Power quality control method based on model prediction

Abstract

The invention relates to the technical field of electric energy quality control, and discloses an electric energy quality control method based on model prediction, a data acquisition module acquires power grid operation parameters including voltage, current, frequency and phase information in real time, indexes such as harmonic content, voltage deviation and three-phase unbalance degree are rapidly calculated according to the input parameters, and the operation is completed within a specified time. The prediction module is connected with the central processing unit, receives indexes and combines historical data through a time sequence algorithm. And the control strategy generation module generates a strategy containing reactive compensation switching and filter parameter adjustment instructions by using an optimization algorithm according to the prediction result. And the execution module drives the adjusting equipment, quickly responds, timely improves the electric energy quality when the threshold value is exceeded, and feeds back a result to the central processing unit to optimize subsequent control. The system has the advantages of prospective accurate control, high adaptability and intelligence, efficient resource utilization and cost control, and real-time performance and reliability guarantee.

Description

Technical Field

[0001] The invention relates to the technical field of power quality control, and in particular to a power quality control method based on model prediction. Background Art

[0002] With the rapid development of modern industry and the widespread application of various power electronic equipment, power quality issues in power grids have become increasingly prominent. Power quality issues such as harmonic pollution, voltage fluctuations and flicker, and three-phase imbalance will not only affect the stable operation of the power system itself and reduce power supply reliability, but may also damage precision equipment in industrial production, causing serious consequences such as production interruptions and product quality degradation, bringing huge economic losses to enterprises.

[0003] Traditional power quality control methods often have lags, and corresponding compensation or adjustment measures are often taken only after power quality problems have occurred. For example, reactive power compensation devices based on fixed parameters can only compensate for reactive power according to preset parameters, and cannot track the dynamic changes in the grid operation status in real time, making it difficult to cope with complex and changeable power quality interference. Moreover, some simple monitoring methods can only measure power quality indicators at the current moment, and lack the ability to predict future power quality trends, making it difficult for power operation and maintenance personnel to adjust control strategies in a timely manner and prevent potential power quality deterioration in advance.

[0004] In addition, the continuous expansion of the power grid and the continuous growth of power load have further aggravated the difficulty of power quality control. The operating conditions of the power grid in different regions and at different times vary greatly. Traditional methods are difficult to take into account various complex situations and cannot achieve accurate and efficient power quality optimization. In the context of large-scale access of new energy to the power grid, the intermittent and volatile characteristics of new energy generation have brought new challenges to power quality. A more intelligent and forward-looking power quality control method is urgently needed. Summary of the invention

[0005] In order to solve the above-mentioned problems, the present invention proposes a power quality control method based on model prediction.

[0006] In order to solve the above technical problems, the technical solution proposed by the present invention is: a power quality control method based on model prediction, comprising:

[0007] The data acquisition module collects the grid operation parameters in real time, including voltage, current, frequency and phase information. The acquisition frequency shall not be lower than the preset high frequency value. After the acquisition is completed, it is transmitted to the central processor in real time;

[0008] The central processor has a built-in power quality analysis model trained based on big data. The model quickly calculates the current power quality indicators based on the input grid operation parameters, including harmonic content, voltage deviation, and three-phase imbalance. The calculation process must be completed within the specified first time interval.

[0009] The prediction module is connected to the central processor, receives the power quality index, uses the prediction algorithm based on time series, combines the current index with the stored historical power quality data, and predicts the change trend of the power quality in the future predetermined time period. The prediction step is adaptively adjusted according to the dynamic characteristics of the power grid. The prediction accuracy must reach the preset high standard. The historical power quality data is stored in the storage unit connected to it;

[0010] The control strategy generation module generates the power quality control strategy according to the change trend of the output of the prediction module using the preset optimization algorithm, which includes the switching scheme of the reactive compensation device and the adjustment instructions of the filter parameters. The optimization algorithm aims to achieve the best comprehensive power quality index while fully considering the control cost and equipment life constraints. After generation, it is immediately transmitted to the execution module;

[0011] The execution module drives power quality regulation equipment such as static VAR generator and active power filter to perform actions according to the received control strategy. The execution response time shall not exceed the preset fast response threshold so as to improve the power quality in time. The execution module has a feedback function and can feed back the execution results to the central processor in real time, thereby realizing the optimization adjustment of subsequent control.

