An artificial intelligence-based power planning optimization method

Abstract

The present application relates to the technical field of data processing, and especially relates to a power planning optimization method based on artificial intelligence, which comprises the following steps: collecting data; determining abnormal events; predicting parameters; determining candidate shutdown equipment; determining high-risk equipment; determining target shutdown equipment; generating a power planning scheme; and adjusting thresholds. The present application collects multi-dimensional data in real time, determines abnormal events according to total power consumption, predicts load loss and residual time in combination with a preset model, determines high-risk equipment according to residual time, heat consumption rate, vibration frequency and rotor speed, determines target shutdown equipment according to active power, climbing rate and load loss of high-risk equipment, and adjusts thresholds dynamically according to total power consumption and rotor speed based on the power planning scheme, thereby effectively solving the problems of low precision of shutdown decision and response lag caused by the complexity of the operation background of the power system and excessive dependence on models.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an artificial intelligence-based method for power planning optimization. Background Technology

[0002] Large-scale shutdowns, production stoppages, and school closures due to various reasons have presented numerous challenges to the power system. On the one hand, rapid changes in electricity demand make traditional power planning and dispatching methods inadequate, potentially leading to wasted power resources or insufficient supply. On the other hand, the shutdown and restart of some generating equipment require careful decision-making to avoid impacting grid stability. Furthermore, how to optimize the operation of power equipment and reduce energy consumption and environmental pollution while ensuring basic electricity needs are met is also an urgent problem to be solved.

[0003] Chinese Patent Application Publication No. CN120387535A discloses a method for intelligent planning and optimization of power systems, comprising: S1: collecting operating status data of the power system, including electrical parameters, generation data, load data, and the charging and discharging status of energy storage devices; S2: based on the load data and historical load data collected in step S1, using an LSTM neural network to predict the load demand of the power system within a preset time range; S3: based on the predicted load demand, the operating status data, preset optimization objectives, and preset constraints, using a deep Q-network to construct a power dispatch model and obtain an optimal planning strategy, wherein the optimal planning strategy includes the output strategy of generator units, the charging and discharging strategy of energy storage systems, and the load adjustment strategy; S4: sending the optimal planning strategy to the dispatch center, and the dispatch center executing the optimal planning strategy.

[0004] Therefore, the aforementioned intelligent planning and optimization method for power systems has the following problems: based on load demand forecasting, it cannot promptly identify rapid changes in electricity demand, leading to waste of power resources or insufficient supply; relying on load data and historical load data, it lacks a multi-dimensional comprehensive assessment of equipment operating status; when electricity demand and equipment operating status change significantly in a short period of time, it is difficult to ensure the stability and adaptability of the power system under complex operating conditions. Summary of the Invention

[0005] To address this, the present invention provides an artificial intelligence-based power planning optimization method, which overcomes the problems of low accuracy and delayed response in outage decisions caused by the complexity of power system operation background and over-reliance on models in the prior art through multi-dimensional parameter analysis and dynamic adjustment mechanisms.

[0006] To achieve the above objectives, the present invention provides an artificial intelligence-based power planning optimization method, comprising:

[0007] Real-time data collection includes the total electricity consumption of the power system within the area under the power plant's responsibility, the heat rate, active power, rotor speed, ramp rate, and bearing vibration frequency of each device under test within the power plant.

[0008] An abnormal event is determined based on the total power consumption and the preset deviation threshold.

[0009] Based on the abnormal event, the load loss and remaining time are predicted according to the preset artificial intelligence model and the total power consumption.

[0010] Based on the remaining time, the heat consumption rate, and the vibration frequency, several candidate shutdown devices are determined;

[0011] A number of high-risk devices are determined based on the vibration frequency, rotor speed, and preset index threshold of each of the candidate shutdown devices.

[0012] Based on the active power, ramp rate, and load loss of each of the high-risk devices, a number of target shutdown devices are determined;

[0013] Generate a power planning scheme based on all the target shutdown equipment;

[0014] Based on the power planning scheme, the preset deviation threshold or the preset index threshold is adjusted according to the total power consumption and the rotor speed within the preset adjustment period.

[0015] Furthermore, the process of determining the occurrence of an abnormal event based on the total power consumption and the preset deviation threshold includes:

[0016] Calculate the electricity consumption deviation based on the total electricity consumption and the historical total electricity consumption;

[0017] The decrease in electricity demand is determined by comparing the electricity consumption deviation with the preset deviation threshold, thereby determining that the abnormal event has occurred.

[0018] Furthermore, the process of determining several candidate shutdown devices based on the remaining time, the heat dissipation rate, and the vibration frequency includes:

[0019] The occurrence of a shutdown requirement event is determined based on the comparison between the remaining duration and the preset downtime duration.

[0020] Based on the shutdown demand event, several candidate shutdown devices are determined according to the heat dissipation rate and the vibration frequency.

[0021] Furthermore, the process of determining several candidate shutdown devices based on the heat dissipation rate and the vibration frequency includes:

[0022] Calculate the rate of change of all heat loss rates from the initial time to each time within the preset determination period to obtain several rates of change of heat loss rates;

[0023] Calculate the rate of change of all vibration frequencies from the initial moment to each moment within the preset determination time to obtain several frequency change rates;

[0024] Several candidate shutdown devices are determined based on all the heat loss rate change rates and all the frequency change rates.

[0025] Furthermore, the process of determining a number of candidate shutdown devices based on all the said heat loss rate change rates and all the said frequency change rates includes:

[0026] Calculate the energy-frequency synergy based on all the heat rate change rates and all the frequency change rates;

[0027] Based on the comparison result between the energy frequency coordination degree and the preset coordination degree threshold, the device under test is determined to be the candidate shutdown device, and several candidate shutdown devices are obtained.

[0028] Furthermore, the process of determining a number of high-risk devices based on the vibration frequency, rotor speed, and preset index threshold of each candidate shutdown device includes:

[0029] Calculate the frequency fluctuation value based on the vibration frequency within a preset time period;

[0030] Calculate the speed fluctuation value based on the rotor speed within the preset time period;

[0031] Several high-risk devices are identified based on the frequency fluctuation value, the rotational speed fluctuation value, and a preset index threshold.

[0032] Furthermore, the process of determining a number of high-risk devices based on the frequency fluctuation value, the rotational speed fluctuation value, and a preset index threshold includes:

[0033] Calculate the risk index based on the frequency fluctuation value and the rotational speed fluctuation value;

[0034] Based on the comparison between the risk index and the preset index threshold, the candidate shutdown equipment is determined to be a high-risk equipment, thereby identifying a number of high-risk equipment.

[0035] Furthermore, the process of determining a number of target shutdown devices based on the active power, ramp rate, and load loss of each of the high-risk devices includes:

[0036] The total capacity that can be reduced is calculated based on the active power of all the aforementioned high-risk devices;

[0037] When the total available capacity is greater than the load loss, a reduction list is determined based on the risk index and ramp rate of each high-risk device, and several target shutdown devices are determined based on the reduction list and the active power of each high-risk device.

