A robust load control method based on model prediction and multi-time scales

Abstract

The present invention discloses a load control method based on model prediction and multi-time scale robustness, and belongs to the field of load control technology. The method performs load prediction on each load device in the control system based on the multi-time scale load prediction model stored in the database to obtain the total load prediction result; establishes a system dynamic model of the control system to be controlled; constructs a minimization cost function based on the system dynamic model and the load prediction result, and determines the load control input, which is the load adjustment amount required for the control system to be controlled. The present invention can accurately adjust the load demand in the system, thereby optimizing energy distribution, reducing energy waste, and coping with the uncertainty and dynamic changes of energy demand, ensuring the economy and reliability of system operation, and can effectively solve the load fluctuation problem in energy management, and improve energy utilization efficiency and system stability.

Description

Technical Field

[0001] The present invention belongs to the technical field of load control, and in particular relates to a load control method based on model prediction and multi-time scale robustness. Background Art

[0002] Load control methods play a key role in power system management, particularly in improving energy efficiency, optimizing grid operations, and implementing demand-side management. These methods can be broadly categorized as direct load control, indirect load control, demand response, and predictive load control. Predictive load control leverages historical and real-time data to forecast power demand. This method collects and analyzes large amounts of historical and real-time data, including electricity consumption, weather conditions, and user behavior. This data can be used to accurately forecast power demand using machine learning and statistical models.

[0003] Existing predictive load control can more accurately predict and manage electricity demand, but it is not convenient to accurately perform multi-time scale load forecasts for each load device to adjust the load demand in the system, thereby optimizing energy distribution and reducing energy waste. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a load control method based on model prediction and multi-time scale robustness, which can effectively solve the load fluctuation problem in energy management and improve energy utilization efficiency and system stability.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0006] The load control method based on model prediction and multi-time scale robustness includes the following steps:

[0007] Based on the multi-time scale load forecasting model stored in the database, load forecasting is performed on each load device in the control system to obtain the total load forecast result;

[0008] Establish a dynamic model of the system to be controlled;

[0009] Based on the system dynamic model and load forecast results, a minimization cost function is constructed to determine the load control input, which is the load adjustment required for the system to be controlled.

[0010] Preferably, the expression of the multi-time scale load forecasting model is:

[0011] ;

[0012] in, is the load forecast result of the i-th load device at time t, i=1, 2, ..., n, n is the number of load devices, is the total load forecast result at time t, is the historical load data of the i-th load device, which is used to capture the time series characteristics of the load. is the environmental condition of the i-th load device, is the market condition of the i-th load device, is the error term of the i-th load device, It is the long short-term memory network LSTM;

[0013] The confidence interval of the load forecast results for each load device is calculated as follows:

[0014] ;

[0015] in, is the quantile of the standard normal distribution of the i-th load device corresponding to the selected confidence level, is the standard deviation of the load forecast results of the i-th load device.

[0016] Preferably, the system dynamic model is used to describe the dynamic change relationship between load and system state, and the expression is:

[0017] ;

[0018] in, 、 They represent the load states of the control system at time t+1 and time t, respectively. ;in, 、 They represent the load status of the i-th load device at time t+1 and time t respectively;

[0019] is the load control input at time t+1, ;in, represents the load control input of the i-th load device at time t+1;

[0020] is the deviation of the control system when it performs load control at time t+1, ;in, It represents the deviation of load control of the i-th load device at time t+1.

[0021] Preferably, the expression for minimizing the cost function J is:

[0022] ;

[0023] Among them, pC and pH are both transit functions, is the total load forecast result at time t+1, is the control cost corresponding to the load control input at time t+1, is the weight factor of the control cost set in the database, is the adjustment smoothness weight factor set in the database, T is the total number of time stages, is the load control input at time t;

[0024] The constraints are:

[0025] ;

[0026] in, is the load control input of the i-th load device at time t, is the minimum allowable load control input of the i-th load device at time t, is the maximum allowable load control input of the i-th load device at time t.

[0027] Preferably, the method for obtaining the control cost corresponding to the load control input is as follows:

[0028] Compare the load control input with various parameterized load control inputs stored in the database, and search the database for the parameterized load control input that is closest to the load control input;

[0029] After determining the closest parameterized load control input, a control cost corresponding to the closest parameterized load control input and stored in the database is obtained.