[0012] Preferably, the data acquisition module adopts a high-precision sensor array, which consists of voltage sensors, current sensors, frequency sensors and phase sensors. These sensors are calibrated in advance to ensure the accuracy of the collected data, have strong anti-interference capabilities, and meet the requirements of relevant industry standards. In addition, the collected raw data can be filtered, amplified, and analog-to-digital converted in sequence, and transmitted to the central processing unit after processing.

[0013] Preferably, when constructing the power quality analysis model, a large amount of grid operation data under different working conditions are widely collected as training samples, and the sample data are first subjected to feature extraction and normalization processing, and then a deep learning algorithm, such as a convolutional neural network or a recurrent neural network, is used to build the model structure, and the model parameters are adjusted through repeated training until the accuracy of the model on the verification set reaches a preset threshold. In addition, the model has an online learning function and can update the model parameters in real time according to the newly connected grid operation data, so as to adapt to the dynamic changes of the grid.

[0014] Preferably, the time series-based prediction algorithm adopted by the prediction module is a hybrid model, which effectively combines the advantages of the autoregressive moving average model ARIMA and the long short-term memory network LSTM. The ARIMA model is responsible for modeling the linear characteristics of the power quality data, while the LSTM model focuses on capturing the nonlinear characteristics and long-term dependencies of the data. Finally, the prediction results of the two are fused in a weighted manner, and the weights are dynamically adjusted according to the historical prediction errors.

[0015] Preferably, the preset optimization algorithm of the control strategy generation module is a multi-objective particle swarm optimization algorithm. During the solution process, the comprehensive power quality index, control cost, and equipment life are set as different objective functions, and with the help of iterative search of the particle swarm, efforts are made to find the optimal power quality control strategy that meets various objective constraints.

[0016] Preferably, the execution module is connected to the power quality regulation device through a high-speed communication interface, such high-speed communication interface includes an industrial Ethernet interface and a fiber optic interface, so as to ensure that the control instructions can be transmitted quickly. The execution module uses pulse width modulation PWM technology to accurately drive the device to operate. At the same time, the execution module also has fault detection and diagnosis functions. Once a fault is detected in the power quality regulation device, an alarm signal is immediately issued, and the fault information is fed back to the central processor so that corresponding emergency measures can be taken in time.

[0017] Preferably, the storage unit adopts a large-capacity solid-state hard drive SSD, which has fast data reading and writing capabilities, can store long-term historical power quality data, and also has a data backup function, which regularly backs up important data to an external storage device to prevent data loss. In addition, encryption technology is used for data transmission between the storage unit and the central processing unit and the prediction module to fully ensure data security and privacy.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] Prospective and precise control: Through the accurate prediction of the power quality change trend by the prediction module, the trend of the power quality of the power grid can be known in advance, so that the control strategy generation module can generate optimization strategies in a targeted manner, changing passive response to active attack. It avoids problems such as increased product defective rate and equipment damage caused by power quality problems, greatly improving production efficiency.

[0020] Highly adaptable and intelligent: The power quality analysis model in the central processor has an online learning function, which can track the changes in the operating conditions of the power grid in real time, continuously optimize its own parameters, and always accurately reflect the actual state of the power grid. Whether it is the intermittent impact caused by the large-scale access of new energy or the fluctuation caused by the rapid increase or decrease of industrial load, it can quickly adapt and give accurate power quality assessment.

[0021] Efficient resource utilization and cost control: The multi-objective particle swarm optimization algorithm used in the control strategy generation module fully takes into account the control cost and equipment life while pursuing the best power quality. Unlike the traditional single-objective control method, it may over-invest in reactive compensation equipment or frequently adjust the filter in order to pursue power quality, resulting in resource waste and premature aging of equipment.

[0022] Real-time and reliability guarantee: The high-precision, high-frequency data acquisition and fast pre-processing transmission of the data acquisition module, combined with the high-speed instruction transmission and fast response execution of the execution module, build a fast response channel for power quality control. Once a small fluctuation occurs in the power grid, it can be quickly detected and adjustment measures can be implemented in a very short time to prevent the problem from expanding. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a principle flow chart of a power quality control method based on model prediction of the present invention. DETAILED DESCRIPTION

[0024] The present invention is further described in detail below with reference to the accompanying drawings.