[0038] When the total reducible capacity is less than or equal to the load loss, all the high-risk equipment is identified as the target shutdown equipment, and the remaining loss load is calculated based on the total reducible capacity and the load loss. The target shutdown equipment is determined based on the remaining loss load, the risk index of each temporary equipment, the ramp rate, and the active power.

[0039] The temporary equipment refers to the equipment other than the high-risk equipment among the candidate shutdown equipment.

[0040] Furthermore, the process of determining the target shutdown equipment based on the remaining loss load, the risk index of each temporary device, the ramp rate, and the active power includes:

[0041] Several matching degrees are calculated based on the active power and the remaining loss load of each of the temporary devices;

[0042] When there is a matching degree greater than zero and less than a preset matching degree threshold, the corresponding temporary device is determined to be the target shutdown device;

[0043] When there is no matching degree greater than zero and less than the preset matching degree threshold, a temporary list is determined based on the risk index and ramp rate of each temporary device, and several target shutdown devices are determined based on the temporary list, the active power, and the remaining loss load.

[0044] Furthermore, the process of adjusting the preset deviation threshold or the preset index threshold based on the total power consumption and the rotor speed within a preset adjustment period includes:

[0045] Calculate the electricity consumption deviation fluctuation value based on the total electricity consumption within the preset adjustment period and the historical total electricity consumption.

[0046] The preset deviation threshold is adjusted based on the first comparison result between the electricity consumption deviation fluctuation value and the preset fluctuation threshold.

[0047] Based on the second comparison result between the power consumption deviation fluctuation value and the preset fluctuation threshold, the rotational speed change rate is calculated according to the rotor speed within the preset adjustment period;

[0048] The preset index threshold is adjusted based on the comparison between the speed change rate and the preset speed change rate threshold.

[0049] Compared with existing technologies, the advantages of this invention lie in its ability to collect multi-dimensional data in real time, which are interconnected and collectively reflect the operating status of the power system. Comparison of total electricity consumption with a preset deviation threshold can quickly determine the occurrence of abnormal events. Based on these abnormal events, combined with a preset artificial intelligence model and the predicted load loss and remaining time based on total electricity consumption, a time basis and load demand reference are provided for subsequent shutdown decisions. The comprehensive consideration of remaining time, heat rate, and vibration frequency is used to determine candidate shutdown equipment. This is because the heat rate reflects the energy utilization efficiency of the equipment, and the vibration frequency is related to the mechanical stability of the equipment. Combining the remaining time allows for the selection of equipment suitable for shutdown under the current abnormal conditions. Furthermore, high-risk equipment is identified through vibration frequency, rotor speed, and a preset index threshold. Fluctuations in rotor speed and abnormal vibration frequency often indicate potential equipment failure risks. The preset index threshold provides a quantitative standard for risk assessment, thereby accurately identifying high-risk equipment. Ultimately, the target shutdown equipment is determined based on the active power, ramp rate, and load loss of high-risk equipment. Active power determines the equipment's contribution to the load, and ramp rate reflects the equipment's adjustment flexibility. By combining these factors, it is possible to rationally determine the equipment that should be shut down while meeting load demand, resulting in a more scientific and feasible power planning scheme. Furthermore, based on the power planning scheme, preset deviation thresholds or preset index thresholds are dynamically adjusted according to the total electricity consumption and rotor speed within a preset adjustment period. This enables continuous monitoring and optimization of the power system's operating status, enhancing the power system's adaptability and stability under complex operating conditions. It effectively solves the problems of low accuracy and delayed response in shutdown decisions caused by the complexity of the power system's operating background and over-reliance on models.

[0050] Furthermore, by combining total electricity consumption with historical total electricity consumption to calculate the electricity consumption deviation, and comparing this deviation with a preset deviation threshold to determine abnormal events, the system addresses several key points. First, total electricity consumption directly reflects the operating status of the power system, while historical total electricity consumption provides a reference benchmark under normal operating conditions. Calculating the electricity consumption deviation between these two metrics accurately captures the difference between current electricity demand and historical norms. This difference may indicate changes in the power system's operating environment or potential anomalies. Comparing the electricity consumption deviation with the preset deviation threshold quantitatively determines whether the change in electricity demand has reached an abnormal level, thus enabling precise identification of abnormal events. This not only allows for the timely detection of reduced electricity demand but also avoids misjudgments caused by random fluctuations or changes within the normal range, improving the accuracy and reliability of abnormal event identification.

[0051] Furthermore, candidate shutdown equipment is determined by comprehensively considering remaining time, heat dissipation rate, and vibration frequency. First, the preset shutdown time is pre-set based on the equipment's characteristics and operational requirements to ensure a smooth shutdown within a safe timeframe. If the remaining time is less than or equal to this preset value, the system must immediately take measures to initiate the shutdown procedure to prevent the equipment from continuing to operate under abnormal events, thereby reducing potential failure risks and energy waste. Based on this, equipment evaluation is conducted using two key parameters: heat dissipation rate and vibration frequency. The heat dissipation rate reflects the equipment's energy utilization efficiency; a lower heat dissipation rate means less energy waste after shutdown. Vibration frequency is related to the equipment's mechanical stability; abnormal vibration frequencies may indicate unstable equipment operation, and shutdown can reduce the risk of equipment failure.

[0052] Furthermore, candidate shutdown equipment is determined by calculating the rate of change of heat dissipation rate and the rate of change of vibration frequency within a preset judgment period. The rate of change of heat dissipation rate reflects the change in energy utilization efficiency of the equipment within a specific time period, which can help identify equipment with declining energy efficiency. These devices may contribute more to energy savings when shut down. The rate of change of vibration frequency reflects the change in the mechanical stability of the equipment over time. Abnormal changes in vibration frequency may indicate instability in the operating state of the equipment, and shutting down these devices can reduce the risk of failure.

[0053] Furthermore, by calculating the energy-frequency synergy between changes in heat rate and vibration frequency, candidate shutdown equipment is determined. Based on the inherent temporal correlation between equipment efficiency degradation and mechanical condition deterioration, i.e., when equipment efficiency decreases (heat rate increases) due to component wear or failure, it is often accompanied by abnormal vibration (vibration frequency change). The more synchronized the trends of the two changes, the more it indicates that the equipment is in an uneconomical and unsafe operating state. Max-min normalization processing eliminates the dimensional differences of the original data, making different monitoring parameters comparable. Then, cosine similarity is used to quantitatively evaluate this synchrony. Finally, preset thresholds are used to accurately screen out equipment with low energy efficiency and poor reliability. This achieves a leap from judging single parameter exceedances to multi-parameter synergistic trend analysis, thereby prioritizing the shutdown of the equipment with the worst overall performance under the premise of ensuring grid safety, achieving the dual optimization goals of economy and safety.