[0030] Preferably, the closest parameterized load control input is determined as follows:

[0031] Determine the similarity between the load control input and each parameter load control input stored in the database based on the Euclidean distance:

[0032] ;

[0033] in, is the similarity between the load control input and the a-th parameter load control input, is the parameterized load control input of the i-th load device in the a-th parameterized load control input;

[0034] The parameterized load control input corresponding to the smallest similarity is determined as the closest parameterized load control input.

[0035] Preferably, before performing load forecasting on each load device in the control system based on the multi-time-scale load forecasting model to obtain a total load forecast result, the method further includes:

[0036] Diagnose the communication equipment to determine whether it is operating normally:

[0037] If the operation is abnormal, the alarm device will send a communication equipment abnormality alarm to the data processing center for manual troubleshooting;

[0038] If it is operating normally, diagnose the load control device to determine whether it is operating normally:

[0039] If the operation is abnormal, the alarm device will send a communication equipment abnormality alarm to the data processing center for manual troubleshooting;

[0040] If the operation is normal, load control is performed on each load device in the control system based on the determined load control input.

[0041] Preferably, determining whether the communication device is operating normally includes the following steps:

[0042] Acquire communication operation data of the communication device based on direct connection. The communication operation data includes instruction packet loss rate, synchronization accuracy, CRC error rate and action response time. CRC stands for cyclic redundancy check.

[0043] Obtain communication operation reference data stored in the database. The communication operation definition data includes the definition of instruction packet loss rate, the definition of synchronization accuracy, the definition of CRC error rate and the definition of action response time;

[0044] Acquire communication operation deviation data, which includes the allowable deviation value of instruction packet loss rate, the allowable deviation value of synchronization accuracy and the allowable deviation value of CRC error rate;

[0045] Determine a communication device operation state value based on the communication operation data, the communication operation reference data, and the communication operation deviation data, and compare the communication device operation state value with a communication device operation state threshold stored in a database:

[0046] If the communication device operation status value is greater than the communication device operation status threshold stored in the database, the communication device is not operating normally;

[0047] If the communication device operation status value is not greater than the communication device operation status threshold stored in the database, the communication device operates normally.

[0048] Preferably, determining whether the load control device is operating normally includes the following steps:

[0049] Obtain equipment operation data of load control equipment based on direct connection, including load adjustment execution delay, output power accuracy and equipment startup time;

[0050] Obtaining equipment operation reference data stored in the database, where the equipment operation definition data includes load adjustment execution reference delay, output power reference accuracy, and equipment startup reference duration;

[0051] Obtaining equipment operation deviation data, including load adjustment execution delay allowable deviation value and output power accuracy allowable deviation value;

[0052] Determine the load control device operation status value based on the device operation data, the device operation reference data, and the device operation deviation data, and compare the load control device operation status value with the load control device operation status threshold stored in the database:

[0053] If the load control device operating state value is greater than the load control device operating state threshold stored in the database, the load control device is operating abnormally;

[0054] If the load control device operating status value is not greater than the load control device operating status threshold stored in the database, the load control device operates normally.

[0055] Preferably, determining whether the communication device is operating normally is based on the operating status value of the communication device, and determining whether the load control device is operating normally is based on the operating status value of the load control device;

[0056] When it is determined that the communication equipment is operating normally and the load control equipment is operating normally, determining the comprehensive stability index of the hardware equipment based on the operating status value of the communication equipment and the operating status value of the load control equipment;

[0057] If the comprehensive stability index of the hardware device is greater than the comprehensive stability threshold of the hardware device stored in the database, a hardware stability warning prompt will be issued.

[0058] The present invention has the following beneficial effects:

[0059] The present invention performs multi-time-scale load forecasting on each load device and constructs a minimization cost function in combination with the system dynamic model. It can accurately adjust the load demand in the system, thereby optimizing energy distribution, reducing energy waste, and responding to the uncertainty and dynamic changes in energy demand, ensuring the economy and reliability of system operation, and can effectively solve the load fluctuation problem in energy management, improve energy utilization efficiency and system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of the method of the present invention;

[0061] Figure 2 A flow chart of the method of the present invention for determining whether a communication device is operating normally;

[0062] Figure 3The flowchart of the method of the present invention is to determine whether the load control device operates normally. DETAILED DESCRIPTION

[0063] The embodiments of the present invention are further described below with reference to the accompanying drawings:

[0064] like Figure 1 As shown, the load control method based on model prediction and multi-time scale robustness includes the following steps: based on the multi-time scale load prediction model stored in the database, load prediction is performed on each load device in the control system to obtain a total load prediction result.