[0025] Example 1

[0026] Power quality control in industrial parks

[0027] In a large industrial park with many manufacturing companies, the power demand is large and the load is complex and variable. The power quality control method of the present invention is deployed as follows:

[0028] The sensors of the data acquisition module are distributed in various substations and main power supply line nodes in the park, capturing parameters such as voltage and current in real time. For example, in the area where a precision machining enterprise is located, the voltage fluctuates frequently due to the frequent start and stop of a large number of machine tools. The sensor collects data at a high frequency of 1,000 times per second and transmits it to the central processor after preprocessing. The power quality analysis model in the central processor is trained based on the operating data of the park in the past few years. It quickly determines that the voltage deviation in the area exceeds the standard range and the harmonic content also has an upward trend. The prediction module combines the current data with the stored historical data to predict that in the next hour, as more companies start work during the morning peak, the voltage fluctuation will intensify and the harmonics may further deteriorate. Based on this prediction, the control strategy generation module uses the multi-objective particle swarm optimization algorithm to determine the start of the park's standby static reactive generator and adjust some active power filter parameters to balance reactive power and filter out harmonics. The execution module transmits instructions to the corresponding equipment within milliseconds through the industrial Ethernet interface. The equipment acts quickly to stabilize the voltage near the rated value and reduce the harmonic content to below the national standard, ensuring the normal production of enterprises in the park and avoiding equipment failure and product scrapping losses caused by power quality problems. According to statistics, the direct economic benefits brought about by improved power quality can reach hundreds of thousands of yuan per month.

[0029] Example 2

[0030] Power quality assurance in new energy access areas

[0031] In a regional power grid where wind power and photovoltaic power are centrally connected, renewable energy accounts for up to 50%. Since renewable energy output is highly intermittent and volatile due to natural conditions, it poses a great challenge to power quality.

[0032] In the system of the present invention, the sensors of the data acquisition module are installed at the key nodes of the new energy power station and the access to the power grid to monitor the operating parameters of the power grid in real time. During the period of strong wind power, the sudden change of wind speed causes a sharp change in output power, causing the voltage flicker of the power grid. The sensor quickly collects data and transmits it to the central processing unit at a frequency of not less than 1200 times per second. The power quality analysis model of the central processing unit uses the continuously updated new energy power generation characteristic data and power grid operation data training and optimization to accurately identify the voltage flicker problem and calculate that the three-phase imbalance is also increasing. With the help of a hybrid prediction algorithm, the prediction module predicts that the power quality will continue to deteriorate, considering that the wind speed will still be unstable in the next half an hour. The control strategy generation module generates a strategy based on this. On the one hand, it adjusts the switching of the reactive compensation device to compensate for the reactive power and stabilize the voltage; on the other hand, it issues fine parameter adjustment instructions to the active power filter to suppress harmonics. The execution module quickly drives the equipment through the optical fiber interface, effectively alleviating the voltage flicker and three-phase imbalance problems in a short time, ensuring the stability of electricity supply for surrounding residents, reducing the negative impact of new energy access on the power quality of the power grid, improving the absorption capacity of new energy electricity, and promoting the efficient use of local clean energy, thus helping to achieve carbon emission reduction targets.

[0033] Example 3

[0034] Power Quality Optimization in Urban Commercial Centers

[0035] Urban commercial centers are home to a large number of commercial facilities such as shopping malls, office buildings, and hotels. The peak and valley differences in electricity load are obvious, and there are a large number of nonlinear loads, such as electronic display screens, elevators, variable-frequency air conditioners, etc., making the power quality issues complex.

[0036] The data acquisition module sets sensors in the power distribution room of the commercial center and the power distribution boxes on each floor to comprehensively collect power grid parameters. During the peak power consumption in the evening, a large number of lighting equipment and air conditioning systems are fully turned on, the current harmonics increase greatly, and the voltage drops. The sensors collect data in real time at a frequency of 800 times per second and transmit it to the central processor. The power quality analysis model in the central processor combines the daily power consumption pattern and real-time data of the commercial center to accurately evaluate the power quality status. Based on historical data and current indicators, the prediction module predicts that in the next 45 minutes, as some shopping malls start promotional activities, the power load will further increase and the power quality will deteriorate sharply. The control strategy generation module uses optimization algorithms to formulate detailed reactive power compensation and filtering plans. The execution module drives the static reactive generator and active power filter to work together through the high-speed communication interface to quickly compensate for reactive power, filter harmonics, and pull the voltage back to the normal range, ensuring the normal operation of all merchants in the commercial center, avoiding problems such as electronic equipment failure and reduced customer experience caused by poor power quality, maintaining the prosperity and stability of the commercial center, and also reducing complaints and disputes caused by power quality problems, and improving the image of urban power supply services.