[0054] Furthermore, by calculating the fluctuation values ​​of vibration frequency and rotor speed, and combining them with preset index thresholds, high-risk equipment is identified. Vibration frequency and rotor speed are key parameters reflecting the mechanical operating state of equipment, and their fluctuation values ​​can intuitively reflect the stability of the equipment during operation. Abnormal fluctuations in vibration frequency may indicate the risk of mechanical failure, while fluctuations in rotor speed may affect the power generation efficiency and stability of the equipment. By calculating the fluctuation values ​​of these two parameters, the operational risk of the equipment can be quantitatively assessed. Comparing these fluctuation values ​​with preset index thresholds allows for the scientific screening of equipment with unstable operating states and high failure risks as high-risk equipment.

[0055] Furthermore, high-risk equipment is identified by calculating a risk index and comparing it with a preset index threshold. Frequency fluctuation and rotational speed fluctuation reflect the stability of equipment under vibration and rotational operation, respectively, and the fluctuation of these two parameters are important indicators for assessing equipment failure risk. By comprehensively calculating the risk index from both frequency and rotational speed fluctuation values, the operational risk of equipment can be quantified more comprehensively, avoiding the one-sidedness that may result from single-parameter assessment. This allows for a more accurate identification of equipment with potential problems in mechanical operation and power generation efficiency. Comparing the risk index with the preset index threshold enables the scientific screening of high-risk equipment, thereby preventing equipment failures in advance while ensuring the stable operation of the power system.

[0056] Furthermore, by prioritizing the supply and demand relationship between the total reducible capacity and the load loss, two decision paths are intelligently differentiated. When capacity is sufficient, i.e., the total reducible capacity is greater than the load loss, a multi-attribute decision model is adopted. The risk index, which represents the health of the equipment, and the ramp rate, which represents the system's adjustment capability, are normalized and weighted to obtain a reduction index. This index is then arranged in descending order to form a shutdown priority list. Finally, the power is accumulated sequentially to just meet the load demand. This ensures that, under the premise of meeting the reduction target, high-risk equipment is prioritized for shutdown to achieve preventive maintenance, while high ramp rate units are retained to maintain system flexibility. When capacity is insufficient, i.e., the total reducible capacity is less than or equal to the load loss, all high-risk equipment is shut down first to maximize safety benefits. Then, the remaining load is calculated and supplemented from temporary equipment. This achieves a dynamic balance between safety and economy, avoiding the lack of flexibility caused by excessive shutdowns while ensuring the completion of critical load reduction tasks.

[0057] Furthermore, by calculating the matching degree, a quantitative correlation between equipment capacity and system requirements is directly established. Priority is given to finding target shutdown equipment that can independently and accurately fill the power gap (with the smallest matching degree and greater than zero), which greatly reduces operational complexity. When there is no single optimal solution, a multi-attribute decision-making process is initiated. The risk index representing the safety status and the ramp rate representing the adjustment capability are normalized and weighted and integrated. The shutdown priority is constructed by arranging the temporary index in descending order. This ensures that when combined shutdowns are necessary, the equipment that should be shut down under comprehensive evaluation (i.e., the equipment with the highest risk and the strongest adjustment capability) is selected first. This satisfies both capacity requirements and achieves the dual goals of safe shutdown and preservation of system flexibility.

[0058] Furthermore, by calculating the fluctuation value of electricity consumption deviation, the stability of external grid demand is perceived. When the fluctuation is large, it indicates that the system is in an unstable state. In this case, proportionally reducing the preset deviation threshold can lower the threshold for judging abnormal events, making the system more sensitive to load changes and thus triggering optimized scheduling more quickly to cope with uncertainties. When the electricity consumption fluctuation is within the normal range, the mechanical operation stability of the generator equipment is further evaluated by the rotor speed change rate. When the speed change rate is small, it indicates that the equipment is in a stable operating state. In this case, proportionally increasing the preset index threshold can raise the judgment standard for high-risk equipment, avoid excessive warnings about equipment status under stable operating conditions, reduce unnecessary shutdown operations, and dynamically optimize its decision sensitivity according to actual operating conditions. This improves the response capability during disturbances and ensures the economic efficiency of operation during stable periods, achieving long-term adaptive optimization. Attached Figure Description

[0059] Figure 1 This is a flowchart of the artificial intelligence-based power planning optimization method in this embodiment;

[0060] Figure 2 This is a logic diagram for determining the occurrence of abnormal events in this embodiment;

[0061] Figure 3 This is the logic diagram for determining candidate shutdown equipment in this embodiment;

[0062] Figure 4 The logic diagram for determining high-risk devices in this embodiment is shown. Detailed Implementation

[0063] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0064] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0065] Please see Figure 1 The diagram shows a flowchart of an AI-based power planning optimization method according to this embodiment. This embodiment provides an AI-based power planning optimization method, including: real-time collection of the total electricity consumption of the power system within the area managed by the power plant, the heat rate, active power, rotor speed, ramp rate, and bearing vibration frequency of each device to be tested within the power plant; determining abnormal events based on the total electricity consumption and a preset deviation threshold; predicting load loss and remaining time based on the abnormal events using a preset AI model and the total electricity consumption; determining several candidate shutdown devices based on the remaining time, heat rate, and vibration frequency; identifying several high-risk devices based on the vibration frequency, rotor speed, and a preset index threshold of each candidate shutdown device; determining several target shutdown devices based on the active power, ramp rate, and load loss of each high-risk device; generating a power planning scheme based on all the target shutdown devices; and adjusting the preset deviation threshold or the preset index threshold based on the power planning scheme and the total electricity consumption and rotor speed within a preset adjustment period.

[0066] In this embodiment, large-scale shutdowns of work, production, and schools frequently occur due to various reasons, such as economic restructuring, industrial upgrading, environmental protection requirements, and responses to public emergencies. These scenarios place special demands on the operation of the power system. Large-scale shutdowns of work, production, and schools can lead to a rapid decline in electricity demand, resulting in a waste of power resources. Even surplus electricity can affect the operation of generator sets and reduce their lifespan. Therefore, precise data analysis can optimize the shutdown decisions of power equipment to reduce load losses and ensure the stable operation of the power system.

[0067] In this embodiment, the power planning scheme covers multiple aspects, from the selection of shutdown equipment, the formulation of shutdown sequence and schedule, load adjustment strategies, equipment operation adjustments, dynamic adjustment mechanisms, and emergency plans. Through this scheme, the power system can scientifically and efficiently adjust its operating status in the event of abnormal events such as large-scale work stoppages, production shutdowns, and school closures, ensuring the stable and reliable operation of the power system, while optimizing energy utilization efficiency and reducing environmental pollution.