[0065] The expression of the multi-time scale load forecasting model is:

[0066] ;

[0067] in, is the load forecast result of the i-th load device at time t, i=1, 2, ..., n, n is the number of load devices, is the total load forecast result at time t, is the historical load data of the i-th load device, which is used to capture the time series characteristics of the load. is the environmental condition of the i-th load device, is the market condition of the i-th load device, is the error term of the i-th load device, It is the long short-term memory network LSTM;

[0068] The confidence interval of the load forecast results for each load device is calculated as follows:

[0069] ;

[0070] in, is the quantile of the standard normal distribution of the i-th load device corresponding to the selected confidence level, is the standard deviation of the load forecast results of the i-th load device.

[0071] Combining data from multiple timescales and comprehensively analyzing historical, environmental, and market factors can significantly improve forecast accuracy and reduce bias and error. Accurate load forecasts enable energy providers to better plan and allocate resources. Confidence intervals quantify forecast uncertainty, helping decision-makers understand the credibility of forecast results and providing a scientific basis for developing risk management strategies and contingency plans. Models that take environmental and market conditions into account make load management more flexible and adaptable, enabling them to respond to rapidly changing market conditions.

[0072] The multi-time-scale load forecasting model stored in the database is pre-trained. It can use the long short-term memory network (LSTM) to train and forecast multi-time-scale load forecasts. The specific logic and steps are as follows:

[0073] Step 1: Data collection and preprocessing: Collect relevant historical load data, environmental condition data (such as temperature and humidity), and market condition data (such as energy prices and demand). Construct features that help model learning, such as converting timestamps into seasonal, weekday / non-workday, and hourly features. Use methods such as MinMaxScaler or StandardScaler to standardize the data to reduce the impact of different dimensions.

[0074] Step 2: Data splitting: Divide the data into training, validation, and test sets. For example, use 70% of the data as the training set, 15% as the validation set, and the remaining 15% as the test set. Organize the data into a sequential data format, with each sample containing historical data within a specific time window as input and data from a future period as output (for example, using data from the past 24 hours to predict load for the next hour).

[0075] Step 3: Build an LSTM model, define one or more LSTM layers, add fully connected layers as appropriate, and design the output layer to correspond to the prediction target (for example, using a single node with a linear activation function to predict future load values). Select an appropriate loss function (such as mean squared error (MSE), optimizer (such as Adam), and evaluation metric (such as MAE).

[0076] Step 4: Train the model. Train the model on the training set and use the validation set to monitor overfitting and adjust model parameters. Adjust parameters such as learning rate, batch size, and number of LSTM units to optimize model performance. After training, save the model and parameters so that they can be loaded and used directly later without retraining.

[0077] Step 5: Performance evaluation: Use the test set to evaluate the performance of the model, focusing on the prediction accuracy and generalization ability of the model. Analyze the errors in the model predictions and look for possible patterns or systematic deviations to further optimize the model.

[0078] Before generating a total load forecast based on a multi-timescale load forecasting model for each load device in the control system, the system checks the proper operation of the communication equipment to ensure reliable data transmission and reception. This is essential for ensuring proper communication between all system components. If a communication device is operating abnormally, an alarm is immediately sent to the data processing center via the alarm device, and manual troubleshooting is performed. This ensures a rapid response to the fault and minimizes system downtime. After the communication equipment is confirmed to be operating normally, the system further checks the status of the load control devices, which are responsible for implementing load adjustment and control strategies. If the load control devices are operating normally, the system regulates and controls each load device based on the preset load control inputs to optimize the energy efficiency and performance of the entire system.

[0079] By ensuring the proper functioning of communication and control equipment, the stability and reliability of the entire system can be significantly improved, preventing erroneous data transmission or load regulation caused by equipment failure. Ensuring the proper functioning of control equipment enables more precise load regulation, helping to optimize energy use, reduce energy waste, and lower energy costs.