[0037] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A power quality control method based on model prediction, characterized in that: include: The data acquisition module collects the grid operation parameters in real time, including voltage, current, frequency and phase information. The acquisition frequency shall not be lower than the preset high frequency value. After the acquisition is completed, it is transmitted to the central processor in real time; The central processor has a built-in power quality analysis model trained based on big data. The model quickly calculates the current power quality indicators based on the input grid operation parameters, including harmonic content, voltage deviation, and three-phase imbalance. The calculation process must be completed within the specified first time interval. The prediction module is connected to the central processor, receives the power quality index, uses the prediction algorithm based on time series, combines the current index with the stored historical power quality data, and predicts the change trend of the power quality in the future predetermined time period. The prediction step is adaptively adjusted according to the dynamic characteristics of the power grid. The prediction accuracy must reach the preset high standard. The historical power quality data is stored in the storage unit connected to it; The control strategy generation module generates the power quality control strategy according to the change trend of the output of the prediction module using the preset optimization algorithm, which includes the switching scheme of the reactive compensation device and the adjustment instructions of the filter parameters. The optimization algorithm aims to achieve the best comprehensive power quality index while fully considering the control cost and equipment life constraints. After generation, it is immediately transmitted to the execution module; The execution module drives power quality regulation equipment such as static VAR generator and active power filter to perform actions according to the received control strategy. The execution response time shall not exceed the preset fast response threshold so as to improve the power quality in time. The execution module has a feedback function and can feed back the execution results to the central processor in real time, thereby realizing the optimization adjustment of subsequent control.

2. The power quality control method based on model prediction according to claim 1, characterized in that: The data acquisition module uses a high-precision sensor array, which consists of voltage sensors, current sensors, frequency sensors and phase sensors. These sensors are calibrated in advance to ensure the accuracy of the collected data, have strong anti-interference capabilities, and meet the requirements of relevant industry standards. In addition, the collected raw data can be filtered, amplified, and analog-to-digital converted in sequence, and then transmitted to the central processing unit after processing.

3. The power quality control method based on model prediction according to claim 1, characterized in that: When constructing the power quality analysis model, a large amount of grid operation data under different working conditions is collected as training samples. The sample data is first subjected to feature extraction and normalization processing, and then a deep learning algorithm, such as a convolutional neural network or a recurrent neural network, is used to build the model structure. The model parameters are adjusted through repeated training until the accuracy of the model on the validation set reaches a preset threshold. The model also has an online learning function and can update the model parameters in real time based on the newly connected grid operation data to adapt to the dynamic changes of the grid.

4. The power quality control method based on model prediction according to claim 1, characterized in that: The time series-based prediction algorithm adopted by the prediction module is a hybrid model, which effectively combines the advantages of the autoregressive moving average model ARIMA and the long short-term memory network LSTM. The ARIMA model is responsible for modeling the linear characteristics of power quality data, while the LSTM model focuses on capturing the nonlinear characteristics and long-term dependencies of the data. Finally, the prediction results of the two are fused in a weighted manner, and the weights are dynamically adjusted according to the historical prediction errors.

5. The power quality control method based on model prediction according to claim 1, characterized in that: The preset optimization algorithm of the control strategy generation module is a multi-objective particle swarm optimization algorithm. During the solution process, the comprehensive power quality indicators, control costs, and equipment life are set as different objective functions. With the help of iterative search of the particle swarm, we strive to find the optimal power quality control strategy that meets various objective constraints.

6. The power quality control method based on model prediction according to claim 1, characterized in that: The execution module is connected to the power quality regulation equipment through a high-speed communication interface, which includes an industrial Ethernet interface and a fiber optic interface, to ensure that control instructions can be transmitted quickly. The execution module uses pulse width modulation (PWM) technology to accurately drive the equipment to operate. At the same time, the execution module also has fault detection and diagnosis functions. Once a fault is detected in the power quality regulation equipment, an alarm signal is immediately issued and the fault information is fed back to the central processor so that appropriate emergency measures can be taken in a timely manner.

7. The power quality control method based on model prediction according to claim 1, characterized in that: The storage unit uses a large-capacity solid-state drive SSD with fast data reading and writing capabilities. It can store long-term historical power quality data and has a data backup function. It regularly backs up important data to an external storage device to prevent data loss. The data transmission between the storage unit and the central processor and prediction module uses encryption technology to fully protect data security and privacy.

Images