[0068] In this embodiment, the power planning scheme is a structured document, including but not limited to: (1) a brief description of the types of abnormal events that are triggered (predicting a sharp drop in regional electricity demand); (2) a clear description of the predicted load loss and remaining time; (3) a determination of the shutdown sequence of each target shutdown equipment based on the active power, ramp rate and load loss of the equipment, prioritizing the shutdown of equipment that has little impact on the load and high adjustment flexibility; (4) a detailed shutdown schedule based on the remaining time and load loss, ensuring that the specific shutdown time of each equipment is completed within the remaining time, while avoiding impact on the stability of the power grid; (5) after shutting down some equipment, redistributing the load based on the active power and ramp rate of the remaining equipment to ensure that the power supply demand of important users is not affected and to maintain the stable operation of the power system; (6) key monitoring of identified high-risk equipment to ensure that the operating status is stable before shutdown and to avoid sudden failures; (7) an emergency plan for dealing with sudden equipment failures to ensure that measures can be taken quickly when a failure occurs during shutdown to reduce the impact on the power system.

[0069] In this embodiment, total electricity consumption refers to the sum of electrical energy consumed by all users (industrial, commercial, residential, etc.) within the power system of the power plant's area of ​​responsibility at a certain moment, reflecting the real-time demand of the power grid. This data is collected through smart meters installed at the power plant. Heat rate refers to the heat consumed by the power generation equipment to produce one kilowatt-hour of electricity. A lower heat rate indicates higher efficiency in converting fuel into electricity, resulting in more economical operation. This is calculated by collecting data on the unit's fuel consumption, lower heating value of fuel, and output active power through the plant-level monitoring information system. Active power refers to the power actually generated, used for work, and consumed by the power generation equipment. It directly determines the generator's output. This is measured by current and voltage transformers installed at the generator outlet, and calculated by a power transmitter. Rotor speed refers to the rotational speed of the generator set's rotor. Fluctuations in rotor speed directly reflect the power balance between the prime mover and the generator, and are a key parameter for assessing the unit's operational stability. This is measured by a magnetoresistive sensor installed on the rotor shaft. The ramp rate refers to the amount of power output that a generator set can increase or decrease per unit time. It reflects the unit's flexibility and adjustment capability in responding to dispatch instructions and adapting to load changes. It is calculated by analyzing the active power change curve (dP / dt) in historical data. The bearing vibration frequency refers to the frequency distribution of the vibration signal generated by the bearings of the power generation equipment during operation. It is the most important indicator for equipment fault prediction and health management, and is collected by vibration acceleration sensors installed on the bearing housing.

[0070] In this embodiment, load loss refers to the total electricity load that the power grid is predicted to lose over a future period due to large-scale work stoppages and production shutdowns. Remaining time refers to the time remaining from the current moment until the predicted load loss reaches its maximum. Both load loss and remaining time are predicted using a pre-set artificial intelligence model.

[0071] The pre-defined artificial intelligence model is a Long Short-Term Memory (LSTM) network model. By using the LSTM model, time series data can be effectively processed to capture the changing trends of total electricity consumption and historical total electricity consumption over time, thereby accurately predicting load loss and remaining duration. This model structure and training method ensure that the model has good adaptability and predictive ability when faced with complex time series data, providing a scientific basis for handling abnormal events in power systems.

[0072] 1. Initial parameters

[0073] Input features: total electricity consumption, historical total electricity consumption, timestamp.

[0074] Output: Load loss (unit: kilowatt-hour, kWh), remaining duration (unit: hour, h).

[0075] Model structure:

[0076] Input layer: Receives total power consumption, historical total power consumption, and timestamp data.

[0077] Hidden layer: Contains multiple LSTM units, responsible for capturing time series features and handling temporal dependencies.

[0078] Output layer: Outputs load loss and remaining duration.

[0079] Hyperparameters: Learning rate: 0.001; Batch size: 32; Training cycles: 100 epochs.

[0080] 2. Training methods

[0081] Data preparation: Collect historical electricity data, including total electricity consumption, historical total electricity consumption, timestamps, and corresponding load loss and remaining duration labels. Load loss and remaining duration labels can be determined by expert assessment or calculated based on historical data.

[0082] Data preprocessing: Normalize the data to fit within the range of 0 to 1 to accelerate model convergence.

[0083] Model training: Train the LSTM model using training data, and update the model parameters using optimization algorithms (such as Adam) to minimize the error between the predicted and actual values.

[0084] Validation and Testing: Use the validation set to tune hyperparameters and prevent overfitting. Use the test set to evaluate the model's performance and ensure it has good generalization ability.

[0085] 3. The model after training

[0086] Model saving: After training is complete, save the model's parameters and structure for later use.

[0087] Model evaluation: Record the model's performance metrics on the training, validation, and test sets, such as mean squared error (MSE) and coefficient of determination (R²), to assess the model's accuracy and reliability.

[0088] The preset deviation threshold is a standard value used to determine whether there has been an abnormal change in current electricity demand. It depends on the operational stability requirements of the power system, load fluctuation characteristics, and sensitivity to abnormal events, and is usually set between 5% and 10%. In this embodiment, it is set to 8%, which can effectively distinguish between normal fluctuations and abnormal events, avoid misjudging small normal fluctuations as abnormal events, and at the same time, can promptly capture large changes in electricity demand, ensuring the stable operation of the power system.

[0089] The preset index threshold is a critical value used to identify truly high-risk devices from candidate devices. It depends on the statistical analysis of historical operating status and risk control strategies, and is usually set between 0.6 and 0.8. In this embodiment, it is set to 0.75 to ensure that the system only issues alerts to devices with a high degree of risk confidence, thereby optimizing the allocation of operation and maintenance resources.

[0090] The preset adjustment duration is the length of the historical data time window used to adjust parameters. It depends on the periodic characteristics of load changes and the stability required for adaptive adjustment of system parameters, and is usually set between 4 and 24 hours. In this embodiment, it is set to 12 hours, which can effectively cover the peak and valley changes of daily load, so that the calculated electricity consumption deviation fluctuation value has sufficient statistical significance.

[0091] By collecting multi-dimensional data in real time, these interconnected data collectively reflect the operating status of the power system. Comparing total electricity consumption with a preset deviation threshold quickly identifies abnormal events. Based on these events, and combining a preset artificial intelligence model with total electricity consumption to predict load loss and remaining time, a time basis and load demand reference are provided for subsequent shutdown decisions. A comprehensive consideration of remaining time, heat rate, and vibration frequency is used to determine candidate shutdown equipment. This is because heat rate reflects the energy utilization efficiency of the equipment, while vibration frequency is related to the mechanical stability of the equipment. Combining remaining time allows for the selection of equipment suitable for shutdown under the current abnormal conditions. Furthermore, high-risk equipment is identified through vibration frequency, rotor speed, and a preset index threshold. Fluctuations in rotor speed and abnormal vibration frequency often indicate potential equipment failure risks. The preset index threshold provides a quantitative standard for risk assessment, thereby accurately identifying high-risk equipment. Ultimately, the target shutdown equipment is determined based on the active power, ramp rate, and load loss of high-risk equipment. Active power determines the equipment's contribution to the load, and ramp rate reflects the equipment's adjustment flexibility. By combining these factors, it is possible to rationally determine the equipment that should be shut down while meeting load demand, resulting in a more scientific and feasible power planning scheme. Furthermore, based on the power planning scheme, preset deviation thresholds or preset index thresholds are dynamically adjusted according to the total electricity consumption and rotor speed within a preset adjustment period. This enables continuous monitoring and optimization of the power system's operating status, enhancing the power system's adaptability and stability under complex operating conditions. It effectively solves the problems of low accuracy and delayed response in shutdown decisions caused by the complexity of the power system's operating background and over-reliance on models.