[0080] like Figure 2 As shown, the communication equipment is diagnosed to determine whether it is operating normally (if it is operating normally, the equipment data collected by the data acquisition device can be sent normally to the data processing center, and the instructions sent by the data processing center can be received and sent to the load control equipment):

[0081] The communication operation data of the communication equipment is obtained based on a direct connection method. The communication operation data includes the instruction packet loss rate Db, synchronization accuracy Tb, CRC error rate Cb and action response time XT. The current operation data, including the instruction packet loss rate, synchronization accuracy, CRC error rate and action response time, are directly obtained from the communication equipment. The direct connection method can be connected through optical fiber and does not require WiFi or other wireless communication methods to connect, so as to avoid the inability to transmit data after the wireless device of the communication equipment is damaged; the communication operation reference data stored in the database is obtained. The communication operation definition data includes the definition of the instruction packet loss rate dDd, the definition of the synchronization accuracy dTd, the definition of the CRC error rate dCd and the definition of the action response time cXT. These data define the baseline conditions for the normal operation of the communication equipment.

[0082] Db measures the number of control commands that fail to transmit successfully during communication. A high packet loss rate typically indicates poor network quality or performance issues with communication equipment, which can prevent control commands from being accurately executed, impacting system responsiveness and reliability. Tb reflects the accuracy of time synchronization between system components. In situations where precise coordination of multiple devices or systems is required, low synchronization accuracy can cause operational disruptions, impacting overall system efficiency and security. Cb detects errors during data storage or transmission. A high CRC error rate indicates frequent data corruption during transmission, which may be caused by physical connection issues, electromagnetic interference, or device failure. XT measures the time it takes from the issuance of a control command to the completion of the corresponding action by the communication equipment. Long response times can lead to sluggish system response, impacting real-time control and overall system performance.

[0083] Db and Cb can be captured in real time using network monitoring tools and software, Tb can be measured using time synchronization technologies such as NTP (Network Time Protocol) or PTP (Precision Time Protocol), and XT can be measured using performance testing software that simulates the sending and receiving of control commands, recording and analyzing response times.

[0084] These parameters work together to assess the overall performance and reliability of communications equipment. Db and Cb directly relate to the integrity and accuracy of data transmission, while Tb and XT reflect the system's processing efficiency and coordination after receiving data. Abnormalities in any of these parameters can affect normal system operation, and they have complex dependencies on each other. For example, a high packet loss rate can lead to reduced synchronization accuracy, while an increase in CRC errors may require more retransmissions, affecting response time.

[0085] Acquire environmental data for the communication device, including electromagnetic interference levels, continuous operating hours, average temperature, and age. Obtain an annualized defined data matching set for the communication device from a database based on age. The annualized defined data matching set includes several environmental matching data sets, including matching values ​​for electromagnetic interference levels, continuous operating hours, and average temperature. Calculate similarity between the communication device's environmental data and each environmental matching data set based on Euclidean distance, and determine the environmental matching data set with the minimum similarity.

[0086] Electromagnetic interference (EMI) can severely impact the data transmission quality of communications equipment, causing signal distortion, packet loss, and communication errors. High interference levels can lead to frequent communication failures and errors. In environments with high EMI, a wider tolerance is required to reflect the actual impact of these environmental conditions on device performance. Continuous operation of equipment can lead to accumulated fatigue, such as physical wear and tear from overheating, which can reduce device stability and responsiveness. For devices operating for extended periods, a wider tolerance may be required to accommodate performance degradation. Temperature significantly impacts electronic equipment, especially communications equipment. Excessively high or low temperatures can affect circuit efficiency, potentially leading to speed drops, hardware damage, and even failure.

[0087] When operating under different temperature conditions, the stability and performance of equipment may vary. Therefore, the tolerance value needs to be adjusted based on the average temperature to ensure accurate evaluation. Longer equipment life typically results in increased wear and aging, which gradually degrades performance. For older equipment, a larger tolerance may be required to account for the actual impact of aging on performance. This comprehensive consideration ensures that the tolerance setting is both scientific and practical, more accurately reflecting the performance of the equipment under specific operating conditions. By adjusting the tolerance, equipment maintenance and evaluation can be more accurately aligned with the actual operating environment and equipment condition, thereby improving the reliability and efficiency of the entire communication system.

