperature generated by the optimal internal thermal state estimation.

6. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 1 is characterized in that: The system diagnosis and adaptation module includes an online diagnosis unit, the workflow of which is constructed as follows: First, obtaining the model residual constructed by the Kalman filter state observation unit in the process of generating the optimal state estimate; Then, the statistical characteristics of the time series of the model residuals are analyzed; Finally, when the statistical characteristic deviates from a preset benchmark, it is determined that the system has experienced performance degradation.

7. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 6 is characterized in that: The system diagnosis and adaptation module further includes a model self-correction unit, which is triggered after the online diagnosis unit determines that the system has experienced performance degradation. The workflow is constructed as follows: constructing and solving a system identification problem using a set of recent real-time operating parameters; Finally, the solution of the system identification problem is used to generate the system parameter matrix in the dynamic thermodynamic model. The correction value is used to update the dynamic thermodynamic model.

8. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 7 is characterized in that: The system diagnosis and self-adaptation module workflow is constructed as follows: First, record the system parameter matrix generated by the model self-correction unit and The cumulative amount of corrections over the years; Then, a normalization function is applied to the Frobenius norm of the cumulative correction To generate a quantitative system health index , and its generation formula is: ； in, No. In subsystem identification, the system parameter matrix The correction amount, that is, the difference between the current estimate and the initial model value; No. In subsystem identification, the system parameter matrix The correction amount, that is, the difference between the current estimate and the initial model value; : Frobenius norm, used to measure the overall correction degree of a matrix, that is, the square root of the sum of the squares of the matrix elements; Indicates that all historical correction steps The sum of the Frobenius norms of the corrections, i.e. the total cumulative deviation of all corrections; Indicates that all historical correction steps The Frobenius norm of the correction quantity is summed.

9. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 1 is characterized in that: The workflow of the human-computer interaction and communication module is constructed as follows: First, receiving the optimal internal thermal state estimate generated by the state estimation and optimization control module and the system operating state determined by the system diagnosis and adaptation module, and graphically displaying them on its human-machine interface; Secondly, it is configured to receive an external power grid signal through its communication interface and providing the signal to the state estimation and optimization control module.

10. The transformer fan PLC control alarm system with integrated soft start and speed regulation according to claim 7, characterized in that: The state estimation and optimization control module further includes a collaborative strategy and dynamic weighting unit, the workflow of which is constructed as follows: First, the external power grid signal provided by the human-computer interaction and communication module is obtained Then the external grid signal Mapping to the cost function Medium weight matrix 、 or weight coefficient A new set of values ​​for Finally, the cost function is reconstructed using the new value , to adjust the optimization direction of the multi-objective optimization problem.

Images