[0092] Please see Figure 2 As shown, this is a logic diagram for determining the occurrence of an abnormal event in this embodiment. In this embodiment, the process of determining the occurrence of an abnormal event based on the total power consumption and the preset deviation threshold includes: calculating the relative deviation between the total power consumption and the historical total power consumption to obtain the power consumption deviation; when the power consumption deviation is greater than the preset deviation threshold, determining that the power demand has decreased, so as to determine that the abnormal event has occurred.

[0093] Historical total electricity consumption refers to the total electricity consumption of the power system under normal operating conditions within the same time period. It depends on the average electricity consumption data over a past period (such as the past week, month, or year) and is used as a reference benchmark for current electricity consumption. In this embodiment, the historical total electricity consumption is selected as the average daily electricity consumption over the past half month, which can smooth out short-term fluctuations and reflect long-term electricity consumption trends.

[0094] By combining total electricity consumption with historical total electricity consumption to calculate the electricity consumption deviation, and comparing this deviation with a preset deviation threshold to determine abnormal events, the system addresses several key issues. First, total electricity consumption directly reflects the operating status of the power system, while historical total electricity consumption provides a reference benchmark under normal operating conditions. Calculating the electricity consumption deviation between these two metrics accurately captures the difference between current electricity demand and historical norms. This difference may indicate changes in the power system's operating environment or potential anomalies. Comparing the electricity consumption deviation with the preset deviation threshold quantitatively determines whether the change in electricity demand has reached an abnormal level, thus enabling precise identification of abnormal events. This not only allows for the timely detection of reduced electricity demand but also avoids misjudgments caused by random fluctuations or changes within the normal range, improving the accuracy and reliability of abnormal event identification.

[0095] Specifically, the process of determining a number of candidate shutdown devices based on the remaining time, the heat dissipation rate, and the vibration frequency includes: determining that a shutdown demand event has occurred when the remaining time is less than or equal to a preset shutdown time; and determining a number of candidate shutdown devices based on the shutdown demand event, according to the heat dissipation rate and the vibration frequency.

[0096] The preset downtime refers to the time required for equipment downtime set in advance based on the characteristics and operational needs of the equipment. It depends on the type of equipment, its operating status, maintenance requirements, and the overall operating strategy of the power system, and is typically set between 1 and 4 hours. In this embodiment, it is set to 2 hours, which ensures that the equipment has sufficient time to safely shut down in the event of an abnormal event, while also avoiding unnecessary energy waste and equipment wear, thereby improving the operating efficiency and reliability of the power system.

[0097] Candidate shutdown equipment is selected by comprehensively considering remaining time, heat dissipation rate, and vibration frequency. First, the preset shutdown time is pre-set based on the equipment's characteristics and operational requirements to ensure a smooth shutdown within a safe timeframe. If the remaining time is less than or equal to this preset value, the system must immediately take measures to initiate the shutdown procedure to prevent continued operation under abnormal events, thereby reducing potential failure risks and energy waste. Based on this, equipment evaluation is conducted using two key parameters: heat dissipation rate and vibration frequency. The heat dissipation rate reflects the equipment's energy utilization efficiency; a lower rate means less energy waste after shutdown. Vibration frequency is related to the equipment's mechanical stability; abnormal vibration frequencies may indicate unstable operation, and shutdown can reduce the risk of equipment failure.

[0098] Specifically, the process of determining several candidate shutdown devices based on the heat dissipation rate and the vibration frequency includes: calculating the rate of change of all the heat dissipation rates from the initial time to each time within a preset determination period to obtain several heat dissipation rate change rates; calculating the rate of change of all the vibration frequencies from the initial time to each time within the preset determination period to obtain several frequency change rates; and determining several candidate shutdown devices based on the total heat dissipation rate change rate and the total frequency change rate.

[0099] Formula for calculating the rate of change of heat loss rate: ;

[0100] Among them, ΔHR t HR is the rate of change of heat rate at time t relative to the heat rate at the initial time; HR0 is the heat rate value at the initial time within the preset judgment duration; HR t It is the heat dissipation rate value at the t-th time point (t=1, 2, 3, ..., n) after the initial time; the preset judgment duration is T minutes; the data sampling interval is Δt minutes; n=T / Δt, that is, the total number of data points in the whole duration minus one.

[0101] Formula for calculating the rate of change of vibration frequency: ;

[0102] Wherein, ΔVF t VF is the rate of change of the vibration frequency at time t relative to the vibration frequency at the initial time; VF0 is the heat dissipation rate at the initial time within the preset judgment duration; VF t It is the heat dissipation rate value at the t-th time point (t=1, 2, 3, ..., n) after the initial time; the preset judgment duration is T minutes; the data sampling interval is Δt minutes; n=T / Δt, that is, the total number of data points in the whole duration minus one.

[0103] The preset judgment duration is the time length used to calculate the rate of change of heat dissipation rate and frequency. It depends on the thermal inertia of the equipment and the minimum observation window required for the system to identify trends, and is usually set between 30 minutes and 5 hours. In this embodiment, it is set to 60 minutes, which can effectively filter out instantaneous noise in the data, capture the trend that can truly reflect the continuous deterioration of equipment efficiency, and synchronize with the common dispatch cycle of the power grid, ensuring the stability of the analysis results and the timeliness of decision-making.

[0104] Candidate equipment for shutdown is determined by calculating the rate of change of heat dissipation rate and the rate of change of vibration frequency within a preset judgment period. The rate of change of heat dissipation rate reflects the change in energy utilization efficiency of the equipment over a specific time period, helping to identify equipment with declining energy efficiency. These devices may contribute more to energy savings when shut down. The rate of change of vibration frequency reflects the change in the mechanical stability of the equipment over time. Abnormal changes in vibration frequency may indicate instability in the operating state of the equipment, and shutting down these devices can reduce the risk of failure.

[0105] Please see Figure 3 As shown, this is the logic diagram for determining candidate shutdown devices in this embodiment. In this embodiment, the process of determining several candidate shutdown devices based on all the heat rate change rates and all the frequency change rates includes: performing maximum-minimum normalization on all the heat rate change rates to obtain several heat rate change normalization values; performing maximum-minimum normalization on all the frequency change rates to obtain several frequency change normalization values; calculating the cosine similarity between all the heat rate change normalization values ​​and all the frequency change normalization values ​​to obtain the energy-frequency synergy; when the energy-frequency synergy is greater than a preset synergy threshold, determining the device under test as a candidate shutdown device, thus obtaining several candidate shutdown devices.