[0088] Obtain the communication operation deviation data stored in the database corresponding to the environment matching data, the communication operation deviation data includes the instruction packet loss rate allowable deviation value , synchronization accuracy allowable deviation value and CRC error rate tolerance Based on the communication operation data, the communication operation reference data and the communication operation deviation data, the communication device operation state value Tz is determined, where the calculation formula of Tz is:

[0089] ;

[0090] In actual applications, communications equipment may be affected by a variety of external factors, such as temperature fluctuations and electromagnetic interference, which can temporarily cause device performance to deviate from ideal conditions. The tolerance value provides a buffer for these environmental variables, preventing misjudgment of device status due to small performance fluctuations. As equipment ages, its performance may gradually decline. Setting the tolerance value can account for the natural aging process of equipment and prevent older equipment from being frequently marked as abnormal due to failure to meet initial performance standards. Even in the absence of external interference, device performance may vary due to factors such as production variations and maintenance status. The tolerance value defines the normal range of performance fluctuations, helping operators distinguish between normal fluctuations and true equipment problems.

[0091] The formula ensures fair and consistent evaluation by standardizing the deviations of each indicator. This means that different performance indicators can be compared on the same scale, regardless of their original dimensions and magnitudes. By monitoring changes in Tz values, potential equipment performance degradation or failure signs can be promptly identified, allowing for early intervention and maintenance, avoiding further operational failures and financial losses.

[0092] At the same time, Tz is compared with the communication equipment operating status threshold stored in the database: if Tz is greater than the communication equipment operating status threshold stored in the database, the communication equipment is operating abnormally; if Tz is not greater than the communication equipment operating status threshold stored in the database, the communication equipment is operating normally. If it is operating abnormally, an abnormal alarm of the communication equipment is sent to the data processing center based on the alarm device for manual troubleshooting; if it is operating normally, the load control equipment is diagnosed to determine whether the load control equipment is operating normally:

[0093] like Figure 3 As shown, the equipment operation data of the load control device is obtained based on the direct connection method, and the equipment operation data includes the load adjustment execution delay Ty, the output power accuracy Cp and the equipment startup time qT; the equipment operation reference data stored in the database is obtained, and the equipment operation definition data includes the load adjustment execution reference delay dTy, the output power reference accuracy dCp and the equipment startup reference time cqT.

[0094] Ty refers to the time from receiving a load adjustment command to actually executing that adjustment. In a dynamically changing system, prolonged execution delays can result in an inability to respond to load changes in a timely manner, impacting the performance and stability of the entire system. Cp refers to the degree of match between the power output of the device and the command or setpoint. High precision ensures efficient energy utilization and precise system operation; lack of precision can lead to energy waste or device overload. qT refers to the time it takes for a device to reach normal operating status from a fully shut-down state. Excessive startup time can affect the system's ability to respond quickly, especially in applications that require frequent starts and stops.

[0095] Ty is measured by measuring the actual time it takes for the device to complete the adjustment from the moment it receives the adjustment command. This can be achieved using a real-time monitoring system or dedicated testing software. Cp is assessed by comparing the actual power output of the device with the preset or expected power. This comparison can be performed using a power meter, calibration instrument, or sensors integrated into the device. qT can be measured by timing the device's startup process, from the moment the startup command is issued until the device reaches normal operating conditions.

[0096] These parameters collectively determine the responsiveness and operational efficiency of load control equipment. Load adjustment execution delay and device startup time are directly related to device response speed, while output power accuracy is related to operational precision. These parameters influence each other and contribute to the overall performance evaluation of the device.

[0097] Obtain load control device impact data, which includes the communication device instruction packet loss rate, the communication device synchronization accuracy, and the load control device power supply voltage mean square deviation; obtain several load control device impact matching data stored in the database, which includes the communication device instruction packet loss rate matching value, the communication device synchronization accuracy matching value, and the load control device power supply voltage mean square deviation matching value.

[0098] The practice of using the communication equipment command packet loss rate, communication equipment synchronization accuracy, and the mean square error of the load control equipment's supply voltage to determine deviations aims to ensure that the load control equipment's operational performance not only meets design standards but also adapts to the challenges of real-world operating environments. These three parameters are directly related to the reliability and efficiency of the load control equipment, and thus have a profound impact on the stability and security of the entire system.

[0099] The packet loss rate is a key indicator of communication quality. In load control systems, if control commands fail to reach their target devices due to packet loss, this can lead to operational delays, execution errors, and even safety incidents. By monitoring the packet loss rate and adjusting deviations accordingly, communication strategies and network configurations can be optimized to reduce command loss, thereby improving the overall responsiveness and reliability of the system. This ensures that every control command is accurately transmitted, reducing operational errors caused by communication issues.

[0100] In load control systems that require the coordinated operation of multiple devices, the synchronization accuracy of communication equipment determines operational coordination. Low synchronization accuracy can lead to timing errors between devices, impacting system efficiency and safety. Supply voltage stability directly impacts the performance and lifespan of load control equipment. Large voltage fluctuations can cause unstable operation and even damage sensitive electronic components. By monitoring the voltage mean square deviation (MSD) and setting an appropriate deviation range, measures can be taken to stabilize the voltage supply and protect equipment from excessively high or low voltages. In summary, by carefully controlling these key parameters and setting appropriate deviation ranges, the reliability, efficiency, and safety of load control systems can be significantly improved, while also reducing economic losses caused by equipment failure or unstable operation.

[0101] Calculate the similarity between the load control device impact data and each load control device impact matching data based on the Euclidean distance, and determine the load control device impact matching data corresponding to the minimum similarity; obtain the device operation deviation data stored in the database corresponding to the load control device impact matching data, and the device operation deviation data includes the load adjustment execution delay allowable deviation value , Output power accuracy allowable deviation value .

[0102] The load control device operating state value Fz is determined based on the equipment operating data, equipment operating reference data and equipment operating deviation data. The calculation formula of Fz is:

[0103] ;

[0104] Introducing allowed deviations is crucial when evaluating load control equipment performance, as it provides a tolerance for normal performance fluctuations under real-world operating conditions. This approach reflects the equipment's adaptability to diverse environments and loads. By standardizing the deviations for each indicator, evaluations of different equipment, or even the same equipment under different conditions, are comparable. This approach provides more impartial evaluation results and helps accurately identify actual equipment issues. By monitoring equipment status values ​​in real time and comparing them with allowed deviations, potential equipment failures can be predicted, allowing for timely maintenance or replacement, reducing system downtime and repair costs.

[0105] At the same time, Fz is compared with the load control device operating status threshold stored in the database: if Fz is greater than the load control device operating status threshold stored in the database, the load control device is operating abnormally; if Fz is not greater than the load control device operating status threshold stored in the database, the load control device is operating normally.

[0106] If the operation is abnormal, the communication equipment abnormality alarm will be sent to the data processing center based on the alarm device for manual troubleshooting; if the operation is normal, the load of each load device in the control system will be regulated based on the determined load control input.

[0107] When it is determined that the communication equipment and the load control equipment are operating normally, the comprehensive stability index W of the hardware equipment is determined based on Tz and Fz. , where a is the weight factor of Tz stored in the database, and b is the weight factor of Fz stored in the database. If W is greater than the comprehensive stability threshold of the hardware device stored in the database, a hardware stability warning prompt is issued.

[0108] The independent performance scores of the two devices are combined into a single composite score. Here, a and b are weighting factors for Tz and Fz, respectively. These weights are determined by the system's design or operational requirements and reflect the importance of each performance factor in overall device stability. Providing a composite metric, W, helps operators or maintenance teams quickly understand the overall device status, simplifying the complexity of evaluating multiple independent metrics. By combining the performance metrics of key devices into a single composite score, a more comprehensive understanding of system health is achieved, especially when these devices are interdependent.

[0109] Establish a dynamic model of the system to be controlled.

[0110] The system dynamic model is used to describe the dynamic relationship between load and system state, and the expression is:

[0111] ;

[0112] in, 、 They represent the load states of the control system at time t+1 and time t, respectively. ;in, 、 They represent the load status of the i-th load device at time t+1 and time t respectively;

[0113] is the load control input at time t+1, ;in, represents the load control input of the i-th load device at time t+1;

[0114] is the deviation of the control system when it performs load control at time t+1, ;in, It represents the deviation of load control of the i-th load device at time t+1.

[0115] Based on the system dynamic model and load forecast results, a minimization cost function J is constructed to determine the load control input, which is the load adjustment required for the system to be controlled.