[0106] The preset coordination threshold is a standard value used to determine the degree of coordination between the energy and frequency responses of devices through cosine similarity. It depends on the system's optimization strategy and historical data analysis, and is usually set between 0.5 and 0.8. In this embodiment, it is set to 0.7, which can effectively distinguish between normal fluctuations in equipment and true inefficient operating trends, thereby ensuring the reliability of the candidate shutdown equipment list.

[0107] Candidate shutdown equipment is determined by calculating the energy-frequency synergy between changes in heat rate and vibration frequency. Based on the inherent temporal correlation between equipment efficiency degradation and mechanical condition deterioration, when equipment efficiency decreases (heat rate increases) due to component wear or failure, it is often accompanied by abnormal vibration (vibration frequency change). The more synchronized the trends of the two changes, the more uneconomical and unsafe the equipment is in operation. Max-min normalization eliminates the dimensional differences of the original data, making different monitoring parameters comparable. Cosine similarity is then used to quantitatively evaluate this synergy. Finally, preset thresholds are used to accurately screen out equipment with low energy efficiency and poor reliability. This achieves a leap from judging single parameter exceedances to multi-parameter synergistic trend analysis, thereby prioritizing the shutdown of the equipment with the worst overall performance while ensuring grid safety, achieving the dual optimization goals of economy and safety.

[0108] Specifically, the process of determining a number of high-risk devices based on the vibration frequency, rotor speed, and preset index threshold of each candidate shutdown device includes: calculating the standard deviation of all vibration frequencies within a preset time period to obtain a frequency fluctuation value; calculating the standard deviation of all rotor speeds within the preset time period to obtain a speed fluctuation value; and determining a number of high-risk devices based on the frequency fluctuation value, the speed fluctuation value, and the preset index threshold.

[0109] High-risk equipment is identified by calculating the fluctuations in vibration frequency and rotor speed, and combining these values ​​with preset index thresholds. Vibration frequency and rotor speed are key parameters reflecting the mechanical operating status of equipment, and their fluctuations directly reflect the stability of the equipment during operation. Abnormal fluctuations in vibration frequency may indicate a risk of mechanical failure, while fluctuations in rotor speed may affect the equipment's power generation efficiency and stability. By calculating the fluctuations of these two parameters, the operational risk of the equipment can be quantitatively assessed. Comparing these fluctuation values ​​with preset index thresholds allows for the scientific screening of equipment with unstable operating conditions and a high risk of failure as high-risk equipment.

[0110] Please see Figure 4 The diagram shown illustrates the logic for determining high-risk equipment in this embodiment. In this embodiment, the process of determining several high-risk equipment based on the frequency fluctuation value, the rotational speed fluctuation value, and a preset index threshold includes: performing maximum-minimum normalization on the frequency fluctuation value to obtain a normalized frequency fluctuation value; performing maximum-minimum normalization on the rotational speed fluctuation value to obtain a normalized rotational speed fluctuation value; performing a weighted summation of the normalized frequency fluctuation value, the normalized rotational speed fluctuation value, a preset frequency weight, and a preset rotational speed weight to obtain a risk index; and determining the candidate shutdown equipment as the high-risk equipment when the risk index is greater than the preset index threshold, thereby identifying several high-risk equipment.

[0111] The preset frequency weight is a coefficient reflecting the relative importance of vibration frequency fluctuation indicators when assessing equipment risk. It depends on the criticality of vibration frequency in predicting mechanical failures (such as bearing damage or rotor imbalance) and the correlation analysis of historical data, and is usually set between 0.5 and 0.8. In this embodiment, it is set to 0.7 to highlight the core role of vibration monitoring in the assessment of equipment mechanical health, thereby improving the accuracy and sensitivity of risk identification.

[0112] The preset speed weight is a coefficient reflecting the relative importance of rotor speed fluctuation when assessing equipment risk. It depends on the criticality of speed fluctuation in indicating problems with the equipment's speed control system or load anomalies, and is usually set between 0.2 and 0.5. In this embodiment, it is set to 0.3, aiming to complement the frequency weight and sum to 1, thus constructing a comprehensive evaluation system that is both primary and secondary and balanced.

[0113] High-risk equipment is identified by calculating a risk index and comparing it with a preset threshold. Frequency fluctuation and rotational speed fluctuation reflect the stability of equipment under vibration and rotational operation, respectively, and the fluctuation of these two parameters are important indicators for assessing equipment failure risk. By comprehensively calculating the risk index from both frequency and rotational speed fluctuations, the operational risk of equipment can be quantified more comprehensively, avoiding the one-sidedness that may result from single-parameter assessment. This allows for a more accurate identification of equipment with potential problems in mechanical operation and power generation efficiency. Comparing the risk index with a preset threshold enables the scientific screening of high-risk equipment, thereby preventing equipment failures in advance while ensuring the stable operation of the power system.

[0114] Specifically, the process of determining several target shutdown devices based on the active power, ramp rate, and load loss of each of the high-risk devices includes: calculating the total reducible capacity based on the active power of all the high-risk devices; when the total reducible capacity is greater than the load loss, normalizing the risk index of each of the high-risk devices to obtain a first index normalized value, normalizing the ramp rate of each high-risk device to obtain a first rate normalized value, and weighting and summing the first index normalized value, the first rate normalized value, the preset index weight, and the preset rate weight to obtain a reduction index; and sorting the reduction indices of each high-risk device in descending order to obtain a reduction index. A reduction list is created, and the active power of each high-risk device is summed sequentially according to the order of the reduction list to obtain a first power sum. When the first power sum is greater than or equal to the load loss for the first time, the corresponding high-risk device is determined as the target shutdown device, thus identifying several target shutdown devices. When the total reducible capacity is less than or equal to the load loss, all the high-risk devices are determined as the target shutdown devices, and the difference between the total reducible capacity and the load loss is calculated to obtain the remaining loss load. The target shutdown devices are determined based on the remaining loss load, the risk index of each temporary device, the ramp rate, and the active power. The temporary devices are the devices other than the high-risk devices among the candidate shutdown devices.

[0115] The preset index weight is a factor used to measure the relative importance of equipment health status in shutdown decisions. It depends on the system operator's trade-off strategy between equipment safety and system flexibility, and is typically set between 0.6 and 0.8. In this embodiment, it is set to 0.7, aiming to prioritize high-risk equipment in shutdown targets, thereby effectively reducing the overall system operational risk and achieving a balance between safety and economy.