[0116] The expression of J is:

[0117] ;

[0118] Among them, pC and pH are both transit functions, is the total load forecast result at time t+1, is the control cost corresponding to the load control input at time t+1, is the weight factor of the control cost set in the database, is the adjustment smoothness weight factor set in the database, T is the total number of time stages, is the load control input at time t;

[0119] The constraints are:

[0120] ;

[0121] in, is the load control input of the i-th load device at time t, is the minimum allowable load control input of the i-th load device at time t, is the maximum allowable load control input of the i-th load device at time t.

[0122] for The solution can be solved by using quadratic programming (QP), defining J as the objective function of the quadratic programming problem, with the constraints unchanged, and applying mathematical optimization software or tools to find the optimal control input to minimize the objective function while satisfying all given constraints.

[0123] The method for obtaining the control cost corresponding to the load control input is as follows: the load control input is compared with each parameter load control input stored in the database, and the parameter load control input closest to the load control input is found in the database.

[0124] Determine the similarity between the load control input and each parameter load control input stored in the database based on the Euclidean distance:

[0125] ;

[0126] in, is the similarity between the load control input and the a-th parameter load control input, is the parameterized load control input of the i-th load device in the a-th parameterized load control input;

[0127] The parameterized load control input corresponding to the smallest similarity is determined as the closest parameterized load control input.

[0128] After determining the closest parameterized load control input, a control cost corresponding to the closest parameterized load control input and stored in the database is obtained.

[0129] In a specific embodiment, load forecasting is performed based on historical load data, environmental conditions, and market conditions. At time t=5, the current environmental conditions E(5) = [temperature 34 degrees, humidity 46%], the market conditions M(5) = [electricity price 0.54 yuan / kWh, market volatility factor 0.14], and the predicted three load devices ZL(6) = [135, 175, 220] kW. Assuming τ(6) is [2, -1, 3], the current state of the system x(5) = [130, 170, 215] kW, setting α to 0.1 and γ to 0.05, and solving through DP, we get u(6) = [5, 5, 6] kW, and update the system state: x(6) = x(5) + u(6) + τ(6) = [137, 174, 224], which is close to the predicted load ZL(6).

[0130] This embodiment can be implemented based on an electronic device and a computer-readable storage medium. The electronic device includes: a processor and a memory, in which computer program instructions are stored. When the computer program instructions are executed by the processor, the processor executes the load control method based on model prediction and multi-time scale robustness as described above.

[0131] A computer-readable storage medium is used to store a program, which, when executed by a processor, implements the load control method based on model prediction and multi-time-scale robustness as described above.

Claims

1. A load control method based on model prediction and multi-time scale robustness, characterized in that: The following steps are involved: Based on the multi-time scale load prediction model stored in the database, load prediction is performed on each load device in the control system to obtain the total load prediction result; Establish the system dynamic model of the control system to be controlled; Based on the system dynamic model and load forecast results, a minimization cost function is constructed to determine the load control input, which is the load adjustment required for the control system to be controlled; The expression of the multi-time scale load forecasting model is: ； in, is the load forecast result of the i-th load device at time t, i=1, 2, …, n, n is the number of load devices, is the total load forecast result at time t, is the historical load data of the i-th load device, which is used to capture the time series characteristics of the load. is the environmental condition of the i-th load device, is the market condition of the ith load device, is the error term of the i-th load device, It is the long short-term memory network LSTM; The calculation method of the confidence interval of the load forecast results of each load device is as follows: ； in, is the quantile of the standard normal distribution of the i-th load device corresponding to the selected confidence level, is the standard deviation of the load forecast result of the i-th load device; The system dynamic model is used to describe the dynamic change relationship between load and system state, and the expression is: ； in, , They represent the load states of the control system at time t+1 and time t respectively. ;in, , They represent the load status of the i-th load device at time t+1 and time t respectively; is the load control input at time t+1, ;in, represents the load control input of the i-th load device at time t+1; is the deviation of the control system when it performs load control at time t+1, ;in, represents the deviation of the load control of the i-th load device at time t+1; The expression of the minimization cost function J is: ； Among them, pC and pH are both transit functions, is the total load forecast result at time t+1, is the control cost corresponding to the load control input at time t+1, is the weight factor of the control cost set in the database, is the adjustment smoothness weight factor set in the database, T is the total number of time stages, is the load control input at time t; The constraints are: ； in, is the load control input of the i-th load device at time t, is the minimum allowable load control input of the ith load device at time t, is the maximum allowable load control input of the i-th load device at time t.

2. The load control method based on model prediction and multi-time scale robustness according to claim 1, characterized in that: The method for obtaining the control cost corresponding to the load control input is as follows: Compare the load control input with each parameter load control input stored in the database, and search the database for the parameter load control input that is closest to the load control input; After determining the closest parameterized load control input, the control cost corresponding to the closest parameterized load control input and stored in the database is obtained.

3. The load control method based on model prediction and multi-time scale robustness according to claim 2 is characterized in that: Determine the closest parameter load control input, the process is as follows: Determine the similarity between the load control input and each parameter load control input stored in the database based on the Euclidean distance: ； in, is the similarity between the load control input and the a-th parameter load control input, is the parameterized load control input of the i-th load device in the a-th parameterized load control input; The parameterized load control input corresponding to the smallest similarity is determined as the closest parameterized load control input.

4. The load control method based on model prediction and multi-time scale robustness according to claim 1, characterized in that: Before performing load forecasting on each load device in the control system based on the multi-time scale load forecasting model and obtaining the total load forecasting result, the following steps are also included: Diagnose the communication device to determine whether it is operating normally: If the operation is abnormal, the abnormal alarm of the communication equipment will be sent to the data processing center based on the alarm device for manual troubleshooting; If it is operating normally, diagnose the load control device to determine whether it is operating normally: If the operation is abnormal, the abnormal alarm of the communication equipment will be sent to the data processing center based on the alarm device for manual troubleshooting; If the operation is normal, load control is performed on each load device in the control system based on the determined load control input.

5. The load control method based on model prediction and multi-time scale robustness according to claim 4 is characterized in that: Determining whether the communication equipment is operating normally includes the following steps: Acquire the communication operation data of the communication device based on direct connection, the communication operation data includes instruction packet loss rate, synchronization accuracy, CRC error rate and action response time, CRC stands for cyclic redundancy check; Obtaining the communication operation reference data stored in the database, the communication operation definition data includes defining the instruction packet loss rate, defining the synchronization accuracy, defining the CRC error rate and defining the action response time; Acquire communication operation deviation data, the communication operation deviation data includes an allowable deviation value of instruction packet loss rate, an allowable deviation value of synchronization accuracy and an allowable deviation value of CRC error rate; Determine a communication device operation state value based on the communication operation data, the communication operation reference data, and the communication operation deviation data, and compare the communication device operation state value with a communication device operation state threshold stored in a database: If the communication device operation status value is greater than the communication device operation status threshold stored in the database, the communication device is not operating normally; If the communication device operation status value is not greater than the communication device operation status threshold stored in the database, the communication device operates normally.

6. The load control method based on model prediction and multi-time scale robustness according to claim 4, characterized in that: Determining whether the load control equipment is operating properly includes the following steps: Obtaining equipment operation data of load control equipment based on direct connection, the equipment operation data including load adjustment execution delay, output power accuracy and equipment startup time; Obtaining equipment operation reference data stored in a database, where the equipment operation definition data includes a reference delay for load adjustment execution, a reference accuracy for output power, and a reference duration for equipment startup; Obtaining equipment operation deviation data, the equipment operation deviation data including load adjustment execution delay allowable deviation value and output power accuracy allowable deviation value; Determine the load control device operation state value based on the device operation data, the device operation reference data and the device operation deviation data, and compare the load control device operation state value with the load control device operation state threshold value stored in the database: If the load control device operation state value is greater than the load control device operation state threshold stored in the database, the load control device is not operating normally; If the load control device operating status value is not greater than the load control device operating status threshold stored in the database, the load control device operates normally.

7. The load control method based on model prediction and multi-time scale robustness according to claim 4 is characterized in that: Determine whether the communication device is operating normally based on the operation status value of the communication device, and determine whether the load control device is operating normally based on the operation status value of the load control device; When it is determined that the communication device is operating normally and the load control device is operating normally, a comprehensive stability index of the hardware device is determined based on the operation status value of the communication device and the operation status value of the load control device; If the comprehensive stability index of the hardware device is greater than the comprehensive stability threshold of the hardware device stored in the database, a hardware stability early warning prompt will be issued.

Images