[0116] The preset rate weight is a factor used to measure the relative importance of equipment regulation flexibility in shutdown decisions. It depends on the system's reliance on rapid regulation capabilities to cope with future uncertainties and is typically set between 0.2 and 0.4. In this embodiment, it is set to 0.3 to ensure that, while meeting capacity reduction and safety objectives, the allocation of system resources is optimized as much as possible, reserving the necessary regulation margin for the stable operation of the power grid.

[0117] By prioritizing the supply and demand relationship between the total reducible capacity and load loss, two decision paths are intelligently differentiated. When capacity is sufficient, i.e., the total reducible capacity is greater than the load loss, a multi-attribute decision model is adopted. The risk index, which represents equipment health, and the ramp rate, which represents system regulation capability, are normalized and weighted to obtain a reduction index. This index is then arranged in descending order to form a shutdown priority list. Finally, the power is accumulated sequentially to meet the load demand. This ensures that, under the premise of meeting the reduction target, high-risk equipment is prioritized for shutdown to achieve preventive maintenance, while high ramp rate units are retained to maintain system flexibility. When capacity is insufficient, i.e., the total reducible capacity is less than or equal to the load loss, all high-risk equipment is shut down first to maximize safety benefits. Then, the remaining load is calculated and supplemented from temporary equipment. This achieves a dynamic balance between safety and economy, avoiding the lack of flexibility caused by excessive shutdowns and ensuring the completion of critical load reduction tasks.

[0118] Specifically, the process of determining the target shutdown equipment based on the remaining loss load, the risk index of each temporary device, the ramp rate, and the active power includes: calculating the difference between the active power and the remaining loss load of each temporary device to obtain several matching degrees; when there is a matching degree greater than zero and less than a preset matching degree threshold, the corresponding temporary device is determined to be the target shutdown equipment, wherein the target shutdown equipment is the temporary device corresponding to the one with the smallest matching degree among the conditions; when there is no matching degree greater than zero and less than the preset matching degree threshold, the risk index of each temporary device is normalized. The system performs a normalization process to obtain a second index normalized value. It then normalizes the ramp rate of each temporary device to obtain a second rate normalized value. A weighted sum is calculated of the second index normalized value, the second rate normalized value, the preset index weight, and the preset rate weight to obtain a temporary index. The temporary indices of each temporary device are then sorted in descending order to obtain a temporary list. The active power of each temporary device is summed sequentially according to the order of the temporary list to obtain a second power sum. When the second power sum is first greater than or equal to the load loss, the corresponding temporary device is determined to be the target shutdown device, thus identifying several target shutdown devices.

[0119] The preset matching threshold is the tolerance for the maximum positive deviation between the power of a single device and the remaining load. It depends on the maximum allowable cost-effectiveness (excessive power reduction) sacrificed for operational simplicity (stopping only one device), and is typically set between 5% and 15% of the remaining lost load. In this embodiment, it is set to 10% of the remaining lost load, which effectively limits excessive capacity reduction caused by selecting a single large device while ensuring that an efficient single solution is not missed, thereby achieving an optimal balance between operational complexity and cost-effectiveness.

[0120] By calculating the matching degree, a quantitative correlation between equipment capacity and system requirements is directly established. Priority is given to finding target shutdown equipment that can independently and accurately fill the power gap (with the smallest matching degree and greater than zero), which greatly reduces operational complexity. When there is no single optimal solution, a multi-attribute decision-making process is initiated. The risk index representing the safety status and the ramp rate representing the adjustment capability are normalized and weighted and integrated. The shutdown priority is constructed by arranging the temporary index in descending order. This ensures that when a combined shutdown is necessary, the equipment that should be shut down under the comprehensive evaluation (i.e., the equipment with the highest risk and the strongest adjustment capability) is selected first. This satisfies both capacity requirements and achieves the dual goals of safe shutdown and preservation of system flexibility.

[0121] Specifically, the process of adjusting the preset deviation threshold or the preset index threshold based on the total electricity consumption and the rotor speed within a preset adjustment period includes: calculating the relative deviation between the total electricity consumption and the historical total electricity consumption to obtain a temporary deviation; calculating the standard deviation of the temporary deviation within the preset adjustment period to obtain an electricity consumption deviation fluctuation value; and when the electricity consumption deviation fluctuation value is greater than a preset fluctuation threshold, reducing the preset deviation threshold based on the relative deviation between the electricity consumption deviation fluctuation value and the preset fluctuation threshold, and a preset adjustment coefficient, where F'=F×[1-a×(E-E') / E'], F' is the adjustment deviation threshold, and F is the preset deviation threshold. 'a' is a preset adjustment coefficient, 'E' is the power consumption deviation fluctuation value, and 'E' is a preset fluctuation threshold. When the power consumption deviation fluctuation value is less than or equal to the preset fluctuation threshold, the rate of change of all rotor speeds within the preset adjustment period is calculated to obtain the speed change rate. When the speed change rate is less than the preset speed change rate threshold, the preset index threshold is increased based on the relative deviation between the speed change rate and the preset speed change rate threshold, as well as the preset adjustment coefficient, where H' = H × [1 + b × (K' - K) / K'], H' is the adjustment index threshold, H is the preset index threshold, b is the preset adjustment coefficient, K is the speed change rate, and K' is the preset speed change rate threshold.

[0122] The preset fluctuation threshold is a critical value used to determine whether the fluctuation of electricity consumption deviation is within the normal range. It depends on the statistical results of the standard deviation of the deviation value in historical electricity consumption data and the system's requirements for anomaly detection sensitivity, and is usually 1.5 to 2 times the historical electricity consumption deviation fluctuation value. In this embodiment, it is set to 1.8 times, which can effectively capture abnormal fluctuations that exceed the normal range, while avoiding frequent erroneous adjustments due to overly sensitive settings.

[0123] The preset adjustment coefficient is a factor used to control the adjustment range of the preset deviation threshold. It depends on the trade-off between the system's desired parameter convergence speed and stability, and is usually set between 0.1 and 0.3. In this embodiment, it is set to 0.2 to ensure that the parameter adjustment process is smooth and stable, avoiding drastic fluctuations in the system's judgment criteria due to an overly aggressive learning strategy.

[0124] The preset speed change rate threshold is a critical value for judging whether the rotor speed of the power generation equipment is stable enough. It depends on the type and capacity of the power generation equipment and the performance indicators of its speed regulation system. It is usually set to 0.5% to 1.5% of the rated speed per minute. When the speed change rate is lower than this threshold, it can be considered that the overall equipment operation is very stable. At this time, the conditions for tightening the judgment criteria for high-risk equipment are met.

[0125] The preset adjustment coefficient is a factor used to control the adjustment range of the preset index threshold. It depends on whether the system's adjustment strategy for changes in risk tolerance is cautious or aggressive, and is usually set between 0.05 and 0.15. In this embodiment, it is set to 0.1 to ensure that the criteria for identifying high-risk devices are only increased slightly and gradually when the system has been proven to be running stably for a long time, which greatly enhances the reliability of the system.

[0126] By calculating the fluctuation value of electricity consumption deviation, the stability of external grid demand is perceived. When the fluctuation is large, it indicates that the system is in an unstable state. In this case, proportionally reducing the preset deviation threshold can lower the threshold for judging abnormal events, making the system more sensitive to load changes and thus triggering optimized scheduling more quickly to cope with uncertainties. When the electricity consumption fluctuation is within the normal range, the mechanical operation stability of the generator equipment is further evaluated by the rotor speed change rate. When the speed change rate is small, it indicates that the equipment is in a stable operating state. In this case, proportionally increasing the preset index threshold can raise the judgment standard for high-risk equipment, avoid excessive warnings about equipment status under stable operating conditions, reduce unnecessary shutdown operations, and dynamically optimize its decision sensitivity according to actual operating conditions. This improves the response capability during disturbances and ensures the economic efficiency of operation during stable periods, achieving long-term adaptive optimization.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An artificial intelligence-based power planning optimization method, characterized in that, The method comprises: collecting total power consumption of a power system in a region responsible for a power plant, heat consumption rate, active power, rotor speed, climbing rate and vibration frequency of each to-be-tested equipment in the power plant in real time; determining an abnormal event according to the total power consumption and a preset deviation threshold; based on the abnormal event, predicting a load loss and a remaining time according to a preset artificial intelligence model and the total power consumption, wherein the load loss refers to a total power consumption that will be lost in the future due to large-scale shutdown and production stoppage, and the remaining time refers to a time remaining from the current time to the maximum load loss; determining a plurality of candidate shutdown equipment according to the remaining time, the heat consumption rate and the vibration frequency; determining a plurality of high-risk equipment according to the vibration frequency, the rotor speed and a preset index threshold of each candidate shutdown equipment, wherein the preset index threshold is a critical value for distinguishing a truly high-risk equipment from the candidate equipment; determining a plurality of target shutdown equipment according to the active power, the climbing rate and the load loss of each high-risk equipment; generating a power planning scheme according to all the target shutdown equipment; adjusting the preset deviation threshold or the preset index threshold according to the total power consumption and the rotor speed in a preset adjustment time based on the power planning scheme; the process of determining an abnormal event according to the total power consumption and a preset deviation threshold comprises: calculating a power consumption deviation according to the total power consumption and historical total power consumption; determining a power consumption demand reduction according to a comparison result of the power consumption deviation and the preset deviation threshold, so as to determine the abnormal event. 2.The artificial intelligence-based power planning optimization method according to claim 1, characterized in that, the process of determining a plurality of candidate shutdown equipment according to the remaining time, the heat consumption rate and the vibration frequency comprises: determining a shutdown demand event according to a comparison result of the remaining time and a preset shutdown time; based on the shutdown demand event, determining a plurality of candidate shutdown equipment according to the heat consumption rate and the vibration frequency. 3.The artificial intelligence-based power planning optimization method of claim 2, wherein, the process of determining a plurality of candidate shutdown equipment according to the heat consumption rate and the vibration frequency comprises: calculating a plurality of heat consumption rate changes by calculating a change rate of all the heat consumption rates from an initial time to each time in a preset determination time; calculating a plurality of frequency changes by calculating a change rate of all the vibration frequencies from the initial time to each time in the preset determination time; determining a plurality of candidate shutdown equipment according to all the heat consumption rate changes and all the frequency changes. 4.The AI-based power planning optimization method of claim 3, wherein, the process of determining a plurality of candidate shutdown equipment according to all the heat consumption rate changes and all the frequency changes comprises: calculating an energy-frequency coordination degree according to all the heat consumption rate changes and all the frequency changes; determining the to-be-tested equipment as the candidate shutdown equipment according to a comparison result of the energy-frequency coordination degree and a preset coordination degree threshold, so as to obtain a plurality of candidate shutdown equipment. 5.The artificial intelligence-based power planning optimization method according to claim 4, characterized in that, the process of determining a plurality of high-risk equipment according to the vibration frequency, the rotor speed and a preset index threshold of each candidate shutdown equipment comprises: calculating a frequency fluctuation value according to the vibration frequency in a preset determination time; calculating a rotor speed fluctuation value according to the rotor speed in the preset determination time; Determine several high-risk devices according to the frequency fluctuation value, the rotation speed fluctuation value, and a preset index threshold. 6.The artificial intelligence-based power planning optimization method according to claim 5, characterized in that, The process of determining several high-risk devices according to the frequency fluctuation value, the rotation speed fluctuation value, and a preset index threshold comprises: Calculate a risk index according to the frequency fluctuation value and the rotation speed fluctuation value; Determine the candidate shutdown device as the high-risk device according to a comparison result of the risk index and the preset index threshold, so as to determine several high-risk devices. 7.The artificial intelligence-based power planning optimization method according to claim 6, characterized in that, The process of determining several target shutdown devices according to the active power, the ramp rate, and the load loss amount of each high-risk device comprises: Calculate a total reducible capacity according to the active power of all the high-risk devices; When the total reducible capacity is greater than the load loss amount, determine a reduction list according to the risk index and the ramp rate of each high-risk device, and determine several target shutdown devices according to the reduction list and the active power of each high-risk device; When the total reducible capacity is less than or equal to the load loss amount, determine all the high-risk devices as the target shutdown devices, calculate a residual loss load according to the total reducible capacity and the load loss amount, and determine target shutdown devices according to the residual loss load, the risk index, the ramp rate, and the active power of each temporary device; The temporary device is a device other than the high-risk device in the candidate shutdown device. 8.The artificial intelligence-based power planning optimization method of claim 7, wherein, The process of determining target shutdown devices according to the residual loss load, the risk index, the ramp rate, and the active power of each temporary device comprises: Calculate several matching degrees according to the active power and the residual loss load of each temporary device; When the matching degree is greater than zero and less than a preset matching degree threshold, determine that the corresponding temporary device is the target shutdown device; When there is no matching degree greater than zero and less than the preset matching degree threshold, determine a temporary list according to the risk index and the ramp rate of each temporary device, and determine several target shutdown devices according to the temporary list, the active power, and the residual loss load. 9.The artificial intelligence-based power planning optimization method of claim 8, wherein, The process of adjusting the preset deviation threshold or adjusting the preset index threshold according to the total power consumption and the rotor rotation speed in a preset adjustment period comprises: Calculate a power consumption deviation fluctuation value according to the total power consumption and the historical total power consumption in the preset adjustment period; Adjust the preset deviation threshold according to a first comparison result of the power consumption deviation fluctuation value and a preset fluctuation threshold; According to a second comparison result of the power consumption deviation fluctuation value and the preset fluctuation threshold, calculate a rotation speed change rate according to the rotor rotation speed in the preset adjustment period; Adjust the preset index threshold according to a comparison result of the rotation speed change rate and a preset rotation speed change rate threshold.

Images